blob: 97b9f82fa3b1cb46c6de1b2bbda4b831235dc175 [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 Tarreaufba74ea2018-12-22 11:19:45 +01005 version 2.0
Willy Tarreau6a06a402007-07-15 20:15:28 +02006 willy tarreau
Willy Tarreaufba74ea2018-12-22 11:19:45 +01007 2018/12/22
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 Faulet99a17a22018-12-11 09:18:27 +01001109.4. Cache
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200111
William Lallemand86d0df02017-11-24 21:36:45 +010011210. Cache
Cyril Bonté7b888f12017-11-26 22:24:31 +010011310.1. Limitation
11410.2. Setup
11510.2.1. Cache section
11610.2.2. Proxy section
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200117
1181. Quick reminder about HTTP
119----------------------------
120
Davor Ocelice9ed2812017-12-25 17:49:28 +0100121When HAProxy is running in HTTP mode, both the request and the response are
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200122fully analyzed and indexed, thus it becomes possible to build matching criteria
123on almost anything found in the contents.
124
125However, it is important to understand how HTTP requests and responses are
126formed, and how HAProxy decomposes them. It will then become easier to write
127correct rules and to debug existing configurations.
128
129
1301.1. The HTTP transaction model
131-------------------------------
132
133The HTTP protocol is transaction-driven. This means that each request will lead
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100134to one and only one response. Traditionally, a TCP connection is established
Davor Ocelice9ed2812017-12-25 17:49:28 +0100135from the client to the server, a request is sent by the client through the
136connection, the server responds, and the connection is closed. A new request
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200137will involve a new connection :
138
139 [CON1] [REQ1] ... [RESP1] [CLO1] [CON2] [REQ2] ... [RESP2] [CLO2] ...
140
141In this mode, called the "HTTP close" mode, there are as many connection
142establishments as there are HTTP transactions. Since the connection is closed
143by the server after the response, the client does not need to know the content
144length.
145
146Due to the transactional nature of the protocol, it was possible to improve it
147to avoid closing a connection between two subsequent transactions. In this mode
148however, it is mandatory that the server indicates the content length for each
149response so that the client does not wait indefinitely. For this, a special
150header is used: "Content-length". This mode is called the "keep-alive" mode :
151
152 [CON] [REQ1] ... [RESP1] [REQ2] ... [RESP2] [CLO] ...
153
154Its advantages are a reduced latency between transactions, and less processing
155power required on the server side. It is generally better than the close mode,
156but not always because the clients often limit their concurrent connections to
Patrick Mezard9ec2ec42010-06-12 17:02:45 +0200157a smaller value.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200158
Willy Tarreau95c4e142017-11-26 12:18:55 +0100159Another improvement in the communications is the pipelining mode. It still uses
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200160keep-alive, but the client does not wait for the first response to send the
161second request. This is useful for fetching large number of images composing a
162page :
163
164 [CON] [REQ1] [REQ2] ... [RESP1] [RESP2] [CLO] ...
165
166This can obviously have a tremendous benefit on performance because the network
167latency is eliminated between subsequent requests. Many HTTP agents do not
168correctly support pipelining since there is no way to associate a response with
169the corresponding request in HTTP. For this reason, it is mandatory for the
Cyril Bonté78caf842010-03-10 22:41:43 +0100170server to reply in the exact same order as the requests were received.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200171
Willy Tarreau95c4e142017-11-26 12:18:55 +0100172The next improvement is the multiplexed mode, as implemented in HTTP/2. This
173time, each transaction is assigned a single stream identifier, and all streams
174are multiplexed over an existing connection. Many requests can be sent in
175parallel by the client, and responses can arrive in any order since they also
176carry the stream identifier.
177
Willy Tarreau70dffda2014-01-30 03:07:23 +0100178By default HAProxy operates in keep-alive mode with regards to persistent
179connections: for each connection it processes each request and response, and
180leaves the connection idle on both sides between the end of a response and the
Willy Tarreau95c4e142017-11-26 12:18:55 +0100181start of a new request. When it receives HTTP/2 connections from a client, it
182processes all the requests in parallel and leaves the connection idling,
183waiting for new requests, just as if it was a keep-alive HTTP connection.
Patrick Mezard9ec2ec42010-06-12 17:02:45 +0200184
Christopher Faulet315b39c2018-09-21 16:26:19 +0200185HAProxy supports 4 connection modes :
Willy Tarreau70dffda2014-01-30 03:07:23 +0100186 - keep alive : all requests and responses are processed (default)
187 - tunnel : only the first request and response are processed,
188 everything else is forwarded with no analysis.
Willy Tarreau70dffda2014-01-30 03:07:23 +0100189 - server close : the server-facing connection is closed after the response.
Christopher Faulet315b39c2018-09-21 16:26:19 +0200190 - close : the connection is actively closed after end of response.
Willy Tarreau70dffda2014-01-30 03:07:23 +0100191
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
William Lallemanddaf4cd22018-04-17 16:46:13 +0200479A special variable $HAPROXY_LOCALPEER is defined at the startup of the process
480which contains the name of the local peer. (See "-L" in the management guide.)
481
William Lallemandb2f07452015-05-12 14:27:13 +0200482
4832.4. Time format
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200484----------------
485
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100486Some parameters involve values representing time, such as timeouts. These
Willy Tarreau0ba27502007-12-24 16:55:16 +0100487values are generally expressed in milliseconds (unless explicitly stated
488otherwise) but may be expressed in any other unit by suffixing the unit to the
489numeric value. It is important to consider this because it will not be repeated
490for every keyword. Supported units are :
491
492 - us : microseconds. 1 microsecond = 1/1000000 second
493 - ms : milliseconds. 1 millisecond = 1/1000 second. This is the default.
494 - s : seconds. 1s = 1000ms
495 - m : minutes. 1m = 60s = 60000ms
496 - h : hours. 1h = 60m = 3600s = 3600000ms
497 - d : days. 1d = 24h = 1440m = 86400s = 86400000ms
498
499
Lukas Tribusaa83a312017-03-21 09:25:09 +00005002.5. Examples
Patrick Mezard35da19c2010-06-12 17:02:47 +0200501-------------
502
503 # Simple configuration for an HTTP proxy listening on port 80 on all
504 # interfaces and forwarding requests to a single backend "servers" with a
505 # single server "server1" listening on 127.0.0.1:8000
506 global
507 daemon
508 maxconn 256
509
510 defaults
511 mode http
512 timeout connect 5000ms
513 timeout client 50000ms
514 timeout server 50000ms
515
516 frontend http-in
517 bind *:80
518 default_backend servers
519
520 backend servers
521 server server1 127.0.0.1:8000 maxconn 32
522
523
524 # The same configuration defined with a single listen block. Shorter but
525 # less expressive, especially in HTTP mode.
526 global
527 daemon
528 maxconn 256
529
530 defaults
531 mode http
532 timeout connect 5000ms
533 timeout client 50000ms
534 timeout server 50000ms
535
536 listen http-in
537 bind *:80
538 server server1 127.0.0.1:8000 maxconn 32
539
540
541Assuming haproxy is in $PATH, test these configurations in a shell with:
542
Willy Tarreauccb289d2010-12-11 20:19:38 +0100543 $ sudo haproxy -f configuration.conf -c
Patrick Mezard35da19c2010-06-12 17:02:47 +0200544
545
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005463. Global parameters
Willy Tarreau6a06a402007-07-15 20:15:28 +0200547--------------------
548
549Parameters in the "global" section are process-wide and often OS-specific. They
550are generally set once for all and do not need being changed once correct. Some
551of them have command-line equivalents.
552
553The following keywords are supported in the "global" section :
554
555 * Process management and security
Emeric Brunc8e8d122012-10-02 18:42:10 +0200556 - ca-base
Willy Tarreau6a06a402007-07-15 20:15:28 +0200557 - chroot
Emeric Brunc8e8d122012-10-02 18:42:10 +0200558 - crt-base
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200559 - cpu-map
Willy Tarreau6a06a402007-07-15 20:15:28 +0200560 - daemon
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200561 - description
562 - deviceatlas-json-file
563 - deviceatlas-log-level
564 - deviceatlas-separator
565 - deviceatlas-properties-cookie
Simon Horman98637e52014-06-20 12:30:16 +0900566 - external-check
Willy Tarreau6a06a402007-07-15 20:15:28 +0200567 - gid
568 - group
Cyril Bonté203ec5a2017-03-23 22:44:13 +0100569 - hard-stop-after
Willy Tarreau6a06a402007-07-15 20:15:28 +0200570 - log
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200571 - log-tag
Joe Williamsdf5b38f2010-12-29 17:05:48 +0100572 - log-send-hostname
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200573 - lua-load
Willy Tarreau6a06a402007-07-15 20:15:28 +0200574 - nbproc
Christopher Fauletbe0faa22017-08-29 15:37:10 +0200575 - nbthread
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200576 - node
Willy Tarreau6a06a402007-07-15 20:15:28 +0200577 - pidfile
Willy Tarreau1d549722016-02-16 12:41:57 +0100578 - presetenv
579 - resetenv
Willy Tarreau6a06a402007-07-15 20:15:28 +0200580 - uid
581 - ulimit-n
582 - user
Willy Tarreau1d549722016-02-16 12:41:57 +0100583 - setenv
Willy Tarreaufbee7132007-10-18 13:53:22 +0200584 - stats
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200585 - ssl-default-bind-ciphers
Dirkjan Bussink415150f2018-09-14 11:14:21 +0200586 - ssl-default-bind-ciphersuites
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200587 - ssl-default-bind-options
588 - ssl-default-server-ciphers
Dirkjan Bussink415150f2018-09-14 11:14:21 +0200589 - ssl-default-server-ciphersuites
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200590 - ssl-default-server-options
591 - ssl-dh-param-file
Emeric Brun850efd52014-01-29 12:24:34 +0100592 - ssl-server-verify
Willy Tarreauceb24bc2010-11-09 12:46:41 +0100593 - unix-bind
Willy Tarreau1d549722016-02-16 12:41:57 +0100594 - unsetenv
Thomas Holmesdb04f192015-05-18 13:21:39 +0100595 - 51degrees-data-file
596 - 51degrees-property-name-list
Dragan Dosen93b38d92015-06-29 16:43:25 +0200597 - 51degrees-property-separator
Dragan Dosenae6d39a2015-06-29 16:43:27 +0200598 - 51degrees-cache-size
scientiamobiled0027ed2016-11-04 10:55:08 +0100599 - wurfl-data-file
600 - wurfl-information-list
601 - wurfl-information-list-separator
602 - wurfl-engine-mode
603 - wurfl-cache-size
604 - wurfl-useragent-priority
Willy Tarreaud72758d2010-01-12 10:42:19 +0100605
Willy Tarreau6a06a402007-07-15 20:15:28 +0200606 * Performance tuning
Willy Tarreau1746eec2014-04-25 10:46:47 +0200607 - max-spread-checks
Willy Tarreau6a06a402007-07-15 20:15:28 +0200608 - maxconn
Willy Tarreau81c25d02011-09-07 15:17:21 +0200609 - maxconnrate
William Lallemandd85f9172012-11-09 17:05:39 +0100610 - maxcomprate
William Lallemand072a2bf2012-11-20 17:01:01 +0100611 - maxcompcpuusage
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100612 - maxpipes
Willy Tarreau93e7c002013-10-07 18:51:07 +0200613 - maxsessrate
Willy Tarreau403edff2012-09-06 11:58:37 +0200614 - maxsslconn
Willy Tarreaue43d5322013-10-07 20:01:52 +0200615 - maxsslrate
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200616 - maxzlibmem
Willy Tarreau6a06a402007-07-15 20:15:28 +0200617 - noepoll
618 - nokqueue
619 - nopoll
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100620 - nosplice
Jarno Huuskonen0e82b922014-04-12 18:22:19 +0300621 - nogetaddrinfo
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +0000622 - noreuseport
Willy Tarreau75c62c22018-11-22 11:02:09 +0100623 - profiling.tasks
Willy Tarreaufe255b72007-10-14 23:09:26 +0200624 - spread-checks
Baptiste Assmann5626f482015-08-23 10:00:10 +0200625 - server-state-base
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +0200626 - server-state-file
Grant Zhang872f9c22017-01-21 01:10:18 +0000627 - ssl-engine
Grant Zhangfa6c7ee2017-01-14 01:42:15 +0000628 - ssl-mode-async
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200629 - tune.buffers.limit
630 - tune.buffers.reserve
Willy Tarreau27a674e2009-08-17 07:23:33 +0200631 - tune.bufsize
Willy Tarreau43961d52010-10-04 20:39:20 +0200632 - tune.chksize
William Lallemandf3747832012-11-09 12:33:10 +0100633 - tune.comp.maxlevel
Willy Tarreaufe20e5b2017-07-27 11:42:14 +0200634 - tune.h2.header-table-size
Willy Tarreaue6baec02017-07-27 11:45:11 +0200635 - tune.h2.initial-window-size
Willy Tarreau5242ef82017-07-27 11:47:28 +0200636 - tune.h2.max-concurrent-streams
Willy Tarreau193b8c62012-11-22 00:17:38 +0100637 - tune.http.cookielen
Stéphane Cottin23e9e932017-05-18 08:58:41 +0200638 - tune.http.logurilen
Willy Tarreauac1932d2011-10-24 19:14:41 +0200639 - tune.http.maxhdr
Willy Tarreau7e312732014-02-12 16:35:14 +0100640 - tune.idletimer
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100641 - tune.lua.forced-yield
Willy Tarreau32f61e22015-03-18 17:54:59 +0100642 - tune.lua.maxmem
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100643 - tune.lua.session-timeout
644 - tune.lua.task-timeout
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +0200645 - tune.lua.service-timeout
Willy Tarreaua0250ba2008-01-06 11:22:57 +0100646 - tune.maxaccept
647 - tune.maxpollevents
Willy Tarreau27a674e2009-08-17 07:23:33 +0200648 - tune.maxrewrite
Willy Tarreauf3045d22015-04-29 16:24:50 +0200649 - tune.pattern.cache-size
Willy Tarreaubd9a0a72011-10-23 21:14:29 +0200650 - tune.pipesize
Willy Tarreaue803de22010-01-21 17:43:04 +0100651 - tune.rcvbuf.client
652 - tune.rcvbuf.server
Willy Tarreaub22fc302015-12-14 12:04:35 +0100653 - tune.recv_enough
Olivier Houchard1599b802018-05-24 18:59:04 +0200654 - tune.runqueue-depth
Willy Tarreaue803de22010-01-21 17:43:04 +0100655 - tune.sndbuf.client
656 - tune.sndbuf.server
Willy Tarreau6ec58db2012-11-16 16:32:15 +0100657 - tune.ssl.cachesize
Willy Tarreaubfd59462013-02-21 07:46:09 +0100658 - tune.ssl.lifetime
Emeric Brun8dc60392014-05-09 13:52:00 +0200659 - tune.ssl.force-private-cache
Willy Tarreaubfd59462013-02-21 07:46:09 +0100660 - tune.ssl.maxrecord
Remi Gacognef46cd6e2014-06-12 14:58:40 +0200661 - tune.ssl.default-dh-param
Christopher Faulet31af49d2015-06-09 17:29:50 +0200662 - tune.ssl.ssl-ctx-cache-size
Thierry FOURNIER5bf77322017-02-25 12:45:22 +0100663 - tune.ssl.capture-cipherlist-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200664 - tune.vars.global-max-size
Christopher Fauletff2613e2016-11-09 11:36:17 +0100665 - tune.vars.proc-max-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200666 - tune.vars.reqres-max-size
667 - tune.vars.sess-max-size
668 - tune.vars.txn-max-size
William Lallemanda509e4c2012-11-07 16:54:34 +0100669 - tune.zlib.memlevel
670 - tune.zlib.windowsize
Willy Tarreaud72758d2010-01-12 10:42:19 +0100671
Willy Tarreau6a06a402007-07-15 20:15:28 +0200672 * Debugging
673 - debug
674 - quiet
Willy Tarreau6a06a402007-07-15 20:15:28 +0200675
676
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006773.1. Process management and security
Willy Tarreau6a06a402007-07-15 20:15:28 +0200678------------------------------------
679
Emeric Brunc8e8d122012-10-02 18:42:10 +0200680ca-base <dir>
681 Assigns a default directory to fetch SSL CA certificates and CRLs from when a
Emeric Brunfd33a262012-10-11 16:28:27 +0200682 relative path is used with "ca-file" or "crl-file" directives. Absolute
683 locations specified in "ca-file" and "crl-file" prevail and ignore "ca-base".
Emeric Brunc8e8d122012-10-02 18:42:10 +0200684
Willy Tarreau6a06a402007-07-15 20:15:28 +0200685chroot <jail dir>
686 Changes current directory to <jail dir> and performs a chroot() there before
687 dropping privileges. This increases the security level in case an unknown
688 vulnerability would be exploited, since it would make it very hard for the
689 attacker to exploit the system. This only works when the process is started
690 with superuser privileges. It is important to ensure that <jail_dir> is both
Davor Ocelice9ed2812017-12-25 17:49:28 +0100691 empty and non-writable to anyone.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100692
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100693cpu-map [auto:]<process-set>[/<thread-set>] <cpu-set>...
694 On Linux 2.6 and above, it is possible to bind a process or a thread to a
695 specific CPU set. This means that the process or the thread will never run on
696 other CPUs. The "cpu-map" directive specifies CPU sets for process or thread
697 sets. The first argument is a process set, eventually followed by a thread
698 set. These sets have the format
699
700 all | odd | even | number[-[number]]
701
702 <number>> must be a number between 1 and 32 or 64, depending on the machine's
Davor Ocelice9ed2812017-12-25 17:49:28 +0100703 word size. Any process IDs above nbproc and any thread IDs above nbthread are
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100704 ignored. It is possible to specify a range with two such number delimited by
705 a dash ('-'). It also is possible to specify all processes at once using
Christopher Faulet1dcb9cb2017-11-22 10:24:40 +0100706 "all", only odd numbers using "odd" or even numbers using "even", just like
707 with the "bind-process" directive. The second and forthcoming arguments are
Davor Ocelice9ed2812017-12-25 17:49:28 +0100708 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 +0100709 range with two such numbers delimited by a dash ('-'). Multiple CPU numbers
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100710 or ranges may be specified, and the processes or threads will be allowed to
Davor Ocelice9ed2812017-12-25 17:49:28 +0100711 bind to all of them. Obviously, multiple "cpu-map" directives may be
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100712 specified. Each "cpu-map" directive will replace the previous ones when they
713 overlap. A thread will be bound on the intersection of its mapping and the
714 one of the process on which it is attached. If the intersection is null, no
715 specific binding will be set for the thread.
Willy Tarreaufc6c0322012-11-16 16:12:27 +0100716
Christopher Fauletff4121f2017-11-22 16:38:49 +0100717 Ranges can be partially defined. The higher bound can be omitted. In such
718 case, it is replaced by the corresponding maximum value, 32 or 64 depending
719 on the machine's word size.
720
Christopher Faulet26028f62017-11-22 15:01:51 +0100721 The prefix "auto:" can be added before the process set to let HAProxy
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100722 automatically bind a process or a thread to a CPU by incrementing
723 process/thread and CPU sets. To be valid, both sets must have the same
724 size. No matter the declaration order of the CPU sets, it will be bound from
725 the lowest to the highest bound. Having a process and a thread range with the
726 "auto:" prefix is not supported. Only one range is supported, the other one
727 must be a fixed number.
Christopher Faulet26028f62017-11-22 15:01:51 +0100728
729 Examples:
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100730 cpu-map 1-4 0-3 # bind processes 1 to 4 on the first 4 CPUs
731
732 cpu-map 1/all 0-3 # bind all threads of the first process on the
733 # first 4 CPUs
734
735 cpu-map 1- 0- # will be replaced by "cpu-map 1-64 0-63"
736 # or "cpu-map 1-32 0-31" depending on the machine's
737 # word size.
738
Christopher Faulet26028f62017-11-22 15:01:51 +0100739 # all these lines bind the process 1 to the cpu 0, the process 2 to cpu 1
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100740 # and so on.
Christopher Faulet26028f62017-11-22 15:01:51 +0100741 cpu-map auto:1-4 0-3
742 cpu-map auto:1-4 0-1 2-3
743 cpu-map auto:1-4 3 2 1 0
744
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100745 # all these lines bind the thread 1 to the cpu 0, the thread 2 to cpu 1
746 # and so on.
747 cpu-map auto:1/1-4 0-3
748 cpu-map auto:1/1-4 0-1 2-3
749 cpu-map auto:1/1-4 3 2 1 0
750
Davor Ocelice9ed2812017-12-25 17:49:28 +0100751 # bind each process to exactly one CPU using all/odd/even keyword
Christopher Faulet26028f62017-11-22 15:01:51 +0100752 cpu-map auto:all 0-63
753 cpu-map auto:even 0-31
754 cpu-map auto:odd 32-63
755
756 # invalid cpu-map because process and CPU sets have different sizes.
757 cpu-map auto:1-4 0 # invalid
758 cpu-map auto:1 0-3 # invalid
759
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100760 # invalid cpu-map because automatic binding is used with a process range
761 # and a thread range.
762 cpu-map auto:all/all 0 # invalid
763 cpu-map auto:all/1-4 0 # invalid
764 cpu-map auto:1-4/all 0 # invalid
765
Emeric Brunc8e8d122012-10-02 18:42:10 +0200766crt-base <dir>
767 Assigns a default directory to fetch SSL certificates from when a relative
768 path is used with "crtfile" directives. Absolute locations specified after
769 "crtfile" prevail and ignore "crt-base".
770
Willy Tarreau6a06a402007-07-15 20:15:28 +0200771daemon
772 Makes the process fork into background. This is the recommended mode of
773 operation. It is equivalent to the command line "-D" argument. It can be
Lukas Tribusf46bf952017-11-21 12:39:34 +0100774 disabled by the command line "-db" argument. This option is ignored in
775 systemd mode.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200776
David Carlier8167f302015-06-01 13:50:06 +0200777deviceatlas-json-file <path>
778 Sets the path of the DeviceAtlas JSON data file to be loaded by the API.
Davor Ocelice9ed2812017-12-25 17:49:28 +0100779 The path must be a valid JSON data file and accessible by HAProxy process.
David Carlier8167f302015-06-01 13:50:06 +0200780
781deviceatlas-log-level <value>
Davor Ocelice9ed2812017-12-25 17:49:28 +0100782 Sets the level of information returned by the API. This directive is
David Carlier8167f302015-06-01 13:50:06 +0200783 optional and set to 0 by default if not set.
784
785deviceatlas-separator <char>
786 Sets the character separator for the API properties results. This directive
787 is optional and set to | by default if not set.
788
Cyril Bonté0306c4a2015-10-26 22:37:38 +0100789deviceatlas-properties-cookie <name>
Cyril Bonté307ee1e2015-09-28 23:16:06 +0200790 Sets the client cookie's name used for the detection if the DeviceAtlas
791 Client-side component was used during the request. This directive is optional
792 and set to DAPROPS by default if not set.
David Carlier29b3ca32015-09-25 14:09:21 +0100793
Simon Horman98637e52014-06-20 12:30:16 +0900794external-check
795 Allows the use of an external agent to perform health checks.
796 This is disabled by default as a security precaution.
797 See "option external-check".
798
Willy Tarreau6a06a402007-07-15 20:15:28 +0200799gid <number>
800 Changes the process' group ID to <number>. It is recommended that the group
801 ID is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
802 be started with a user belonging to this group, or with superuser privileges.
Michael Schererab012dd2013-01-12 18:35:19 +0100803 Note that if haproxy is started from a user having supplementary groups, it
804 will only be able to drop these groups if started with superuser privileges.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200805 See also "group" and "uid".
Willy Tarreaud72758d2010-01-12 10:42:19 +0100806
Cyril Bonté203ec5a2017-03-23 22:44:13 +0100807hard-stop-after <time>
808 Defines the maximum time allowed to perform a clean soft-stop.
809
810 Arguments :
811 <time> is the maximum time (by default in milliseconds) for which the
812 instance will remain alive when a soft-stop is received via the
813 SIGUSR1 signal.
814
815 This may be used to ensure that the instance will quit even if connections
816 remain opened during a soft-stop (for example with long timeouts for a proxy
817 in tcp mode). It applies both in TCP and HTTP mode.
818
819 Example:
820 global
821 hard-stop-after 30s
822
Willy Tarreau6a06a402007-07-15 20:15:28 +0200823group <group name>
824 Similar to "gid" but uses the GID of group name <group name> from /etc/group.
825 See also "gid" and "user".
Willy Tarreaud72758d2010-01-12 10:42:19 +0100826
Dragan Dosen7ad31542015-09-28 17:16:47 +0200827log <address> [len <length>] [format <format>] <facility> [max level [min level]]
Cyril Bonté3e954872018-03-20 23:30:27 +0100828 Adds a global syslog server. Several global servers can be defined. They
Davor Ocelice9ed2812017-12-25 17:49:28 +0100829 will receive logs for starts and exits, as well as all logs from proxies
Robert Tsai81ae1952007-12-05 10:47:29 +0100830 configured with "log global".
831
832 <address> can be one of:
833
Willy Tarreau2769aa02007-12-27 18:26:09 +0100834 - An IPv4 address optionally followed by a colon and a UDP port. If
Robert Tsai81ae1952007-12-05 10:47:29 +0100835 no port is specified, 514 is used by default (the standard syslog
836 port).
837
David du Colombier24bb5f52011-03-17 10:40:23 +0100838 - An IPv6 address followed by a colon and optionally a UDP port. If
839 no port is specified, 514 is used by default (the standard syslog
840 port).
841
Willy Tarreau5a32ecc2018-11-12 07:34:59 +0100842 - A filesystem path to a datagram UNIX domain socket, keeping in mind
Robert Tsai81ae1952007-12-05 10:47:29 +0100843 considerations for chroot (be sure the path is accessible inside
844 the chroot) and uid/gid (be sure the path is appropriately
Davor Ocelice9ed2812017-12-25 17:49:28 +0100845 writable).
Robert Tsai81ae1952007-12-05 10:47:29 +0100846
Willy Tarreau5a32ecc2018-11-12 07:34:59 +0100847 - A file descriptor number in the form "fd@<number>", which may point
848 to a pipe, terminal, or socket. In this case unbuffered logs are used
849 and one writev() call per log is performed. This is a bit expensive
850 but acceptable for most workloads. Messages sent this way will not be
851 truncated but may be dropped, in which case the DroppedLogs counter
852 will be incremented. The writev() call is atomic even on pipes for
853 messages up to PIPE_BUF size, which POSIX recommends to be at least
854 512 and which is 4096 bytes on most modern operating systems. Any
855 larger message may be interleaved with messages from other processes.
856 Exceptionally for debugging purposes the file descriptor may also be
857 directed to a file, but doing so will significantly slow haproxy down
858 as non-blocking calls will be ignored. Also there will be no way to
859 purge nor rotate this file without restarting the process. Note that
860 the configured syslog format is preserved, so the output is suitable
Willy Tarreauc1b06452018-11-12 11:57:56 +0100861 for use with a TCP syslog server. See also the "short" and "raw"
862 format below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +0100863
864 - "stdout" / "stderr", which are respectively aliases for "fd@1" and
865 "fd@2", see above.
866
William Lallemandb2f07452015-05-12 14:27:13 +0200867 You may want to reference some environment variables in the address
868 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +0100869
Willy Tarreau18324f52014-06-27 18:10:07 +0200870 <length> is an optional maximum line length. Log lines larger than this value
871 will be truncated before being sent. The reason is that syslog
872 servers act differently on log line length. All servers support the
873 default value of 1024, but some servers simply drop larger lines
874 while others do log them. If a server supports long lines, it may
875 make sense to set this value here in order to avoid truncating long
876 lines. Similarly, if a server drops long lines, it is preferable to
877 truncate them before sending them. Accepted values are 80 to 65535
878 inclusive. The default value of 1024 is generally fine for all
879 standard usages. Some specific cases of long captures or
Davor Ocelice9ed2812017-12-25 17:49:28 +0100880 JSON-formatted logs may require larger values. You may also need to
881 increase "tune.http.logurilen" if your request URIs are truncated.
Willy Tarreau18324f52014-06-27 18:10:07 +0200882
Dragan Dosen7ad31542015-09-28 17:16:47 +0200883 <format> is the log format used when generating syslog messages. It may be
884 one of the following :
885
886 rfc3164 The RFC3164 syslog message format. This is the default.
887 (https://tools.ietf.org/html/rfc3164)
888
889 rfc5424 The RFC5424 syslog message format.
890 (https://tools.ietf.org/html/rfc5424)
891
Willy Tarreaue8746a02018-11-12 08:45:00 +0100892 short A message containing only a level between angle brackets such as
893 '<3>', followed by the text. The PID, date, time, process name
894 and system name are omitted. This is designed to be used with a
895 local log server. This format is compatible with what the systemd
896 logger consumes.
897
Willy Tarreauc1b06452018-11-12 11:57:56 +0100898 raw A message containing only the text. The level, PID, date, time,
899 process name and system name are omitted. This is designed to be
900 used in containers or during development, where the severity only
901 depends on the file descriptor used (stdout/stderr).
902
Robert Tsai81ae1952007-12-05 10:47:29 +0100903 <facility> must be one of the 24 standard syslog facilities :
Willy Tarreau6a06a402007-07-15 20:15:28 +0200904
Willy Tarreaue8746a02018-11-12 08:45:00 +0100905 kern user mail daemon auth syslog lpr news
906 uucp cron auth2 ftp ntp audit alert cron2
907 local0 local1 local2 local3 local4 local5 local6 local7
908
Willy Tarreauc1b06452018-11-12 11:57:56 +0100909 Note that the facility is ignored for the "short" and "raw"
910 formats, but still required as a positional field. It is
911 recommended to use "daemon" in this case to make it clear that
912 it's only supposed to be used locally.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200913
914 An optional level can be specified to filter outgoing messages. By default,
Willy Tarreauf7edefa2009-05-10 17:20:05 +0200915 all messages are sent. If a maximum level is specified, only messages with a
916 severity at least as important as this level will be sent. An optional minimum
917 level can be specified. If it is set, logs emitted with a more severe level
918 than this one will be capped to this level. This is used to avoid sending
919 "emerg" messages on all terminals on some default syslog configurations.
920 Eight levels are known :
Willy Tarreau6a06a402007-07-15 20:15:28 +0200921
Cyril Bontédc4d9032012-04-08 21:57:39 +0200922 emerg alert crit err warning notice info debug
Willy Tarreau6a06a402007-07-15 20:15:28 +0200923
Joe Williamsdf5b38f2010-12-29 17:05:48 +0100924log-send-hostname [<string>]
925 Sets the hostname field in the syslog header. If optional "string" parameter
926 is set the header is set to the string contents, otherwise uses the hostname
927 of the system. Generally used if one is not relaying logs through an
928 intermediate syslog server or for simply customizing the hostname printed in
929 the logs.
930
Kevinm48936af2010-12-22 16:08:21 +0000931log-tag <string>
932 Sets the tag field in the syslog header to this string. It defaults to the
933 program name as launched from the command line, which usually is "haproxy".
934 Sometimes it can be useful to differentiate between multiple processes
Willy Tarreau094af4e2015-01-07 15:03:42 +0100935 running on the same host. See also the per-proxy "log-tag" directive.
Kevinm48936af2010-12-22 16:08:21 +0000936
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100937lua-load <file>
938 This global directive loads and executes a Lua file. This directive can be
939 used multiple times.
940
William Lallemand4cfede82017-11-24 22:02:34 +0100941master-worker [no-exit-on-failure]
William Lallemande202b1e2017-06-01 17:38:56 +0200942 Master-worker mode. It is equivalent to the command line "-W" argument.
943 This mode will launch a "master" which will monitor the "workers". Using
944 this mode, you can reload HAProxy directly by sending a SIGUSR2 signal to
Davor Ocelice9ed2812017-12-25 17:49:28 +0100945 the master. The master-worker mode is compatible either with the foreground
William Lallemande202b1e2017-06-01 17:38:56 +0200946 or daemon mode. It is recommended to use this mode with multiprocess and
947 systemd.
William Lallemand4cfede82017-11-24 22:02:34 +0100948 By default, if a worker exits with a bad return code, in the case of a
949 segfault for example, all workers will be killed, and the master will leave.
950 It is convenient to combine this behavior with Restart=on-failure in a
951 systemd unit file in order to relaunch the whole process. If you don't want
952 this behavior, you must use the keyword "no-exit-on-failure".
William Lallemande202b1e2017-06-01 17:38:56 +0200953
William Lallemand4cfede82017-11-24 22:02:34 +0100954 See also "-W" in the management guide.
William Lallemande202b1e2017-06-01 17:38:56 +0200955
Willy Tarreau6a06a402007-07-15 20:15:28 +0200956nbproc <number>
957 Creates <number> processes when going daemon. This requires the "daemon"
958 mode. By default, only one process is created, which is the recommended mode
959 of operation. For systems limited to small sets of file descriptors per
960 process, it may be needed to fork multiple daemons. USING MULTIPLE PROCESSES
961 IS HARDER TO DEBUG AND IS REALLY DISCOURAGED. See also "daemon".
962
Christopher Fauletbe0faa22017-08-29 15:37:10 +0200963nbthread <number>
964 This setting is only available when support for threads was built in. It
965 creates <number> threads for each created processes. It means if HAProxy is
966 started in foreground, it only creates <number> threads for the first
967 process. FOR NOW, THREADS SUPPORT IN HAPROXY IS HIGHLY EXPERIMENTAL AND IT
968 MUST BE ENABLED WITH CAUTION AND AT YOUR OWN RISK. See also "nbproc".
969
Willy Tarreau6a06a402007-07-15 20:15:28 +0200970pidfile <pidfile>
Davor Ocelice9ed2812017-12-25 17:49:28 +0100971 Writes PIDs of all daemons into file <pidfile>. This option is equivalent to
Willy Tarreau6a06a402007-07-15 20:15:28 +0200972 the "-p" command line argument. The file must be accessible to the user
973 starting the process. See also "daemon".
974
Willy Tarreau1d549722016-02-16 12:41:57 +0100975presetenv <name> <value>
976 Sets environment variable <name> to value <value>. If the variable exists, it
977 is NOT overwritten. The changes immediately take effect so that the next line
978 in the configuration file sees the new value. See also "setenv", "resetenv",
979 and "unsetenv".
980
981resetenv [<name> ...]
982 Removes all environment variables except the ones specified in argument. It
983 allows to use a clean controlled environment before setting new values with
984 setenv or unsetenv. Please note that some internal functions may make use of
985 some environment variables, such as time manipulation functions, but also
986 OpenSSL or even external checks. This must be used with extreme care and only
987 after complete validation. The changes immediately take effect so that the
988 next line in the configuration file sees the new environment. See also
989 "setenv", "presetenv", and "unsetenv".
990
Christopher Fauletff4121f2017-11-22 16:38:49 +0100991stats bind-process [ all | odd | even | <process_num>[-[process_num>]] ] ...
Willy Tarreau35b7b162012-10-22 23:17:18 +0200992 Limits the stats socket to a certain set of processes numbers. By default the
993 stats socket is bound to all processes, causing a warning to be emitted when
994 nbproc is greater than 1 because there is no way to select the target process
995 when connecting. However, by using this setting, it becomes possible to pin
996 the stats socket to a specific set of processes, typically the first one. The
997 warning will automatically be disabled when this setting is used, whatever
Willy Tarreaua9db57e2013-01-18 11:29:29 +0100998 the number of processes used. The maximum process ID depends on the machine's
Christopher Fauletff4121f2017-11-22 16:38:49 +0100999 word size (32 or 64). Ranges can be partially defined. The higher bound can
1000 be omitted. In such case, it is replaced by the corresponding maximum
1001 value. A better option consists in using the "process" setting of the "stats
1002 socket" line to force the process on each line.
Willy Tarreau35b7b162012-10-22 23:17:18 +02001003
Baptiste Assmann5626f482015-08-23 10:00:10 +02001004server-state-base <directory>
1005 Specifies the directory prefix to be prepended in front of all servers state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001006 file names which do not start with a '/'. See also "server-state-file",
1007 "load-server-state-from-file" and "server-state-file-name".
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +02001008
1009server-state-file <file>
1010 Specifies the path to the file containing state of servers. If the path starts
1011 with a slash ('/'), it is considered absolute, otherwise it is considered
1012 relative to the directory specified using "server-state-base" (if set) or to
1013 the current directory. Before reloading HAProxy, it is possible to save the
1014 servers' current state using the stats command "show servers state". The
1015 output of this command must be written in the file pointed by <file>. When
1016 starting up, before handling traffic, HAProxy will read, load and apply state
1017 for each server found in the file and available in its current running
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001018 configuration. See also "server-state-base" and "show servers state",
1019 "load-server-state-from-file" and "server-state-file-name"
Baptiste Assmann5626f482015-08-23 10:00:10 +02001020
Willy Tarreau1d549722016-02-16 12:41:57 +01001021setenv <name> <value>
1022 Sets environment variable <name> to value <value>. If the variable exists, it
1023 is overwritten. The changes immediately take effect so that the next line in
1024 the configuration file sees the new value. See also "presetenv", "resetenv",
1025 and "unsetenv".
1026
Willy Tarreau610f04b2014-02-13 11:36:41 +01001027ssl-default-bind-ciphers <ciphers>
1028 This setting is only available when support for OpenSSL was built in. It sets
1029 the default string describing the list of cipher algorithms ("cipher suite")
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001030 that are negotiated during the SSL/TLS handshake except for TLSv1.3 for all
1031 "bind" lines which do not explicitly define theirs. The format of the string
1032 is defined in "man 1 ciphers" from OpenSSL man pages, and can be for instance
1033 a string such as "AES:ALL:!aNULL:!eNULL:+RC4:@STRENGTH" (without quotes). For
1034 TLSv1.3 cipher configuration, please check the "ssl-default-bind-ciphersuites"
1035 keyword. Please check the "bind" keyword for more information.
1036
1037ssl-default-bind-ciphersuites <ciphersuites>
1038 This setting is only available when support for OpenSSL was built in and
1039 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default string
1040 describing the list of cipher algorithms ("cipher suite") that are negotiated
1041 during the TLSv1.3 handshake for all "bind" lines which do not explicitly define
1042 theirs. The format of the string is defined in
1043 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites", and can
1044 be for instance a string such as
1045 "TLS_AES_256_GCM_SHA384:TLS_CHACHA20_POLY1305_SHA256:TLS_AES_128_GCM_SHA256"
1046 (without quotes). For cipher configuration for TLSv1.2 and earlier, please check
1047 the "ssl-default-bind-ciphers" keyword. Please check the "bind" keyword for more
1048 information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001049
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001050ssl-default-bind-options [<option>]...
1051 This setting is only available when support for OpenSSL was built in. It sets
1052 default ssl-options to force on all "bind" lines. Please check the "bind"
1053 keyword to see available options.
1054
1055 Example:
1056 global
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +02001057 ssl-default-bind-options ssl-min-ver TLSv1.0 no-tls-tickets
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001058
Willy Tarreau610f04b2014-02-13 11:36:41 +01001059ssl-default-server-ciphers <ciphers>
1060 This setting is only available when support for OpenSSL was built in. It
1061 sets the default string describing the list of cipher algorithms that are
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001062 negotiated during the SSL/TLS handshake except for TLSv1.3 with the server,
1063 for all "server" lines which do not explicitly define theirs. The format of
1064 the string is defined in "man 1 ciphers". For TLSv1.3 cipher configuration,
1065 please check the "ssl-default-server-ciphersuites" keyword. Please check the
1066 "server" keyword for more information.
1067
1068ssl-default-server-ciphersuites <ciphersuites>
1069 This setting is only available when support for OpenSSL was built in and
1070 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default
1071 string describing the list of cipher algorithms that are negotiated during
1072 the TLSv1.3 handshake with the server, for all "server" lines which do not
1073 explicitly define theirs. The format of the string is defined in
1074 "man 1 ciphers" under the "ciphersuites" section. For cipher configuration for
1075 TLSv1.2 and earlier, please check the "ssl-default-server-ciphers" keyword.
1076 Please check the "server" keyword for more information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001077
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001078ssl-default-server-options [<option>]...
1079 This setting is only available when support for OpenSSL was built in. It sets
1080 default ssl-options to force on all "server" lines. Please check the "server"
1081 keyword to see available options.
1082
Remi Gacogne47783ef2015-05-29 15:53:22 +02001083ssl-dh-param-file <file>
1084 This setting is only available when support for OpenSSL was built in. It sets
1085 the default DH parameters that are used during the SSL/TLS handshake when
1086 ephemeral Diffie-Hellman (DHE) key exchange is used, for all "bind" lines
Davor Ocelice9ed2812017-12-25 17:49:28 +01001087 which do not explicitly define theirs. It will be overridden by custom DH
Remi Gacogne47783ef2015-05-29 15:53:22 +02001088 parameters found in a bind certificate file if any. If custom DH parameters
Cyril Bonté307ee1e2015-09-28 23:16:06 +02001089 are not specified either by using ssl-dh-param-file or by setting them
1090 directly in the certificate file, pre-generated DH parameters of the size
1091 specified by tune.ssl.default-dh-param will be used. Custom parameters are
1092 known to be more secure and therefore their use is recommended.
Remi Gacogne47783ef2015-05-29 15:53:22 +02001093 Custom DH parameters may be generated by using the OpenSSL command
1094 "openssl dhparam <size>", where size should be at least 2048, as 1024-bit DH
1095 parameters should not be considered secure anymore.
1096
Emeric Brun850efd52014-01-29 12:24:34 +01001097ssl-server-verify [none|required]
1098 The default behavior for SSL verify on servers side. If specified to 'none',
1099 servers certificates are not verified. The default is 'required' except if
1100 forced using cmdline option '-dV'.
1101
Willy Tarreauabb175f2012-09-24 12:43:26 +02001102stats socket [<address:port>|<path>] [param*]
1103 Binds a UNIX socket to <path> or a TCPv4/v6 address to <address:port>.
1104 Connections to this socket will return various statistics outputs and even
1105 allow some commands to be issued to change some runtime settings. Please
Willy Tarreau1af20c72017-06-23 16:01:14 +02001106 consult section 9.3 "Unix Socket commands" of Management Guide for more
Kevin Decherf949c7202015-10-13 23:26:44 +02001107 details.
Willy Tarreau6162db22009-10-10 17:13:00 +02001108
Willy Tarreauabb175f2012-09-24 12:43:26 +02001109 All parameters supported by "bind" lines are supported, for instance to
1110 restrict access to some users or their access rights. Please consult
1111 section 5.1 for more information.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001112
1113stats timeout <timeout, in milliseconds>
1114 The default timeout on the stats socket is set to 10 seconds. It is possible
1115 to change this value with "stats timeout". The value must be passed in
Willy Tarreaubefdff12007-12-02 22:27:38 +01001116 milliseconds, or be suffixed by a time unit among { us, ms, s, m, h, d }.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001117
1118stats maxconn <connections>
1119 By default, the stats socket is limited to 10 concurrent connections. It is
1120 possible to change this value with "stats maxconn".
1121
Willy Tarreau6a06a402007-07-15 20:15:28 +02001122uid <number>
1123 Changes the process' user ID to <number>. It is recommended that the user ID
1124 is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
1125 be started with superuser privileges in order to be able to switch to another
1126 one. See also "gid" and "user".
1127
1128ulimit-n <number>
1129 Sets the maximum number of per-process file-descriptors to <number>. By
1130 default, it is automatically computed, so it is recommended not to use this
1131 option.
1132
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001133unix-bind [ prefix <prefix> ] [ mode <mode> ] [ user <user> ] [ uid <uid> ]
1134 [ group <group> ] [ gid <gid> ]
1135
1136 Fixes common settings to UNIX listening sockets declared in "bind" statements.
1137 This is mainly used to simplify declaration of those UNIX sockets and reduce
1138 the risk of errors, since those settings are most commonly required but are
1139 also process-specific. The <prefix> setting can be used to force all socket
1140 path to be relative to that directory. This might be needed to access another
1141 component's chroot. Note that those paths are resolved before haproxy chroots
1142 itself, so they are absolute. The <mode>, <user>, <uid>, <group> and <gid>
1143 all have the same meaning as their homonyms used by the "bind" statement. If
1144 both are specified, the "bind" statement has priority, meaning that the
1145 "unix-bind" settings may be seen as process-wide default settings.
1146
Willy Tarreau1d549722016-02-16 12:41:57 +01001147unsetenv [<name> ...]
1148 Removes environment variables specified in arguments. This can be useful to
1149 hide some sensitive information that are occasionally inherited from the
1150 user's environment during some operations. Variables which did not exist are
1151 silently ignored so that after the operation, it is certain that none of
1152 these variables remain. The changes immediately take effect so that the next
1153 line in the configuration file will not see these variables. See also
1154 "setenv", "presetenv", and "resetenv".
1155
Willy Tarreau6a06a402007-07-15 20:15:28 +02001156user <user name>
1157 Similar to "uid" but uses the UID of user name <user name> from /etc/passwd.
1158 See also "uid" and "group".
1159
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02001160node <name>
1161 Only letters, digits, hyphen and underscore are allowed, like in DNS names.
1162
1163 This statement is useful in HA configurations where two or more processes or
1164 servers share the same IP address. By setting a different node-name on all
1165 nodes, it becomes easy to immediately spot what server is handling the
1166 traffic.
1167
1168description <text>
1169 Add a text that describes the instance.
1170
1171 Please note that it is required to escape certain characters (# for example)
1172 and this text is inserted into a html page so you should avoid using
1173 "<" and ">" characters.
1174
Thomas Holmesdb04f192015-05-18 13:21:39 +0100117551degrees-data-file <file path>
1176 The path of the 51Degrees data file to provide device detection services. The
Davor Ocelice9ed2812017-12-25 17:49:28 +01001177 file should be unzipped and accessible by HAProxy with relevant permissions.
Thomas Holmesdb04f192015-05-18 13:21:39 +01001178
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001179 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001180 compiled with USE_51DEGREES.
1181
Ben Shillitof25e8e52016-12-02 14:25:37 +0000118251degrees-property-name-list [<string> ...]
Thomas Holmesdb04f192015-05-18 13:21:39 +01001183 A list of 51Degrees property names to be load from the dataset. A full list
1184 of names is available on the 51Degrees website:
1185 https://51degrees.com/resources/property-dictionary
1186
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001187 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001188 compiled with USE_51DEGREES.
1189
Dragan Dosen93b38d92015-06-29 16:43:25 +0200119051degrees-property-separator <char>
Thomas Holmesdb04f192015-05-18 13:21:39 +01001191 A char that will be appended to every property value in a response header
1192 containing 51Degrees results. If not set that will be set as ','.
1193
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001194 Please note that this option is only available when haproxy has been
1195 compiled with USE_51DEGREES.
1196
119751degrees-cache-size <number>
1198 Sets the size of the 51Degrees converter cache to <number> entries. This
1199 is an LRU cache which reminds previous device detections and their results.
1200 By default, this cache is disabled.
1201
1202 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001203 compiled with USE_51DEGREES.
1204
scientiamobiled0027ed2016-11-04 10:55:08 +01001205wurfl-data-file <file path>
1206 The path of the WURFL data file to provide device detection services. The
1207 file should be accessible by HAProxy with relevant permissions.
1208
1209 Please note that this option is only available when haproxy has been compiled
1210 with USE_WURFL=1.
1211
1212wurfl-information-list [<capability>]*
1213 A space-delimited list of WURFL capabilities, virtual capabilities, property
1214 names we plan to use in injected headers. A full list of capability and
1215 virtual capability names is available on the Scientiamobile website :
1216
1217 https://www.scientiamobile.com/wurflCapability
1218
1219 Valid WURFL properties are:
1220 - wurfl_id Contains the device ID of the matched device.
1221
1222 - wurfl_root_id Contains the device root ID of the matched
1223 device.
1224
1225 - wurfl_isdevroot Tells if the matched device is a root device.
1226 Possible values are "TRUE" or "FALSE".
1227
1228 - wurfl_useragent The original useragent coming with this
1229 particular web request.
1230
1231 - wurfl_api_version Contains a string representing the currently
1232 used Libwurfl API version.
1233
1234 - wurfl_engine_target Contains a string representing the currently
1235 set WURFL Engine Target. Possible values are
1236 "HIGH_ACCURACY", "HIGH_PERFORMANCE", "INVALID".
1237
1238 - wurfl_info A string containing information on the parsed
1239 wurfl.xml and its full path.
1240
1241 - wurfl_last_load_time Contains the UNIX timestamp of the last time
1242 WURFL has been loaded successfully.
1243
1244 - wurfl_normalized_useragent The normalized useragent.
1245
1246 - wurfl_useragent_priority The user agent priority used by WURFL.
1247
1248 Please note that this option is only available when haproxy has been compiled
1249 with USE_WURFL=1.
1250
1251wurfl-information-list-separator <char>
1252 A char that will be used to separate values in a response header containing
1253 WURFL results. If not set that a comma (',') will be used by default.
1254
1255 Please note that this option is only available when haproxy has been compiled
1256 with USE_WURFL=1.
1257
1258wurfl-patch-file [<file path>]
1259 A list of WURFL patch file paths. Note that patches are loaded during startup
1260 thus before the chroot.
1261
1262 Please note that this option is only available when haproxy has been compiled
1263 with USE_WURFL=1.
1264
1265wurfl-engine-mode { accuracy | performance }
1266 Sets the WURFL engine target. You can choose between 'accuracy' or
1267 'performance' targets. In performance mode, desktop web browser detection is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001268 done programmatically without referencing the WURFL data. As a result, most
scientiamobiled0027ed2016-11-04 10:55:08 +01001269 desktop web browsers are returned as generic_web_browser WURFL ID for
Davor Ocelice9ed2812017-12-25 17:49:28 +01001270 performance. If either performance or accuracy are not defined, performance
scientiamobiled0027ed2016-11-04 10:55:08 +01001271 mode is enabled by default.
1272
1273 Please note that this option is only available when haproxy has been compiled
1274 with USE_WURFL=1.
1275
1276wurfl-cache-size <U>[,<D>]
1277 Sets the WURFL caching strategy. Here <U> is the Useragent cache size, and
1278 <D> is the internal device cache size. There are three possibilities here :
1279 - "0" : no cache is used.
1280 - <U> : the Single LRU cache is used, the size is expressed in elements.
1281 - <U>,<D> : the Double LRU cache is used, both sizes are in elements. This is
1282 the highest performing option.
1283
1284 Please note that this option is only available when haproxy has been compiled
1285 with USE_WURFL=1.
1286
1287wurfl-useragent-priority { plain | sideloaded_browser }
1288 Tells WURFL if it should prioritize use of the plain user agent ('plain')
1289 over the default sideloaded browser user agent ('sideloaded_browser').
1290
1291 Please note that this option is only available when haproxy has been compiled
1292 with USE_WURFL=1.
1293
Willy Tarreau6a06a402007-07-15 20:15:28 +02001294
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012953.2. Performance tuning
Willy Tarreau6a06a402007-07-15 20:15:28 +02001296-----------------------
1297
Willy Tarreaubeb859a2018-11-22 18:07:59 +01001298busy-polling
1299 In some situations, especially when dealing with low latency on processors
1300 supporting a variable frequency or when running inside virtual machines, each
1301 time the process waits for an I/O using the poller, the processor goes back
1302 to sleep or is offered to another VM for a long time, and it causes
1303 excessively high latencies. This option provides a solution preventing the
1304 processor from sleeping by always using a null timeout on the pollers. This
1305 results in a significant latency reduction (30 to 100 microseconds observed)
1306 at the expense of a risk to overheat the processor. It may even be used with
1307 threads, in which case improperly bound threads may heavily conflict,
1308 resulting in a worse performance and high values for the CPU stolen fields
1309 in "show info" output, indicating which threads are misconfigured. It is
1310 important not to let the process run on the same processor as the network
1311 interrupts when this option is used. It is also better to avoid using it on
1312 multiple CPU threads sharing the same core. This option is disabled by
1313 default. If it has been enabled, it may still be forcibly disabled by
1314 prefixing it with the "no" keyword. It is ignored by the "select" and
1315 "poll" pollers.
1316
Willy Tarreau1746eec2014-04-25 10:46:47 +02001317max-spread-checks <delay in milliseconds>
1318 By default, haproxy tries to spread the start of health checks across the
1319 smallest health check interval of all the servers in a farm. The principle is
1320 to avoid hammering services running on the same server. But when using large
1321 check intervals (10 seconds or more), the last servers in the farm take some
1322 time before starting to be tested, which can be a problem. This parameter is
1323 used to enforce an upper bound on delay between the first and the last check,
1324 even if the servers' check intervals are larger. When servers run with
1325 shorter intervals, their intervals will be respected though.
1326
Willy Tarreau6a06a402007-07-15 20:15:28 +02001327maxconn <number>
1328 Sets the maximum per-process number of concurrent connections to <number>. It
1329 is equivalent to the command-line argument "-n". Proxies will stop accepting
1330 connections when this limit is reached. The "ulimit-n" parameter is
Willy Tarreau8274e102014-06-19 15:31:25 +02001331 automatically adjusted according to this value. See also "ulimit-n". Note:
1332 the "select" poller cannot reliably use more than 1024 file descriptors on
1333 some platforms. If your platform only supports select and reports "select
1334 FAILED" on startup, you need to reduce maxconn until it works (slightly
Willy Tarreaud0256482015-01-15 21:45:22 +01001335 below 500 in general). If this value is not set, it will default to the value
1336 set in DEFAULT_MAXCONN at build time (reported in haproxy -vv) if no memory
1337 limit is enforced, or will be computed based on the memory limit, the buffer
1338 size, memory allocated to compression, SSL cache size, and use or not of SSL
1339 and the associated maxsslconn (which can also be automatic).
Willy Tarreau6a06a402007-07-15 20:15:28 +02001340
Willy Tarreau81c25d02011-09-07 15:17:21 +02001341maxconnrate <number>
1342 Sets the maximum per-process number of connections per second to <number>.
1343 Proxies will stop accepting connections when this limit is reached. It can be
1344 used to limit the global capacity regardless of each frontend capacity. It is
1345 important to note that this can only be used as a service protection measure,
1346 as there will not necessarily be a fair share between frontends when the
1347 limit is reached, so it's a good idea to also limit each frontend to some
1348 value close to its expected share. Also, lowering tune.maxaccept can improve
1349 fairness.
1350
William Lallemandd85f9172012-11-09 17:05:39 +01001351maxcomprate <number>
1352 Sets the maximum per-process input compression rate to <number> kilobytes
Davor Ocelice9ed2812017-12-25 17:49:28 +01001353 per second. For each session, if the maximum is reached, the compression
William Lallemandd85f9172012-11-09 17:05:39 +01001354 level will be decreased during the session. If the maximum is reached at the
1355 beginning of a session, the session will not compress at all. If the maximum
1356 is not reached, the compression level will be increased up to
Davor Ocelice9ed2812017-12-25 17:49:28 +01001357 tune.comp.maxlevel. A value of zero means there is no limit, this is the
William Lallemandd85f9172012-11-09 17:05:39 +01001358 default value.
1359
William Lallemand072a2bf2012-11-20 17:01:01 +01001360maxcompcpuusage <number>
1361 Sets the maximum CPU usage HAProxy can reach before stopping the compression
1362 for new requests or decreasing the compression level of current requests.
1363 It works like 'maxcomprate' but measures CPU usage instead of incoming data
1364 bandwidth. The value is expressed in percent of the CPU used by haproxy. In
1365 case of multiple processes (nbproc > 1), each process manages its individual
1366 usage. A value of 100 disable the limit. The default value is 100. Setting
1367 a lower value will prevent the compression work from slowing the whole
1368 process down and from introducing high latencies.
1369
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001370maxpipes <number>
1371 Sets the maximum per-process number of pipes to <number>. Currently, pipes
1372 are only used by kernel-based tcp splicing. Since a pipe contains two file
1373 descriptors, the "ulimit-n" value will be increased accordingly. The default
1374 value is maxconn/4, which seems to be more than enough for most heavy usages.
1375 The splice code dynamically allocates and releases pipes, and can fall back
1376 to standard copy, so setting this value too low may only impact performance.
1377
Willy Tarreau93e7c002013-10-07 18:51:07 +02001378maxsessrate <number>
1379 Sets the maximum per-process number of sessions per second to <number>.
1380 Proxies will stop accepting connections when this limit is reached. It can be
1381 used to limit the global capacity regardless of each frontend capacity. It is
1382 important to note that this can only be used as a service protection measure,
1383 as there will not necessarily be a fair share between frontends when the
1384 limit is reached, so it's a good idea to also limit each frontend to some
1385 value close to its expected share. Also, lowering tune.maxaccept can improve
1386 fairness.
1387
Willy Tarreau403edff2012-09-06 11:58:37 +02001388maxsslconn <number>
1389 Sets the maximum per-process number of concurrent SSL connections to
1390 <number>. By default there is no SSL-specific limit, which means that the
1391 global maxconn setting will apply to all connections. Setting this limit
1392 avoids having openssl use too much memory and crash when malloc returns NULL
1393 (since it unfortunately does not reliably check for such conditions). Note
1394 that the limit applies both to incoming and outgoing connections, so one
1395 connection which is deciphered then ciphered accounts for 2 SSL connections.
Willy Tarreaud0256482015-01-15 21:45:22 +01001396 If this value is not set, but a memory limit is enforced, this value will be
1397 automatically computed based on the memory limit, maxconn, the buffer size,
1398 memory allocated to compression, SSL cache size, and use of SSL in either
1399 frontends, backends or both. If neither maxconn nor maxsslconn are specified
1400 when there is a memory limit, haproxy will automatically adjust these values
1401 so that 100% of the connections can be made over SSL with no risk, and will
1402 consider the sides where it is enabled (frontend, backend, both).
Willy Tarreau403edff2012-09-06 11:58:37 +02001403
Willy Tarreaue43d5322013-10-07 20:01:52 +02001404maxsslrate <number>
1405 Sets the maximum per-process number of SSL sessions per second to <number>.
1406 SSL listeners will stop accepting connections when this limit is reached. It
1407 can be used to limit the global SSL CPU usage regardless of each frontend
1408 capacity. It is important to note that this can only be used as a service
1409 protection measure, as there will not necessarily be a fair share between
1410 frontends when the limit is reached, so it's a good idea to also limit each
1411 frontend to some value close to its expected share. It is also important to
1412 note that the sessions are accounted before they enter the SSL stack and not
1413 after, which also protects the stack against bad handshakes. Also, lowering
1414 tune.maxaccept can improve fairness.
1415
William Lallemand9d5f5482012-11-07 16:12:57 +01001416maxzlibmem <number>
1417 Sets the maximum amount of RAM in megabytes per process usable by the zlib.
1418 When the maximum amount is reached, future sessions will not compress as long
1419 as RAM is unavailable. When sets to 0, there is no limit.
William Lallemande3a7d992012-11-20 11:25:20 +01001420 The default value is 0. The value is available in bytes on the UNIX socket
1421 with "show info" on the line "MaxZlibMemUsage", the memory used by zlib is
1422 "ZlibMemUsage" in bytes.
1423
Willy Tarreau6a06a402007-07-15 20:15:28 +02001424noepoll
1425 Disables the use of the "epoll" event polling system on Linux. It is
1426 equivalent to the command-line argument "-de". The next polling system
Willy Tarreaue9f49e72012-11-11 17:42:00 +01001427 used will generally be "poll". See also "nopoll".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001428
1429nokqueue
1430 Disables the use of the "kqueue" event polling system on BSD. It is
1431 equivalent to the command-line argument "-dk". The next polling system
1432 used will generally be "poll". See also "nopoll".
1433
1434nopoll
1435 Disables the use of the "poll" event polling system. It is equivalent to the
1436 command-line argument "-dp". The next polling system used will be "select".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001437 It should never be needed to disable "poll" since it's available on all
Willy Tarreaue9f49e72012-11-11 17:42:00 +01001438 platforms supported by HAProxy. See also "nokqueue" and "noepoll".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001439
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001440nosplice
1441 Disables the use of kernel tcp splicing between sockets on Linux. It is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001442 equivalent to the command line argument "-dS". Data will then be copied
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001443 using conventional and more portable recv/send calls. Kernel tcp splicing is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001444 limited to some very recent instances of kernel 2.6. Most versions between
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001445 2.6.25 and 2.6.28 are buggy and will forward corrupted data, so they must not
1446 be used. This option makes it easier to globally disable kernel splicing in
1447 case of doubt. See also "option splice-auto", "option splice-request" and
1448 "option splice-response".
1449
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001450nogetaddrinfo
1451 Disables the use of getaddrinfo(3) for name resolving. It is equivalent to
1452 the command line argument "-dG". Deprecated gethostbyname(3) will be used.
1453
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +00001454noreuseport
1455 Disables the use of SO_REUSEPORT - see socket(7). It is equivalent to the
1456 command line argument "-dR".
1457
Willy Tarreau75c62c22018-11-22 11:02:09 +01001458profiling.tasks { on | off }
1459 Enables ('on') or disables ('off') per-task CPU profiling. CPU profiling per
1460 task can be very convenient to report where the time is spent and which
1461 requests have what effect on which other request. It is not enabled by
1462 default as it may consume a little bit extra CPU. This requires a system
1463 supporting the clock_gettime(2) syscall with clock identifiers
1464 CLOCK_MONOTONIC and CLOCK_THREAD_CPUTIME_ID, otherwise the reported time will
1465 be zero. This option may be changed at run time using "set profiling" on the
1466 CLI.
1467
Willy Tarreaufe255b72007-10-14 23:09:26 +02001468spread-checks <0..50, in percent>
Simon Hormand60d6912013-11-25 10:46:36 +09001469 Sometimes it is desirable to avoid sending agent and health checks to
1470 servers at exact intervals, for instance when many logical servers are
1471 located on the same physical server. With the help of this parameter, it
1472 becomes possible to add some randomness in the check interval between 0
1473 and +/- 50%. A value between 2 and 5 seems to show good results. The
1474 default value remains at 0.
Willy Tarreaufe255b72007-10-14 23:09:26 +02001475
Davor Ocelice9ed2812017-12-25 17:49:28 +01001476ssl-engine <name> [algo <comma-separated list of algorithms>]
Grant Zhang872f9c22017-01-21 01:10:18 +00001477 Sets the OpenSSL engine to <name>. List of valid values for <name> may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01001478 obtained using the command "openssl engine". This statement may be used
Grant Zhang872f9c22017-01-21 01:10:18 +00001479 multiple times, it will simply enable multiple crypto engines. Referencing an
1480 unsupported engine will prevent haproxy from starting. Note that many engines
1481 will lead to lower HTTPS performance than pure software with recent
1482 processors. The optional command "algo" sets the default algorithms an ENGINE
1483 will supply using the OPENSSL function ENGINE_set_default_string(). A value
Davor Ocelice9ed2812017-12-25 17:49:28 +01001484 of "ALL" uses the engine for all cryptographic operations. If no list of
1485 algo is specified then the value of "ALL" is used. A comma-separated list
Grant Zhang872f9c22017-01-21 01:10:18 +00001486 of different algorithms may be specified, including: RSA, DSA, DH, EC, RAND,
1487 CIPHERS, DIGESTS, PKEY, PKEY_CRYPTO, PKEY_ASN1. This is the same format that
1488 openssl configuration file uses:
1489 https://www.openssl.org/docs/man1.0.2/apps/config.html
1490
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00001491ssl-mode-async
1492 Adds SSL_MODE_ASYNC mode to the SSL context. This enables asynchronous TLS
Emeric Brun3854e012017-05-17 20:42:48 +02001493 I/O operations if asynchronous capable SSL engines are used. The current
Emeric Brunb5e42a82017-06-06 12:35:14 +00001494 implementation supports a maximum of 32 engines. The Openssl ASYNC API
1495 doesn't support moving read/write buffers and is not compliant with
1496 haproxy's buffer management. So the asynchronous mode is disabled on
1497 read/write operations (it is only enabled during initial and reneg
1498 handshakes).
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00001499
Willy Tarreau33cb0652014-12-23 22:52:37 +01001500tune.buffers.limit <number>
1501 Sets a hard limit on the number of buffers which may be allocated per process.
1502 The default value is zero which means unlimited. The minimum non-zero value
1503 will always be greater than "tune.buffers.reserve" and should ideally always
1504 be about twice as large. Forcing this value can be particularly useful to
1505 limit the amount of memory a process may take, while retaining a sane
Davor Ocelice9ed2812017-12-25 17:49:28 +01001506 behavior. When this limit is reached, sessions which need a buffer wait for
Willy Tarreau33cb0652014-12-23 22:52:37 +01001507 another one to be released by another session. Since buffers are dynamically
1508 allocated and released, the waiting time is very short and not perceptible
1509 provided that limits remain reasonable. In fact sometimes reducing the limit
1510 may even increase performance by increasing the CPU cache's efficiency. Tests
1511 have shown good results on average HTTP traffic with a limit to 1/10 of the
1512 expected global maxconn setting, which also significantly reduces memory
1513 usage. The memory savings come from the fact that a number of connections
1514 will not allocate 2*tune.bufsize. It is best not to touch this value unless
1515 advised to do so by an haproxy core developer.
1516
Willy Tarreau1058ae72014-12-23 22:40:40 +01001517tune.buffers.reserve <number>
1518 Sets the number of buffers which are pre-allocated and reserved for use only
1519 during memory shortage conditions resulting in failed memory allocations. The
1520 minimum value is 2 and is also the default. There is no reason a user would
1521 want to change this value, it's mostly aimed at haproxy core developers.
1522
Willy Tarreau27a674e2009-08-17 07:23:33 +02001523tune.bufsize <number>
1524 Sets the buffer size to this size (in bytes). Lower values allow more
1525 sessions to coexist in the same amount of RAM, and higher values allow some
1526 applications with very large cookies to work. The default value is 16384 and
1527 can be changed at build time. It is strongly recommended not to change this
1528 from the default value, as very low values will break some services such as
1529 statistics, and values larger than default size will increase memory usage,
1530 possibly causing the system to run out of memory. At least the global maxconn
Willy Tarreau45a66cc2017-11-24 11:28:00 +01001531 parameter should be decreased by the same factor as this one is increased. In
1532 addition, use of HTTP/2 mandates that this value must be 16384 or more. If an
1533 HTTP request is larger than (tune.bufsize - tune.maxrewrite), haproxy will
Dmitry Sivachenkof6f4f7b2012-10-21 18:10:25 +04001534 return HTTP 400 (Bad Request) error. Similarly if an HTTP response is larger
Willy Tarreauc77d3642018-12-12 06:19:42 +01001535 than this size, haproxy will return HTTP 502 (Bad Gateway). Note that the
1536 value set using this parameter will automatically be rounded up to the next
1537 multiple of 8 on 32-bit machines and 16 on 64-bit machines.
Willy Tarreau27a674e2009-08-17 07:23:33 +02001538
Willy Tarreau43961d52010-10-04 20:39:20 +02001539tune.chksize <number>
1540 Sets the check buffer size to this size (in bytes). Higher values may help
1541 find string or regex patterns in very large pages, though doing so may imply
1542 more memory and CPU usage. The default value is 16384 and can be changed at
1543 build time. It is not recommended to change this value, but to use better
1544 checks whenever possible.
1545
William Lallemandf3747832012-11-09 12:33:10 +01001546tune.comp.maxlevel <number>
1547 Sets the maximum compression level. The compression level affects CPU
1548 usage during compression. This value affects CPU usage during compression.
1549 Each session using compression initializes the compression algorithm with
1550 this value. The default value is 1.
1551
Willy Tarreaufe20e5b2017-07-27 11:42:14 +02001552tune.h2.header-table-size <number>
1553 Sets the HTTP/2 dynamic header table size. It defaults to 4096 bytes and
1554 cannot be larger than 65536 bytes. A larger value may help certain clients
1555 send more compact requests, depending on their capabilities. This amount of
1556 memory is consumed for each HTTP/2 connection. It is recommended not to
1557 change it.
1558
Willy Tarreaue6baec02017-07-27 11:45:11 +02001559tune.h2.initial-window-size <number>
1560 Sets the HTTP/2 initial window size, which is the number of bytes the client
Davor Ocelice9ed2812017-12-25 17:49:28 +01001561 can upload before waiting for an acknowledgment from haproxy. This setting
1562 only affects payload contents (i.e. the body of POST requests), not headers.
Willy Tarreaue6baec02017-07-27 11:45:11 +02001563 The default value is 65535, which roughly allows up to 5 Mbps of upload
1564 bandwidth per client over a network showing a 100 ms ping time, or 500 Mbps
1565 over a 1-ms local network. It can make sense to increase this value to allow
1566 faster uploads, or to reduce it to increase fairness when dealing with many
1567 clients. It doesn't affect resource usage.
1568
Willy Tarreau5242ef82017-07-27 11:47:28 +02001569tune.h2.max-concurrent-streams <number>
1570 Sets the HTTP/2 maximum number of concurrent streams per connection (ie the
1571 number of outstanding requests on a single connection). The default value is
1572 100. A larger one may slightly improve page load time for complex sites when
1573 visited over high latency networks, but increases the amount of resources a
1574 single client may allocate. A value of zero disables the limit so a single
1575 client may create as many streams as allocatable by haproxy. It is highly
1576 recommended not to change this value.
1577
Willy Tarreau193b8c62012-11-22 00:17:38 +01001578tune.http.cookielen <number>
1579 Sets the maximum length of captured cookies. This is the maximum value that
1580 the "capture cookie xxx len yyy" will be allowed to take, and any upper value
1581 will automatically be truncated to this one. It is important not to set too
1582 high a value because all cookie captures still allocate this size whatever
1583 their configured value (they share a same pool). This value is per request
1584 per response, so the memory allocated is twice this value per connection.
1585 When not specified, the limit is set to 63 characters. It is recommended not
1586 to change this value.
1587
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001588tune.http.logurilen <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001589 Sets the maximum length of request URI in logs. This prevents truncating long
1590 request URIs with valuable query strings in log lines. This is not related
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001591 to syslog limits. If you increase this limit, you may also increase the
Davor Ocelice9ed2812017-12-25 17:49:28 +01001592 'log ... len yyy' parameter. Your syslog daemon may also need specific
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001593 configuration directives too.
1594 The default value is 1024.
1595
Willy Tarreauac1932d2011-10-24 19:14:41 +02001596tune.http.maxhdr <number>
1597 Sets the maximum number of headers in a request. When a request comes with a
1598 number of headers greater than this value (including the first line), it is
1599 rejected with a "400 Bad Request" status code. Similarly, too large responses
1600 are blocked with "502 Bad Gateway". The default value is 101, which is enough
1601 for all usages, considering that the widely deployed Apache server uses the
1602 same limit. It can be useful to push this limit further to temporarily allow
Christopher Faulet50174f32017-06-21 16:31:35 +02001603 a buggy application to work by the time it gets fixed. The accepted range is
1604 1..32767. Keep in mind that each new header consumes 32bits of memory for
1605 each session, so don't push this limit too high.
Willy Tarreauac1932d2011-10-24 19:14:41 +02001606
Willy Tarreau7e312732014-02-12 16:35:14 +01001607tune.idletimer <timeout>
1608 Sets the duration after which haproxy will consider that an empty buffer is
1609 probably associated with an idle stream. This is used to optimally adjust
1610 some packet sizes while forwarding large and small data alternatively. The
1611 decision to use splice() or to send large buffers in SSL is modulated by this
1612 parameter. The value is in milliseconds between 0 and 65535. A value of zero
1613 means that haproxy will not try to detect idle streams. The default is 1000,
Davor Ocelice9ed2812017-12-25 17:49:28 +01001614 which seems to correctly detect end user pauses (e.g. read a page before
Willy Tarreau7e312732014-02-12 16:35:14 +01001615 clicking). There should be not reason for changing this value. Please check
1616 tune.ssl.maxrecord below.
1617
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001618tune.lua.forced-yield <number>
1619 This directive forces the Lua engine to execute a yield each <number> of
Tim Düsterhus4896c442016-11-29 02:15:19 +01001620 instructions executed. This permits interrupting a long script and allows the
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001621 HAProxy scheduler to process other tasks like accepting connections or
1622 forwarding traffic. The default value is 10000 instructions. If HAProxy often
Davor Ocelice9ed2812017-12-25 17:49:28 +01001623 executes some Lua code but more responsiveness is required, this value can be
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001624 lowered. If the Lua code is quite long and its result is absolutely required
1625 to process the data, the <number> can be increased.
1626
Willy Tarreau32f61e22015-03-18 17:54:59 +01001627tune.lua.maxmem
1628 Sets the maximum amount of RAM in megabytes per process usable by Lua. By
1629 default it is zero which means unlimited. It is important to set a limit to
1630 ensure that a bug in a script will not result in the system running out of
1631 memory.
1632
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001633tune.lua.session-timeout <timeout>
1634 This is the execution timeout for the Lua sessions. This is useful for
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001635 preventing infinite loops or spending too much time in Lua. This timeout
1636 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001637 not taken in account. The default timeout is 4s.
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001638
1639tune.lua.task-timeout <timeout>
1640 Purpose is the same as "tune.lua.session-timeout", but this timeout is
1641 dedicated to the tasks. By default, this timeout isn't set because a task may
1642 remain alive during of the lifetime of HAProxy. For example, a task used to
1643 check servers.
1644
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001645tune.lua.service-timeout <timeout>
1646 This is the execution timeout for the Lua services. This is useful for
1647 preventing infinite loops or spending too much time in Lua. This timeout
1648 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001649 not taken in account. The default timeout is 4s.
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001650
Willy Tarreaua0250ba2008-01-06 11:22:57 +01001651tune.maxaccept <number>
Willy Tarreau16a21472012-11-19 12:39:59 +01001652 Sets the maximum number of consecutive connections a process may accept in a
1653 row before switching to other work. In single process mode, higher numbers
1654 give better performance at high connection rates. However in multi-process
1655 modes, keeping a bit of fairness between processes generally is better to
1656 increase performance. This value applies individually to each listener, so
1657 that the number of processes a listener is bound to is taken into account.
1658 This value defaults to 64. In multi-process mode, it is divided by twice
1659 the number of processes the listener is bound to. Setting this value to -1
1660 completely disables the limitation. It should normally not be needed to tweak
1661 this value.
Willy Tarreaua0250ba2008-01-06 11:22:57 +01001662
1663tune.maxpollevents <number>
1664 Sets the maximum amount of events that can be processed at once in a call to
1665 the polling system. The default value is adapted to the operating system. It
1666 has been noticed that reducing it below 200 tends to slightly decrease
1667 latency at the expense of network bandwidth, and increasing it above 200
1668 tends to trade latency for slightly increased bandwidth.
1669
Willy Tarreau27a674e2009-08-17 07:23:33 +02001670tune.maxrewrite <number>
1671 Sets the reserved buffer space to this size in bytes. The reserved space is
1672 used for header rewriting or appending. The first reads on sockets will never
1673 fill more than bufsize-maxrewrite. Historically it has defaulted to half of
1674 bufsize, though that does not make much sense since there are rarely large
1675 numbers of headers to add. Setting it too high prevents processing of large
1676 requests or responses. Setting it too low prevents addition of new headers
1677 to already large requests or to POST requests. It is generally wise to set it
1678 to about 1024. It is automatically readjusted to half of bufsize if it is
1679 larger than that. This means you don't have to worry about it when changing
1680 bufsize.
1681
Willy Tarreauf3045d22015-04-29 16:24:50 +02001682tune.pattern.cache-size <number>
1683 Sets the size of the pattern lookup cache to <number> entries. This is an LRU
1684 cache which reminds previous lookups and their results. It is used by ACLs
1685 and maps on slow pattern lookups, namely the ones using the "sub", "reg",
1686 "dir", "dom", "end", "bin" match methods as well as the case-insensitive
1687 strings. It applies to pattern expressions which means that it will be able
1688 to memorize the result of a lookup among all the patterns specified on a
1689 configuration line (including all those loaded from files). It automatically
1690 invalidates entries which are updated using HTTP actions or on the CLI. The
1691 default cache size is set to 10000 entries, which limits its footprint to
1692 about 5 MB on 32-bit systems and 8 MB on 64-bit systems. There is a very low
1693 risk of collision in this cache, which is in the order of the size of the
1694 cache divided by 2^64. Typically, at 10000 requests per second with the
1695 default cache size of 10000 entries, there's 1% chance that a brute force
1696 attack could cause a single collision after 60 years, or 0.1% after 6 years.
1697 This is considered much lower than the risk of a memory corruption caused by
1698 aging components. If this is not acceptable, the cache can be disabled by
1699 setting this parameter to 0.
1700
Willy Tarreaubd9a0a72011-10-23 21:14:29 +02001701tune.pipesize <number>
1702 Sets the kernel pipe buffer size to this size (in bytes). By default, pipes
1703 are the default size for the system. But sometimes when using TCP splicing,
1704 it can improve performance to increase pipe sizes, especially if it is
1705 suspected that pipes are not filled and that many calls to splice() are
1706 performed. This has an impact on the kernel's memory footprint, so this must
1707 not be changed if impacts are not understood.
1708
Willy Tarreaue803de22010-01-21 17:43:04 +01001709tune.rcvbuf.client <number>
1710tune.rcvbuf.server <number>
1711 Forces the kernel socket receive buffer size on the client or the server side
1712 to the specified value in bytes. This value applies to all TCP/HTTP frontends
1713 and backends. It should normally never be set, and the default size (0) lets
1714 the kernel autotune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01001715 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01001716 order to save kernel memory by preventing it from buffering too large amounts
1717 of received data. Lower values will significantly increase CPU usage though.
1718
Willy Tarreaub22fc302015-12-14 12:04:35 +01001719tune.recv_enough <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001720 HAProxy uses some hints to detect that a short read indicates the end of the
Willy Tarreaub22fc302015-12-14 12:04:35 +01001721 socket buffers. One of them is that a read returns more than <recv_enough>
1722 bytes, which defaults to 10136 (7 segments of 1448 each). This default value
1723 may be changed by this setting to better deal with workloads involving lots
1724 of short messages such as telnet or SSH sessions.
1725
Olivier Houchard1599b802018-05-24 18:59:04 +02001726tune.runqueue-depth <number>
1727 Sets the maxinum amount of task that can be processed at once when running
1728 tasks. The default value is 200. Increasing it may incur latency when
1729 dealing with I/Os, making it too small can incur extra overhead.
1730
Willy Tarreaue803de22010-01-21 17:43:04 +01001731tune.sndbuf.client <number>
1732tune.sndbuf.server <number>
1733 Forces the kernel socket send buffer size on the client or the server side to
1734 the specified value in bytes. This value applies to all TCP/HTTP frontends
1735 and backends. It should normally never be set, and the default size (0) lets
1736 the kernel autotune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01001737 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01001738 order to save kernel memory by preventing it from buffering too large amounts
1739 of received data. Lower values will significantly increase CPU usage though.
1740 Another use case is to prevent write timeouts with extremely slow clients due
1741 to the kernel waiting for a large part of the buffer to be read before
1742 notifying haproxy again.
1743
Willy Tarreau6ec58db2012-11-16 16:32:15 +01001744tune.ssl.cachesize <number>
Emeric Brunaf9619d2012-11-28 18:47:52 +01001745 Sets the size of the global SSL session cache, in a number of blocks. A block
1746 is large enough to contain an encoded session without peer certificate.
1747 An encoded session with peer certificate is stored in multiple blocks
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001748 depending on the size of the peer certificate. A block uses approximately
Emeric Brunaf9619d2012-11-28 18:47:52 +01001749 200 bytes of memory. The default value may be forced at build time, otherwise
Davor Ocelice9ed2812017-12-25 17:49:28 +01001750 defaults to 20000. When the cache is full, the most idle entries are purged
Emeric Brunaf9619d2012-11-28 18:47:52 +01001751 and reassigned. Higher values reduce the occurrence of such a purge, hence
1752 the number of CPU-intensive SSL handshakes by ensuring that all users keep
1753 their session as long as possible. All entries are pre-allocated upon startup
Emeric Brun22890a12012-12-28 14:41:32 +01001754 and are shared between all processes if "nbproc" is greater than 1. Setting
1755 this value to 0 disables the SSL session cache.
Willy Tarreau6ec58db2012-11-16 16:32:15 +01001756
Emeric Brun8dc60392014-05-09 13:52:00 +02001757tune.ssl.force-private-cache
Lukas Tribus27935782018-10-01 02:00:16 +02001758 This option disables SSL session cache sharing between all processes. It
Emeric Brun8dc60392014-05-09 13:52:00 +02001759 should normally not be used since it will force many renegotiations due to
1760 clients hitting a random process. But it may be required on some operating
1761 systems where none of the SSL cache synchronization method may be used. In
1762 this case, adding a first layer of hash-based load balancing before the SSL
1763 layer might limit the impact of the lack of session sharing.
1764
Emeric Brun4f65bff2012-11-16 15:11:00 +01001765tune.ssl.lifetime <timeout>
1766 Sets how long a cached SSL session may remain valid. This time is expressed
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001767 in seconds and defaults to 300 (5 min). It is important to understand that it
Emeric Brun4f65bff2012-11-16 15:11:00 +01001768 does not guarantee that sessions will last that long, because if the cache is
1769 full, the longest idle sessions will be purged despite their configured
1770 lifetime. The real usefulness of this setting is to prevent sessions from
1771 being used for too long.
1772
Willy Tarreaubfd59462013-02-21 07:46:09 +01001773tune.ssl.maxrecord <number>
1774 Sets the maximum amount of bytes passed to SSL_write() at a time. Default
1775 value 0 means there is no limit. Over SSL/TLS, the client can decipher the
1776 data only once it has received a full record. With large records, it means
1777 that clients might have to download up to 16kB of data before starting to
1778 process them. Limiting the value can improve page load times on browsers
1779 located over high latency or low bandwidth networks. It is suggested to find
1780 optimal values which fit into 1 or 2 TCP segments (generally 1448 bytes over
1781 Ethernet with TCP timestamps enabled, or 1460 when timestamps are disabled),
1782 keeping in mind that SSL/TLS add some overhead. Typical values of 1419 and
1783 2859 gave good results during tests. Use "strace -e trace=write" to find the
Davor Ocelice9ed2812017-12-25 17:49:28 +01001784 best value. HAProxy will automatically switch to this setting after an idle
Willy Tarreau7e312732014-02-12 16:35:14 +01001785 stream has been detected (see tune.idletimer above).
Willy Tarreaubfd59462013-02-21 07:46:09 +01001786
Remi Gacognef46cd6e2014-06-12 14:58:40 +02001787tune.ssl.default-dh-param <number>
1788 Sets the maximum size of the Diffie-Hellman parameters used for generating
1789 the ephemeral/temporary Diffie-Hellman key in case of DHE key exchange. The
1790 final size will try to match the size of the server's RSA (or DSA) key (e.g,
1791 a 2048 bits temporary DH key for a 2048 bits RSA key), but will not exceed
1792 this maximum value. Default value if 1024. Only 1024 or higher values are
1793 allowed. Higher values will increase the CPU load, and values greater than
1794 1024 bits are not supported by Java 7 and earlier clients. This value is not
Remi Gacogne47783ef2015-05-29 15:53:22 +02001795 used if static Diffie-Hellman parameters are supplied either directly
1796 in the certificate file or by using the ssl-dh-param-file parameter.
Remi Gacognef46cd6e2014-06-12 14:58:40 +02001797
Christopher Faulet31af49d2015-06-09 17:29:50 +02001798tune.ssl.ssl-ctx-cache-size <number>
1799 Sets the size of the cache used to store generated certificates to <number>
1800 entries. This is a LRU cache. Because generating a SSL certificate
1801 dynamically is expensive, they are cached. The default cache size is set to
1802 1000 entries.
1803
Thierry FOURNIER5bf77322017-02-25 12:45:22 +01001804tune.ssl.capture-cipherlist-size <number>
1805 Sets the maximum size of the buffer used for capturing client-hello cipher
1806 list. If the value is 0 (default value) the capture is disabled, otherwise
1807 a buffer is allocated for each SSL/TLS connection.
1808
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001809tune.vars.global-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01001810tune.vars.proc-max-size <size>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001811tune.vars.reqres-max-size <size>
1812tune.vars.sess-max-size <size>
1813tune.vars.txn-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01001814 These five tunes help to manage the maximum amount of memory used by the
1815 variables system. "global" limits the overall amount of memory available for
1816 all scopes. "proc" limits the memory for the process scope, "sess" limits the
1817 memory for the session scope, "txn" for the transaction scope, and "reqres"
1818 limits the memory for each request or response processing.
1819 Memory accounting is hierarchical, meaning more coarse grained limits include
1820 the finer grained ones: "proc" includes "sess", "sess" includes "txn", and
1821 "txn" includes "reqres".
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001822
Daniel Schneller0b547052016-03-21 20:46:57 +01001823 For example, when "tune.vars.sess-max-size" is limited to 100,
1824 "tune.vars.txn-max-size" and "tune.vars.reqres-max-size" cannot exceed
1825 100 either. If we create a variable "txn.var" that contains 100 bytes,
1826 all available space is consumed.
1827 Notice that exceeding the limits at runtime will not result in an error
1828 message, but values might be cut off or corrupted. So make sure to accurately
1829 plan for the amount of space needed to store all your variables.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001830
William Lallemanda509e4c2012-11-07 16:54:34 +01001831tune.zlib.memlevel <number>
1832 Sets the memLevel parameter in zlib initialization for each session. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001833 defines how much memory should be allocated for the internal compression
William Lallemanda509e4c2012-11-07 16:54:34 +01001834 state. A value of 1 uses minimum memory but is slow and reduces compression
Davor Ocelice9ed2812017-12-25 17:49:28 +01001835 ratio, a value of 9 uses maximum memory for optimal speed. Can be a value
William Lallemanda509e4c2012-11-07 16:54:34 +01001836 between 1 and 9. The default value is 8.
1837
1838tune.zlib.windowsize <number>
1839 Sets the window size (the size of the history buffer) as a parameter of the
1840 zlib initialization for each session. Larger values of this parameter result
Davor Ocelice9ed2812017-12-25 17:49:28 +01001841 in better compression at the expense of memory usage. Can be a value between
1842 8 and 15. The default value is 15.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001843
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018443.3. Debugging
1845--------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02001846
1847debug
1848 Enables debug mode which dumps to stdout all exchanges, and disables forking
1849 into background. It is the equivalent of the command-line argument "-d". It
1850 should never be used in a production configuration since it may prevent full
1851 system startup.
1852
1853quiet
1854 Do not display any message during startup. It is equivalent to the command-
1855 line argument "-q".
1856
Emeric Brunf099e792010-09-27 12:05:28 +02001857
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010018583.4. Userlists
1859--------------
1860It is possible to control access to frontend/backend/listen sections or to
1861http stats by allowing only authenticated and authorized users. To do this,
1862it is required to create at least one userlist and to define users.
1863
1864userlist <listname>
Cyril Bonté78caf842010-03-10 22:41:43 +01001865 Creates new userlist with name <listname>. Many independent userlists can be
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001866 used to store authentication & authorization data for independent customers.
1867
1868group <groupname> [users <user>,<user>,(...)]
Cyril Bonté78caf842010-03-10 22:41:43 +01001869 Adds group <groupname> to the current userlist. It is also possible to
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001870 attach users to this group by using a comma separated list of names
1871 proceeded by "users" keyword.
1872
Cyril Bontéf0c60612010-02-06 14:44:47 +01001873user <username> [password|insecure-password <password>]
1874 [groups <group>,<group>,(...)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001875 Adds user <username> to the current userlist. Both secure (encrypted) and
1876 insecure (unencrypted) passwords can be used. Encrypted passwords are
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01001877 evaluated using the crypt(3) function, so depending on the system's
1878 capabilities, different algorithms are supported. For example, modern Glibc
1879 based Linux systems support MD5, SHA-256, SHA-512, and, of course, the
1880 classic DES-based method of encrypting passwords.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001881
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01001882 Attention: Be aware that using encrypted passwords might cause significantly
1883 increased CPU usage, depending on the number of requests, and the algorithm
1884 used. For any of the hashed variants, the password for each request must
1885 be processed through the chosen algorithm, before it can be compared to the
1886 value specified in the config file. Most current algorithms are deliberately
1887 designed to be expensive to compute to achieve resistance against brute
1888 force attacks. They do not simply salt/hash the clear text password once,
1889 but thousands of times. This can quickly become a major factor in haproxy's
1890 overall CPU consumption!
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001891
1892 Example:
Cyril Bontéf0c60612010-02-06 14:44:47 +01001893 userlist L1
1894 group G1 users tiger,scott
1895 group G2 users xdb,scott
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001896
Cyril Bontéf0c60612010-02-06 14:44:47 +01001897 user tiger password $6$k6y3o.eP$JlKBx9za9667qe4(...)xHSwRv6J.C0/D7cV91
1898 user scott insecure-password elgato
1899 user xdb insecure-password hello
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001900
Cyril Bontéf0c60612010-02-06 14:44:47 +01001901 userlist L2
1902 group G1
1903 group G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001904
Cyril Bontéf0c60612010-02-06 14:44:47 +01001905 user tiger password $6$k6y3o.eP$JlKBx(...)xHSwRv6J.C0/D7cV91 groups G1
1906 user scott insecure-password elgato groups G1,G2
1907 user xdb insecure-password hello groups G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001908
1909 Please note that both lists are functionally identical.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001910
Emeric Brunf099e792010-09-27 12:05:28 +02001911
19123.5. Peers
Cyril Bontédc4d9032012-04-08 21:57:39 +02001913----------
Emeric Brun94900952015-06-11 18:25:54 +02001914It is possible to propagate entries of any data-types in stick-tables between
1915several haproxy instances over TCP connections in a multi-master fashion. Each
1916instance pushes its local updates and insertions to remote peers. The pushed
1917values overwrite remote ones without aggregation. Interrupted exchanges are
1918automatically detected and recovered from the last known point.
1919In addition, during a soft restart, the old process connects to the new one
1920using such a TCP connection to push all its entries before the new process
1921tries to connect to other peers. That ensures very fast replication during a
1922reload, it typically takes a fraction of a second even for large tables.
1923Note that Server IDs are used to identify servers remotely, so it is important
1924that configurations look similar or at least that the same IDs are forced on
1925each server on all participants.
Emeric Brunf099e792010-09-27 12:05:28 +02001926
1927peers <peersect>
Jamie Gloudon801a0a32012-08-25 00:18:33 -04001928 Creates a new peer list with name <peersect>. It is an independent section,
Emeric Brunf099e792010-09-27 12:05:28 +02001929 which is referenced by one or more stick-tables.
1930
Willy Tarreau77e4bd12015-05-01 20:02:17 +02001931disabled
1932 Disables a peers section. It disables both listening and any synchronization
1933 related to this section. This is provided to disable synchronization of stick
1934 tables without having to comment out all "peers" references.
1935
1936enable
1937 This re-enables a disabled peers section which was previously disabled.
1938
Emeric Brunf099e792010-09-27 12:05:28 +02001939peer <peername> <ip>:<port>
1940 Defines a peer inside a peers section.
1941 If <peername> is set to the local peer name (by default hostname, or forced
1942 using "-L" command line option), haproxy will listen for incoming remote peer
1943 connection on <ip>:<port>. Otherwise, <ip>:<port> defines where to connect to
1944 to join the remote peer, and <peername> is used at the protocol level to
1945 identify and validate the remote peer on the server side.
1946
1947 During a soft restart, local peer <ip>:<port> is used by the old instance to
1948 connect the new one and initiate a complete replication (teaching process).
1949
1950 It is strongly recommended to have the exact same peers declaration on all
1951 peers and to only rely on the "-L" command line argument to change the local
1952 peer name. This makes it easier to maintain coherent configuration files
1953 across all peers.
1954
William Lallemandb2f07452015-05-12 14:27:13 +02001955 You may want to reference some environment variables in the address
1956 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01001957
Cyril Bontédc4d9032012-04-08 21:57:39 +02001958 Example:
Emeric Brunf099e792010-09-27 12:05:28 +02001959 peers mypeers
Willy Tarreauf7b30a92010-12-06 22:59:17 +01001960 peer haproxy1 192.168.0.1:1024
1961 peer haproxy2 192.168.0.2:1024
1962 peer haproxy3 10.2.0.1:1024
Emeric Brunf099e792010-09-27 12:05:28 +02001963
1964 backend mybackend
1965 mode tcp
1966 balance roundrobin
1967 stick-table type ip size 20k peers mypeers
1968 stick on src
1969
Willy Tarreauf7b30a92010-12-06 22:59:17 +01001970 server srv1 192.168.0.30:80
1971 server srv2 192.168.0.31:80
Emeric Brunf099e792010-09-27 12:05:28 +02001972
1973
Simon Horman51a1cf62015-02-03 13:00:44 +090019743.6. Mailers
1975------------
1976It is possible to send email alerts when the state of servers changes.
1977If configured email alerts are sent to each mailer that is configured
1978in a mailers section. Email is sent to mailers using SMTP.
1979
Pieter Baauw386a1272015-08-16 15:26:24 +02001980mailers <mailersect>
Simon Horman51a1cf62015-02-03 13:00:44 +09001981 Creates a new mailer list with the name <mailersect>. It is an
1982 independent section which is referenced by one or more proxies.
1983
1984mailer <mailername> <ip>:<port>
1985 Defines a mailer inside a mailers section.
1986
1987 Example:
1988 mailers mymailers
1989 mailer smtp1 192.168.0.1:587
1990 mailer smtp2 192.168.0.2:587
1991
1992 backend mybackend
1993 mode tcp
1994 balance roundrobin
1995
1996 email-alert mailers mymailers
1997 email-alert from test1@horms.org
1998 email-alert to test2@horms.org
1999
2000 server srv1 192.168.0.30:80
2001 server srv2 192.168.0.31:80
2002
Pieter Baauw235fcfc2016-02-13 15:33:40 +01002003timeout mail <time>
2004 Defines the time available for a mail/connection to be made and send to
2005 the mail-server. If not defined the default value is 10 seconds. To allow
2006 for at least two SYN-ACK packets to be send during initial TCP handshake it
2007 is advised to keep this value above 4 seconds.
2008
2009 Example:
2010 mailers mymailers
2011 timeout mail 20s
2012 mailer smtp1 192.168.0.1:587
Simon Horman51a1cf62015-02-03 13:00:44 +09002013
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020144. Proxies
Willy Tarreau6a06a402007-07-15 20:15:28 +02002015----------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002016
Willy Tarreau6a06a402007-07-15 20:15:28 +02002017Proxy configuration can be located in a set of sections :
William Lallemand6e62fb62015-04-28 16:55:23 +02002018 - defaults [<name>]
Willy Tarreau6a06a402007-07-15 20:15:28 +02002019 - frontend <name>
2020 - backend <name>
2021 - listen <name>
2022
2023A "defaults" section sets default parameters for all other sections following
2024its declaration. Those default parameters are reset by the next "defaults"
2025section. See below for the list of parameters which can be set in a "defaults"
Willy Tarreau0ba27502007-12-24 16:55:16 +01002026section. The name is optional but its use is encouraged for better readability.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002027
2028A "frontend" section describes a set of listening sockets accepting client
2029connections.
2030
2031A "backend" section describes a set of servers to which the proxy will connect
2032to forward incoming connections.
2033
2034A "listen" section defines a complete proxy with its frontend and backend
2035parts combined in one section. It is generally useful for TCP-only traffic.
2036
Willy Tarreau0ba27502007-12-24 16:55:16 +01002037All proxy names must be formed from upper and lower case letters, digits,
2038'-' (dash), '_' (underscore) , '.' (dot) and ':' (colon). ACL names are
2039case-sensitive, which means that "www" and "WWW" are two different proxies.
2040
2041Historically, all proxy names could overlap, it just caused troubles in the
2042logs. Since the introduction of content switching, it is mandatory that two
2043proxies with overlapping capabilities (frontend/backend) have different names.
2044However, it is still permitted that a frontend and a backend share the same
2045name, as this configuration seems to be commonly encountered.
2046
2047Right now, two major proxy modes are supported : "tcp", also known as layer 4,
2048and "http", also known as layer 7. In layer 4 mode, HAProxy simply forwards
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002049bidirectional traffic between two sides. In layer 7 mode, HAProxy analyzes the
Willy Tarreau0ba27502007-12-24 16:55:16 +01002050protocol, and can interact with it by allowing, blocking, switching, adding,
2051modifying, or removing arbitrary contents in requests or responses, based on
2052arbitrary criteria.
2053
Willy Tarreau70dffda2014-01-30 03:07:23 +01002054In HTTP mode, the processing applied to requests and responses flowing over
2055a connection depends in the combination of the frontend's HTTP options and
Christopher Faulet315b39c2018-09-21 16:26:19 +02002056the backend's. HAProxy supports 4 connection modes :
Willy Tarreau70dffda2014-01-30 03:07:23 +01002057
2058 - KAL : keep alive ("option http-keep-alive") which is the default mode : all
2059 requests and responses are processed, and connections remain open but idle
2060 between responses and new requests.
2061
2062 - TUN: tunnel ("option http-tunnel") : this was the default mode for versions
2063 1.0 to 1.5-dev21 : only the first request and response are processed, and
2064 everything else is forwarded with no analysis at all. This mode should not
Christopher Faulet4212a302018-09-21 10:42:19 +02002065 be used as it creates lots of trouble with logging and HTTP processing. It
2066 is supported only on frontends.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002067
Willy Tarreau70dffda2014-01-30 03:07:23 +01002068 - SCL: server close ("option http-server-close") : the server-facing
2069 connection is closed after the end of the response is received, but the
2070 client-facing connection remains open.
2071
Christopher Faulet315b39c2018-09-21 16:26:19 +02002072 - CLO: close ("option httpclose"): the connection is closed after the end of
2073 the response and "Connection: close" appended in both directions.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002074
2075The effective mode that will be applied to a connection passing through a
2076frontend and a backend can be determined by both proxy modes according to the
2077following matrix, but in short, the modes are symmetric, keep-alive is the
Christopher Faulet315b39c2018-09-21 16:26:19 +02002078weakest option and close is the strongest.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002079
Christopher Faulet315b39c2018-09-21 16:26:19 +02002080 Backend mode
Willy Tarreau70dffda2014-01-30 03:07:23 +01002081
Christopher Faulet315b39c2018-09-21 16:26:19 +02002082 | KAL | SCL | CLO
2083 ----+-----+-----+----
2084 KAL | KAL | SCL | CLO
2085 ----+-----+-----+----
2086 TUN | TUN | SCL | CLO
2087 Frontend ----+-----+-----+----
2088 mode SCL | SCL | SCL | CLO
2089 ----+-----+-----+----
2090 CLO | CLO | CLO | CLO
Willy Tarreau70dffda2014-01-30 03:07:23 +01002091
Willy Tarreau0ba27502007-12-24 16:55:16 +01002092
Willy Tarreau70dffda2014-01-30 03:07:23 +01002093
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020944.1. Proxy keywords matrix
2095--------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002096
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002097The following list of keywords is supported. Most of them may only be used in a
2098limited set of section types. Some of them are marked as "deprecated" because
2099they are inherited from an old syntax which may be confusing or functionally
2100limited, and there are new recommended keywords to replace them. Keywords
Davor Ocelice9ed2812017-12-25 17:49:28 +01002101marked with "(*)" can be optionally inverted using the "no" prefix, e.g. "no
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002102option contstats". This makes sense when the option has been enabled by default
Willy Tarreau3842f002009-06-14 11:39:52 +02002103and must be disabled for a specific instance. Such options may also be prefixed
2104with "default" in order to restore default settings regardless of what has been
2105specified in a previous "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002106
Willy Tarreau6a06a402007-07-15 20:15:28 +02002107
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002108 keyword defaults frontend listen backend
2109------------------------------------+----------+----------+---------+---------
2110acl - X X X
Willy Tarreau294d0f02015-08-10 19:40:12 +02002111appsession - - - -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002112backlog X X X -
2113balance X - X X
2114bind - X X -
2115bind-process X X X X
Jarno Huuskonen8c8c3492016-12-28 18:50:29 +02002116block (deprecated) - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002117capture cookie - X X -
2118capture request header - X X -
2119capture response header - X X -
2120clitimeout (deprecated) X X X -
William Lallemand82fe75c2012-10-23 10:25:10 +02002121compression X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002122contimeout (deprecated) X - X X
2123cookie X - X X
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02002124declare capture - X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002125default-server X - X X
2126default_backend X X X -
2127description - X X X
2128disabled X X X X
2129dispatch - - X X
Simon Horman51a1cf62015-02-03 13:00:44 +09002130email-alert from X X X X
Simon Horman64e34162015-02-06 11:11:57 +09002131email-alert level X X X X
Simon Horman51a1cf62015-02-03 13:00:44 +09002132email-alert mailers X X X X
2133email-alert myhostname X X X X
2134email-alert to X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002135enabled X X X X
2136errorfile X X X X
2137errorloc X X X X
2138errorloc302 X X X X
2139-- keyword -------------------------- defaults - frontend - listen -- backend -
2140errorloc303 X X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01002141force-persist - - X X
Christopher Fauletc3fe5332016-04-07 15:30:10 +02002142filter - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002143fullconn X - X X
2144grace X X X X
2145hash-type X - X X
2146http-check disable-on-404 X - X X
Willy Tarreaubd741542010-03-16 18:46:54 +01002147http-check expect - - X X
Willy Tarreau7ab6aff2010-10-12 06:30:16 +02002148http-check send-state X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002149http-request - X X X
Willy Tarreaue365c0b2013-06-11 16:06:12 +02002150http-response - X X X
Willy Tarreau30631952015-08-06 15:05:24 +02002151http-reuse X - X X
Baptiste Assmann2c42ef52013-10-09 21:57:02 +02002152http-send-name-header - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002153id - X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01002154ignore-persist - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02002155load-server-state-from-file X - X X
William Lallemand0f99e342011-10-12 17:50:54 +02002156log (*) X X X X
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01002157log-format X X X -
Dragan Dosen7ad31542015-09-28 17:16:47 +02002158log-format-sd X X X -
Willy Tarreau094af4e2015-01-07 15:03:42 +01002159log-tag X X X X
Willy Tarreauc35362a2014-04-25 13:58:37 +02002160max-keep-alive-queue X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002161maxconn X X X -
2162mode X X X X
2163monitor fail - X X -
2164monitor-net X X X -
2165monitor-uri X X X -
2166option abortonclose (*) X - X X
2167option accept-invalid-http-request (*) X X X -
2168option accept-invalid-http-response (*) X - X X
2169option allbackups (*) X - X X
2170option checkcache (*) X - X X
2171option clitcpka (*) X X X -
2172option contstats (*) X X X -
2173option dontlog-normal (*) X X X -
2174option dontlognull (*) X X X -
Christopher Faulet315b39c2018-09-21 16:26:19 +02002175option forceclose (deprectated) (*) X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002176-- keyword -------------------------- defaults - frontend - listen -- backend -
2177option forwardfor X X X X
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02002178option http-buffer-request (*) X X X X
Willy Tarreau82649f92015-05-01 22:40:51 +02002179option http-ignore-probes (*) X X X -
Willy Tarreau16bfb022010-01-16 19:48:41 +01002180option http-keep-alive (*) X X X X
Willy Tarreau96e31212011-05-30 18:10:30 +02002181option http-no-delay (*) X X X X
Christopher Faulet98db9762018-09-21 10:25:19 +02002182option http-pretend-keepalive (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002183option http-server-close (*) X X X X
Christopher Faulet4212a302018-09-21 10:42:19 +02002184option http-tunnel (*) X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002185option http-use-proxy-header (*) X X X -
Willy Tarreau68ad3a42018-10-22 11:49:15 +02002186option http-use-htx (*) X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002187option httpchk X - X X
2188option httpclose (*) X X X X
2189option httplog X X X X
2190option http_proxy (*) X X X X
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002191option independent-streams (*) X X X X
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02002192option ldap-check X - X X
Simon Horman98637e52014-06-20 12:30:16 +09002193option external-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002194option log-health-checks (*) X - X X
2195option log-separate-errors (*) X X X -
2196option logasap (*) X X X -
2197option mysql-check X - X X
2198option nolinger (*) X X X X
2199option originalto X X X X
2200option persist (*) X - X X
Baptiste Assmann809e22a2015-10-12 20:22:55 +02002201option pgsql-check X - X X
2202option prefer-last-server (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002203option redispatch (*) X - X X
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02002204option redis-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002205option smtpchk X - X X
2206option socket-stats (*) X X X -
2207option splice-auto (*) X X X X
2208option splice-request (*) X X X X
2209option splice-response (*) X X X X
Christopher Fauletba7bc162016-11-07 21:07:38 +01002210option spop-check - - - X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002211option srvtcpka (*) X - X X
2212option ssl-hello-chk X - X X
2213-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreaued179852013-12-16 01:07:00 +01002214option tcp-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002215option tcp-smart-accept (*) X X X -
2216option tcp-smart-connect (*) X - X X
2217option tcpka X X X X
2218option tcplog X X X X
2219option transparent (*) X - X X
Simon Horman98637e52014-06-20 12:30:16 +09002220external-check command X - X X
2221external-check path X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002222persist rdp-cookie X - X X
2223rate-limit sessions X X X -
2224redirect - X X X
2225redisp (deprecated) X - X X
2226redispatch (deprecated) X - X X
2227reqadd - X X X
2228reqallow - X X X
2229reqdel - X X X
2230reqdeny - X X X
2231reqiallow - X X X
2232reqidel - X X X
2233reqideny - X X X
2234reqipass - X X X
2235reqirep - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002236reqitarpit - X X X
2237reqpass - X X X
2238reqrep - X X X
2239-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002240reqtarpit - X X X
2241retries X - X X
2242rspadd - X X X
2243rspdel - X X X
2244rspdeny - X X X
2245rspidel - X X X
2246rspideny - X X X
2247rspirep - X X X
2248rsprep - X X X
2249server - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02002250server-state-file-name X - X X
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02002251server-template - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002252source X - X X
2253srvtimeout (deprecated) X - X X
Baptiste Assmann5a549212015-10-12 20:30:24 +02002254stats admin - X X X
2255stats auth X X X X
2256stats enable X X X X
2257stats hide-version X X X X
2258stats http-request - X X X
2259stats realm X X X X
2260stats refresh X X X X
2261stats scope X X X X
2262stats show-desc X X X X
2263stats show-legends X X X X
2264stats show-node X X X X
2265stats uri X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002266-- keyword -------------------------- defaults - frontend - listen -- backend -
2267stick match - - X X
2268stick on - - X X
2269stick store-request - - X X
Willy Tarreaud8dc99f2011-07-01 11:33:25 +02002270stick store-response - - X X
Adam Spiers68af3c12017-04-06 16:31:39 +01002271stick-table - X X X
Willy Tarreau938c7fe2014-04-25 14:21:39 +02002272tcp-check connect - - X X
2273tcp-check expect - - X X
2274tcp-check send - - X X
2275tcp-check send-binary - - X X
Willy Tarreaue9656522010-08-17 15:40:09 +02002276tcp-request connection - X X -
2277tcp-request content - X X X
Willy Tarreaua56235c2010-09-14 11:31:36 +02002278tcp-request inspect-delay - X X X
Willy Tarreau4f614292016-10-21 17:49:36 +02002279tcp-request session - X X -
Emeric Brun0a3b67f2010-09-24 15:34:53 +02002280tcp-response content - - X X
2281tcp-response inspect-delay - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002282timeout check X - X X
2283timeout client X X X -
Willy Tarreau05cdd962014-05-10 14:30:07 +02002284timeout client-fin X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002285timeout clitimeout (deprecated) X X X -
2286timeout connect X - X X
2287timeout contimeout (deprecated) X - X X
2288timeout http-keep-alive X X X X
2289timeout http-request X X X X
2290timeout queue X - X X
2291timeout server X - X X
Willy Tarreau05cdd962014-05-10 14:30:07 +02002292timeout server-fin X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002293timeout srvtimeout (deprecated) X - X X
2294timeout tarpit X X X X
Willy Tarreauce887fd2012-05-12 12:50:00 +02002295timeout tunnel X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002296transparent (deprecated) X - X X
William Lallemanda73203e2012-03-12 12:48:57 +01002297unique-id-format X X X -
2298unique-id-header X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002299use_backend - X X -
Willy Tarreau4a5cade2012-04-05 21:09:48 +02002300use-server - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002301------------------------------------+----------+----------+---------+---------
2302 keyword defaults frontend listen backend
Willy Tarreau6a06a402007-07-15 20:15:28 +02002303
Willy Tarreau0ba27502007-12-24 16:55:16 +01002304
Willy Tarreauc57f0e22009-05-10 13:12:33 +020023054.2. Alphabetically sorted keywords reference
2306---------------------------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002307
2308This section provides a description of each keyword and its usage.
2309
2310
2311acl <aclname> <criterion> [flags] [operator] <value> ...
2312 Declare or complete an access list.
2313 May be used in sections : defaults | frontend | listen | backend
2314 no | yes | yes | yes
2315 Example:
2316 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
2317 acl invalid_src src_port 0:1023
2318 acl local_dst hdr(host) -i localhost
2319
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002320 See section 7 about ACL usage.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002321
2322
Cyril Bontéb21570a2009-11-29 20:04:48 +01002323appsession <cookie> len <length> timeout <holdtime>
2324 [request-learn] [prefix] [mode <path-parameters|query-string>]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002325 Define session stickiness on an existing application cookie.
2326 May be used in sections : defaults | frontend | listen | backend
2327 no | no | yes | yes
2328 Arguments :
2329 <cookie> this is the name of the cookie used by the application and which
2330 HAProxy will have to learn for each new session.
2331
Cyril Bontéb21570a2009-11-29 20:04:48 +01002332 <length> this is the max number of characters that will be memorized and
Willy Tarreau0ba27502007-12-24 16:55:16 +01002333 checked in each cookie value.
2334
2335 <holdtime> this is the time after which the cookie will be removed from
2336 memory if unused. If no unit is specified, this time is in
2337 milliseconds.
2338
Cyril Bontébf47aeb2009-10-15 00:15:40 +02002339 request-learn
2340 If this option is specified, then haproxy will be able to learn
2341 the cookie found in the request in case the server does not
2342 specify any in response. This is typically what happens with
2343 PHPSESSID cookies, or when haproxy's session expires before
2344 the application's session and the correct server is selected.
2345 It is recommended to specify this option to improve reliability.
2346
Cyril Bontéb21570a2009-11-29 20:04:48 +01002347 prefix When this option is specified, haproxy will match on the cookie
2348 prefix (or URL parameter prefix). The appsession value is the
2349 data following this prefix.
2350
2351 Example :
2352 appsession ASPSESSIONID len 64 timeout 3h prefix
2353
Davor Ocelice9ed2812017-12-25 17:49:28 +01002354 This will match the cookie ASPSESSIONIDXXX=XXXX,
2355 the appsession value will be XXX=XXXX.
Cyril Bontéb21570a2009-11-29 20:04:48 +01002356
2357 mode This option allows to change the URL parser mode.
2358 2 modes are currently supported :
2359 - path-parameters :
2360 The parser looks for the appsession in the path parameters
2361 part (each parameter is separated by a semi-colon), which is
2362 convenient for JSESSIONID for example.
2363 This is the default mode if the option is not set.
2364 - query-string :
2365 In this mode, the parser will look for the appsession in the
2366 query string.
2367
Willy Tarreau294d0f02015-08-10 19:40:12 +02002368 As of version 1.6, appsessions was removed. It is more flexible and more
2369 convenient to use stick-tables instead, and stick-tables support multi-master
2370 replication and data conservation across reloads, which appsessions did not.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002371
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01002372 See also : "cookie", "capture cookie", "balance", "stick", "stick-table",
2373 "ignore-persist", "nbproc" and "bind-process".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002374
2375
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01002376backlog <conns>
2377 Give hints to the system about the approximate listen backlog desired size
2378 May be used in sections : defaults | frontend | listen | backend
2379 yes | yes | yes | no
2380 Arguments :
2381 <conns> is the number of pending connections. Depending on the operating
2382 system, it may represent the number of already acknowledged
Cyril Bontédc4d9032012-04-08 21:57:39 +02002383 connections, of non-acknowledged ones, or both.
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01002384
2385 In order to protect against SYN flood attacks, one solution is to increase
2386 the system's SYN backlog size. Depending on the system, sometimes it is just
2387 tunable via a system parameter, sometimes it is not adjustable at all, and
2388 sometimes the system relies on hints given by the application at the time of
2389 the listen() syscall. By default, HAProxy passes the frontend's maxconn value
2390 to the listen() syscall. On systems which can make use of this value, it can
2391 sometimes be useful to be able to specify a different value, hence this
2392 backlog parameter.
2393
2394 On Linux 2.4, the parameter is ignored by the system. On Linux 2.6, it is
2395 used as a hint and the system accepts up to the smallest greater power of
2396 two, and never more than some limits (usually 32768).
2397
2398 See also : "maxconn" and the target operating system's tuning guide.
2399
2400
Willy Tarreau0ba27502007-12-24 16:55:16 +01002401balance <algorithm> [ <arguments> ]
Willy Tarreau226071e2014-04-10 11:55:45 +02002402balance url_param <param> [check_post]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002403 Define the load balancing algorithm to be used in a backend.
2404 May be used in sections : defaults | frontend | listen | backend
2405 yes | no | yes | yes
2406 Arguments :
2407 <algorithm> is the algorithm used to select a server when doing load
2408 balancing. This only applies when no persistence information
2409 is available, or when a connection is redispatched to another
2410 server. <algorithm> may be one of the following :
2411
2412 roundrobin Each server is used in turns, according to their weights.
2413 This is the smoothest and fairest algorithm when the server's
2414 processing time remains equally distributed. This algorithm
2415 is dynamic, which means that server weights may be adjusted
Willy Tarreau9757a382009-10-03 12:56:50 +02002416 on the fly for slow starts for instance. It is limited by
Godbacha34bdc02013-07-22 07:44:53 +08002417 design to 4095 active servers per backend. Note that in some
Willy Tarreau9757a382009-10-03 12:56:50 +02002418 large farms, when a server becomes up after having been down
2419 for a very short time, it may sometimes take a few hundreds
2420 requests for it to be re-integrated into the farm and start
2421 receiving traffic. This is normal, though very rare. It is
2422 indicated here in case you would have the chance to observe
2423 it, so that you don't worry.
2424
2425 static-rr Each server is used in turns, according to their weights.
2426 This algorithm is as similar to roundrobin except that it is
2427 static, which means that changing a server's weight on the
2428 fly will have no effect. On the other hand, it has no design
2429 limitation on the number of servers, and when a server goes
2430 up, it is always immediately reintroduced into the farm, once
2431 the full map is recomputed. It also uses slightly less CPU to
2432 run (around -1%).
Willy Tarreau0ba27502007-12-24 16:55:16 +01002433
Willy Tarreau2d2a7f82008-03-17 12:07:56 +01002434 leastconn The server with the lowest number of connections receives the
2435 connection. Round-robin is performed within groups of servers
2436 of the same load to ensure that all servers will be used. Use
2437 of this algorithm is recommended where very long sessions are
2438 expected, such as LDAP, SQL, TSE, etc... but is not very well
2439 suited for protocols using short sessions such as HTTP. This
2440 algorithm is dynamic, which means that server weights may be
2441 adjusted on the fly for slow starts for instance.
2442
Willy Tarreauf09c6602012-02-13 17:12:08 +01002443 first The first server with available connection slots receives the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002444 connection. The servers are chosen from the lowest numeric
Willy Tarreauf09c6602012-02-13 17:12:08 +01002445 identifier to the highest (see server parameter "id"), which
2446 defaults to the server's position in the farm. Once a server
Willy Tarreau64559c52012-04-07 09:08:45 +02002447 reaches its maxconn value, the next server is used. It does
Willy Tarreauf09c6602012-02-13 17:12:08 +01002448 not make sense to use this algorithm without setting maxconn.
2449 The purpose of this algorithm is to always use the smallest
2450 number of servers so that extra servers can be powered off
2451 during non-intensive hours. This algorithm ignores the server
2452 weight, and brings more benefit to long session such as RDP
Willy Tarreau64559c52012-04-07 09:08:45 +02002453 or IMAP than HTTP, though it can be useful there too. In
2454 order to use this algorithm efficiently, it is recommended
2455 that a cloud controller regularly checks server usage to turn
2456 them off when unused, and regularly checks backend queue to
2457 turn new servers on when the queue inflates. Alternatively,
2458 using "http-check send-state" may inform servers on the load.
Willy Tarreauf09c6602012-02-13 17:12:08 +01002459
Willy Tarreau0ba27502007-12-24 16:55:16 +01002460 source The source IP address is hashed and divided by the total
2461 weight of the running servers to designate which server will
2462 receive the request. This ensures that the same client IP
2463 address will always reach the same server as long as no
2464 server goes down or up. If the hash result changes due to the
2465 number of running servers changing, many clients will be
2466 directed to a different server. This algorithm is generally
2467 used in TCP mode where no cookie may be inserted. It may also
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002468 be used on the Internet to provide a best-effort stickiness
Willy Tarreau0ba27502007-12-24 16:55:16 +01002469 to clients which refuse session cookies. This algorithm is
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002470 static by default, which means that changing a server's
2471 weight on the fly will have no effect, but this can be
2472 changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002473
Oskar Stolc8dc41842012-05-19 10:19:54 +01002474 uri This algorithm hashes either the left part of the URI (before
2475 the question mark) or the whole URI (if the "whole" parameter
2476 is present) and divides the hash value by the total weight of
2477 the running servers. The result designates which server will
2478 receive the request. This ensures that the same URI will
2479 always be directed to the same server as long as no server
2480 goes up or down. This is used with proxy caches and
2481 anti-virus proxies in order to maximize the cache hit rate.
2482 Note that this algorithm may only be used in an HTTP backend.
2483 This algorithm is static by default, which means that
2484 changing a server's weight on the fly will have no effect,
2485 but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002486
Oskar Stolc8dc41842012-05-19 10:19:54 +01002487 This algorithm supports two optional parameters "len" and
Marek Majkowski9c30fc12008-04-27 23:25:55 +02002488 "depth", both followed by a positive integer number. These
2489 options may be helpful when it is needed to balance servers
2490 based on the beginning of the URI only. The "len" parameter
2491 indicates that the algorithm should only consider that many
2492 characters at the beginning of the URI to compute the hash.
2493 Note that having "len" set to 1 rarely makes sense since most
2494 URIs start with a leading "/".
2495
2496 The "depth" parameter indicates the maximum directory depth
2497 to be used to compute the hash. One level is counted for each
2498 slash in the request. If both parameters are specified, the
2499 evaluation stops when either is reached.
2500
Willy Tarreau0ba27502007-12-24 16:55:16 +01002501 url_param The URL parameter specified in argument will be looked up in
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002502 the query string of each HTTP GET request.
2503
2504 If the modifier "check_post" is used, then an HTTP POST
Cyril Bontédc4d9032012-04-08 21:57:39 +02002505 request entity will be searched for the parameter argument,
2506 when it is not found in a query string after a question mark
Willy Tarreau226071e2014-04-10 11:55:45 +02002507 ('?') in the URL. The message body will only start to be
2508 analyzed once either the advertised amount of data has been
2509 received or the request buffer is full. In the unlikely event
2510 that chunked encoding is used, only the first chunk is
Cyril Bontédc4d9032012-04-08 21:57:39 +02002511 scanned. Parameter values separated by a chunk boundary, may
Willy Tarreau226071e2014-04-10 11:55:45 +02002512 be randomly balanced if at all. This keyword used to support
2513 an optional <max_wait> parameter which is now ignored.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002514
2515 If the parameter is found followed by an equal sign ('=') and
2516 a value, then the value is hashed and divided by the total
2517 weight of the running servers. The result designates which
2518 server will receive the request.
2519
2520 This is used to track user identifiers in requests and ensure
2521 that a same user ID will always be sent to the same server as
2522 long as no server goes up or down. If no value is found or if
2523 the parameter is not found, then a round robin algorithm is
2524 applied. Note that this algorithm may only be used in an HTTP
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002525 backend. This algorithm is static by default, which means
2526 that changing a server's weight on the fly will have no
2527 effect, but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002528
Cyril Bontédc4d9032012-04-08 21:57:39 +02002529 hdr(<name>) The HTTP header <name> will be looked up in each HTTP
2530 request. Just as with the equivalent ACL 'hdr()' function,
2531 the header name in parenthesis is not case sensitive. If the
2532 header is absent or if it does not contain any value, the
2533 roundrobin algorithm is applied instead.
Benoitaffb4812009-03-25 13:02:10 +01002534
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002535 An optional 'use_domain_only' parameter is available, for
Benoitaffb4812009-03-25 13:02:10 +01002536 reducing the hash algorithm to the main domain part with some
2537 specific headers such as 'Host'. For instance, in the Host
2538 value "haproxy.1wt.eu", only "1wt" will be considered.
2539
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002540 This algorithm is static by default, which means that
2541 changing a server's weight on the fly will have no effect,
2542 but this can be changed using "hash-type".
2543
Willy Tarreau760e81d2018-05-03 07:20:40 +02002544 random A random number will be used as the key for the consistent
2545 hashing function. This means that the servers' weights are
2546 respected, dynamic weight changes immediately take effect, as
2547 well as new server additions. Random load balancing can be
2548 useful with large farms or when servers are frequently added
2549 or removed. The hash-balance-factor directive can be used to
2550 further improve fairness of the load balancing, especially
2551 in situations where servers show highly variable response
2552 times.
2553
Emeric Brun736aa232009-06-30 17:56:00 +02002554 rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02002555 rdp-cookie(<name>)
Emeric Brun736aa232009-06-30 17:56:00 +02002556 The RDP cookie <name> (or "mstshash" if omitted) will be
2557 looked up and hashed for each incoming TCP request. Just as
2558 with the equivalent ACL 'req_rdp_cookie()' function, the name
2559 is not case-sensitive. This mechanism is useful as a degraded
2560 persistence mode, as it makes it possible to always send the
2561 same user (or the same session ID) to the same server. If the
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002562 cookie is not found, the normal roundrobin algorithm is
Emeric Brun736aa232009-06-30 17:56:00 +02002563 used instead.
2564
2565 Note that for this to work, the frontend must ensure that an
2566 RDP cookie is already present in the request buffer. For this
2567 you must use 'tcp-request content accept' rule combined with
2568 a 'req_rdp_cookie_cnt' ACL.
2569
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002570 This algorithm is static by default, which means that
2571 changing a server's weight on the fly will have no effect,
2572 but this can be changed using "hash-type".
2573
Cyril Bontédc4d9032012-04-08 21:57:39 +02002574 See also the rdp_cookie pattern fetch function.
Simon Hormanab814e02011-06-24 14:50:20 +09002575
Willy Tarreau0ba27502007-12-24 16:55:16 +01002576 <arguments> is an optional list of arguments which may be needed by some
Marek Majkowski9c30fc12008-04-27 23:25:55 +02002577 algorithms. Right now, only "url_param" and "uri" support an
2578 optional argument.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002579
Willy Tarreau3cd9af22009-03-15 14:06:41 +01002580 The load balancing algorithm of a backend is set to roundrobin when no other
2581 algorithm, mode nor option have been set. The algorithm may only be set once
2582 for each backend.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002583
Lukas Tribus80512b12018-10-27 20:07:40 +02002584 With authentication schemes that require the same connection like NTLM, URI
2585 based alghoritms must not be used, as they would cause subsequent requests
2586 to be routed to different backend servers, breaking the invalid assumptions
2587 NTLM relies on.
2588
Willy Tarreau0ba27502007-12-24 16:55:16 +01002589 Examples :
2590 balance roundrobin
2591 balance url_param userid
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002592 balance url_param session_id check_post 64
Benoitaffb4812009-03-25 13:02:10 +01002593 balance hdr(User-Agent)
2594 balance hdr(host)
2595 balance hdr(Host) use_domain_only
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002596
2597 Note: the following caveats and limitations on using the "check_post"
2598 extension with "url_param" must be considered :
2599
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002600 - all POST requests are eligible for consideration, because there is no way
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002601 to determine if the parameters will be found in the body or entity which
2602 may contain binary data. Therefore another method may be required to
2603 restrict consideration of POST requests that have no URL parameters in
2604 the body. (see acl reqideny http_end)
2605
2606 - using a <max_wait> value larger than the request buffer size does not
2607 make sense and is useless. The buffer size is set at build time, and
2608 defaults to 16 kB.
2609
2610 - Content-Encoding is not supported, the parameter search will probably
2611 fail; and load balancing will fall back to Round Robin.
2612
2613 - Expect: 100-continue is not supported, load balancing will fall back to
2614 Round Robin.
2615
Lukas Tribus23953682017-04-28 13:24:30 +00002616 - Transfer-Encoding (RFC7230 3.3.1) is only supported in the first chunk.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002617 If the entire parameter value is not present in the first chunk, the
2618 selection of server is undefined (actually, defined by how little
2619 actually appeared in the first chunk).
2620
2621 - This feature does not support generation of a 100, 411 or 501 response.
2622
2623 - In some cases, requesting "check_post" MAY attempt to scan the entire
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002624 contents of a message body. Scanning normally terminates when linear
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002625 white space or control characters are found, indicating the end of what
2626 might be a URL parameter list. This is probably not a concern with SGML
2627 type message bodies.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002628
Willy Tarreau294d0f02015-08-10 19:40:12 +02002629 See also : "dispatch", "cookie", "transparent", "hash-type" and "http_proxy".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002630
2631
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002632bind [<address>]:<port_range> [, ...] [param*]
2633bind /<path> [, ...] [param*]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002634 Define one or several listening addresses and/or ports in a frontend.
2635 May be used in sections : defaults | frontend | listen | backend
2636 no | yes | yes | no
2637 Arguments :
Willy Tarreaub1e52e82008-01-13 14:49:51 +01002638 <address> is optional and can be a host name, an IPv4 address, an IPv6
2639 address, or '*'. It designates the address the frontend will
2640 listen on. If unset, all IPv4 addresses of the system will be
2641 listened on. The same will apply for '*' or the system's
David du Colombier9c938da2011-03-17 10:40:27 +01002642 special address "0.0.0.0". The IPv6 equivalent is '::'.
Willy Tarreau24709282013-03-10 21:32:12 +01002643 Optionally, an address family prefix may be used before the
2644 address to force the family regardless of the address format,
2645 which can be useful to specify a path to a unix socket with
2646 no slash ('/'). Currently supported prefixes are :
2647 - 'ipv4@' -> address is always IPv4
2648 - 'ipv6@' -> address is always IPv6
2649 - 'unix@' -> address is a path to a local unix socket
Willy Tarreau70f72e02014-07-08 00:37:50 +02002650 - 'abns@' -> address is in abstract namespace (Linux only).
2651 Note: since abstract sockets are not "rebindable", they
2652 do not cope well with multi-process mode during
2653 soft-restart, so it is better to avoid them if
2654 nbproc is greater than 1. The effect is that if the
2655 new process fails to start, only one of the old ones
2656 will be able to rebind to the socket.
Willy Tarreau40aa0702013-03-10 23:51:38 +01002657 - 'fd@<n>' -> use file descriptor <n> inherited from the
2658 parent. The fd must be bound and may or may not already
2659 be listening.
William Lallemand2fe7dd02018-09-11 16:51:29 +02002660 - 'sockpair@<n>'-> like fd@ but you must use the fd of a
2661 connected unix socket or of a socketpair. The bind waits
2662 to receive a FD over the unix socket and uses it as if it
2663 was the FD of an accept(). Should be used carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02002664 You may want to reference some environment variables in the
2665 address parameter, see section 2.3 about environment
2666 variables.
Willy Tarreaub1e52e82008-01-13 14:49:51 +01002667
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002668 <port_range> is either a unique TCP port, or a port range for which the
2669 proxy will accept connections for the IP address specified
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002670 above. The port is mandatory for TCP listeners. Note that in
2671 the case of an IPv6 address, the port is always the number
2672 after the last colon (':'). A range can either be :
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002673 - a numerical port (ex: '80')
2674 - a dash-delimited ports range explicitly stating the lower
2675 and upper bounds (ex: '2000-2100') which are included in
2676 the range.
2677
2678 Particular care must be taken against port ranges, because
2679 every <address:port> couple consumes one socket (= a file
2680 descriptor), so it's easy to consume lots of descriptors
2681 with a simple range, and to run out of sockets. Also, each
2682 <address:port> couple must be used only once among all
2683 instances running on a same system. Please note that binding
2684 to ports lower than 1024 generally require particular
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002685 privileges to start the program, which are independent of
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002686 the 'uid' parameter.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002687
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002688 <path> is a UNIX socket path beginning with a slash ('/'). This is
Davor Ocelice9ed2812017-12-25 17:49:28 +01002689 alternative to the TCP listening port. HAProxy will then
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002690 receive UNIX connections on the socket located at this place.
2691 The path must begin with a slash and by default is absolute.
2692 It can be relative to the prefix defined by "unix-bind" in
2693 the global section. Note that the total length of the prefix
2694 followed by the socket path cannot exceed some system limits
2695 for UNIX sockets, which commonly are set to 107 characters.
2696
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002697 <param*> is a list of parameters common to all sockets declared on the
2698 same line. These numerous parameters depend on OS and build
2699 options and have a complete section dedicated to them. Please
2700 refer to section 5 to for more details.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002701
Willy Tarreau0ba27502007-12-24 16:55:16 +01002702 It is possible to specify a list of address:port combinations delimited by
2703 commas. The frontend will then listen on all of these addresses. There is no
2704 fixed limit to the number of addresses and ports which can be listened on in
2705 a frontend, as well as there is no limit to the number of "bind" statements
2706 in a frontend.
2707
2708 Example :
2709 listen http_proxy
2710 bind :80,:443
2711 bind 10.0.0.1:10080,10.0.0.1:10443
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002712 bind /var/run/ssl-frontend.sock user root mode 600 accept-proxy
Willy Tarreau0ba27502007-12-24 16:55:16 +01002713
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002714 listen http_https_proxy
2715 bind :80
Cyril Bonté0d44fc62012-10-09 22:45:33 +02002716 bind :443 ssl crt /etc/haproxy/site.pem
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002717
Willy Tarreau24709282013-03-10 21:32:12 +01002718 listen http_https_proxy_explicit
2719 bind ipv6@:80
2720 bind ipv4@public_ssl:443 ssl crt /etc/haproxy/site.pem
2721 bind unix@ssl-frontend.sock user root mode 600 accept-proxy
2722
Willy Tarreaudad36a32013-03-11 01:20:04 +01002723 listen external_bind_app1
William Lallemandb2f07452015-05-12 14:27:13 +02002724 bind "fd@${FD_APP1}"
Willy Tarreaudad36a32013-03-11 01:20:04 +01002725
Willy Tarreau55dcaf62015-09-27 15:03:15 +02002726 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
2727 sun_path length is used for the address length. Some other programs
2728 such as socat use the string length only by default. Pass the option
2729 ",unix-tightsocklen=0" to any abstract socket definition in socat to
2730 make it compatible with HAProxy's.
2731
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002732 See also : "source", "option forwardfor", "unix-bind" and the PROXY protocol
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002733 documentation, and section 5 about bind options.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002734
2735
Christopher Fauletff4121f2017-11-22 16:38:49 +01002736bind-process [ all | odd | even | <process_num>[-[<process_num>]] ] ...
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002737 Limit visibility of an instance to a certain set of processes numbers.
2738 May be used in sections : defaults | frontend | listen | backend
2739 yes | yes | yes | yes
2740 Arguments :
2741 all All process will see this instance. This is the default. It
2742 may be used to override a default value.
2743
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002744 odd This instance will be enabled on processes 1,3,5,...63. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002745 option may be combined with other numbers.
2746
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002747 even This instance will be enabled on processes 2,4,6,...64. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002748 option may be combined with other numbers. Do not use it
2749 with less than 2 processes otherwise some instances might be
2750 missing from all processes.
2751
Christopher Fauletff4121f2017-11-22 16:38:49 +01002752 process_num The instance will be enabled on this process number or range,
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002753 whose values must all be between 1 and 32 or 64 depending on
Christopher Fauletff4121f2017-11-22 16:38:49 +01002754 the machine's word size. Ranges can be partially defined. The
2755 higher bound can be omitted. In such case, it is replaced by
2756 the corresponding maximum value. If a proxy is bound to
2757 process numbers greater than the configured global.nbproc, it
2758 will either be forced to process #1 if a single process was
Willy Tarreau102df612014-05-07 23:56:38 +02002759 specified, or to all processes otherwise.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002760
2761 This keyword limits binding of certain instances to certain processes. This
2762 is useful in order not to have too many processes listening to the same
2763 ports. For instance, on a dual-core machine, it might make sense to set
2764 'nbproc 2' in the global section, then distributes the listeners among 'odd'
2765 and 'even' instances.
2766
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002767 At the moment, it is not possible to reference more than 32 or 64 processes
2768 using this keyword, but this should be more than enough for most setups.
2769 Please note that 'all' really means all processes regardless of the machine's
2770 word size, and is not limited to the first 32 or 64.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002771
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02002772 Each "bind" line may further be limited to a subset of the proxy's processes,
2773 please consult the "process" bind keyword in section 5.1.
2774
Willy Tarreaub369a042014-09-16 13:21:03 +02002775 When a frontend has no explicit "bind-process" line, it tries to bind to all
2776 the processes referenced by its "bind" lines. That means that frontends can
2777 easily adapt to their listeners' processes.
2778
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002779 If some backends are referenced by frontends bound to other processes, the
2780 backend automatically inherits the frontend's processes.
2781
2782 Example :
2783 listen app_ip1
2784 bind 10.0.0.1:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02002785 bind-process odd
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002786
2787 listen app_ip2
2788 bind 10.0.0.2:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02002789 bind-process even
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002790
2791 listen management
2792 bind 10.0.0.3:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02002793 bind-process 1 2 3 4
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002794
Willy Tarreau110ecc12012-11-15 17:50:01 +01002795 listen management
2796 bind 10.0.0.4:80
2797 bind-process 1-4
2798
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02002799 See also : "nbproc" in global section, and "process" in section 5.1.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002800
2801
Jarno Huuskonen8c8c3492016-12-28 18:50:29 +02002802block { if | unless } <condition> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01002803 Block a layer 7 request if/unless a condition is matched
2804 May be used in sections : defaults | frontend | listen | backend
2805 no | yes | yes | yes
2806
2807 The HTTP request will be blocked very early in the layer 7 processing
2808 if/unless <condition> is matched. A 403 error will be returned if the request
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002809 is blocked. The condition has to reference ACLs (see section 7). This is
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02002810 typically used to deny access to certain sensitive resources if some
Willy Tarreau0ba27502007-12-24 16:55:16 +01002811 conditions are met or not met. There is no fixed limit to the number of
Jarno Huuskonen95b012b2017-04-06 13:59:14 +03002812 "block" statements per instance. To block connections at layer 4 (without
2813 sending a 403 error) see "tcp-request connection reject" and
2814 "tcp-request content reject" rules.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002815
Jarno Huuskonen8c8c3492016-12-28 18:50:29 +02002816 This form is deprecated, do not use it in any new configuration, use the new
2817 "http-request deny" instead.
2818
Willy Tarreau0ba27502007-12-24 16:55:16 +01002819 Example:
2820 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
2821 acl invalid_src src_port 0:1023
2822 acl local_dst hdr(host) -i localhost
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +03002823 # block is deprecated. Use http-request deny instead:
2824 #block if invalid_src || local_dst
2825 http-request deny if invalid_src || local_dst
Willy Tarreau0ba27502007-12-24 16:55:16 +01002826
Jarno Huuskonen95b012b2017-04-06 13:59:14 +03002827 See also : section 7 about ACL usage, "http-request deny",
2828 "http-response deny", "tcp-request connection reject" and
2829 "tcp-request content reject".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002830
2831capture cookie <name> len <length>
2832 Capture and log a cookie in the request and in the response.
2833 May be used in sections : defaults | frontend | listen | backend
2834 no | yes | yes | no
2835 Arguments :
2836 <name> is the beginning of the name of the cookie to capture. In order
2837 to match the exact name, simply suffix the name with an equal
2838 sign ('='). The full name will appear in the logs, which is
2839 useful with application servers which adjust both the cookie name
Davor Ocelice9ed2812017-12-25 17:49:28 +01002840 and value (e.g. ASPSESSIONXXX).
Willy Tarreau0ba27502007-12-24 16:55:16 +01002841
2842 <length> is the maximum number of characters to report in the logs, which
2843 include the cookie name, the equal sign and the value, all in the
2844 standard "name=value" form. The string will be truncated on the
2845 right if it exceeds <length>.
2846
2847 Only the first cookie is captured. Both the "cookie" request headers and the
2848 "set-cookie" response headers are monitored. This is particularly useful to
2849 check for application bugs causing session crossing or stealing between
2850 users, because generally the user's cookies can only change on a login page.
2851
2852 When the cookie was not presented by the client, the associated log column
2853 will report "-". When a request does not cause a cookie to be assigned by the
2854 server, a "-" is reported in the response column.
2855
2856 The capture is performed in the frontend only because it is necessary that
2857 the log format does not change for a given frontend depending on the
2858 backends. This may change in the future. Note that there can be only one
Willy Tarreau193b8c62012-11-22 00:17:38 +01002859 "capture cookie" statement in a frontend. The maximum capture length is set
2860 by the global "tune.http.cookielen" setting and defaults to 63 characters. It
2861 is not possible to specify a capture in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002862
2863 Example:
2864 capture cookie ASPSESSION len 32
2865
2866 See also : "capture request header", "capture response header" as well as
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002867 section 8 about logging.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002868
2869
2870capture request header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01002871 Capture and log the last occurrence of the specified request header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002872 May be used in sections : defaults | frontend | listen | backend
2873 no | yes | yes | no
2874 Arguments :
2875 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002876 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01002877 appear in the requests, with the first letter of each word in
2878 upper case. The header name will not appear in the logs, only the
2879 value is reported, but the position in the logs is respected.
2880
2881 <length> is the maximum number of characters to extract from the value and
2882 report in the logs. The string will be truncated on the right if
2883 it exceeds <length>.
2884
Willy Tarreau4460d032012-11-21 23:37:37 +01002885 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01002886 value will be added to the logs between braces ('{}'). If multiple headers
2887 are captured, they will be delimited by a vertical bar ('|') and will appear
Willy Tarreaucc6c8912009-02-22 10:53:55 +01002888 in the same order they were declared in the configuration. Non-existent
2889 headers will be logged just as an empty string. Common uses for request
2890 header captures include the "Host" field in virtual hosting environments, the
2891 "Content-length" when uploads are supported, "User-agent" to quickly
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002892 differentiate between real users and robots, and "X-Forwarded-For" in proxied
Willy Tarreaucc6c8912009-02-22 10:53:55 +01002893 environments to find where the request came from.
2894
2895 Note that when capturing headers such as "User-agent", some spaces may be
2896 logged, making the log analysis more difficult. Thus be careful about what
2897 you log if you know your log parser is not smart enough to rely on the
2898 braces.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002899
Willy Tarreau0900abb2012-11-22 00:21:46 +01002900 There is no limit to the number of captured request headers nor to their
2901 length, though it is wise to keep them low to limit memory usage per session.
2902 In order to keep log format consistent for a same frontend, header captures
2903 can only be declared in a frontend. It is not possible to specify a capture
2904 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002905
2906 Example:
2907 capture request header Host len 15
2908 capture request header X-Forwarded-For len 15
Cyril Bontéd1b0f7c2015-10-26 22:37:39 +01002909 capture request header Referer len 15
Willy Tarreau0ba27502007-12-24 16:55:16 +01002910
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002911 See also : "capture cookie", "capture response header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01002912 about logging.
2913
2914
2915capture response header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01002916 Capture and log the last occurrence of the specified response header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002917 May be used in sections : defaults | frontend | listen | backend
2918 no | yes | yes | no
2919 Arguments :
2920 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002921 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01002922 appear in the response, with the first letter of each word in
2923 upper case. The header name will not appear in the logs, only the
2924 value is reported, but the position in the logs is respected.
2925
2926 <length> is the maximum number of characters to extract from the value and
2927 report in the logs. The string will be truncated on the right if
2928 it exceeds <length>.
2929
Willy Tarreau4460d032012-11-21 23:37:37 +01002930 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01002931 result will be added to the logs between braces ('{}') after the captured
2932 request headers. If multiple headers are captured, they will be delimited by
2933 a vertical bar ('|') and will appear in the same order they were declared in
Willy Tarreaucc6c8912009-02-22 10:53:55 +01002934 the configuration. Non-existent headers will be logged just as an empty
2935 string. Common uses for response header captures include the "Content-length"
2936 header which indicates how many bytes are expected to be returned, the
2937 "Location" header to track redirections.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002938
Willy Tarreau0900abb2012-11-22 00:21:46 +01002939 There is no limit to the number of captured response headers nor to their
2940 length, though it is wise to keep them low to limit memory usage per session.
2941 In order to keep log format consistent for a same frontend, header captures
2942 can only be declared in a frontend. It is not possible to specify a capture
2943 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002944
2945 Example:
2946 capture response header Content-length len 9
2947 capture response header Location len 15
2948
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002949 See also : "capture cookie", "capture request header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01002950 about logging.
2951
2952
Cyril Bontéf0c60612010-02-06 14:44:47 +01002953clitimeout <timeout> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01002954 Set the maximum inactivity time on the client side.
2955 May be used in sections : defaults | frontend | listen | backend
2956 yes | yes | yes | no
2957 Arguments :
2958 <timeout> is the timeout value is specified in milliseconds by default, but
2959 can be in any other unit if the number is suffixed by the unit,
2960 as explained at the top of this document.
2961
2962 The inactivity timeout applies when the client is expected to acknowledge or
2963 send data. In HTTP mode, this timeout is particularly important to consider
2964 during the first phase, when the client sends the request, and during the
2965 response while it is reading data sent by the server. The value is specified
2966 in milliseconds by default, but can be in any other unit if the number is
2967 suffixed by the unit, as specified at the top of this document. In TCP mode
2968 (and to a lesser extent, in HTTP mode), it is highly recommended that the
2969 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002970 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +01002971 losses by specifying timeouts that are slightly above multiples of 3 seconds
Davor Ocelice9ed2812017-12-25 17:49:28 +01002972 (e.g. 4 or 5 seconds).
Willy Tarreau0ba27502007-12-24 16:55:16 +01002973
2974 This parameter is specific to frontends, but can be specified once for all in
2975 "defaults" sections. This is in fact one of the easiest solutions not to
2976 forget about it. An unspecified timeout results in an infinite timeout, which
2977 is not recommended. Such a usage is accepted and works but reports a warning
2978 during startup because it may results in accumulation of expired sessions in
2979 the system if the system's timeouts are not configured either.
2980
2981 This parameter is provided for compatibility but is currently deprecated.
2982 Please use "timeout client" instead.
2983
Willy Tarreau036fae02008-01-06 13:24:40 +01002984 See also : "timeout client", "timeout http-request", "timeout server", and
2985 "srvtimeout".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002986
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002987compression algo <algorithm> ...
2988compression type <mime type> ...
Willy Tarreau70737d12012-10-27 00:34:28 +02002989compression offload
William Lallemand82fe75c2012-10-23 10:25:10 +02002990 Enable HTTP compression.
2991 May be used in sections : defaults | frontend | listen | backend
2992 yes | yes | yes | yes
2993 Arguments :
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002994 algo is followed by the list of supported compression algorithms.
2995 type is followed by the list of MIME types that will be compressed.
2996 offload makes haproxy work as a compression offloader only (see notes).
2997
2998 The currently supported algorithms are :
Willy Tarreauc91840a2015-03-28 17:00:39 +01002999 identity this is mostly for debugging, and it was useful for developing
3000 the compression feature. Identity does not apply any change on
3001 data.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003002
Willy Tarreauc91840a2015-03-28 17:00:39 +01003003 gzip applies gzip compression. This setting is only available when
Baptiste Assmannf085d632015-12-21 17:57:32 +01003004 support for zlib or libslz was built in.
Willy Tarreauc91840a2015-03-28 17:00:39 +01003005
3006 deflate same as "gzip", but with deflate algorithm and zlib format.
3007 Note that this algorithm has ambiguous support on many
3008 browsers and no support at all from recent ones. It is
3009 strongly recommended not to use it for anything else than
3010 experimentation. This setting is only available when support
Baptiste Assmannf085d632015-12-21 17:57:32 +01003011 for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003012
Willy Tarreauc91840a2015-03-28 17:00:39 +01003013 raw-deflate same as "deflate" without the zlib wrapper, and used as an
3014 alternative when the browser wants "deflate". All major
3015 browsers understand it and despite violating the standards,
3016 it is known to work better than "deflate", at least on MSIE
3017 and some versions of Safari. Do not use it in conjunction
3018 with "deflate", use either one or the other since both react
3019 to the same Accept-Encoding token. This setting is only
Baptiste Assmannf085d632015-12-21 17:57:32 +01003020 available when support for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003021
Dmitry Sivachenko87c208b2012-11-22 20:03:26 +04003022 Compression will be activated depending on the Accept-Encoding request
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003023 header. With identity, it does not take care of that header.
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04003024 If backend servers support HTTP compression, these directives
3025 will be no-op: haproxy will see the compressed response and will not
3026 compress again. If backend servers do not support HTTP compression and
3027 there is Accept-Encoding header in request, haproxy will compress the
3028 matching response.
Willy Tarreau70737d12012-10-27 00:34:28 +02003029
3030 The "offload" setting makes haproxy remove the Accept-Encoding header to
3031 prevent backend servers from compressing responses. It is strongly
3032 recommended not to do this because this means that all the compression work
3033 will be done on the single point where haproxy is located. However in some
3034 deployment scenarios, haproxy may be installed in front of a buggy gateway
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04003035 with broken HTTP compression implementation which can't be turned off.
3036 In that case haproxy can be used to prevent that gateway from emitting
3037 invalid payloads. In this case, simply removing the header in the
3038 configuration does not work because it applies before the header is parsed,
3039 so that prevents haproxy from compressing. The "offload" setting should
Willy Tarreauffea9fd2014-07-12 16:37:02 +02003040 then be used for such scenarios. Note: for now, the "offload" setting is
3041 ignored when set in a defaults section.
William Lallemand82fe75c2012-10-23 10:25:10 +02003042
William Lallemand05097442012-11-20 12:14:28 +01003043 Compression is disabled when:
Baptiste Assmann650d53d2013-01-05 15:44:44 +01003044 * the request does not advertise a supported compression algorithm in the
3045 "Accept-Encoding" header
3046 * the response message is not HTTP/1.1
William Lallemandd3002612012-11-26 14:34:47 +01003047 * HTTP status code is not 200
William Lallemand8bb4e342013-12-10 17:28:48 +01003048 * response header "Transfer-Encoding" contains "chunked" (Temporary
3049 Workaround)
Baptiste Assmann650d53d2013-01-05 15:44:44 +01003050 * response contain neither a "Content-Length" header nor a
3051 "Transfer-Encoding" whose last value is "chunked"
3052 * response contains a "Content-Type" header whose first value starts with
3053 "multipart"
3054 * the response contains the "no-transform" value in the "Cache-control"
3055 header
3056 * User-Agent matches "Mozilla/4" unless it is MSIE 6 with XP SP2, or MSIE 7
3057 and later
3058 * The response contains a "Content-Encoding" header, indicating that the
3059 response is already compressed (see compression offload)
William Lallemand05097442012-11-20 12:14:28 +01003060
Baptiste Assmann650d53d2013-01-05 15:44:44 +01003061 Note: The compression does not rewrite Etag headers, and does not emit the
3062 Warning header.
William Lallemand05097442012-11-20 12:14:28 +01003063
William Lallemand82fe75c2012-10-23 10:25:10 +02003064 Examples :
3065 compression algo gzip
3066 compression type text/html text/plain
Willy Tarreau0ba27502007-12-24 16:55:16 +01003067
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003068
Cyril Bontéf0c60612010-02-06 14:44:47 +01003069contimeout <timeout> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01003070 Set the maximum time to wait for a connection attempt to a server to succeed.
3071 May be used in sections : defaults | frontend | listen | backend
3072 yes | no | yes | yes
3073 Arguments :
3074 <timeout> is the timeout value is specified in milliseconds by default, but
3075 can be in any other unit if the number is suffixed by the unit,
3076 as explained at the top of this document.
3077
3078 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003079 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +01003080 cover one or several TCP packet losses by specifying timeouts that are
Davor Ocelice9ed2812017-12-25 17:49:28 +01003081 slightly above multiples of 3 seconds (e.g. 4 or 5 seconds). By default, the
Willy Tarreau0ba27502007-12-24 16:55:16 +01003082 connect timeout also presets the queue timeout to the same value if this one
3083 has not been specified. Historically, the contimeout was also used to set the
3084 tarpit timeout in a listen section, which is not possible in a pure frontend.
3085
3086 This parameter is specific to backends, but can be specified once for all in
3087 "defaults" sections. This is in fact one of the easiest solutions not to
3088 forget about it. An unspecified timeout results in an infinite timeout, which
3089 is not recommended. Such a usage is accepted and works but reports a warning
3090 during startup because it may results in accumulation of failed sessions in
3091 the system if the system's timeouts are not configured either.
3092
3093 This parameter is provided for backwards compatibility but is currently
3094 deprecated. Please use "timeout connect", "timeout queue" or "timeout tarpit"
3095 instead.
3096
3097 See also : "timeout connect", "timeout queue", "timeout tarpit",
3098 "timeout server", "contimeout".
3099
3100
Willy Tarreau55165fe2009-05-10 12:02:55 +02003101cookie <name> [ rewrite | insert | prefix ] [ indirect ] [ nocache ]
Willy Tarreau4992dd22012-05-31 21:02:17 +02003102 [ postonly ] [ preserve ] [ httponly ] [ secure ]
3103 [ domain <domain> ]* [ maxidle <idle> ] [ maxlife <life> ]
Olivier Houchard4e694042017-03-14 20:01:29 +01003104 [ dynamic ]
Willy Tarreau0ba27502007-12-24 16:55:16 +01003105 Enable cookie-based persistence in a backend.
3106 May be used in sections : defaults | frontend | listen | backend
3107 yes | no | yes | yes
3108 Arguments :
3109 <name> is the name of the cookie which will be monitored, modified or
3110 inserted in order to bring persistence. This cookie is sent to
3111 the client via a "Set-Cookie" header in the response, and is
3112 brought back by the client in a "Cookie" header in all requests.
3113 Special care should be taken to choose a name which does not
3114 conflict with any likely application cookie. Also, if the same
Davor Ocelice9ed2812017-12-25 17:49:28 +01003115 backends are subject to be used by the same clients (e.g.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003116 HTTP/HTTPS), care should be taken to use different cookie names
3117 between all backends if persistence between them is not desired.
3118
3119 rewrite This keyword indicates that the cookie will be provided by the
3120 server and that haproxy will have to modify its value to set the
3121 server's identifier in it. This mode is handy when the management
3122 of complex combinations of "Set-cookie" and "Cache-control"
3123 headers is left to the application. The application can then
3124 decide whether or not it is appropriate to emit a persistence
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003125 cookie. Since all responses should be monitored, this mode
3126 doesn't work in HTTP tunnel mode. Unless the application
Davor Ocelice9ed2812017-12-25 17:49:28 +01003127 behavior is very complex and/or broken, it is advised not to
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003128 start with this mode for new deployments. This keyword is
3129 incompatible with "insert" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003130
3131 insert This keyword indicates that the persistence cookie will have to
Willy Tarreaua79094d2010-08-31 22:54:15 +02003132 be inserted by haproxy in server responses if the client did not
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003133
Willy Tarreaua79094d2010-08-31 22:54:15 +02003134 already have a cookie that would have permitted it to access this
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003135 server. When used without the "preserve" option, if the server
3136 emits a cookie with the same name, it will be remove before
Davor Ocelice9ed2812017-12-25 17:49:28 +01003137 processing. For this reason, this mode can be used to upgrade
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003138 existing configurations running in the "rewrite" mode. The cookie
3139 will only be a session cookie and will not be stored on the
3140 client's disk. By default, unless the "indirect" option is added,
3141 the server will see the cookies emitted by the client. Due to
3142 caching effects, it is generally wise to add the "nocache" or
3143 "postonly" keywords (see below). The "insert" keyword is not
3144 compatible with "rewrite" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003145
3146 prefix This keyword indicates that instead of relying on a dedicated
3147 cookie for the persistence, an existing one will be completed.
3148 This may be needed in some specific environments where the client
3149 does not support more than one single cookie and the application
3150 already needs it. In this case, whenever the server sets a cookie
3151 named <name>, it will be prefixed with the server's identifier
3152 and a delimiter. The prefix will be removed from all client
3153 requests so that the server still finds the cookie it emitted.
3154 Since all requests and responses are subject to being modified,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003155 this mode doesn't work with tunnel mode. The "prefix" keyword is
Willy Tarreau37229df2011-10-17 12:24:55 +02003156 not compatible with "rewrite" and "insert". Note: it is highly
3157 recommended not to use "indirect" with "prefix", otherwise server
3158 cookie updates would not be sent to clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003159
Willy Tarreaua79094d2010-08-31 22:54:15 +02003160 indirect When this option is specified, no cookie will be emitted to a
3161 client which already has a valid one for the server which has
3162 processed the request. If the server sets such a cookie itself,
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003163 it will be removed, unless the "preserve" option is also set. In
3164 "insert" mode, this will additionally remove cookies from the
3165 requests transmitted to the server, making the persistence
3166 mechanism totally transparent from an application point of view.
Willy Tarreau37229df2011-10-17 12:24:55 +02003167 Note: it is highly recommended not to use "indirect" with
3168 "prefix", otherwise server cookie updates would not be sent to
3169 clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003170
3171 nocache This option is recommended in conjunction with the insert mode
3172 when there is a cache between the client and HAProxy, as it
3173 ensures that a cacheable response will be tagged non-cacheable if
3174 a cookie needs to be inserted. This is important because if all
3175 persistence cookies are added on a cacheable home page for
3176 instance, then all customers will then fetch the page from an
3177 outer cache and will all share the same persistence cookie,
3178 leading to one server receiving much more traffic than others.
3179 See also the "insert" and "postonly" options.
3180
3181 postonly This option ensures that cookie insertion will only be performed
3182 on responses to POST requests. It is an alternative to the
3183 "nocache" option, because POST responses are not cacheable, so
3184 this ensures that the persistence cookie will never get cached.
3185 Since most sites do not need any sort of persistence before the
3186 first POST which generally is a login request, this is a very
3187 efficient method to optimize caching without risking to find a
3188 persistence cookie in the cache.
3189 See also the "insert" and "nocache" options.
3190
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003191 preserve This option may only be used with "insert" and/or "indirect". It
3192 allows the server to emit the persistence cookie itself. In this
3193 case, if a cookie is found in the response, haproxy will leave it
3194 untouched. This is useful in order to end persistence after a
3195 logout request for instance. For this, the server just has to
Davor Ocelice9ed2812017-12-25 17:49:28 +01003196 emit a cookie with an invalid value (e.g. empty) or with a date in
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003197 the past. By combining this mechanism with the "disable-on-404"
3198 check option, it is possible to perform a completely graceful
3199 shutdown because users will definitely leave the server after
3200 they logout.
3201
Willy Tarreau4992dd22012-05-31 21:02:17 +02003202 httponly This option tells haproxy to add an "HttpOnly" cookie attribute
3203 when a cookie is inserted. This attribute is used so that a
3204 user agent doesn't share the cookie with non-HTTP components.
3205 Please check RFC6265 for more information on this attribute.
3206
3207 secure This option tells haproxy to add a "Secure" cookie attribute when
3208 a cookie is inserted. This attribute is used so that a user agent
3209 never emits this cookie over non-secure channels, which means
3210 that a cookie learned with this flag will be presented only over
3211 SSL/TLS connections. Please check RFC6265 for more information on
3212 this attribute.
3213
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02003214 domain This option allows to specify the domain at which a cookie is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003215 inserted. It requires exactly one parameter: a valid domain
Willy Tarreau68a897b2009-12-03 23:28:34 +01003216 name. If the domain begins with a dot, the browser is allowed to
3217 use it for any host ending with that name. It is also possible to
3218 specify several domain names by invoking this option multiple
3219 times. Some browsers might have small limits on the number of
3220 domains, so be careful when doing that. For the record, sending
3221 10 domains to MSIE 6 or Firefox 2 works as expected.
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02003222
Willy Tarreau996a92c2010-10-13 19:30:47 +02003223 maxidle This option allows inserted cookies to be ignored after some idle
3224 time. It only works with insert-mode cookies. When a cookie is
3225 sent to the client, the date this cookie was emitted is sent too.
3226 Upon further presentations of this cookie, if the date is older
3227 than the delay indicated by the parameter (in seconds), it will
3228 be ignored. Otherwise, it will be refreshed if needed when the
3229 response is sent to the client. This is particularly useful to
3230 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01003231 too long on the same server (e.g. after a farm size change). When
Willy Tarreau996a92c2010-10-13 19:30:47 +02003232 this option is set and a cookie has no date, it is always
3233 accepted, but gets refreshed in the response. This maintains the
3234 ability for admins to access their sites. Cookies that have a
3235 date in the future further than 24 hours are ignored. Doing so
3236 lets admins fix timezone issues without risking kicking users off
3237 the site.
3238
3239 maxlife This option allows inserted cookies to be ignored after some life
3240 time, whether they're in use or not. It only works with insert
3241 mode cookies. When a cookie is first sent to the client, the date
3242 this cookie was emitted is sent too. Upon further presentations
3243 of this cookie, if the date is older than the delay indicated by
3244 the parameter (in seconds), it will be ignored. If the cookie in
3245 the request has no date, it is accepted and a date will be set.
3246 Cookies that have a date in the future further than 24 hours are
3247 ignored. Doing so lets admins fix timezone issues without risking
3248 kicking users off the site. Contrary to maxidle, this value is
3249 not refreshed, only the first visit date counts. Both maxidle and
3250 maxlife may be used at the time. This is particularly useful to
3251 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01003252 too long on the same server (e.g. after a farm size change). This
Willy Tarreau996a92c2010-10-13 19:30:47 +02003253 is stronger than the maxidle method in that it forces a
3254 redispatch after some absolute delay.
3255
Olivier Houchard4e694042017-03-14 20:01:29 +01003256 dynamic Activate dynamic cookies. When used, a session cookie is
3257 dynamically created for each server, based on the IP and port
3258 of the server, and a secret key, specified in the
3259 "dynamic-cookie-key" backend directive.
3260 The cookie will be regenerated each time the IP address change,
3261 and is only generated for IPv4/IPv6.
3262
Willy Tarreau0ba27502007-12-24 16:55:16 +01003263 There can be only one persistence cookie per HTTP backend, and it can be
3264 declared in a defaults section. The value of the cookie will be the value
3265 indicated after the "cookie" keyword in a "server" statement. If no cookie
3266 is declared for a given server, the cookie is not set.
Willy Tarreau6a06a402007-07-15 20:15:28 +02003267
Willy Tarreau0ba27502007-12-24 16:55:16 +01003268 Examples :
3269 cookie JSESSIONID prefix
3270 cookie SRV insert indirect nocache
3271 cookie SRV insert postonly indirect
Willy Tarreau996a92c2010-10-13 19:30:47 +02003272 cookie SRV insert indirect nocache maxidle 30m maxlife 8h
Willy Tarreau0ba27502007-12-24 16:55:16 +01003273
Willy Tarreau294d0f02015-08-10 19:40:12 +02003274 See also : "balance source", "capture cookie", "server" and "ignore-persist".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003275
Willy Tarreau983e01e2010-01-11 18:42:06 +01003276
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02003277declare capture [ request | response ] len <length>
3278 Declares a capture slot.
3279 May be used in sections : defaults | frontend | listen | backend
3280 no | yes | yes | no
3281 Arguments:
3282 <length> is the length allowed for the capture.
3283
3284 This declaration is only available in the frontend or listen section, but the
3285 reserved slot can be used in the backends. The "request" keyword allocates a
3286 capture slot for use in the request, and "response" allocates a capture slot
3287 for use in the response.
3288
3289 See also: "capture-req", "capture-res" (sample converters),
Baptiste Assmann5ac425c2015-10-21 23:13:46 +02003290 "capture.req.hdr", "capture.res.hdr" (sample fetches),
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02003291 "http-request capture" and "http-response capture".
3292
3293
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003294default-server [param*]
3295 Change default options for a server in a backend
3296 May be used in sections : defaults | frontend | listen | backend
3297 yes | no | yes | yes
3298 Arguments:
Willy Tarreau983e01e2010-01-11 18:42:06 +01003299 <param*> is a list of parameters for this server. The "default-server"
3300 keyword accepts an important number of options and has a complete
3301 section dedicated to it. Please refer to section 5 for more
3302 details.
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003303
Willy Tarreau983e01e2010-01-11 18:42:06 +01003304 Example :
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003305 default-server inter 1000 weight 13
3306
3307 See also: "server" and section 5 about server options
Willy Tarreau0ba27502007-12-24 16:55:16 +01003308
Willy Tarreau983e01e2010-01-11 18:42:06 +01003309
Willy Tarreau0ba27502007-12-24 16:55:16 +01003310default_backend <backend>
3311 Specify the backend to use when no "use_backend" rule has been matched.
3312 May be used in sections : defaults | frontend | listen | backend
3313 yes | yes | yes | no
3314 Arguments :
3315 <backend> is the name of the backend to use.
3316
3317 When doing content-switching between frontend and backends using the
3318 "use_backend" keyword, it is often useful to indicate which backend will be
3319 used when no rule has matched. It generally is the dynamic backend which
3320 will catch all undetermined requests.
3321
Willy Tarreau0ba27502007-12-24 16:55:16 +01003322 Example :
3323
3324 use_backend dynamic if url_dyn
3325 use_backend static if url_css url_img extension_img
3326 default_backend dynamic
3327
Willy Tarreau98d04852015-05-26 12:18:29 +02003328 See also : "use_backend"
Willy Tarreau2769aa02007-12-27 18:26:09 +01003329
Willy Tarreau0ba27502007-12-24 16:55:16 +01003330
Baptiste Assmann27f51342013-10-09 06:51:49 +02003331description <string>
3332 Describe a listen, frontend or backend.
3333 May be used in sections : defaults | frontend | listen | backend
3334 no | yes | yes | yes
3335 Arguments : string
3336
3337 Allows to add a sentence to describe the related object in the HAProxy HTML
3338 stats page. The description will be printed on the right of the object name
3339 it describes.
3340 No need to backslash spaces in the <string> arguments.
3341
3342
Willy Tarreau0ba27502007-12-24 16:55:16 +01003343disabled
3344 Disable a proxy, frontend or backend.
3345 May be used in sections : defaults | frontend | listen | backend
3346 yes | yes | yes | yes
3347 Arguments : none
3348
3349 The "disabled" keyword is used to disable an instance, mainly in order to
3350 liberate a listening port or to temporarily disable a service. The instance
3351 will still be created and its configuration will be checked, but it will be
3352 created in the "stopped" state and will appear as such in the statistics. It
3353 will not receive any traffic nor will it send any health-checks or logs. It
3354 is possible to disable many instances at once by adding the "disabled"
3355 keyword in a "defaults" section.
3356
3357 See also : "enabled"
3358
3359
Willy Tarreau5ce94572010-06-07 14:35:41 +02003360dispatch <address>:<port>
3361 Set a default server address
3362 May be used in sections : defaults | frontend | listen | backend
3363 no | no | yes | yes
Cyril Bonté108cf6e2012-04-21 23:30:29 +02003364 Arguments :
Willy Tarreau5ce94572010-06-07 14:35:41 +02003365
3366 <address> is the IPv4 address of the default server. Alternatively, a
3367 resolvable hostname is supported, but this name will be resolved
3368 during start-up.
3369
3370 <ports> is a mandatory port specification. All connections will be sent
3371 to this port, and it is not permitted to use port offsets as is
3372 possible with normal servers.
3373
Willy Tarreau787aed52011-04-15 06:45:37 +02003374 The "dispatch" keyword designates a default server for use when no other
Willy Tarreau5ce94572010-06-07 14:35:41 +02003375 server can take the connection. In the past it was used to forward non
3376 persistent connections to an auxiliary load balancer. Due to its simple
3377 syntax, it has also been used for simple TCP relays. It is recommended not to
3378 use it for more clarity, and to use the "server" directive instead.
3379
3380 See also : "server"
3381
Olivier Houchard4e694042017-03-14 20:01:29 +01003382
3383dynamic-cookie-key <string>
3384 Set the dynamic cookie secret key for a backend.
3385 May be used in sections : defaults | frontend | listen | backend
3386 yes | no | yes | yes
3387 Arguments : The secret key to be used.
3388
3389 When dynamic cookies are enabled (see the "dynamic" directive for cookie),
Davor Ocelice9ed2812017-12-25 17:49:28 +01003390 a dynamic cookie is created for each server (unless one is explicitly
Olivier Houchard4e694042017-03-14 20:01:29 +01003391 specified on the "server" line), using a hash of the IP address of the
3392 server, the TCP port, and the secret key.
Davor Ocelice9ed2812017-12-25 17:49:28 +01003393 That way, we can ensure session persistence across multiple load-balancers,
Olivier Houchard4e694042017-03-14 20:01:29 +01003394 even if servers are dynamically added or removed.
Willy Tarreau5ce94572010-06-07 14:35:41 +02003395
Willy Tarreau0ba27502007-12-24 16:55:16 +01003396enabled
3397 Enable a proxy, frontend or backend.
3398 May be used in sections : defaults | frontend | listen | backend
3399 yes | yes | yes | yes
3400 Arguments : none
3401
3402 The "enabled" keyword is used to explicitly enable an instance, when the
3403 defaults has been set to "disabled". This is very rarely used.
3404
3405 See also : "disabled"
3406
3407
3408errorfile <code> <file>
3409 Return a file contents instead of errors generated by HAProxy
3410 May be used in sections : defaults | frontend | listen | backend
3411 yes | yes | yes | yes
3412 Arguments :
3413 <code> is the HTTP status code. Currently, HAProxy is capable of
Olivier Houchard51a76d82017-10-02 16:12:07 +02003414 generating codes 200, 400, 403, 405, 408, 425, 429, 500, 502,
3415 503, and 504.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003416
3417 <file> designates a file containing the full HTTP response. It is
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003418 recommended to follow the common practice of appending ".http" to
Willy Tarreau0ba27502007-12-24 16:55:16 +01003419 the filename so that people do not confuse the response with HTML
Willy Tarreau59140a22009-02-22 12:02:30 +01003420 error pages, and to use absolute paths, since files are read
3421 before any chroot is performed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003422
3423 It is important to understand that this keyword is not meant to rewrite
3424 errors returned by the server, but errors detected and returned by HAProxy.
3425 This is why the list of supported errors is limited to a small set.
3426
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003427 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3428
Willy Tarreau0ba27502007-12-24 16:55:16 +01003429 The files are returned verbatim on the TCP socket. This allows any trick such
3430 as redirections to another URL or site, as well as tricks to clean cookies,
3431 force enable or disable caching, etc... The package provides default error
3432 files returning the same contents as default errors.
3433
Willy Tarreau59140a22009-02-22 12:02:30 +01003434 The files should not exceed the configured buffer size (BUFSIZE), which
3435 generally is 8 or 16 kB, otherwise they will be truncated. It is also wise
Davor Ocelice9ed2812017-12-25 17:49:28 +01003436 not to put any reference to local contents (e.g. images) in order to avoid
Willy Tarreau59140a22009-02-22 12:02:30 +01003437 loops between the client and HAProxy when all servers are down, causing an
3438 error to be returned instead of an image. For better HTTP compliance, it is
3439 recommended that all header lines end with CR-LF and not LF alone.
3440
Willy Tarreau0ba27502007-12-24 16:55:16 +01003441 The files are read at the same time as the configuration and kept in memory.
3442 For this reason, the errors continue to be returned even when the process is
3443 chrooted, and no file change is considered while the process is running. A
Willy Tarreauc27debf2008-01-06 08:57:02 +01003444 simple method for developing those files consists in associating them to the
Willy Tarreau0ba27502007-12-24 16:55:16 +01003445 403 status code and interrogating a blocked URL.
3446
3447 See also : "errorloc", "errorloc302", "errorloc303"
3448
Willy Tarreau59140a22009-02-22 12:02:30 +01003449 Example :
3450 errorfile 400 /etc/haproxy/errorfiles/400badreq.http
Willy Tarreau989222a2016-01-15 10:26:26 +01003451 errorfile 408 /dev/null # work around Chrome pre-connect bug
Willy Tarreau59140a22009-02-22 12:02:30 +01003452 errorfile 403 /etc/haproxy/errorfiles/403forbid.http
3453 errorfile 503 /etc/haproxy/errorfiles/503sorry.http
3454
Willy Tarreau2769aa02007-12-27 18:26:09 +01003455
3456errorloc <code> <url>
3457errorloc302 <code> <url>
3458 Return an HTTP redirection to a URL instead of errors generated by HAProxy
3459 May be used in sections : defaults | frontend | listen | backend
3460 yes | yes | yes | yes
3461 Arguments :
3462 <code> is the HTTP status code. Currently, HAProxy is capable of
Olivier Houchard51a76d82017-10-02 16:12:07 +02003463 generating codes 200, 400, 403, 405, 408, 425, 429, 500, 502,
3464 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003465
3466 <url> it is the exact contents of the "Location" header. It may contain
3467 either a relative URI to an error page hosted on the same site,
3468 or an absolute URI designating an error page on another site.
3469 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01003470 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01003471
3472 It is important to understand that this keyword is not meant to rewrite
3473 errors returned by the server, but errors detected and returned by HAProxy.
3474 This is why the list of supported errors is limited to a small set.
3475
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003476 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3477
Willy Tarreau2769aa02007-12-27 18:26:09 +01003478 Note that both keyword return the HTTP 302 status code, which tells the
3479 client to fetch the designated URL using the same HTTP method. This can be
3480 quite problematic in case of non-GET methods such as POST, because the URL
3481 sent to the client might not be allowed for something other than GET. To
Willy Tarreau989222a2016-01-15 10:26:26 +01003482 work around this problem, please use "errorloc303" which send the HTTP 303
Willy Tarreau2769aa02007-12-27 18:26:09 +01003483 status code, indicating to the client that the URL must be fetched with a GET
3484 request.
3485
3486 See also : "errorfile", "errorloc303"
3487
3488
3489errorloc303 <code> <url>
3490 Return an HTTP redirection to a URL instead of errors generated by HAProxy
3491 May be used in sections : defaults | frontend | listen | backend
3492 yes | yes | yes | yes
3493 Arguments :
3494 <code> is the HTTP status code. Currently, HAProxy is capable of
Olivier Houchard51a76d82017-10-02 16:12:07 +02003495 generating codes 200, 400, 403, 405, 408, 425, 429, 500, 502,
3496 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003497
3498 <url> it is the exact contents of the "Location" header. It may contain
3499 either a relative URI to an error page hosted on the same site,
3500 or an absolute URI designating an error page on another site.
3501 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01003502 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01003503
3504 It is important to understand that this keyword is not meant to rewrite
3505 errors returned by the server, but errors detected and returned by HAProxy.
3506 This is why the list of supported errors is limited to a small set.
3507
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003508 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3509
Willy Tarreau2769aa02007-12-27 18:26:09 +01003510 Note that both keyword return the HTTP 303 status code, which tells the
3511 client to fetch the designated URL using the same HTTP GET method. This
3512 solves the usual problems associated with "errorloc" and the 302 code. It is
3513 possible that some very old browsers designed before HTTP/1.1 do not support
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003514 it, but no such problem has been reported till now.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003515
3516 See also : "errorfile", "errorloc", "errorloc302"
3517
3518
Simon Horman51a1cf62015-02-03 13:00:44 +09003519email-alert from <emailaddr>
3520 Declare the from email address to be used in both the envelope and header
Davor Ocelice9ed2812017-12-25 17:49:28 +01003521 of email alerts. This is the address that email alerts are sent from.
Simon Horman51a1cf62015-02-03 13:00:44 +09003522 May be used in sections: defaults | frontend | listen | backend
3523 yes | yes | yes | yes
3524
3525 Arguments :
3526
3527 <emailaddr> is the from email address to use when sending email alerts
3528
3529 Also requires "email-alert mailers" and "email-alert to" to be set
3530 and if so sending email alerts is enabled for the proxy.
3531
Simon Horman64e34162015-02-06 11:11:57 +09003532 See also : "email-alert level", "email-alert mailers",
Cyril Bonté307ee1e2015-09-28 23:16:06 +02003533 "email-alert myhostname", "email-alert to", section 3.6 about
3534 mailers.
Simon Horman64e34162015-02-06 11:11:57 +09003535
3536
3537email-alert level <level>
3538 Declare the maximum log level of messages for which email alerts will be
3539 sent. This acts as a filter on the sending of email alerts.
3540 May be used in sections: defaults | frontend | listen | backend
3541 yes | yes | yes | yes
3542
3543 Arguments :
3544
3545 <level> One of the 8 syslog levels:
3546 emerg alert crit err warning notice info debug
3547 The above syslog levels are ordered from lowest to highest.
3548
3549 By default level is alert
3550
3551 Also requires "email-alert from", "email-alert mailers" and
3552 "email-alert to" to be set and if so sending email alerts is enabled
3553 for the proxy.
3554
Simon Horman1421e212015-04-30 13:10:35 +09003555 Alerts are sent when :
3556
3557 * An un-paused server is marked as down and <level> is alert or lower
3558 * A paused server is marked as down and <level> is notice or lower
3559 * A server is marked as up or enters the drain state and <level>
3560 is notice or lower
3561 * "option log-health-checks" is enabled, <level> is info or lower,
3562 and a health check status update occurs
3563
Simon Horman64e34162015-02-06 11:11:57 +09003564 See also : "email-alert from", "email-alert mailers",
3565 "email-alert myhostname", "email-alert to",
Simon Horman51a1cf62015-02-03 13:00:44 +09003566 section 3.6 about mailers.
3567
3568
3569email-alert mailers <mailersect>
3570 Declare the mailers to be used when sending email alerts
3571 May be used in sections: defaults | frontend | listen | backend
3572 yes | yes | yes | yes
3573
3574 Arguments :
3575
3576 <mailersect> is the name of the mailers section to send email alerts.
3577
3578 Also requires "email-alert from" and "email-alert to" to be set
3579 and if so sending email alerts is enabled for the proxy.
3580
Simon Horman64e34162015-02-06 11:11:57 +09003581 See also : "email-alert from", "email-alert level", "email-alert myhostname",
3582 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09003583
3584
3585email-alert myhostname <hostname>
3586 Declare the to hostname address to be used when communicating with
3587 mailers.
3588 May be used in sections: defaults | frontend | listen | backend
3589 yes | yes | yes | yes
3590
3591 Arguments :
3592
Baptiste Assmann738bad92015-12-21 15:27:53 +01003593 <hostname> is the hostname to use when communicating with mailers
Simon Horman51a1cf62015-02-03 13:00:44 +09003594
3595 By default the systems hostname is used.
3596
3597 Also requires "email-alert from", "email-alert mailers" and
3598 "email-alert to" to be set and if so sending email alerts is enabled
3599 for the proxy.
3600
Simon Horman64e34162015-02-06 11:11:57 +09003601 See also : "email-alert from", "email-alert level", "email-alert mailers",
3602 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09003603
3604
3605email-alert to <emailaddr>
Davor Ocelice9ed2812017-12-25 17:49:28 +01003606 Declare both the recipient address in the envelope and to address in the
Simon Horman51a1cf62015-02-03 13:00:44 +09003607 header of email alerts. This is the address that email alerts are sent to.
3608 May be used in sections: defaults | frontend | listen | backend
3609 yes | yes | yes | yes
3610
3611 Arguments :
3612
3613 <emailaddr> is the to email address to use when sending email alerts
3614
3615 Also requires "email-alert mailers" and "email-alert to" to be set
3616 and if so sending email alerts is enabled for the proxy.
3617
Simon Horman64e34162015-02-06 11:11:57 +09003618 See also : "email-alert from", "email-alert level", "email-alert mailers",
Simon Horman51a1cf62015-02-03 13:00:44 +09003619 "email-alert myhostname", section 3.6 about mailers.
3620
3621
Willy Tarreau4de91492010-01-22 19:10:05 +01003622force-persist { if | unless } <condition>
3623 Declare a condition to force persistence on down servers
3624 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01003625 no | no | yes | yes
Willy Tarreau4de91492010-01-22 19:10:05 +01003626
3627 By default, requests are not dispatched to down servers. It is possible to
3628 force this using "option persist", but it is unconditional and redispatches
3629 to a valid server if "option redispatch" is set. That leaves with very little
3630 possibilities to force some requests to reach a server which is artificially
3631 marked down for maintenance operations.
3632
3633 The "force-persist" statement allows one to declare various ACL-based
3634 conditions which, when met, will cause a request to ignore the down status of
3635 a server and still try to connect to it. That makes it possible to start a
3636 server, still replying an error to the health checks, and run a specially
3637 configured browser to test the service. Among the handy methods, one could
3638 use a specific source IP address, or a specific cookie. The cookie also has
3639 the advantage that it can easily be added/removed on the browser from a test
3640 page. Once the service is validated, it is then possible to open the service
3641 to the world by returning a valid response to health checks.
3642
3643 The forced persistence is enabled when an "if" condition is met, or unless an
3644 "unless" condition is met. The final redispatch is always disabled when this
3645 is used.
3646
Cyril Bonté0d4bf012010-04-25 23:21:46 +02003647 See also : "option redispatch", "ignore-persist", "persist",
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +02003648 and section 7 about ACL usage.
Willy Tarreau4de91492010-01-22 19:10:05 +01003649
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003650
3651filter <name> [param*]
3652 Add the filter <name> in the filter list attached to the proxy.
3653 May be used in sections : defaults | frontend | listen | backend
3654 no | yes | yes | yes
3655 Arguments :
3656 <name> is the name of the filter. Officially supported filters are
3657 referenced in section 9.
3658
Tim Düsterhus4896c442016-11-29 02:15:19 +01003659 <param*> is a list of parameters accepted by the filter <name>. The
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003660 parsing of these parameters are the responsibility of the
Tim Düsterhus4896c442016-11-29 02:15:19 +01003661 filter. Please refer to the documentation of the corresponding
3662 filter (section 9) for all details on the supported parameters.
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003663
3664 Multiple occurrences of the filter line can be used for the same proxy. The
3665 same filter can be referenced many times if needed.
3666
3667 Example:
3668 listen
3669 bind *:80
3670
3671 filter trace name BEFORE-HTTP-COMP
3672 filter compression
3673 filter trace name AFTER-HTTP-COMP
3674
3675 compression algo gzip
3676 compression offload
3677
3678 server srv1 192.168.0.1:80
3679
3680 See also : section 9.
3681
Willy Tarreau4de91492010-01-22 19:10:05 +01003682
Willy Tarreau2769aa02007-12-27 18:26:09 +01003683fullconn <conns>
3684 Specify at what backend load the servers will reach their maxconn
3685 May be used in sections : defaults | frontend | listen | backend
3686 yes | no | yes | yes
3687 Arguments :
3688 <conns> is the number of connections on the backend which will make the
3689 servers use the maximal number of connections.
3690
Willy Tarreau198a7442008-01-17 12:05:32 +01003691 When a server has a "maxconn" parameter specified, it means that its number
Willy Tarreau2769aa02007-12-27 18:26:09 +01003692 of concurrent connections will never go higher. Additionally, if it has a
Willy Tarreau198a7442008-01-17 12:05:32 +01003693 "minconn" parameter, it indicates a dynamic limit following the backend's
Willy Tarreau2769aa02007-12-27 18:26:09 +01003694 load. The server will then always accept at least <minconn> connections,
3695 never more than <maxconn>, and the limit will be on the ramp between both
3696 values when the backend has less than <conns> concurrent connections. This
3697 makes it possible to limit the load on the servers during normal loads, but
3698 push it further for important loads without overloading the servers during
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003699 exceptional loads.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003700
Willy Tarreaufbb78422011-06-05 15:38:35 +02003701 Since it's hard to get this value right, haproxy automatically sets it to
3702 10% of the sum of the maxconns of all frontends that may branch to this
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01003703 backend (based on "use_backend" and "default_backend" rules). That way it's
3704 safe to leave it unset. However, "use_backend" involving dynamic names are
3705 not counted since there is no way to know if they could match or not.
Willy Tarreaufbb78422011-06-05 15:38:35 +02003706
Willy Tarreau2769aa02007-12-27 18:26:09 +01003707 Example :
3708 # The servers will accept between 100 and 1000 concurrent connections each
3709 # and the maximum of 1000 will be reached when the backend reaches 10000
3710 # connections.
3711 backend dynamic
3712 fullconn 10000
3713 server srv1 dyn1:80 minconn 100 maxconn 1000
3714 server srv2 dyn2:80 minconn 100 maxconn 1000
3715
3716 See also : "maxconn", "server"
3717
3718
3719grace <time>
3720 Maintain a proxy operational for some time after a soft stop
3721 May be used in sections : defaults | frontend | listen | backend
Cyril Bonté99ed3272010-01-24 23:29:44 +01003722 yes | yes | yes | yes
Willy Tarreau2769aa02007-12-27 18:26:09 +01003723 Arguments :
3724 <time> is the time (by default in milliseconds) for which the instance
3725 will remain operational with the frontend sockets still listening
3726 when a soft-stop is received via the SIGUSR1 signal.
3727
3728 This may be used to ensure that the services disappear in a certain order.
3729 This was designed so that frontends which are dedicated to monitoring by an
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003730 external equipment fail immediately while other ones remain up for the time
Willy Tarreau2769aa02007-12-27 18:26:09 +01003731 needed by the equipment to detect the failure.
3732
3733 Note that currently, there is very little benefit in using this parameter,
3734 and it may in fact complicate the soft-reconfiguration process more than
3735 simplify it.
3736
Willy Tarreau0ba27502007-12-24 16:55:16 +01003737
Andrew Rodland17be45e2016-10-25 17:04:12 -04003738hash-balance-factor <factor>
3739 Specify the balancing factor for bounded-load consistent hashing
3740 May be used in sections : defaults | frontend | listen | backend
3741 yes | no | no | yes
3742 Arguments :
3743 <factor> is the control for the maximum number of concurrent requests to
3744 send to a server, expressed as a percentage of the average number
3745 of concurrent requests across all of the active servers.
3746
3747 Specifying a "hash-balance-factor" for a server with "hash-type consistent"
3748 enables an algorithm that prevents any one server from getting too many
3749 requests at once, even if some hash buckets receive many more requests than
3750 others. Setting <factor> to 0 (the default) disables the feature. Otherwise,
3751 <factor> is a percentage greater than 100. For example, if <factor> is 150,
3752 then no server will be allowed to have a load more than 1.5 times the average.
3753 If server weights are used, they will be respected.
3754
3755 If the first-choice server is disqualified, the algorithm will choose another
3756 server based on the request hash, until a server with additional capacity is
3757 found. A higher <factor> allows more imbalance between the servers, while a
3758 lower <factor> means that more servers will be checked on average, affecting
3759 performance. Reasonable values are from 125 to 200.
3760
Willy Tarreau760e81d2018-05-03 07:20:40 +02003761 This setting is also used by "balance random" which internally relies on the
3762 consistent hashing mechanism.
3763
Andrew Rodland17be45e2016-10-25 17:04:12 -04003764 See also : "balance" and "hash-type".
3765
3766
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003767hash-type <method> <function> <modifier>
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003768 Specify a method to use for mapping hashes to servers
3769 May be used in sections : defaults | frontend | listen | backend
3770 yes | no | yes | yes
3771 Arguments :
Bhaskar98634f02013-10-29 23:30:51 -04003772 <method> is the method used to select a server from the hash computed by
3773 the <function> :
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003774
Bhaskar98634f02013-10-29 23:30:51 -04003775 map-based the hash table is a static array containing all alive servers.
3776 The hashes will be very smooth, will consider weights, but
3777 will be static in that weight changes while a server is up
3778 will be ignored. This means that there will be no slow start.
3779 Also, since a server is selected by its position in the array,
3780 most mappings are changed when the server count changes. This
3781 means that when a server goes up or down, or when a server is
3782 added to a farm, most connections will be redistributed to
3783 different servers. This can be inconvenient with caches for
3784 instance.
Willy Tarreau798a39c2010-11-24 15:04:29 +01003785
Bhaskar98634f02013-10-29 23:30:51 -04003786 consistent the hash table is a tree filled with many occurrences of each
3787 server. The hash key is looked up in the tree and the closest
3788 server is chosen. This hash is dynamic, it supports changing
3789 weights while the servers are up, so it is compatible with the
3790 slow start feature. It has the advantage that when a server
3791 goes up or down, only its associations are moved. When a
3792 server is added to the farm, only a few part of the mappings
3793 are redistributed, making it an ideal method for caches.
3794 However, due to its principle, the distribution will never be
3795 very smooth and it may sometimes be necessary to adjust a
3796 server's weight or its ID to get a more balanced distribution.
3797 In order to get the same distribution on multiple load
3798 balancers, it is important that all servers have the exact
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003799 same IDs. Note: consistent hash uses sdbm and avalanche if no
3800 hash function is specified.
Bhaskar98634f02013-10-29 23:30:51 -04003801
3802 <function> is the hash function to be used :
3803
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03003804 sdbm this function was created initially for sdbm (a public-domain
Bhaskar98634f02013-10-29 23:30:51 -04003805 reimplementation of ndbm) database library. It was found to do
3806 well in scrambling bits, causing better distribution of the keys
3807 and fewer splits. It also happens to be a good general hashing
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003808 function with good distribution, unless the total server weight
3809 is a multiple of 64, in which case applying the avalanche
3810 modifier may help.
Bhaskar98634f02013-10-29 23:30:51 -04003811
3812 djb2 this function was first proposed by Dan Bernstein many years ago
3813 on comp.lang.c. Studies have shown that for certain workload this
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003814 function provides a better distribution than sdbm. It generally
3815 works well with text-based inputs though it can perform extremely
3816 poorly with numeric-only input or when the total server weight is
3817 a multiple of 33, unless the avalanche modifier is also used.
3818
Willy Tarreaua0f42712013-11-14 14:30:35 +01003819 wt6 this function was designed for haproxy while testing other
3820 functions in the past. It is not as smooth as the other ones, but
3821 is much less sensible to the input data set or to the number of
3822 servers. It can make sense as an alternative to sdbm+avalanche or
3823 djb2+avalanche for consistent hashing or when hashing on numeric
3824 data such as a source IP address or a visitor identifier in a URL
3825 parameter.
3826
Willy Tarreau324f07f2015-01-20 19:44:50 +01003827 crc32 this is the most common CRC32 implementation as used in Ethernet,
3828 gzip, PNG, etc. It is slower than the other ones but may provide
3829 a better distribution or less predictable results especially when
3830 used on strings.
3831
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003832 <modifier> indicates an optional method applied after hashing the key :
3833
3834 avalanche This directive indicates that the result from the hash
3835 function above should not be used in its raw form but that
3836 a 4-byte full avalanche hash must be applied first. The
3837 purpose of this step is to mix the resulting bits from the
3838 previous hash in order to avoid any undesired effect when
3839 the input contains some limited values or when the number of
3840 servers is a multiple of one of the hash's components (64
3841 for SDBM, 33 for DJB2). Enabling avalanche tends to make the
3842 result less predictable, but it's also not as smooth as when
3843 using the original function. Some testing might be needed
3844 with some workloads. This hash is one of the many proposed
3845 by Bob Jenkins.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003846
Bhaskar98634f02013-10-29 23:30:51 -04003847 The default hash type is "map-based" and is recommended for most usages. The
3848 default function is "sdbm", the selection of a function should be based on
3849 the range of the values being hashed.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003850
Andrew Rodland17be45e2016-10-25 17:04:12 -04003851 See also : "balance", "hash-balance-factor", "server"
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003852
3853
Willy Tarreau0ba27502007-12-24 16:55:16 +01003854http-check disable-on-404
3855 Enable a maintenance mode upon HTTP/404 response to health-checks
3856 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau2769aa02007-12-27 18:26:09 +01003857 yes | no | yes | yes
Willy Tarreau0ba27502007-12-24 16:55:16 +01003858 Arguments : none
3859
3860 When this option is set, a server which returns an HTTP code 404 will be
3861 excluded from further load-balancing, but will still receive persistent
3862 connections. This provides a very convenient method for Web administrators
3863 to perform a graceful shutdown of their servers. It is also important to note
3864 that a server which is detected as failed while it was in this mode will not
3865 generate an alert, just a notice. If the server responds 2xx or 3xx again, it
3866 will immediately be reinserted into the farm. The status on the stats page
3867 reports "NOLB" for a server in this mode. It is important to note that this
Willy Tarreaubd741542010-03-16 18:46:54 +01003868 option only works in conjunction with the "httpchk" option. If this option
3869 is used with "http-check expect", then it has precedence over it so that 404
3870 responses will still be considered as soft-stop.
3871
3872 See also : "option httpchk", "http-check expect"
3873
3874
3875http-check expect [!] <match> <pattern>
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003876 Make HTTP health checks consider response contents or specific status codes
Willy Tarreaubd741542010-03-16 18:46:54 +01003877 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau1ee51a62011-08-19 20:04:17 +02003878 yes | no | yes | yes
Willy Tarreaubd741542010-03-16 18:46:54 +01003879 Arguments :
3880 <match> is a keyword indicating how to look for a specific pattern in the
3881 response. The keyword may be one of "status", "rstatus",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003882 "string", or "rstring". The keyword may be preceded by an
Willy Tarreaubd741542010-03-16 18:46:54 +01003883 exclamation mark ("!") to negate the match. Spaces are allowed
3884 between the exclamation mark and the keyword. See below for more
3885 details on the supported keywords.
3886
3887 <pattern> is the pattern to look for. It may be a string or a regular
3888 expression. If the pattern contains spaces, they must be escaped
3889 with the usual backslash ('\').
3890
3891 By default, "option httpchk" considers that response statuses 2xx and 3xx
3892 are valid, and that others are invalid. When "http-check expect" is used,
3893 it defines what is considered valid or invalid. Only one "http-check"
3894 statement is supported in a backend. If a server fails to respond or times
3895 out, the check obviously fails. The available matches are :
3896
3897 status <string> : test the exact string match for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003898 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01003899 response's status code is exactly this string. If the
3900 "status" keyword is prefixed with "!", then the response
3901 will be considered invalid if the status code matches.
3902
3903 rstatus <regex> : test a regular expression for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003904 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01003905 response's status code matches the expression. If the
3906 "rstatus" keyword is prefixed with "!", then the response
3907 will be considered invalid if the status code matches.
3908 This is mostly used to check for multiple codes.
3909
3910 string <string> : test the exact string match in the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003911 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01003912 response's body contains this exact string. If the
3913 "string" keyword is prefixed with "!", then the response
3914 will be considered invalid if the body contains this
3915 string. This can be used to look for a mandatory word at
3916 the end of a dynamic page, or to detect a failure when a
Davor Ocelice9ed2812017-12-25 17:49:28 +01003917 specific error appears on the check page (e.g. a stack
Willy Tarreaubd741542010-03-16 18:46:54 +01003918 trace).
3919
3920 rstring <regex> : test a regular expression on the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003921 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01003922 response's body matches this expression. If the "rstring"
3923 keyword is prefixed with "!", then the response will be
3924 considered invalid if the body matches the expression.
3925 This can be used to look for a mandatory word at the end
3926 of a dynamic page, or to detect a failure when a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01003927 error appears on the check page (e.g. a stack trace).
Willy Tarreaubd741542010-03-16 18:46:54 +01003928
3929 It is important to note that the responses will be limited to a certain size
3930 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
3931 Thus, too large responses may not contain the mandatory pattern when using
3932 "string" or "rstring". If a large response is absolutely required, it is
3933 possible to change the default max size by setting the global variable.
3934 However, it is worth keeping in mind that parsing very large responses can
3935 waste some CPU cycles, especially when regular expressions are used, and that
3936 it is always better to focus the checks on smaller resources.
3937
Cyril Bonté32602d22015-01-30 00:07:07 +01003938 Also "http-check expect" doesn't support HTTP keep-alive. Keep in mind that it
3939 will automatically append a "Connection: close" header, meaning that this
3940 header should not be present in the request provided by "option httpchk".
3941
Willy Tarreaubd741542010-03-16 18:46:54 +01003942 Last, if "http-check expect" is combined with "http-check disable-on-404",
3943 then this last one has precedence when the server responds with 404.
3944
3945 Examples :
3946 # only accept status 200 as valid
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01003947 http-check expect status 200
Willy Tarreaubd741542010-03-16 18:46:54 +01003948
3949 # consider SQL errors as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01003950 http-check expect ! string SQL\ Error
Willy Tarreaubd741542010-03-16 18:46:54 +01003951
3952 # consider status 5xx only as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01003953 http-check expect ! rstatus ^5
Willy Tarreaubd741542010-03-16 18:46:54 +01003954
3955 # check that we have a correct hexadecimal tag before /html
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03003956 http-check expect rstring <!--tag:[0-9a-f]*--></html>
Willy Tarreau0ba27502007-12-24 16:55:16 +01003957
Willy Tarreaubd741542010-03-16 18:46:54 +01003958 See also : "option httpchk", "http-check disable-on-404"
Willy Tarreau2769aa02007-12-27 18:26:09 +01003959
3960
Willy Tarreauef781042010-01-27 11:53:01 +01003961http-check send-state
3962 Enable emission of a state header with HTTP health checks
3963 May be used in sections : defaults | frontend | listen | backend
3964 yes | no | yes | yes
3965 Arguments : none
3966
3967 When this option is set, haproxy will systematically send a special header
3968 "X-Haproxy-Server-State" with a list of parameters indicating to each server
3969 how they are seen by haproxy. This can be used for instance when a server is
3970 manipulated without access to haproxy and the operator needs to know whether
3971 haproxy still sees it up or not, or if the server is the last one in a farm.
3972
3973 The header is composed of fields delimited by semi-colons, the first of which
3974 is a word ("UP", "DOWN", "NOLB"), possibly followed by a number of valid
3975 checks on the total number before transition, just as appears in the stats
3976 interface. Next headers are in the form "<variable>=<value>", indicating in
3977 no specific order some values available in the stats interface :
Joseph Lynch514061c2015-01-15 17:52:59 -08003978 - a variable "address", containing the address of the backend server.
3979 This corresponds to the <address> field in the server declaration. For
3980 unix domain sockets, it will read "unix".
3981
3982 - a variable "port", containing the port of the backend server. This
3983 corresponds to the <port> field in the server declaration. For unix
3984 domain sockets, it will read "unix".
3985
Willy Tarreauef781042010-01-27 11:53:01 +01003986 - a variable "name", containing the name of the backend followed by a slash
3987 ("/") then the name of the server. This can be used when a server is
3988 checked in multiple backends.
3989
3990 - a variable "node" containing the name of the haproxy node, as set in the
3991 global "node" variable, otherwise the system's hostname if unspecified.
3992
3993 - a variable "weight" indicating the weight of the server, a slash ("/")
3994 and the total weight of the farm (just counting usable servers). This
3995 helps to know if other servers are available to handle the load when this
3996 one fails.
3997
3998 - a variable "scur" indicating the current number of concurrent connections
3999 on the server, followed by a slash ("/") then the total number of
4000 connections on all servers of the same backend.
4001
4002 - a variable "qcur" indicating the current number of requests in the
4003 server's queue.
4004
4005 Example of a header received by the application server :
4006 >>> X-Haproxy-Server-State: UP 2/3; name=bck/srv2; node=lb1; weight=1/2; \
4007 scur=13/22; qcur=0
4008
4009 See also : "option httpchk", "http-check disable-on-404"
4010
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004011
4012http-request <action> [options...] [ { if | unless } <condition> ]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004013 Access control for Layer 7 requests
4014
4015 May be used in sections: defaults | frontend | listen | backend
4016 no | yes | yes | yes
4017
Willy Tarreau20b0de52012-12-24 15:45:22 +01004018 The http-request statement defines a set of rules which apply to layer 7
4019 processing. The rules are evaluated in their declaration order when they are
4020 met in a frontend, listen or backend section. Any rule may optionally be
4021 followed by an ACL-based condition, in which case it will only be evaluated
4022 if the condition is true.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004023
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004024 The first keyword is the rule's action. The supported actions are described
4025 below.
Willy Tarreau20b0de52012-12-24 15:45:22 +01004026
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004027 There is no limit to the number of http-request statements per instance.
Willy Tarreau20b0de52012-12-24 15:45:22 +01004028
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004029 It is important to know that http-request rules are processed very early in
4030 the HTTP processing, just after "block" rules and before "reqdel" or "reqrep"
4031 or "reqadd" rules. That way, headers added by "add-header"/"set-header" are
4032 visible by almost all further ACL rules.
Willy Tarreau53275e82017-11-24 07:52:01 +01004033
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004034 Using "reqadd"/"reqdel"/"reqrep" to manipulate request headers is discouraged
4035 in newer versions (>= 1.5). But if you need to use regular expression to
4036 delete headers, you can still use "reqdel". Also please use
4037 "http-request deny/allow/tarpit" instead of "reqdeny"/"reqpass"/"reqtarpit".
Willy Tarreauccbcc372012-12-27 12:37:57 +01004038
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004039 Example:
4040 acl nagios src 192.168.129.3
4041 acl local_net src 192.168.0.0/16
4042 acl auth_ok http_auth(L1)
Willy Tarreau20b0de52012-12-24 15:45:22 +01004043
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004044 http-request allow if nagios
4045 http-request allow if local_net auth_ok
4046 http-request auth realm Gimme if local_net auth_ok
4047 http-request deny
Willy Tarreau81499eb2012-12-27 12:19:02 +01004048
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004049 Example:
4050 acl key req.hdr(X-Add-Acl-Key) -m found
4051 acl add path /addacl
4052 acl del path /delacl
Willy Tarreau20b0de52012-12-24 15:45:22 +01004053
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004054 acl myhost hdr(Host) -f myhost.lst
Willy Tarreau20b0de52012-12-24 15:45:22 +01004055
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004056 http-request add-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key add
4057 http-request del-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key del
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02004058
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004059 Example:
4060 acl value req.hdr(X-Value) -m found
4061 acl setmap path /setmap
4062 acl delmap path /delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06004063
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004064 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06004065
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004066 http-request set-map(map.lst) %[src] %[req.hdr(X-Value)] if setmap value
4067 http-request del-map(map.lst) %[src] if delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06004068
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004069 See also : "stats http-request", section 3.4 about userlists and section 7
4070 about ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06004071
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004072http-request add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004073
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004074 This is used to add a new entry into an ACL. The ACL must be loaded from a
4075 file (even a dummy empty file). The file name of the ACL to be updated is
4076 passed between parentheses. It takes one argument: <key fmt>, which follows
4077 log-format rules, to collect content of the new entry. It performs a lookup
4078 in the ACL before insertion, to avoid duplicated (or more) values. This
4079 lookup is done by a linear search and can be expensive with large lists!
4080 It is the equivalent of the "add acl" command from the stats socket, but can
4081 be triggered by an HTTP request.
Sasha Pachev218f0642014-06-16 12:05:59 -06004082
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004083http-request add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004084
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004085 This appends an HTTP header field whose name is specified in <name> and
4086 whose value is defined by <fmt> which follows the log-format rules (see
4087 Custom Log Format in section 8.2.4). This is particularly useful to pass
4088 connection-specific information to the server (e.g. the client's SSL
4089 certificate), or to combine several headers into one. This rule is not
4090 final, so it is possible to add other similar rules. Note that header
4091 addition is performed immediately, so one rule might reuse the resulting
4092 header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06004093
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004094http-request allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004095
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004096 This stops the evaluation of the rules and lets the request pass the check.
4097 No further "http-request" rules are evaluated.
Sasha Pachev218f0642014-06-16 12:05:59 -06004098
Sasha Pachev218f0642014-06-16 12:05:59 -06004099
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004100http-request auth [realm <realm>] [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004101
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004102 This stops the evaluation of the rules and immediately responds with an
4103 HTTP 401 or 407 error code to invite the user to present a valid user name
4104 and password. No further "http-request" rules are evaluated. An optional
4105 "realm" parameter is supported, it sets the authentication realm that is
4106 returned with the response (typically the application's name).
Sasha Pachev218f0642014-06-16 12:05:59 -06004107
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004108 Example:
4109 acl auth_ok http_auth_group(L1) G1
4110 http-request auth unless auth_ok
Sasha Pachev218f0642014-06-16 12:05:59 -06004111
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004112http-request cache-use [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004113
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004114 See section 10.2 about cache setup.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004115
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004116http-request capture <sample> [ len <length> | id <id> ]
4117 [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004118
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004119 This captures sample expression <sample> from the request buffer, and
4120 converts it to a string of at most <len> characters. The resulting string is
4121 stored into the next request "capture" slot, so it will possibly appear next
4122 to some captured HTTP headers. It will then automatically appear in the logs,
4123 and it will be possible to extract it using sample fetch rules to feed it
4124 into headers or anything. The length should be limited given that this size
4125 will be allocated for each capture during the whole session life.
4126 Please check section 7.3 (Fetching samples) and "capture request header" for
4127 more information.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004128
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004129 If the keyword "id" is used instead of "len", the action tries to store the
4130 captured string in a previously declared capture slot. This is useful to run
4131 captures in backends. The slot id can be declared by a previous directive
4132 "http-request capture" or with the "declare capture" keyword. If the slot
4133 <id> doesn't exist, then HAProxy fails parsing the configuration to prevent
4134 unexpected behavior at run time.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004135
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004136http-request del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004137
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004138 This is used to delete an entry from an ACL. The ACL must be loaded from a
4139 file (even a dummy empty file). The file name of the ACL to be updated is
4140 passed between parentheses. It takes one argument: <key fmt>, which follows
4141 log-format rules, to collect content of the entry to delete.
4142 It is the equivalent of the "del acl" command from the stats socket, but can
4143 be triggered by an HTTP request.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004144
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004145http-request del-header <name> [ { if | unless } <condition> ]
Willy Tarreauf4c43c12013-06-11 17:01:13 +02004146
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004147 This removes all HTTP header fields whose name is specified in <name>.
Willy Tarreau9a355ec2013-06-11 17:45:46 +02004148
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004149http-request del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau42cf39e2013-06-11 18:51:32 +02004150
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004151 This is used to delete an entry from a MAP. The MAP must be loaded from a
4152 file (even a dummy empty file). The file name of the MAP to be updated is
4153 passed between parentheses. It takes one argument: <key fmt>, which follows
4154 log-format rules, to collect content of the entry to delete.
4155 It takes one argument: "file name" It is the equivalent of the "del map"
4156 command from the stats socket, but can be triggered by an HTTP request.
Willy Tarreau51347ed2013-06-11 19:34:13 +02004157
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004158http-request deny [deny_status <status>] [ { if | unless } <condition> ]
Patrick Hemmer268a7072018-05-11 12:52:31 -04004159
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004160 This stops the evaluation of the rules and immediately rejects the request
4161 and emits an HTTP 403 error, or optionally the status code specified as an
4162 argument to "deny_status". The list of permitted status codes is limited to
4163 those that can be overridden by the "errorfile" directive.
4164 No further "http-request" rules are evaluated.
Patrick Hemmer268a7072018-05-11 12:52:31 -04004165
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01004166http-request early-hint <name> <fmt> [ { if | unless } <condition> ]
4167
4168 This is used to build an HTTP 103 Early Hints response prior to any other one.
4169 This appends an HTTP header field to this response whose name is specified in
4170 <name> and whose value is defined by <fmt> which follows the log-format rules
4171 (see Custom Log Format in section 8.2.4). This is particularly useful to pass
Frédéric Lécaille3aac1062018-11-13 09:42:13 +01004172 to the client some Link headers to preload resources required to render the
4173 HTML documents.
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01004174
4175 See RFC 8297 for more information.
4176
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004177http-request redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004178
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004179 This performs an HTTP redirection based on a redirect rule. This is exactly
4180 the same as the "redirect" statement except that it inserts a redirect rule
4181 which can be processed in the middle of other "http-request" rules and that
4182 these rules use the "log-format" strings. See the "redirect" keyword for the
4183 rule's syntax.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004184
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004185http-request reject [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004186
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004187 This stops the evaluation of the rules and immediately closes the connection
4188 without sending any response. It acts similarly to the
4189 "tcp-request content reject" rules. It can be useful to force an immediate
4190 connection closure on HTTP/2 connections.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004191
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004192http-request replace-header <name> <match-regex> <replace-fmt>
4193 [ { if | unless } <condition> ]
Willy Tarreaua9083d02015-05-08 15:27:59 +02004194
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004195 This matches the regular expression in all occurrences of header field
4196 <name> according to <match-regex>, and replaces them with the
4197 <replace-fmt> argument. Format characters are allowed in replace-fmt and
4198 work like in <fmt> arguments in "http-request add-header". The match is
4199 only case-sensitive. It is important to understand that this action only
4200 considers whole header lines, regardless of the number of values they may
4201 contain. This usage is suited to headers naturally containing commas in
4202 their value, such as If-Modified-Since and so on.
Thierry FOURNIER82bf70d2015-05-26 17:58:29 +02004203
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004204 Example:
4205 http-request replace-header Cookie foo=([^;]*);(.*) foo=\1;ip=%bi;\2
William Lallemand86d0df02017-11-24 21:36:45 +01004206
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004207 # applied to:
4208 Cookie: foo=foobar; expires=Tue, 14-Jun-2016 01:40:45 GMT;
Willy Tarreau09448f72014-06-25 18:12:15 +02004209
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004210 # outputs:
4211 Cookie: foo=foobar;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT;
Willy Tarreau09448f72014-06-25 18:12:15 +02004212
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004213 # assuming the backend IP is 192.168.1.20
Willy Tarreau09448f72014-06-25 18:12:15 +02004214
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004215http-request replace-value <name> <match-regex> <replace-fmt>
4216 [ { if | unless } <condition> ]
Willy Tarreau09448f72014-06-25 18:12:15 +02004217
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004218 This works like "replace-header" except that it matches the regex against
4219 every comma-delimited value of the header field <name> instead of the
4220 entire header. This is suited for all headers which are allowed to carry
4221 more than one value. An example could be the Accept header.
Willy Tarreau09448f72014-06-25 18:12:15 +02004222
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004223 Example:
4224 http-request replace-value X-Forwarded-For ^192\.168\.(.*)$ 172.16.\1
Thierry FOURNIER236657b2015-08-19 08:25:14 +02004225
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004226 # applied to:
4227 X-Forwarded-For: 192.168.10.1, 192.168.13.24, 10.0.0.37
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02004228
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004229 # outputs:
4230 X-Forwarded-For: 172.16.10.1, 172.16.13.24, 10.0.0.37
Frédéric Lécaille6778b272018-01-29 15:22:53 +01004231
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004232http-request sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
4233http-request sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004234
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004235 This actions increments the GPC0 or GPC1 counter according with the sticky
4236 counter designated by <sc-id>. If an error occurs, this action silently fails
4237 and the actions evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004238
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004239http-request sc-set-gpt0(<sc-id>) <int> [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004240
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004241 This action sets the GPT0 tag according to the sticky counter designated by
4242 <sc-id> and the value of <int>. The expected result is a boolean. If an error
4243 occurs, this action silently fails and the actions evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004244
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004245http-request set-dst <expr> [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004246
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004247 This is used to set the destination IP address to the value of specified
4248 expression. Useful when a proxy in front of HAProxy rewrites destination IP,
4249 but provides the correct IP in a HTTP header; or you want to mask the IP for
4250 privacy. If you want to connect to the new address/port, use '0.0.0.0:0' as a
4251 server address in the backend.
Christopher Faulet85d79c92016-11-09 16:54:56 +01004252
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004253 Arguments:
4254 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
4255 by some converters.
Christopher Faulet85d79c92016-11-09 16:54:56 +01004256
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004257 Example:
4258 http-request set-dst hdr(x-dst)
4259 http-request set-dst dst,ipmask(24)
Christopher Faulet85d79c92016-11-09 16:54:56 +01004260
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004261 When possible, set-dst preserves the original destination port as long as the
4262 address family allows it, otherwise the destination port is set to 0.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004263
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004264http-request set-dst-port <expr> [ { if | unless } <condition> ]
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004265
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004266 This is used to set the destination port address to the value of specified
4267 expression. If you want to connect to the new address/port, use '0.0.0.0:0'
4268 as a server address in the backend.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004269
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004270 Arguments:
4271 <expr> Is a standard HAProxy expression formed by a sample-fetch
4272 followed by some converters.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004273
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004274 Example:
4275 http-request set-dst-port hdr(x-port)
4276 http-request set-dst-port int(4000)
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004277
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004278 When possible, set-dst-port preserves the original destination address as
4279 long as the address family supports a port, otherwise it forces the
4280 destination address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +02004281
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004282http-request set-header <name> <fmt> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02004283
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004284 This does the same as "http-request add-header" except that the header name
4285 is first removed if it existed. This is useful when passing security
4286 information to the server, where the header must not be manipulated by
4287 external users. Note that the new value is computed before the removal so it
4288 is possible to concatenate a value to an existing header.
William Lallemand44be6402016-05-25 01:51:35 +02004289
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004290 Example:
4291 http-request set-header X-Haproxy-Current-Date %T
4292 http-request set-header X-SSL %[ssl_fc]
4293 http-request set-header X-SSL-Session_ID %[ssl_fc_session_id,hex]
4294 http-request set-header X-SSL-Client-Verify %[ssl_c_verify]
4295 http-request set-header X-SSL-Client-DN %{+Q}[ssl_c_s_dn]
4296 http-request set-header X-SSL-Client-CN %{+Q}[ssl_c_s_dn(cn)]
4297 http-request set-header X-SSL-Issuer %{+Q}[ssl_c_i_dn]
4298 http-request set-header X-SSL-Client-NotBefore %{+Q}[ssl_c_notbefore]
4299 http-request set-header X-SSL-Client-NotAfter %{+Q}[ssl_c_notafter]
William Lallemand44be6402016-05-25 01:51:35 +02004300
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004301http-request set-log-level <level> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02004302
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004303 This is used to change the log level of the current request when a certain
4304 condition is met. Valid levels are the 8 syslog levels (see the "log"
4305 keyword) plus the special level "silent" which disables logging for this
4306 request. This rule is not final so the last matching rule wins. This rule
4307 can be useful to disable health checks coming from another equipment.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004308
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004309http-request set-map(<file-name>) <key fmt> <value fmt>
4310 [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004311
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004312 This is used to add a new entry into a MAP. The MAP must be loaded from a
4313 file (even a dummy empty file). The file name of the MAP to be updated is
4314 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
4315 log-format rules, used to collect MAP key, and <value fmt>, which follows
4316 log-format rules, used to collect content for the new entry.
4317 It performs a lookup in the MAP before insertion, to avoid duplicated (or
4318 more) values. This lookup is done by a linear search and can be expensive
4319 with large lists! It is the equivalent of the "set map" command from the
4320 stats socket, but can be triggered by an HTTP request.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004321
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004322http-request set-mark <mark> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004323
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004324 This is used to set the Netfilter MARK on all packets sent to the client to
4325 the value passed in <mark> on platforms which support it. This value is an
4326 unsigned 32 bit value which can be matched by netfilter and by the routing
4327 table. It can be expressed both in decimal or hexadecimal format (prefixed by
4328 "0x"). This can be useful to force certain packets to take a different route
4329 (for example a cheaper network path for bulk downloads). This works on Linux
4330 kernels 2.6.32 and above and requires admin privileges.
Willy Tarreau00005ce2016-10-21 15:07:45 +02004331
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004332http-request set-method <fmt> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004333
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004334 This rewrites the request method with the result of the evaluation of format
4335 string <fmt>. There should be very few valid reasons for having to do so as
4336 this is more likely to break something than to fix it.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004337
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004338http-request set-nice <nice> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004339
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004340 This sets the "nice" factor of the current request being processed. It only
4341 has effect against the other requests being processed at the same time.
4342 The default value is 0, unless altered by the "nice" setting on the "bind"
4343 line. The accepted range is -1024..1024. The higher the value, the nicest
4344 the request will be. Lower values will make the request more important than
4345 other ones. This can be useful to improve the speed of some requests, or
4346 lower the priority of non-important requests. Using this setting without
4347 prior experimentation can cause some major slowdown.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004348
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004349http-request set-path <fmt> [ { if | unless } <condition> ]
Willy Tarreau00005ce2016-10-21 15:07:45 +02004350
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004351 This rewrites the request path with the result of the evaluation of format
4352 string <fmt>. The query string, if any, is left intact. If a scheme and
4353 authority is found before the path, they are left intact as well. If the
4354 request doesn't have a path ("*"), this one is replaced with the format.
4355 This can be used to prepend a directory component in front of a path for
4356 example. See also "http-request set-query" and "http-request set-uri".
Willy Tarreau2d392c22015-08-24 01:43:45 +02004357
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004358 Example :
4359 # prepend the host name before the path
4360 http-request set-path /%[hdr(host)]%[path]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004361
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004362http-request set-priority-class <expr> [ { if | unless } <condition> ]
Olivier Houchardccaa7de2017-10-02 11:51:03 +02004363
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004364 This is used to set the queue priority class of the current request.
4365 The value must be a sample expression which converts to an integer in the
4366 range -2047..2047. Results outside this range will be truncated.
4367 The priority class determines the order in which queued requests are
4368 processed. Lower values have higher priority.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004369
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004370http-request set-priority-offset <expr> [ { if | unless } <condition> ]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004371
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004372 This is used to set the queue priority timestamp offset of the current
4373 request. The value must be a sample expression which converts to an integer
4374 in the range -524287..524287. Results outside this range will be truncated.
4375 When a request is queued, it is ordered first by the priority class, then by
4376 the current timestamp adjusted by the given offset in milliseconds. Lower
4377 values have higher priority.
4378 Note that the resulting timestamp is is only tracked with enough precision
4379 for 524,287ms (8m44s287ms). If the request is queued long enough to where the
4380 adjusted timestamp exceeds this value, it will be misidentified as highest
4381 priority. Thus it is important to set "timeout queue" to a value, where when
4382 combined with the offset, does not exceed this limit.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004383
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004384http-request set-query <fmt> [ { if | unless } <condition> ]
Willy Tarreau20b0de52012-12-24 15:45:22 +01004385
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004386 This rewrites the request's query string which appears after the first
4387 question mark ("?") with the result of the evaluation of format string <fmt>.
4388 The part prior to the question mark is left intact. If the request doesn't
4389 contain a question mark and the new value is not empty, then one is added at
4390 the end of the URI, followed by the new value. If a question mark was
4391 present, it will never be removed even if the value is empty. This can be
4392 used to add or remove parameters from the query string.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08004393
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004394 See also "http-request set-query" and "http-request set-uri".
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004395
4396 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004397 # replace "%3D" with "=" in the query string
4398 http-request set-query %[query,regsub(%3D,=,g)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004399
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004400http-request set-src <expr> [ { if | unless } <condition> ]
4401 This is used to set the source IP address to the value of specified
4402 expression. Useful when a proxy in front of HAProxy rewrites source IP, but
4403 provides the correct IP in a HTTP header; or you want to mask source IP for
4404 privacy.
4405
4406 Arguments :
4407 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
4408 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004409
Cyril Bonté78caf842010-03-10 22:41:43 +01004410 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004411 http-request set-src hdr(x-forwarded-for)
4412 http-request set-src src,ipmask(24)
4413
4414 When possible, set-src preserves the original source port as long as the
4415 address family allows it, otherwise the source port is set to 0.
4416
4417http-request set-src-port <expr> [ { if | unless } <condition> ]
4418
4419 This is used to set the source port address to the value of specified
4420 expression.
4421
4422 Arguments:
4423 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
4424 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004425
Willy Tarreau20b0de52012-12-24 15:45:22 +01004426 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004427 http-request set-src-port hdr(x-port)
4428 http-request set-src-port int(4000)
4429
4430 When possible, set-src-port preserves the original source address as long as
4431 the address family supports a port, otherwise it forces the source address to
4432 IPv4 "0.0.0.0" before rewriting the port.
4433
4434http-request set-tos <tos> [ { if | unless } <condition> ]
4435
4436 This is used to set the TOS or DSCP field value of packets sent to the client
4437 to the value passed in <tos> on platforms which support this. This value
4438 represents the whole 8 bits of the IP TOS field, and can be expressed both in
4439 decimal or hexadecimal format (prefixed by "0x"). Note that only the 6 higher
4440 bits are used in DSCP or TOS, and the two lower bits are always 0. This can
4441 be used to adjust some routing behavior on border routers based on some
4442 information from the request.
4443
4444 See RFC 2474, 2597, 3260 and 4594 for more information.
4445
4446http-request set-uri <fmt> [ { if | unless } <condition> ]
4447
4448 This rewrites the request URI with the result of the evaluation of format
4449 string <fmt>. The scheme, authority, path and query string are all replaced
4450 at once. This can be used to rewrite hosts in front of proxies, or to
4451 perform complex modifications to the URI such as moving parts between the
4452 path and the query string.
4453 See also "http-request set-path" and "http-request set-query".
4454
4455http-request set-var(<var-name>) <expr> [ { if | unless } <condition> ]
4456
4457 This is used to set the contents of a variable. The variable is declared
4458 inline.
4459
4460 Arguments:
4461 <var-name> The name of the variable starts with an indication about its
4462 scope. The scopes allowed are:
4463 "proc" : the variable is shared with the whole process
4464 "sess" : the variable is shared with the whole session
4465 "txn" : the variable is shared with the transaction
4466 (request and response)
4467 "req" : the variable is shared only during request
4468 processing
4469 "res" : the variable is shared only during response
4470 processing
4471 This prefix is followed by a name. The separator is a '.'.
4472 The name may only contain characters 'a-z', 'A-Z', '0-9'
4473 and '_'.
4474
4475 <expr> Is a standard HAProxy expression formed by a sample-fetch
4476 followed by some converters.
Willy Tarreau20b0de52012-12-24 15:45:22 +01004477
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004478 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004479 http-request set-var(req.my_var) req.fhdr(user-agent),lower
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004480
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004481http-request send-spoe-group <engine-name> <group-name>
4482 [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004483
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004484 This action is used to trigger sending of a group of SPOE messages. To do so,
4485 the SPOE engine used to send messages must be defined, as well as the SPOE
4486 group to send. Of course, the SPOE engine must refer to an existing SPOE
4487 filter. If not engine name is provided on the SPOE filter line, the SPOE
4488 agent name must be used.
4489
4490 Arguments:
4491 <engine-name> The SPOE engine name.
4492
4493 <group-name> The SPOE group name as specified in the engine
4494 configuration.
4495
4496http-request silent-drop [ { if | unless } <condition> ]
4497
4498 This stops the evaluation of the rules and makes the client-facing connection
4499 suddenly disappear using a system-dependent way that tries to prevent the
4500 client from being notified. The effect it then that the client still sees an
4501 established connection while there's none on HAProxy. The purpose is to
4502 achieve a comparable effect to "tarpit" except that it doesn't use any local
4503 resource at all on the machine running HAProxy. It can resist much higher
4504 loads than "tarpit", and slow down stronger attackers. It is important to
4505 understand the impact of using this mechanism. All stateful equipment placed
4506 between the client and HAProxy (firewalls, proxies, load balancers) will also
4507 keep the established connection for a long time and may suffer from this
4508 action.
4509 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
4510 option is used to block the emission of a TCP reset. On other systems, the
4511 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
4512 router, though it's still delivered to local networks. Do not use it unless
4513 you fully understand how it works.
4514
4515http-request tarpit [deny_status <status>] [ { if | unless } <condition> ]
4516
4517 This stops the evaluation of the rules and immediately blocks the request
4518 without responding for a delay specified by "timeout tarpit" or
4519 "timeout connect" if the former is not set. After that delay, if the client
4520 is still connected, an HTTP error 500 (or optionally the status code
4521 specified as an argument to "deny_status") is returned so that the client
4522 does not suspect it has been tarpitted. Logs will report the flags "PT".
4523 The goal of the tarpit rule is to slow down robots during an attack when
4524 they're limited on the number of concurrent requests. It can be very
4525 efficient against very dumb robots, and will significantly reduce the load
4526 on firewalls compared to a "deny" rule. But when facing "correctly"
4527 developed robots, it can make things worse by forcing haproxy and the front
4528 firewall to support insane number of concurrent connections.
4529 See also the "silent-drop" action.
4530
4531http-request track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
4532http-request track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
4533http-request track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
4534
4535 This enables tracking of sticky counters from current request. These rules do
4536 not stop evaluation and do not change default action. The number of counters
4537 that may be simultaneously tracked by the same connection is set in
4538 MAX_SESS_STKCTR at build time (reported in haproxy -vv) which defaults to 3,
4539 so the track-sc number is between 0 and (MAX_SESS_STCKTR-1). The first
4540 "track-sc0" rule executed enables tracking of the counters of the specified
4541 table as the first set. The first "track-sc1" rule executed enables tracking
4542 of the counters of the specified table as the second set. The first
4543 "track-sc2" rule executed enables tracking of the counters of the specified
4544 table as the third set. It is a recommended practice to use the first set of
4545 counters for the per-frontend counters and the second set for the per-backend
4546 ones. But this is just a guideline, all may be used everywhere.
4547
4548 Arguments :
4549 <key> is mandatory, and is a sample expression rule as described in
4550 section 7.3. It describes what elements of the incoming request or
4551 connection will be analyzed, extracted, combined, and used to
4552 select which table entry to update the counters.
4553
4554 <table> is an optional table to be used instead of the default one, which
4555 is the stick-table declared in the current proxy. All the counters
4556 for the matches and updates for the key will then be performed in
4557 that table until the session ends.
4558
4559 Once a "track-sc*" rule is executed, the key is looked up in the table and if
4560 it is not found, an entry is allocated for it. Then a pointer to that entry
4561 is kept during all the session's life, and this entry's counters are updated
4562 as often as possible, every time the session's counters are updated, and also
4563 systematically when the session ends. Counters are only updated for events
4564 that happen after the tracking has been started. As an exception, connection
4565 counters and request counters are systematically updated so that they reflect
4566 useful information.
4567
4568 If the entry tracks concurrent connection counters, one connection is counted
4569 for as long as the entry is tracked, and the entry will not expire during
4570 that time. Tracking counters also provides a performance advantage over just
4571 checking the keys, because only one table lookup is performed for all ACL
4572 checks that make use of it.
4573
4574http-request unset-var(<var-name>) [ { if | unless } <condition> ]
4575
4576 This is used to unset a variable. See above for details about <var-name>.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004577
4578 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004579 http-request unset-var(req.my_var)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004580
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004581http-request wait-for-handshake [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004582
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004583 This will delay the processing of the request until the SSL handshake
4584 happened. This is mostly useful to delay processing early data until we're
4585 sure they are valid.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004586
Willy Tarreauef781042010-01-27 11:53:01 +01004587
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004588http-response <action> <options...> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004589 Access control for Layer 7 responses
4590
4591 May be used in sections: defaults | frontend | listen | backend
4592 no | yes | yes | yes
4593
4594 The http-response statement defines a set of rules which apply to layer 7
4595 processing. The rules are evaluated in their declaration order when they are
4596 met in a frontend, listen or backend section. Any rule may optionally be
4597 followed by an ACL-based condition, in which case it will only be evaluated
4598 if the condition is true. Since these rules apply on responses, the backend
4599 rules are applied first, followed by the frontend's rules.
4600
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004601 The first keyword is the rule's action. The supported actions are described
4602 below.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004603
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004604 There is no limit to the number of http-response statements per instance.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004605
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004606 It is important to know that http-response rules are processed very early in
4607 the HTTP processing, before "rspdel" or "rsprep" or "rspadd" rules. That way,
4608 headers added by "add-header"/"set-header" are visible by almost all further
4609 ACL rules.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004610
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004611 Using "rspadd"/"rspdel"/"rsprep" to manipulate request headers is discouraged
4612 in newer versions (>= 1.5). But if you need to use regular expression to
4613 delete headers, you can still use "rspdel". Also please use
4614 "http-response deny" instead of "rspdeny".
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004615
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004616 Example:
4617 acl key_acl res.hdr(X-Acl-Key) -m found
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02004618
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004619 acl myhost hdr(Host) -f myhost.lst
Sasha Pachev218f0642014-06-16 12:05:59 -06004620
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004621 http-response add-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
4622 http-response del-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
Sasha Pachev218f0642014-06-16 12:05:59 -06004623
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004624 Example:
4625 acl value res.hdr(X-Value) -m found
Sasha Pachev218f0642014-06-16 12:05:59 -06004626
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004627 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06004628
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004629 http-response set-map(map.lst) %[src] %[res.hdr(X-Value)] if value
4630 http-response del-map(map.lst) %[src] if ! value
Sasha Pachev218f0642014-06-16 12:05:59 -06004631
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004632 See also : "http-request", section 3.4 about userlists and section 7 about
4633 ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06004634
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004635http-response add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004636
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004637 This is used to add a new entry into an ACL. The ACL must be loaded from a
4638 file (even a dummy empty file). The file name of the ACL to be updated is
4639 passed between parentheses. It takes one argument: <key fmt>, which follows
4640 log-format rules, to collect content of the new entry. It performs a lookup
4641 in the ACL before insertion, to avoid duplicated (or more) values.
4642 This lookup is done by a linear search and can be expensive with large lists!
4643 It is the equivalent of the "add acl" command from the stats socket, but can
4644 be triggered by an HTTP response.
Sasha Pachev218f0642014-06-16 12:05:59 -06004645
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004646http-response add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004647
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004648 This appends an HTTP header field whose name is specified in <name> and whose
4649 value is defined by <fmt> which follows the log-format rules (see Custom Log
4650 Format in section 8.2.4). This may be used to send a cookie to a client for
4651 example, or to pass some internal information.
4652 This rule is not final, so it is possible to add other similar rules.
4653 Note that header addition is performed immediately, so one rule might reuse
4654 the resulting header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06004655
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004656http-response allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004657
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004658 This stops the evaluation of the rules and lets the response pass the check.
4659 No further "http-response" rules are evaluated for the current section.
Sasha Pachev218f0642014-06-16 12:05:59 -06004660
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004661http-response cache-store [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004662
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004663 See section 10.2 about cache setup.
Sasha Pachev218f0642014-06-16 12:05:59 -06004664
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004665http-response capture <sample> id <id> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004666
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004667 This captures sample expression <sample> from the response buffer, and
4668 converts it to a string. The resulting string is stored into the next request
4669 "capture" slot, so it will possibly appear next to some captured HTTP
4670 headers. It will then automatically appear in the logs, and it will be
4671 possible to extract it using sample fetch rules to feed it into headers or
4672 anything. Please check section 7.3 (Fetching samples) and
4673 "capture response header" for more information.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02004674
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004675 The keyword "id" is the id of the capture slot which is used for storing the
4676 string. The capture slot must be defined in an associated frontend.
4677 This is useful to run captures in backends. The slot id can be declared by a
4678 previous directive "http-response capture" or with the "declare capture"
4679 keyword.
4680 If the slot <id> doesn't exist, then HAProxy fails parsing the configuration
4681 to prevent unexpected behavior at run time.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02004682
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004683http-response del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02004684
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004685 This is used to delete an entry from an ACL. The ACL must be loaded from a
4686 file (even a dummy empty file). The file name of the ACL to be updated is
4687 passed between parentheses. It takes one argument: <key fmt>, which follows
4688 log-format rules, to collect content of the entry to delete.
4689 It is the equivalent of the "del acl" command from the stats socket, but can
4690 be triggered by an HTTP response.
Willy Tarreauf4c43c12013-06-11 17:01:13 +02004691
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004692http-response del-header <name> [ { if | unless } <condition> ]
Willy Tarreau9a355ec2013-06-11 17:45:46 +02004693
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004694 This removes all HTTP header fields whose name is specified in <name>.
Willy Tarreau42cf39e2013-06-11 18:51:32 +02004695
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004696http-response del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau51347ed2013-06-11 19:34:13 +02004697
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004698 This is used to delete an entry from a MAP. The MAP must be loaded from a
4699 file (even a dummy empty file). The file name of the MAP to be updated is
4700 passed between parentheses. It takes one argument: <key fmt>, which follows
4701 log-format rules, to collect content of the entry to delete.
4702 It takes one argument: "file name" It is the equivalent of the "del map"
4703 command from the stats socket, but can be triggered by an HTTP response.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004704
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004705http-response deny [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004706
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004707 This stops the evaluation of the rules and immediately rejects the response
4708 and emits an HTTP 502 error. No further "http-response" rules are evaluated.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004709
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004710http-response redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004711
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004712 This performs an HTTP redirection based on a redirect rule.
4713 This supports a format string similarly to "http-request redirect" rules,
4714 with the exception that only the "location" type of redirect is possible on
4715 the response. See the "redirect" keyword for the rule's syntax. When a
4716 redirect rule is applied during a response, connections to the server are
4717 closed so that no data can be forwarded from the server to the client.
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02004718
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004719http-response replace-header <name> <regex-match> <replace-fmt>
4720 [ { if | unless } <condition> ]
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02004721
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004722 This matches the regular expression in all occurrences of header field <name>
4723 according to <match-regex>, and replaces them with the <replace-fmt> argument.
4724 Format characters are allowed in replace-fmt and work like in <fmt> arguments
4725 in "add-header". The match is only case-sensitive. It is important to
4726 understand that this action only considers whole header lines, regardless of
4727 the number of values they may contain. This usage is suited to headers
4728 naturally containing commas in their value, such as Set-Cookie, Expires and
4729 so on.
William Lallemand86d0df02017-11-24 21:36:45 +01004730
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004731 Example:
4732 http-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2
Willy Tarreau51d861a2015-05-22 17:30:48 +02004733
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004734 # applied to:
4735 Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004736
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004737 # outputs:
4738 Set-Cookie: C=1;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004739
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004740 # assuming the backend IP is 192.168.1.20.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004741
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004742http-response replace-value <name> <regex-match> <replace-fmt>
4743 [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004744
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004745 This works like "replace-header" except that it matches the regex against
4746 every comma-delimited value of the header field <name> instead of the entire
4747 header. This is suited for all headers which are allowed to carry more than
4748 one value. An example could be the Accept header.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004749
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004750 Example:
4751 http-response replace-value Cache-control ^public$ private
Christopher Faulet85d79c92016-11-09 16:54:56 +01004752
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004753 # applied to:
4754 Cache-Control: max-age=3600, public
Christopher Faulet85d79c92016-11-09 16:54:56 +01004755
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004756 # outputs:
4757 Cache-Control: max-age=3600, private
Christopher Faulet85d79c92016-11-09 16:54:56 +01004758
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004759http-response sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
4760http-response sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Ruoshan Huange4edc6b2016-07-14 15:07:45 +08004761
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004762 This action increments the GPC0 or GPC1 counter according with the sticky
4763 counter designated by <sc-id>. If an error occurs, this action silently fails
4764 and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +02004765
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004766http-response sc-set-gpt0(<sc-id>) <int> [ { if | unless } <condition> ]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02004767
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004768 This action sets the GPT0 tag according to the sticky counter designated by
4769 <sc-id> and the value of <int>. The expected result is a boolean. If an error
4770 occurs, this action silently fails and the actions evaluation continues.
Frédéric Lécaille6778b272018-01-29 15:22:53 +01004771
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004772http-response send-spoe-group [ { if | unless } <condition> ]
Willy Tarreau2d392c22015-08-24 01:43:45 +02004773
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004774 This action is used to trigger sending of a group of SPOE messages. To do so,
4775 the SPOE engine used to send messages must be defined, as well as the SPOE
4776 group to send. Of course, the SPOE engine must refer to an existing SPOE
4777 filter. If not engine name is provided on the SPOE filter line, the SPOE
4778 agent name must be used.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004779
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004780 Arguments:
4781 <engine-name> The SPOE engine name.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004782
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004783 <group-name> The SPOE group name as specified in the engine
4784 configuration.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004785
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004786http-response set-header <name> <fmt> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004787
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004788 This does the same as "add-header" except that the header name is first
4789 removed if it existed. This is useful when passing security information to
4790 the server, where the header must not be manipulated by external users.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004791
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004792http-response set-log-level <level> [ { if | unless } <condition> ]
4793
4794 This is used to change the log level of the current request when a certain
4795 condition is met. Valid levels are the 8 syslog levels (see the "log"
4796 keyword) plus the special level "silent" which disables logging for this
4797 request. This rule is not final so the last matching rule wins. This rule can
4798 be useful to disable health checks coming from another equipment.
4799
4800http-response set-map(<file-name>) <key fmt> <value fmt>
4801
4802 This is used to add a new entry into a MAP. The MAP must be loaded from a
4803 file (even a dummy empty file). The file name of the MAP to be updated is
4804 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
4805 log-format rules, used to collect MAP key, and <value fmt>, which follows
4806 log-format rules, used to collect content for the new entry. It performs a
4807 lookup in the MAP before insertion, to avoid duplicated (or more) values.
4808 This lookup is done by a linear search and can be expensive with large lists!
4809 It is the equivalent of the "set map" command from the stats socket, but can
4810 be triggered by an HTTP response.
4811
4812http-response set-mark <mark> [ { if | unless } <condition> ]
4813
4814 This is used to set the Netfilter MARK on all packets sent to the client to
4815 the value passed in <mark> on platforms which support it. This value is an
4816 unsigned 32 bit value which can be matched by netfilter and by the routing
4817 table. It can be expressed both in decimal or hexadecimal format (prefixed
4818 by "0x"). This can be useful to force certain packets to take a different
4819 route (for example a cheaper network path for bulk downloads). This works on
4820 Linux kernels 2.6.32 and above and requires admin privileges.
4821
4822http-response set-nice <nice> [ { if | unless } <condition> ]
4823
4824 This sets the "nice" factor of the current request being processed.
4825 It only has effect against the other requests being processed at the same
4826 time. The default value is 0, unless altered by the "nice" setting on the
4827 "bind" line. The accepted range is -1024..1024. The higher the value, the
4828 nicest the request will be. Lower values will make the request more important
4829 than other ones. This can be useful to improve the speed of some requests, or
4830 lower the priority of non-important requests. Using this setting without
4831 prior experimentation can cause some major slowdown.
4832
4833http-response set-status <status> [reason <str>]
4834 [ { if | unless } <condition> ]
4835
4836 This replaces the response status code with <status> which must be an integer
4837 between 100 and 999. Optionally, a custom reason text can be provided defined
4838 by <str>, or the default reason for the specified code will be used as a
4839 fallback.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08004840
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004841 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004842 # return "431 Request Header Fields Too Large"
4843 http-response set-status 431
4844 # return "503 Slow Down", custom reason
4845 http-response set-status 503 reason "Slow Down".
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004846
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004847http-response set-tos <tos> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004848
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004849 This is used to set the TOS or DSCP field value of packets sent to the client
4850 to the value passed in <tos> on platforms which support this.
4851 This value represents the whole 8 bits of the IP TOS field, and can be
4852 expressed both in decimal or hexadecimal format (prefixed by "0x"). Note that
4853 only the 6 higher bits are used in DSCP or TOS, and the two lower bits are
4854 always 0. This can be used to adjust some routing behavior on border routers
4855 based on some information from the request.
4856
4857 See RFC 2474, 2597, 3260 and 4594 for more information.
4858
4859http-response set-var(<var-name>) <expr> [ { if | unless } <condition> ]
4860
4861 This is used to set the contents of a variable. The variable is declared
4862 inline.
4863
4864 Arguments:
4865 <var-name> The name of the variable starts with an indication about its
4866 scope. The scopes allowed are:
4867 "proc" : the variable is shared with the whole process
4868 "sess" : the variable is shared with the whole session
4869 "txn" : the variable is shared with the transaction
4870 (request and response)
4871 "req" : the variable is shared only during request
4872 processing
4873 "res" : the variable is shared only during response
4874 processing
4875 This prefix is followed by a name. The separator is a '.'.
4876 The name may only contain characters 'a-z', 'A-Z', '0-9', '.'
4877 and '_'.
4878
4879 <expr> Is a standard HAProxy expression formed by a sample-fetch
4880 followed by some converters.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004881
4882 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004883 http-response set-var(sess.last_redir) res.hdr(location)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004884
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004885http-response silent-drop [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004886
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004887 This stops the evaluation of the rules and makes the client-facing connection
4888 suddenly disappear using a system-dependent way that tries to prevent the
4889 client from being notified. The effect it then that the client still sees an
4890 established connection while there's none on HAProxy. The purpose is to
4891 achieve a comparable effect to "tarpit" except that it doesn't use any local
4892 resource at all on the machine running HAProxy. It can resist much higher
4893 loads than "tarpit", and slow down stronger attackers. It is important to
4894 understand the impact of using this mechanism. All stateful equipment placed
4895 between the client and HAProxy (firewalls, proxies, load balancers) will also
4896 keep the established connection for a long time and may suffer from this
4897 action.
4898 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
4899 option is used to block the emission of a TCP reset. On other systems, the
4900 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
4901 router, though it's still delivered to local networks. Do not use it unless
4902 you fully understand how it works.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004903
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004904http-response track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
4905http-response track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
4906http-response track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004907
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004908 This enables tracking of sticky counters from current response. Please refer
4909 to "http-request track-sc" for a complete description. The only difference
4910 from "http-request track-sc" is the <key> sample expression can only make use
4911 of samples in response (e.g. res.*, status etc.) and samples below Layer 6
4912 (e.g. SSL-related samples, see section 7.3.4). If the sample is not
4913 supported, haproxy will fail and warn while parsing the config.
4914
4915http-response unset-var(<var-name>) [ { if | unless } <condition> ]
4916
4917 This is used to unset a variable. See "http-response set-var" for details
4918 about <var-name>.
4919
4920 Example:
4921 http-response unset-var(sess.last_redir)
4922
Baptiste Assmann5ecb77f2013-10-06 23:24:13 +02004923
Willy Tarreau30631952015-08-06 15:05:24 +02004924http-reuse { never | safe | aggressive | always }
4925 Declare how idle HTTP connections may be shared between requests
4926
4927 May be used in sections: defaults | frontend | listen | backend
4928 yes | no | yes | yes
4929
4930 By default, a connection established between haproxy and the backend server
Olivier Houchard86006a52018-12-14 19:37:49 +01004931 which is considered safe for reuse is moved back to the server's idle
4932 connections pool so that any other request can make use of it. This is the
4933 "safe" strategy below.
Willy Tarreau30631952015-08-06 15:05:24 +02004934
4935 The argument indicates the desired connection reuse strategy :
4936
Olivier Houchard86006a52018-12-14 19:37:49 +01004937 - "never" : idle connections are never shared between sessions. This mode
4938 may be enforced to cancel a different strategy inherited from
4939 a defaults section or for troubleshooting. For example, if an
4940 old bogus application considers that multiple requests over
4941 the same connection come from the same client and it is not
4942 possible to fix the application, it may be desirable to
4943 disable connection sharing in a single backend. An example of
4944 such an application could be an old haproxy using cookie
4945 insertion in tunnel mode and not checking any request past the
4946 first one.
Willy Tarreau30631952015-08-06 15:05:24 +02004947
Olivier Houchard86006a52018-12-14 19:37:49 +01004948 - "safe" : this is the default and the recommended strategy. The first
4949 request of a session is always sent over its own connection,
4950 and only subsequent requests may be dispatched over other
4951 existing connections. This ensures that in case the server
4952 closes the connection when the request is being sent, the
4953 browser can decide to silently retry it. Since it is exactly
4954 equivalent to regular keep-alive, there should be no side
4955 effects.
Willy Tarreau30631952015-08-06 15:05:24 +02004956
4957 - "aggressive" : this mode may be useful in webservices environments where
4958 all servers are not necessarily known and where it would be
4959 appreciable to deliver most first requests over existing
4960 connections. In this case, first requests are only delivered
4961 over existing connections that have been reused at least once,
4962 proving that the server correctly supports connection reuse.
4963 It should only be used when it's sure that the client can
4964 retry a failed request once in a while and where the benefit
4965 of aggressive connection reuse significantly outweights the
4966 downsides of rare connection failures.
4967
4968 - "always" : this mode is only recommended when the path to the server is
4969 known for never breaking existing connections quickly after
4970 releasing them. It allows the first request of a session to be
4971 sent to an existing connection. This can provide a significant
4972 performance increase over the "safe" strategy when the backend
4973 is a cache farm, since such components tend to show a
Davor Ocelice9ed2812017-12-25 17:49:28 +01004974 consistent behavior and will benefit from the connection
Willy Tarreau30631952015-08-06 15:05:24 +02004975 sharing. It is recommended that the "http-keep-alive" timeout
4976 remains low in this mode so that no dead connections remain
4977 usable. In most cases, this will lead to the same performance
4978 gains as "aggressive" but with more risks. It should only be
4979 used when it improves the situation over "aggressive".
4980
4981 When http connection sharing is enabled, a great care is taken to respect the
Davor Ocelice9ed2812017-12-25 17:49:28 +01004982 connection properties and compatibility. Specifically :
4983 - connections made with "usesrc" followed by a client-dependent value
4984 ("client", "clientip", "hdr_ip") are marked private and never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02004985
4986 - connections sent to a server with a TLS SNI extension are marked private
Davor Ocelice9ed2812017-12-25 17:49:28 +01004987 and are never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02004988
Lukas Tribusfd9b68c2018-10-27 20:06:59 +02004989 - connections with certain bogus authentication schemes (relying on the
4990 connection) like NTLM are detected, marked private and are never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02004991
4992 No connection pool is involved, once a session dies, the last idle connection
4993 it was attached to is deleted at the same time. This ensures that connections
4994 may not last after all sessions are closed.
4995
4996 Note: connection reuse improves the accuracy of the "server maxconn" setting,
4997 because almost no new connection will be established while idle connections
4998 remain available. This is particularly true with the "always" strategy.
4999
5000 See also : "option http-keep-alive", "server maxconn"
5001
5002
Mark Lamourinec2247f02012-01-04 13:02:01 -05005003http-send-name-header [<header>]
5004 Add the server name to a request. Use the header string given by <header>
5005
5006 May be used in sections: defaults | frontend | listen | backend
5007 yes | no | yes | yes
5008
5009 Arguments :
5010
5011 <header> The header string to use to send the server name
5012
5013 The "http-send-name-header" statement causes the name of the target
Davor Ocelice9ed2812017-12-25 17:49:28 +01005014 server to be added to the headers of an HTTP request. The name
Mark Lamourinec2247f02012-01-04 13:02:01 -05005015 is added with the header string proved.
5016
5017 See also : "server"
5018
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01005019id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +02005020 Set a persistent ID to a proxy.
5021 May be used in sections : defaults | frontend | listen | backend
5022 no | yes | yes | yes
5023 Arguments : none
5024
5025 Set a persistent ID for the proxy. This ID must be unique and positive.
5026 An unused ID will automatically be assigned if unset. The first assigned
5027 value will be 1. This ID is currently only returned in statistics.
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01005028
5029
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005030ignore-persist { if | unless } <condition>
5031 Declare a condition to ignore persistence
5032 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01005033 no | no | yes | yes
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005034
5035 By default, when cookie persistence is enabled, every requests containing
5036 the cookie are unconditionally persistent (assuming the target server is up
5037 and running).
5038
5039 The "ignore-persist" statement allows one to declare various ACL-based
5040 conditions which, when met, will cause a request to ignore persistence.
5041 This is sometimes useful to load balance requests for static files, which
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005042 often don't require persistence. This can also be used to fully disable
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005043 persistence for a specific User-Agent (for example, some web crawler bots).
5044
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005045 The persistence is ignored when an "if" condition is met, or unless an
5046 "unless" condition is met.
5047
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03005048 Example:
5049 acl url_static path_beg /static /images /img /css
5050 acl url_static path_end .gif .png .jpg .css .js
5051 ignore-persist if url_static
5052
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005053 See also : "force-persist", "cookie", and section 7 about ACL usage.
5054
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005055load-server-state-from-file { global | local | none }
5056 Allow seamless reload of HAProxy
5057 May be used in sections: defaults | frontend | listen | backend
5058 yes | no | yes | yes
5059
5060 This directive points HAProxy to a file where server state from previous
5061 running process has been saved. That way, when starting up, before handling
5062 traffic, the new process can apply old states to servers exactly has if no
Davor Ocelice9ed2812017-12-25 17:49:28 +01005063 reload occurred. The purpose of the "load-server-state-from-file" directive is
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005064 to tell haproxy which file to use. For now, only 2 arguments to either prevent
5065 loading state or load states from a file containing all backends and servers.
5066 The state file can be generated by running the command "show servers state"
5067 over the stats socket and redirect output.
5068
Davor Ocelice9ed2812017-12-25 17:49:28 +01005069 The format of the file is versioned and is very specific. To understand it,
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005070 please read the documentation of the "show servers state" command (chapter
Willy Tarreau1af20c72017-06-23 16:01:14 +02005071 9.3 of Management Guide).
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005072
5073 Arguments:
5074 global load the content of the file pointed by the global directive
5075 named "server-state-file".
5076
5077 local load the content of the file pointed by the directive
5078 "server-state-file-name" if set. If not set, then the backend
5079 name is used as a file name.
5080
5081 none don't load any stat for this backend
5082
5083 Notes:
Willy Tarreaue5a60682016-11-09 14:54:53 +01005084 - server's IP address is preserved across reloads by default, but the
5085 order can be changed thanks to the server's "init-addr" setting. This
5086 means that an IP address change performed on the CLI at run time will
Davor Ocelice9ed2812017-12-25 17:49:28 +01005087 be preserved, and that any change to the local resolver (e.g. /etc/hosts)
Willy Tarreaue5a60682016-11-09 14:54:53 +01005088 will possibly not have any effect if the state file is in use.
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005089
5090 - server's weight is applied from previous running process unless it has
5091 has changed between previous and new configuration files.
5092
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005093 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005094
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005095 global
5096 stats socket /tmp/socket
5097 server-state-file /tmp/server_state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005098
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005099 defaults
5100 load-server-state-from-file global
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005101
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005102 backend bk
5103 server s1 127.0.0.1:22 check weight 11
5104 server s2 127.0.0.1:22 check weight 12
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005105
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005106
5107 Then one can run :
5108
5109 socat /tmp/socket - <<< "show servers state" > /tmp/server_state
5110
5111 Content of the file /tmp/server_state would be like this:
5112
5113 1
5114 # <field names skipped for the doc example>
5115 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
5116 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
5117
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005118 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005119
5120 global
5121 stats socket /tmp/socket
5122 server-state-base /etc/haproxy/states
5123
5124 defaults
5125 load-server-state-from-file local
5126
5127 backend bk
5128 server s1 127.0.0.1:22 check weight 11
5129 server s2 127.0.0.1:22 check weight 12
5130
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005131
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005132 Then one can run :
5133
5134 socat /tmp/socket - <<< "show servers state bk" > /etc/haproxy/states/bk
5135
5136 Content of the file /etc/haproxy/states/bk would be like this:
5137
5138 1
5139 # <field names skipped for the doc example>
5140 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
5141 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
5142
5143 See also: "server-state-file", "server-state-file-name", and
5144 "show servers state"
5145
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005146
Willy Tarreau2769aa02007-12-27 18:26:09 +01005147log global
Willy Tarreauadb345d2018-11-12 07:56:13 +01005148log <address> [len <length>] [format <format>] <facility> [<level> [<minlevel>]]
William Lallemand0f99e342011-10-12 17:50:54 +02005149no log
Willy Tarreau2769aa02007-12-27 18:26:09 +01005150 Enable per-instance logging of events and traffic.
5151 May be used in sections : defaults | frontend | listen | backend
5152 yes | yes | yes | yes
William Lallemand0f99e342011-10-12 17:50:54 +02005153
5154 Prefix :
5155 no should be used when the logger list must be flushed. For example,
5156 if you don't want to inherit from the default logger list. This
5157 prefix does not allow arguments.
5158
Willy Tarreau2769aa02007-12-27 18:26:09 +01005159 Arguments :
5160 global should be used when the instance's logging parameters are the
5161 same as the global ones. This is the most common usage. "global"
5162 replaces <address>, <facility> and <level> with those of the log
5163 entries found in the "global" section. Only one "log global"
5164 statement may be used per instance, and this form takes no other
5165 parameter.
5166
5167 <address> indicates where to send the logs. It takes the same format as
5168 for the "global" section's logs, and can be one of :
5169
5170 - An IPv4 address optionally followed by a colon (':') and a UDP
5171 port. If no port is specified, 514 is used by default (the
5172 standard syslog port).
5173
David du Colombier24bb5f52011-03-17 10:40:23 +01005174 - An IPv6 address followed by a colon (':') and optionally a UDP
5175 port. If no port is specified, 514 is used by default (the
5176 standard syslog port).
5177
Willy Tarreau2769aa02007-12-27 18:26:09 +01005178 - A filesystem path to a UNIX domain socket, keeping in mind
5179 considerations for chroot (be sure the path is accessible
5180 inside the chroot) and uid/gid (be sure the path is
Davor Ocelice9ed2812017-12-25 17:49:28 +01005181 appropriately writable).
Willy Tarreau2769aa02007-12-27 18:26:09 +01005182
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01005183 - A file descriptor number in the form "fd@<number>", which may
5184 point to a pipe, terminal, or socket. In this case unbuffered
5185 logs are used and one writev() call per log is performed. This
5186 is a bit expensive but acceptable for most workloads. Messages
5187 sent this way will not be truncated but may be dropped, in
5188 which case the DroppedLogs counter will be incremented. The
5189 writev() call is atomic even on pipes for messages up to
5190 PIPE_BUF size, which POSIX recommends to be at least 512 and
5191 which is 4096 bytes on most modern operating systems. Any
5192 larger message may be interleaved with messages from other
5193 processes. Exceptionally for debugging purposes the file
5194 descriptor may also be directed to a file, but doing so will
5195 significantly slow haproxy down as non-blocking calls will be
5196 ignored. Also there will be no way to purge nor rotate this
5197 file without restarting the process. Note that the configured
5198 syslog format is preserved, so the output is suitable for use
Willy Tarreauc1b06452018-11-12 11:57:56 +01005199 with a TCP syslog server. See also the "short" and "raw"
5200 formats below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01005201
5202 - "stdout" / "stderr", which are respectively aliases for "fd@1"
5203 and "fd@2", see above.
5204
5205 You may want to reference some environment variables in the
5206 address parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01005207
Willy Tarreau18324f52014-06-27 18:10:07 +02005208 <length> is an optional maximum line length. Log lines larger than this
5209 value will be truncated before being sent. The reason is that
5210 syslog servers act differently on log line length. All servers
5211 support the default value of 1024, but some servers simply drop
5212 larger lines while others do log them. If a server supports long
5213 lines, it may make sense to set this value here in order to avoid
5214 truncating long lines. Similarly, if a server drops long lines,
5215 it is preferable to truncate them before sending them. Accepted
5216 values are 80 to 65535 inclusive. The default value of 1024 is
5217 generally fine for all standard usages. Some specific cases of
Davor Ocelice9ed2812017-12-25 17:49:28 +01005218 long captures or JSON-formatted logs may require larger values.
Willy Tarreau18324f52014-06-27 18:10:07 +02005219
Willy Tarreauadb345d2018-11-12 07:56:13 +01005220 <format> is the log format used when generating syslog messages. It may be
5221 one of the following :
5222
5223 rfc3164 The RFC3164 syslog message format. This is the default.
5224 (https://tools.ietf.org/html/rfc3164)
5225
5226 rfc5424 The RFC5424 syslog message format.
5227 (https://tools.ietf.org/html/rfc5424)
5228
Willy Tarreaue8746a02018-11-12 08:45:00 +01005229 short A message containing only a level between angle brackets such as
5230 '<3>', followed by the text. The PID, date, time, process name
5231 and system name are omitted. This is designed to be used with a
5232 local log server. This format is compatible with what the
5233 systemd logger consumes.
5234
Willy Tarreauc1b06452018-11-12 11:57:56 +01005235 raw A message containing only the text. The level, PID, date, time,
5236 process name and system name are omitted. This is designed to
5237 be used in containers or during development, where the severity
5238 only depends on the file descriptor used (stdout/stderr).
5239
Willy Tarreau2769aa02007-12-27 18:26:09 +01005240 <facility> must be one of the 24 standard syslog facilities :
5241
Willy Tarreaue8746a02018-11-12 08:45:00 +01005242 kern user mail daemon auth syslog lpr news
5243 uucp cron auth2 ftp ntp audit alert cron2
5244 local0 local1 local2 local3 local4 local5 local6 local7
5245
Willy Tarreauc1b06452018-11-12 11:57:56 +01005246 Note that the facility is ignored for the "short" and "raw"
5247 formats, but still required as a positional field. It is
5248 recommended to use "daemon" in this case to make it clear that
5249 it's only supposed to be used locally.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005250
5251 <level> is optional and can be specified to filter outgoing messages. By
5252 default, all messages are sent. If a level is specified, only
5253 messages with a severity at least as important as this level
Willy Tarreauf7edefa2009-05-10 17:20:05 +02005254 will be sent. An optional minimum level can be specified. If it
5255 is set, logs emitted with a more severe level than this one will
5256 be capped to this level. This is used to avoid sending "emerg"
5257 messages on all terminals on some default syslog configurations.
5258 Eight levels are known :
Willy Tarreau2769aa02007-12-27 18:26:09 +01005259
5260 emerg alert crit err warning notice info debug
5261
William Lallemand0f99e342011-10-12 17:50:54 +02005262 It is important to keep in mind that it is the frontend which decides what to
5263 log from a connection, and that in case of content switching, the log entries
5264 from the backend will be ignored. Connections are logged at level "info".
Willy Tarreaucc6c8912009-02-22 10:53:55 +01005265
5266 However, backend log declaration define how and where servers status changes
5267 will be logged. Level "notice" will be used to indicate a server going up,
5268 "warning" will be used for termination signals and definitive service
5269 termination, and "alert" will be used for when a server goes down.
5270
5271 Note : According to RFC3164, messages are truncated to 1024 bytes before
5272 being emitted.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005273
5274 Example :
5275 log global
Willy Tarreauc1b06452018-11-12 11:57:56 +01005276 log stdout format short daemon # send log to systemd
5277 log stdout format raw daemon # send everything to stdout
5278 log stderr format raw daemon notice # send important events to stderr
Willy Tarreauf7edefa2009-05-10 17:20:05 +02005279 log 127.0.0.1:514 local0 notice # only send important events
5280 log 127.0.0.1:514 local0 notice notice # same but limit output level
William Lallemandb2f07452015-05-12 14:27:13 +02005281 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
Willy Tarreaudad36a32013-03-11 01:20:04 +01005282
Willy Tarreau2769aa02007-12-27 18:26:09 +01005283
William Lallemand48940402012-01-30 16:47:22 +01005284log-format <string>
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01005285 Specifies the log format string to use for traffic logs
5286 May be used in sections: defaults | frontend | listen | backend
5287 yes | yes | yes | no
William Lallemand48940402012-01-30 16:47:22 +01005288
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01005289 This directive specifies the log format string that will be used for all logs
5290 resulting from traffic passing through the frontend using this line. If the
5291 directive is used in a defaults section, all subsequent frontends will use
5292 the same log format. Please see section 8.2.4 which covers the log format
5293 string in depth.
William Lallemand48940402012-01-30 16:47:22 +01005294
Guillaume de Lafond29f45602017-03-31 19:52:15 +02005295 "log-format" directive overrides previous "option tcplog", "log-format" and
5296 "option httplog" directives.
5297
Dragan Dosen7ad31542015-09-28 17:16:47 +02005298log-format-sd <string>
5299 Specifies the RFC5424 structured-data log format string
5300 May be used in sections: defaults | frontend | listen | backend
5301 yes | yes | yes | no
5302
5303 This directive specifies the RFC5424 structured-data log format string that
5304 will be used for all logs resulting from traffic passing through the frontend
5305 using this line. If the directive is used in a defaults section, all
5306 subsequent frontends will use the same log format. Please see section 8.2.4
5307 which covers the log format string in depth.
5308
5309 See https://tools.ietf.org/html/rfc5424#section-6.3 for more information
5310 about the RFC5424 structured-data part.
5311
5312 Note : This log format string will be used only for loggers that have set
5313 log format to "rfc5424".
5314
5315 Example :
5316 log-format-sd [exampleSDID@1234\ bytes=\"%B\"\ status=\"%ST\"]
5317
5318
Willy Tarreau094af4e2015-01-07 15:03:42 +01005319log-tag <string>
5320 Specifies the log tag to use for all outgoing logs
5321 May be used in sections: defaults | frontend | listen | backend
5322 yes | yes | yes | yes
5323
5324 Sets the tag field in the syslog header to this string. It defaults to the
5325 log-tag set in the global section, otherwise the program name as launched
5326 from the command line, which usually is "haproxy". Sometimes it can be useful
5327 to differentiate between multiple processes running on the same host, or to
5328 differentiate customer instances running in the same process. In the backend,
5329 logs about servers up/down will use this tag. As a hint, it can be convenient
5330 to set a log-tag related to a hosted customer in a defaults section then put
5331 all the frontends and backends for that customer, then start another customer
5332 in a new defaults section. See also the global "log-tag" directive.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005333
Willy Tarreauc35362a2014-04-25 13:58:37 +02005334max-keep-alive-queue <value>
5335 Set the maximum server queue size for maintaining keep-alive connections
5336 May be used in sections: defaults | frontend | listen | backend
5337 yes | no | yes | yes
5338
5339 HTTP keep-alive tries to reuse the same server connection whenever possible,
5340 but sometimes it can be counter-productive, for example if a server has a lot
5341 of connections while other ones are idle. This is especially true for static
5342 servers.
5343
5344 The purpose of this setting is to set a threshold on the number of queued
5345 connections at which haproxy stops trying to reuse the same server and prefers
5346 to find another one. The default value, -1, means there is no limit. A value
5347 of zero means that keep-alive requests will never be queued. For very close
5348 servers which can be reached with a low latency and which are not sensible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01005349 breaking keep-alive, a low value is recommended (e.g. local static server can
Willy Tarreauc35362a2014-04-25 13:58:37 +02005350 use a value of 10 or less). For remote servers suffering from a high latency,
5351 higher values might be needed to cover for the latency and/or the cost of
5352 picking a different server.
5353
5354 Note that this has no impact on responses which are maintained to the same
5355 server consecutively to a 401 response. They will still go to the same server
5356 even if they have to be queued.
5357
5358 See also : "option http-server-close", "option prefer-last-server", server
5359 "maxconn" and cookie persistence.
5360
Olivier Houcharda4d4fdf2018-12-14 19:27:06 +01005361max-session-srv-conns <nb>
5362 Set the maximum number of outgoing connections we can keep idling for a given
5363 client session. The default is 5 (it precisely equals MAX_SRV_LIST which is
5364 defined at build time).
Willy Tarreauc35362a2014-04-25 13:58:37 +02005365
Willy Tarreau2769aa02007-12-27 18:26:09 +01005366maxconn <conns>
5367 Fix the maximum number of concurrent connections on a frontend
5368 May be used in sections : defaults | frontend | listen | backend
5369 yes | yes | yes | no
5370 Arguments :
5371 <conns> is the maximum number of concurrent connections the frontend will
5372 accept to serve. Excess connections will be queued by the system
5373 in the socket's listen queue and will be served once a connection
5374 closes.
5375
5376 If the system supports it, it can be useful on big sites to raise this limit
5377 very high so that haproxy manages connection queues, instead of leaving the
5378 clients with unanswered connection attempts. This value should not exceed the
5379 global maxconn. Also, keep in mind that a connection contains two buffers
Baptiste Assmann79fb45d2016-03-06 23:34:31 +01005380 of tune.bufsize (16kB by default) each, as well as some other data resulting
5381 in about 33 kB of RAM being consumed per established connection. That means
5382 that a medium system equipped with 1GB of RAM can withstand around
5383 20000-25000 concurrent connections if properly tuned.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005384
5385 Also, when <conns> is set to large values, it is possible that the servers
5386 are not sized to accept such loads, and for this reason it is generally wise
5387 to assign them some reasonable connection limits.
5388
Vincent Bernat6341be52012-06-27 17:18:30 +02005389 By default, this value is set to 2000.
5390
Willy Tarreau2769aa02007-12-27 18:26:09 +01005391 See also : "server", global section's "maxconn", "fullconn"
5392
5393
5394mode { tcp|http|health }
5395 Set the running mode or protocol of the instance
5396 May be used in sections : defaults | frontend | listen | backend
5397 yes | yes | yes | yes
5398 Arguments :
5399 tcp The instance will work in pure TCP mode. A full-duplex connection
5400 will be established between clients and servers, and no layer 7
5401 examination will be performed. This is the default mode. It
5402 should be used for SSL, SSH, SMTP, ...
5403
5404 http The instance will work in HTTP mode. The client request will be
5405 analyzed in depth before connecting to any server. Any request
5406 which is not RFC-compliant will be rejected. Layer 7 filtering,
5407 processing and switching will be possible. This is the mode which
5408 brings HAProxy most of its value.
5409
5410 health The instance will work in "health" mode. It will just reply "OK"
Willy Tarreau82569f92012-09-27 23:48:56 +02005411 to incoming connections and close the connection. Alternatively,
5412 If the "httpchk" option is set, "HTTP/1.0 200 OK" will be sent
5413 instead. Nothing will be logged in either case. This mode is used
5414 to reply to external components health checks. This mode is
5415 deprecated and should not be used anymore as it is possible to do
5416 the same and even better by combining TCP or HTTP modes with the
5417 "monitor" keyword.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005418
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005419 When doing content switching, it is mandatory that the frontend and the
5420 backend are in the same mode (generally HTTP), otherwise the configuration
5421 will be refused.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005422
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005423 Example :
Willy Tarreau2769aa02007-12-27 18:26:09 +01005424 defaults http_instances
5425 mode http
5426
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005427 See also : "monitor", "monitor-net"
Willy Tarreau2769aa02007-12-27 18:26:09 +01005428
Willy Tarreau0ba27502007-12-24 16:55:16 +01005429
Cyril Bontéf0c60612010-02-06 14:44:47 +01005430monitor fail { if | unless } <condition>
Willy Tarreau2769aa02007-12-27 18:26:09 +01005431 Add a condition to report a failure to a monitor HTTP request.
Willy Tarreau0ba27502007-12-24 16:55:16 +01005432 May be used in sections : defaults | frontend | listen | backend
5433 no | yes | yes | no
Willy Tarreau0ba27502007-12-24 16:55:16 +01005434 Arguments :
5435 if <cond> the monitor request will fail if the condition is satisfied,
5436 and will succeed otherwise. The condition should describe a
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005437 combined test which must induce a failure if all conditions
Willy Tarreau0ba27502007-12-24 16:55:16 +01005438 are met, for instance a low number of servers both in a
5439 backend and its backup.
5440
5441 unless <cond> the monitor request will succeed only if the condition is
5442 satisfied, and will fail otherwise. Such a condition may be
5443 based on a test on the presence of a minimum number of active
5444 servers in a list of backends.
5445
5446 This statement adds a condition which can force the response to a monitor
5447 request to report a failure. By default, when an external component queries
5448 the URI dedicated to monitoring, a 200 response is returned. When one of the
5449 conditions above is met, haproxy will return 503 instead of 200. This is
5450 very useful to report a site failure to an external component which may base
5451 routing advertisements between multiple sites on the availability reported by
5452 haproxy. In this case, one would rely on an ACL involving the "nbsrv"
Willy Tarreauae94d4d2011-05-11 16:28:49 +02005453 criterion. Note that "monitor fail" only works in HTTP mode. Both status
5454 messages may be tweaked using "errorfile" or "errorloc" if needed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01005455
5456 Example:
5457 frontend www
Willy Tarreau2769aa02007-12-27 18:26:09 +01005458 mode http
Willy Tarreau0ba27502007-12-24 16:55:16 +01005459 acl site_dead nbsrv(dynamic) lt 2
5460 acl site_dead nbsrv(static) lt 2
5461 monitor-uri /site_alive
5462 monitor fail if site_dead
5463
Willy Tarreauae94d4d2011-05-11 16:28:49 +02005464 See also : "monitor-net", "monitor-uri", "errorfile", "errorloc"
Willy Tarreau2769aa02007-12-27 18:26:09 +01005465
5466
5467monitor-net <source>
5468 Declare a source network which is limited to monitor requests
5469 May be used in sections : defaults | frontend | listen | backend
5470 yes | yes | yes | no
5471 Arguments :
5472 <source> is the source IPv4 address or network which will only be able to
5473 get monitor responses to any request. It can be either an IPv4
5474 address, a host name, or an address followed by a slash ('/')
5475 followed by a mask.
5476
5477 In TCP mode, any connection coming from a source matching <source> will cause
5478 the connection to be immediately closed without any log. This allows another
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005479 equipment to probe the port and verify that it is still listening, without
Willy Tarreau2769aa02007-12-27 18:26:09 +01005480 forwarding the connection to a remote server.
5481
5482 In HTTP mode, a connection coming from a source matching <source> will be
5483 accepted, the following response will be sent without waiting for a request,
5484 then the connection will be closed : "HTTP/1.0 200 OK". This is normally
5485 enough for any front-end HTTP probe to detect that the service is UP and
Willy Tarreau82569f92012-09-27 23:48:56 +02005486 running without forwarding the request to a backend server. Note that this
5487 response is sent in raw format, without any transformation. This is important
5488 as it means that it will not be SSL-encrypted on SSL listeners.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005489
Willy Tarreau82569f92012-09-27 23:48:56 +02005490 Monitor requests are processed very early, just after tcp-request connection
5491 ACLs which are the only ones able to block them. These connections are short
5492 lived and never wait for any data from the client. They cannot be logged, and
5493 it is the intended purpose. They are only used to report HAProxy's health to
5494 an upper component, nothing more. Please note that "monitor fail" rules do
5495 not apply to connections intercepted by "monitor-net".
Willy Tarreau2769aa02007-12-27 18:26:09 +01005496
Willy Tarreau95cd2832010-03-04 23:36:33 +01005497 Last, please note that only one "monitor-net" statement can be specified in
5498 a frontend. If more than one is found, only the last one will be considered.
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005499
Willy Tarreau2769aa02007-12-27 18:26:09 +01005500 Example :
5501 # addresses .252 and .253 are just probing us.
5502 frontend www
5503 monitor-net 192.168.0.252/31
5504
5505 See also : "monitor fail", "monitor-uri"
5506
5507
5508monitor-uri <uri>
5509 Intercept a URI used by external components' monitor requests
5510 May be used in sections : defaults | frontend | listen | backend
5511 yes | yes | yes | no
5512 Arguments :
5513 <uri> is the exact URI which we want to intercept to return HAProxy's
5514 health status instead of forwarding the request.
5515
5516 When an HTTP request referencing <uri> will be received on a frontend,
5517 HAProxy will not forward it nor log it, but instead will return either
5518 "HTTP/1.0 200 OK" or "HTTP/1.0 503 Service unavailable", depending on failure
5519 conditions defined with "monitor fail". This is normally enough for any
5520 front-end HTTP probe to detect that the service is UP and running without
5521 forwarding the request to a backend server. Note that the HTTP method, the
5522 version and all headers are ignored, but the request must at least be valid
5523 at the HTTP level. This keyword may only be used with an HTTP-mode frontend.
5524
Willy Tarreau721d8e02017-12-01 18:25:08 +01005525 Monitor requests are processed very early, just after the request is parsed
5526 and even before any "http-request" or "block" rulesets. The only rulesets
5527 applied before are the tcp-request ones. They cannot be logged either, and it
5528 is the intended purpose. They are only used to report HAProxy's health to an
5529 upper component, nothing more. However, it is possible to add any number of
5530 conditions using "monitor fail" and ACLs so that the result can be adjusted
5531 to whatever check can be imagined (most often the number of available servers
5532 in a backend).
Willy Tarreau2769aa02007-12-27 18:26:09 +01005533
5534 Example :
5535 # Use /haproxy_test to report haproxy's status
5536 frontend www
5537 mode http
5538 monitor-uri /haproxy_test
5539
5540 See also : "monitor fail", "monitor-net"
5541
Willy Tarreau0ba27502007-12-24 16:55:16 +01005542
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005543option abortonclose
5544no option abortonclose
5545 Enable or disable early dropping of aborted requests pending in queues.
5546 May be used in sections : defaults | frontend | listen | backend
5547 yes | no | yes | yes
5548 Arguments : none
5549
5550 In presence of very high loads, the servers will take some time to respond.
5551 The per-instance connection queue will inflate, and the response time will
5552 increase respective to the size of the queue times the average per-session
5553 response time. When clients will wait for more than a few seconds, they will
Willy Tarreau198a7442008-01-17 12:05:32 +01005554 often hit the "STOP" button on their browser, leaving a useless request in
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005555 the queue, and slowing down other users, and the servers as well, because the
5556 request will eventually be served, then aborted at the first error
5557 encountered while delivering the response.
5558
5559 As there is no way to distinguish between a full STOP and a simple output
5560 close on the client side, HTTP agents should be conservative and consider
5561 that the client might only have closed its output channel while waiting for
5562 the response. However, this introduces risks of congestion when lots of users
5563 do the same, and is completely useless nowadays because probably no client at
5564 all will close the session while waiting for the response. Some HTTP agents
Davor Ocelice9ed2812017-12-25 17:49:28 +01005565 support this behavior (Squid, Apache, HAProxy), and others do not (TUX, most
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005566 hardware-based load balancers). So the probability for a closed input channel
Willy Tarreau198a7442008-01-17 12:05:32 +01005567 to represent a user hitting the "STOP" button is close to 100%, and the risk
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005568 of being the single component to break rare but valid traffic is extremely
5569 low, which adds to the temptation to be able to abort a session early while
5570 still not served and not pollute the servers.
5571
Davor Ocelice9ed2812017-12-25 17:49:28 +01005572 In HAProxy, the user can choose the desired behavior using the option
5573 "abortonclose". By default (without the option) the behavior is HTTP
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005574 compliant and aborted requests will be served. But when the option is
5575 specified, a session with an incoming channel closed will be aborted while
5576 it is still possible, either pending in the queue for a connection slot, or
5577 during the connection establishment if the server has not yet acknowledged
5578 the connection request. This considerably reduces the queue size and the load
5579 on saturated servers when users are tempted to click on STOP, which in turn
Willy Tarreaud72758d2010-01-12 10:42:19 +01005580 reduces the response time for other users.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005581
5582 If this option has been enabled in a "defaults" section, it can be disabled
5583 in a specific instance by prepending the "no" keyword before it.
5584
5585 See also : "timeout queue" and server's "maxconn" and "maxqueue" parameters
5586
5587
Willy Tarreau4076a152009-04-02 15:18:36 +02005588option accept-invalid-http-request
5589no option accept-invalid-http-request
5590 Enable or disable relaxing of HTTP request parsing
5591 May be used in sections : defaults | frontend | listen | backend
5592 yes | yes | yes | no
5593 Arguments : none
5594
Willy Tarreau91852eb2015-05-01 13:26:00 +02005595 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02005596 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01005597 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02005598 forbidden characters are essentially used to build attacks exploiting server
5599 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
5600 server will emit invalid header names for whatever reason (configuration,
5601 implementation) and the issue will not be immediately fixed. In such a case,
5602 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau422246e2012-01-07 23:54:13 +01005603 even if that does not make sense, by specifying this option. Similarly, the
5604 list of characters allowed to appear in a URI is well defined by RFC3986, and
5605 chars 0-31, 32 (space), 34 ('"'), 60 ('<'), 62 ('>'), 92 ('\'), 94 ('^'), 96
5606 ('`'), 123 ('{'), 124 ('|'), 125 ('}'), 127 (delete) and anything above are
Davor Ocelice9ed2812017-12-25 17:49:28 +01005607 not allowed at all. HAProxy always blocks a number of them (0..32, 127). The
Willy Tarreau91852eb2015-05-01 13:26:00 +02005608 remaining ones are blocked by default unless this option is enabled. This
Willy Tarreau13317662015-05-01 13:47:08 +02005609 option also relaxes the test on the HTTP version, it allows HTTP/0.9 requests
5610 to pass through (no version specified) and multiple digits for both the major
5611 and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02005612
5613 This option should never be enabled by default as it hides application bugs
5614 and open security breaches. It should only be deployed after a problem has
5615 been confirmed.
5616
5617 When this option is enabled, erroneous header names will still be accepted in
5618 requests, but the complete request will be captured in order to permit later
Willy Tarreau422246e2012-01-07 23:54:13 +01005619 analysis using the "show errors" request on the UNIX stats socket. Similarly,
5620 requests containing invalid chars in the URI part will be logged. Doing this
Willy Tarreau4076a152009-04-02 15:18:36 +02005621 also helps confirming that the issue has been solved.
5622
5623 If this option has been enabled in a "defaults" section, it can be disabled
5624 in a specific instance by prepending the "no" keyword before it.
5625
5626 See also : "option accept-invalid-http-response" and "show errors" on the
5627 stats socket.
5628
5629
5630option accept-invalid-http-response
5631no option accept-invalid-http-response
5632 Enable or disable relaxing of HTTP response parsing
5633 May be used in sections : defaults | frontend | listen | backend
5634 yes | no | yes | yes
5635 Arguments : none
5636
Willy Tarreau91852eb2015-05-01 13:26:00 +02005637 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02005638 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01005639 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02005640 forbidden characters are essentially used to build attacks exploiting server
5641 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
5642 server will emit invalid header names for whatever reason (configuration,
5643 implementation) and the issue will not be immediately fixed. In such a case,
5644 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau91852eb2015-05-01 13:26:00 +02005645 even if that does not make sense, by specifying this option. This option also
5646 relaxes the test on the HTTP version format, it allows multiple digits for
5647 both the major and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02005648
5649 This option should never be enabled by default as it hides application bugs
5650 and open security breaches. It should only be deployed after a problem has
5651 been confirmed.
5652
5653 When this option is enabled, erroneous header names will still be accepted in
5654 responses, but the complete response will be captured in order to permit
5655 later analysis using the "show errors" request on the UNIX stats socket.
5656 Doing this also helps confirming that the issue has been solved.
5657
5658 If this option has been enabled in a "defaults" section, it can be disabled
5659 in a specific instance by prepending the "no" keyword before it.
5660
5661 See also : "option accept-invalid-http-request" and "show errors" on the
5662 stats socket.
5663
5664
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005665option allbackups
5666no option allbackups
5667 Use either all backup servers at a time or only the first one
5668 May be used in sections : defaults | frontend | listen | backend
5669 yes | no | yes | yes
5670 Arguments : none
5671
5672 By default, the first operational backup server gets all traffic when normal
5673 servers are all down. Sometimes, it may be preferred to use multiple backups
5674 at once, because one will not be enough. When "option allbackups" is enabled,
5675 the load balancing will be performed among all backup servers when all normal
5676 ones are unavailable. The same load balancing algorithm will be used and the
5677 servers' weights will be respected. Thus, there will not be any priority
5678 order between the backup servers anymore.
5679
5680 This option is mostly used with static server farms dedicated to return a
5681 "sorry" page when an application is completely offline.
5682
5683 If this option has been enabled in a "defaults" section, it can be disabled
5684 in a specific instance by prepending the "no" keyword before it.
5685
5686
5687option checkcache
5688no option checkcache
Godbach7056a352013-12-11 20:01:07 +08005689 Analyze all server responses and block responses with cacheable cookies
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005690 May be used in sections : defaults | frontend | listen | backend
5691 yes | no | yes | yes
5692 Arguments : none
5693
5694 Some high-level frameworks set application cookies everywhere and do not
5695 always let enough control to the developer to manage how the responses should
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005696 be cached. When a session cookie is returned on a cacheable object, there is a
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005697 high risk of session crossing or stealing between users traversing the same
5698 caches. In some situations, it is better to block the response than to let
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02005699 some sensitive session information go in the wild.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005700
5701 The option "checkcache" enables deep inspection of all server responses for
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005702 strict compliance with HTTP specification in terms of cacheability. It
Willy Tarreau198a7442008-01-17 12:05:32 +01005703 carefully checks "Cache-control", "Pragma" and "Set-cookie" headers in server
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005704 response to check if there's a risk of caching a cookie on a client-side
5705 proxy. When this option is enabled, the only responses which can be delivered
Willy Tarreau198a7442008-01-17 12:05:32 +01005706 to the client are :
Davor Ocelice9ed2812017-12-25 17:49:28 +01005707 - all those without "Set-Cookie" header;
Willy Tarreauc55ddce2017-12-21 11:41:38 +01005708 - all those with a return code other than 200, 203, 204, 206, 300, 301,
5709 404, 405, 410, 414, 501, provided that the server has not set a
Davor Ocelice9ed2812017-12-25 17:49:28 +01005710 "Cache-control: public" header field;
Willy Tarreau24ea0bc2017-12-21 11:32:55 +01005711 - all those that result from a request using a method other than GET, HEAD,
5712 OPTIONS, TRACE, provided that the server has not set a 'Cache-Control:
Davor Ocelice9ed2812017-12-25 17:49:28 +01005713 public' header field;
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005714 - those with a 'Pragma: no-cache' header
5715 - those with a 'Cache-control: private' header
5716 - those with a 'Cache-control: no-store' header
5717 - those with a 'Cache-control: max-age=0' header
5718 - those with a 'Cache-control: s-maxage=0' header
5719 - those with a 'Cache-control: no-cache' header
5720 - those with a 'Cache-control: no-cache="set-cookie"' header
5721 - those with a 'Cache-control: no-cache="set-cookie,' header
5722 (allowing other fields after set-cookie)
5723
5724 If a response doesn't respect these requirements, then it will be blocked
Willy Tarreau198a7442008-01-17 12:05:32 +01005725 just as if it was from an "rspdeny" filter, with an "HTTP 502 bad gateway".
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005726 The session state shows "PH--" meaning that the proxy blocked the response
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005727 during headers processing. Additionally, an alert will be sent in the logs so
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005728 that admins are informed that there's something to be fixed.
5729
5730 Due to the high impact on the application, the application should be tested
5731 in depth with the option enabled before going to production. It is also a
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01005732 good practice to always activate it during tests, even if it is not used in
Davor Ocelice9ed2812017-12-25 17:49:28 +01005733 production, as it will report potentially dangerous application behaviors.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005734
5735 If this option has been enabled in a "defaults" section, it can be disabled
5736 in a specific instance by prepending the "no" keyword before it.
5737
5738
5739option clitcpka
5740no option clitcpka
5741 Enable or disable the sending of TCP keepalive packets on the client side
5742 May be used in sections : defaults | frontend | listen | backend
5743 yes | yes | yes | no
5744 Arguments : none
5745
5746 When there is a firewall or any session-aware component between a client and
5747 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01005748 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005749 components decides to expire a session which has remained idle for too long.
5750
5751 Enabling socket-level TCP keep-alives makes the system regularly send packets
5752 to the other end of the connection, leaving it active. The delay between
5753 keep-alive probes is controlled by the system only and depends both on the
5754 operating system and its tuning parameters.
5755
5756 It is important to understand that keep-alive packets are neither emitted nor
5757 received at the application level. It is only the network stacks which sees
5758 them. For this reason, even if one side of the proxy already uses keep-alives
5759 to maintain its connection alive, those keep-alive packets will not be
5760 forwarded to the other side of the proxy.
5761
5762 Please note that this has nothing to do with HTTP keep-alive.
5763
5764 Using option "clitcpka" enables the emission of TCP keep-alive probes on the
5765 client side of a connection, which should help when session expirations are
5766 noticed between HAProxy and a client.
5767
5768 If this option has been enabled in a "defaults" section, it can be disabled
5769 in a specific instance by prepending the "no" keyword before it.
5770
5771 See also : "option srvtcpka", "option tcpka"
5772
5773
Willy Tarreau0ba27502007-12-24 16:55:16 +01005774option contstats
5775 Enable continuous traffic statistics updates
5776 May be used in sections : defaults | frontend | listen | backend
5777 yes | yes | yes | no
5778 Arguments : none
5779
5780 By default, counters used for statistics calculation are incremented
5781 only when a session finishes. It works quite well when serving small
5782 objects, but with big ones (for example large images or archives) or
5783 with A/V streaming, a graph generated from haproxy counters looks like
Willy Tarreaudef0d222016-11-08 22:03:00 +01005784 a hedgehog. With this option enabled counters get incremented frequently
5785 along the session, typically every 5 seconds, which is often enough to
5786 produce clean graphs. Recounting touches a hotpath directly so it is not
5787 not enabled by default, as it can cause a lot of wakeups for very large
5788 session counts and cause a small performance drop.
Willy Tarreau0ba27502007-12-24 16:55:16 +01005789
5790
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02005791option dontlog-normal
5792no option dontlog-normal
5793 Enable or disable logging of normal, successful connections
5794 May be used in sections : defaults | frontend | listen | backend
5795 yes | yes | yes | no
5796 Arguments : none
5797
5798 There are large sites dealing with several thousand connections per second
5799 and for which logging is a major pain. Some of them are even forced to turn
5800 logs off and cannot debug production issues. Setting this option ensures that
5801 normal connections, those which experience no error, no timeout, no retry nor
5802 redispatch, will not be logged. This leaves disk space for anomalies. In HTTP
5803 mode, the response status code is checked and return codes 5xx will still be
5804 logged.
5805
5806 It is strongly discouraged to use this option as most of the time, the key to
5807 complex issues is in the normal logs which will not be logged here. If you
5808 need to separate logs, see the "log-separate-errors" option instead.
5809
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005810 See also : "log", "dontlognull", "log-separate-errors" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02005811 logging.
5812
5813
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005814option dontlognull
5815no option dontlognull
5816 Enable or disable logging of null connections
5817 May be used in sections : defaults | frontend | listen | backend
5818 yes | yes | yes | no
5819 Arguments : none
5820
5821 In certain environments, there are components which will regularly connect to
5822 various systems to ensure that they are still alive. It can be the case from
5823 another load balancer as well as from monitoring systems. By default, even a
5824 simple port probe or scan will produce a log. If those connections pollute
5825 the logs too much, it is possible to enable option "dontlognull" to indicate
5826 that a connection on which no data has been transferred will not be logged,
Willy Tarreau0f228a02015-05-01 15:37:53 +02005827 which typically corresponds to those probes. Note that errors will still be
5828 returned to the client and accounted for in the stats. If this is not what is
5829 desired, option http-ignore-probes can be used instead.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005830
5831 It is generally recommended not to use this option in uncontrolled
Davor Ocelice9ed2812017-12-25 17:49:28 +01005832 environments (e.g. internet), otherwise scans and other malicious activities
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005833 would not be logged.
5834
5835 If this option has been enabled in a "defaults" section, it can be disabled
5836 in a specific instance by prepending the "no" keyword before it.
5837
Willy Tarreau0f228a02015-05-01 15:37:53 +02005838 See also : "log", "http-ignore-probes", "monitor-net", "monitor-uri", and
5839 section 8 about logging.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005840
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005841
Christopher Faulet315b39c2018-09-21 16:26:19 +02005842option forceclose (deprecated)
5843no option forceclose (deprecated)
5844 This is an alias for "option httpclose". Thus this option is deprecated.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005845
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02005846 See also : "option httpclose" and "option http-pretend-keepalive"
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005847
5848
Willy Tarreau87cf5142011-08-19 22:57:24 +02005849option forwardfor [ except <network> ] [ header <name> ] [ if-none ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01005850 Enable insertion of the X-Forwarded-For header to requests sent to servers
5851 May be used in sections : defaults | frontend | listen | backend
5852 yes | yes | yes | yes
5853 Arguments :
5854 <network> is an optional argument used to disable this option for sources
5855 matching <network>
Ross Westaf72a1d2008-08-03 10:51:45 +02005856 <name> an optional argument to specify a different "X-Forwarded-For"
Willy Tarreaud72758d2010-01-12 10:42:19 +01005857 header name.
Willy Tarreauc27debf2008-01-06 08:57:02 +01005858
5859 Since HAProxy works in reverse-proxy mode, the servers see its IP address as
5860 their client address. This is sometimes annoying when the client's IP address
5861 is expected in server logs. To solve this problem, the well-known HTTP header
5862 "X-Forwarded-For" may be added by HAProxy to all requests sent to the server.
5863 This header contains a value representing the client's IP address. Since this
5864 header is always appended at the end of the existing header list, the server
5865 must be configured to always use the last occurrence of this header only. See
Ross Westaf72a1d2008-08-03 10:51:45 +02005866 the server's manual to find how to enable use of this standard header. Note
5867 that only the last occurrence of the header must be used, since it is really
5868 possible that the client has already brought one.
5869
Willy Tarreaud72758d2010-01-12 10:42:19 +01005870 The keyword "header" may be used to supply a different header name to replace
Ross Westaf72a1d2008-08-03 10:51:45 +02005871 the default "X-Forwarded-For". This can be useful where you might already
Davor Ocelice9ed2812017-12-25 17:49:28 +01005872 have a "X-Forwarded-For" header from a different application (e.g. stunnel),
Willy Tarreaud72758d2010-01-12 10:42:19 +01005873 and you need preserve it. Also if your backend server doesn't use the
Davor Ocelice9ed2812017-12-25 17:49:28 +01005874 "X-Forwarded-For" header and requires different one (e.g. Zeus Web Servers
Ross Westaf72a1d2008-08-03 10:51:45 +02005875 require "X-Cluster-Client-IP").
Willy Tarreauc27debf2008-01-06 08:57:02 +01005876
5877 Sometimes, a same HAProxy instance may be shared between a direct client
5878 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
5879 used to decrypt HTTPS traffic). It is possible to disable the addition of the
5880 header for a known source address or network by adding the "except" keyword
5881 followed by the network address. In this case, any source IP matching the
5882 network will not cause an addition of this header. Most common uses are with
5883 private networks or 127.0.0.1.
5884
Willy Tarreau87cf5142011-08-19 22:57:24 +02005885 Alternatively, the keyword "if-none" states that the header will only be
5886 added if it is not present. This should only be used in perfectly trusted
5887 environment, as this might cause a security issue if headers reaching haproxy
5888 are under the control of the end-user.
5889
Willy Tarreauc27debf2008-01-06 08:57:02 +01005890 This option may be specified either in the frontend or in the backend. If at
Ross Westaf72a1d2008-08-03 10:51:45 +02005891 least one of them uses it, the header will be added. Note that the backend's
5892 setting of the header subargument takes precedence over the frontend's if
Willy Tarreau87cf5142011-08-19 22:57:24 +02005893 both are defined. In the case of the "if-none" argument, if at least one of
5894 the frontend or the backend does not specify it, it wants the addition to be
5895 mandatory, so it wins.
Willy Tarreauc27debf2008-01-06 08:57:02 +01005896
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005897 Example :
Willy Tarreauc27debf2008-01-06 08:57:02 +01005898 # Public HTTP address also used by stunnel on the same machine
5899 frontend www
5900 mode http
5901 option forwardfor except 127.0.0.1 # stunnel already adds the header
5902
Ross Westaf72a1d2008-08-03 10:51:45 +02005903 # Those servers want the IP Address in X-Client
5904 backend www
5905 mode http
5906 option forwardfor header X-Client
5907
Willy Tarreau87cf5142011-08-19 22:57:24 +02005908 See also : "option httpclose", "option http-server-close",
Christopher Faulet315b39c2018-09-21 16:26:19 +02005909 "option http-keep-alive"
Willy Tarreauc27debf2008-01-06 08:57:02 +01005910
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02005911
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02005912option http-buffer-request
5913no option http-buffer-request
5914 Enable or disable waiting for whole HTTP request body before proceeding
5915 May be used in sections : defaults | frontend | listen | backend
5916 yes | yes | yes | yes
5917 Arguments : none
5918
5919 It is sometimes desirable to wait for the body of an HTTP request before
5920 taking a decision. This is what is being done by "balance url_param" for
5921 example. The first use case is to buffer requests from slow clients before
5922 connecting to the server. Another use case consists in taking the routing
5923 decision based on the request body's contents. This option placed in a
5924 frontend or backend forces the HTTP processing to wait until either the whole
5925 body is received, or the request buffer is full, or the first chunk is
5926 complete in case of chunked encoding. It can have undesired side effects with
Tim Düsterhus4896c442016-11-29 02:15:19 +01005927 some applications abusing HTTP by expecting unbuffered transmissions between
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02005928 the frontend and the backend, so this should definitely not be used by
5929 default.
5930
Baptiste Assmanneccdf432015-10-28 13:49:01 +01005931 See also : "option http-no-delay", "timeout http-request"
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02005932
5933
Willy Tarreau0f228a02015-05-01 15:37:53 +02005934option http-ignore-probes
5935no option http-ignore-probes
5936 Enable or disable logging of null connections and request timeouts
5937 May be used in sections : defaults | frontend | listen | backend
5938 yes | yes | yes | no
5939 Arguments : none
5940
5941 Recently some browsers started to implement a "pre-connect" feature
5942 consisting in speculatively connecting to some recently visited web sites
5943 just in case the user would like to visit them. This results in many
5944 connections being established to web sites, which end up in 408 Request
5945 Timeout if the timeout strikes first, or 400 Bad Request when the browser
5946 decides to close them first. These ones pollute the log and feed the error
5947 counters. There was already "option dontlognull" but it's insufficient in
5948 this case. Instead, this option does the following things :
5949 - prevent any 400/408 message from being sent to the client if nothing
Davor Ocelice9ed2812017-12-25 17:49:28 +01005950 was received over a connection before it was closed;
5951 - prevent any log from being emitted in this situation;
Willy Tarreau0f228a02015-05-01 15:37:53 +02005952 - prevent any error counter from being incremented
5953
5954 That way the empty connection is silently ignored. Note that it is better
5955 not to use this unless it is clear that it is needed, because it will hide
5956 real problems. The most common reason for not receiving a request and seeing
5957 a 408 is due to an MTU inconsistency between the client and an intermediary
5958 element such as a VPN, which blocks too large packets. These issues are
5959 generally seen with POST requests as well as GET with large cookies. The logs
5960 are often the only way to detect them.
5961
5962 If this option has been enabled in a "defaults" section, it can be disabled
5963 in a specific instance by prepending the "no" keyword before it.
5964
5965 See also : "log", "dontlognull", "errorfile", and section 8 about logging.
5966
5967
Willy Tarreau16bfb022010-01-16 19:48:41 +01005968option http-keep-alive
5969no option http-keep-alive
5970 Enable or disable HTTP keep-alive from client to server
5971 May be used in sections : defaults | frontend | listen | backend
5972 yes | yes | yes | yes
5973 Arguments : none
5974
Willy Tarreau70dffda2014-01-30 03:07:23 +01005975 By default HAProxy operates in keep-alive mode with regards to persistent
5976 connections: for each connection it processes each request and response, and
Christopher Faulet315b39c2018-09-21 16:26:19 +02005977 leaves the connection idle on both sides between the end of a response and
5978 the start of a new request. This mode may be changed by several options such
5979 as "option http-server-close", "option httpclose" or "option http-tunnel".
5980 This option allows to set back the keep-alive mode, which can be useful when
5981 another mode was used in a defaults section.
Willy Tarreau70dffda2014-01-30 03:07:23 +01005982
5983 Setting "option http-keep-alive" enables HTTP keep-alive mode on the client-
5984 and server- sides. This provides the lowest latency on the client side (slow
Willy Tarreau16bfb022010-01-16 19:48:41 +01005985 network) and the fastest session reuse on the server side at the expense
5986 of maintaining idle connections to the servers. In general, it is possible
5987 with this option to achieve approximately twice the request rate that the
5988 "http-server-close" option achieves on small objects. There are mainly two
5989 situations where this option may be useful :
5990
5991 - when the server is non-HTTP compliant and authenticates the connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01005992 instead of requests (e.g. NTLM authentication)
Willy Tarreau16bfb022010-01-16 19:48:41 +01005993
5994 - when the cost of establishing the connection to the server is significant
5995 compared to the cost of retrieving the associated object from the server.
5996
5997 This last case can happen when the server is a fast static server of cache.
5998 In this case, the server will need to be properly tuned to support high enough
5999 connection counts because connections will last until the client sends another
6000 request.
6001
6002 If the client request has to go to another backend or another server due to
6003 content switching or the load balancing algorithm, the idle connection will
Willy Tarreau9420b122013-12-15 18:58:25 +01006004 immediately be closed and a new one re-opened. Option "prefer-last-server" is
6005 available to try optimize server selection so that if the server currently
6006 attached to an idle connection is usable, it will be used.
Willy Tarreau16bfb022010-01-16 19:48:41 +01006007
Willy Tarreau16bfb022010-01-16 19:48:41 +01006008 At the moment, logs will not indicate whether requests came from the same
6009 session or not. The accept date reported in the logs corresponds to the end
6010 of the previous request, and the request time corresponds to the time spent
6011 waiting for a new request. The keep-alive request time is still bound to the
6012 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
6013 not set.
6014
Cyril Bonté653dcd62014-02-20 00:13:15 +01006015 This option disables and replaces any previous "option httpclose", "option
Christopher Faulet315b39c2018-09-21 16:26:19 +02006016 http-server-close" or "option http-tunnel". When backend and frontend options
6017 differ, all of these 4 options have precedence over "option http-keep-alive".
Willy Tarreau16bfb022010-01-16 19:48:41 +01006018
Christopher Faulet315b39c2018-09-21 16:26:19 +02006019 See also : "option httpclose",, "option http-server-close",
Willy Tarreau9420b122013-12-15 18:58:25 +01006020 "option prefer-last-server", "option http-pretend-keepalive",
Christopher Faulet315b39c2018-09-21 16:26:19 +02006021 and "1.1. The HTTP transaction model".
Willy Tarreau16bfb022010-01-16 19:48:41 +01006022
6023
Willy Tarreau96e31212011-05-30 18:10:30 +02006024option http-no-delay
6025no option http-no-delay
6026 Instruct the system to favor low interactive delays over performance in HTTP
6027 May be used in sections : defaults | frontend | listen | backend
6028 yes | yes | yes | yes
6029 Arguments : none
6030
6031 In HTTP, each payload is unidirectional and has no notion of interactivity.
6032 Any agent is expected to queue data somewhat for a reasonably low delay.
6033 There are some very rare server-to-server applications that abuse the HTTP
6034 protocol and expect the payload phase to be highly interactive, with many
6035 interleaved data chunks in both directions within a single request. This is
6036 absolutely not supported by the HTTP specification and will not work across
6037 most proxies or servers. When such applications attempt to do this through
6038 haproxy, it works but they will experience high delays due to the network
6039 optimizations which favor performance by instructing the system to wait for
6040 enough data to be available in order to only send full packets. Typical
6041 delays are around 200 ms per round trip. Note that this only happens with
6042 abnormal uses. Normal uses such as CONNECT requests nor WebSockets are not
6043 affected.
6044
6045 When "option http-no-delay" is present in either the frontend or the backend
6046 used by a connection, all such optimizations will be disabled in order to
6047 make the exchanges as fast as possible. Of course this offers no guarantee on
6048 the functionality, as it may break at any other place. But if it works via
6049 HAProxy, it will work as fast as possible. This option should never be used
6050 by default, and should never be used at all unless such a buggy application
6051 is discovered. The impact of using this option is an increase of bandwidth
6052 usage and CPU usage, which may significantly lower performance in high
6053 latency environments.
6054
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02006055 See also : "option http-buffer-request"
6056
Willy Tarreau96e31212011-05-30 18:10:30 +02006057
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006058option http-pretend-keepalive
6059no option http-pretend-keepalive
6060 Define whether haproxy will announce keepalive to the server or not
6061 May be used in sections : defaults | frontend | listen | backend
Christopher Faulet98db9762018-09-21 10:25:19 +02006062 yes | no | yes | yes
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006063 Arguments : none
6064
Christopher Faulet315b39c2018-09-21 16:26:19 +02006065 When running with "option http-server-close" or "option httpclose", haproxy
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006066 adds a "Connection: close" header to the request forwarded to the server.
6067 Unfortunately, when some servers see this header, they automatically refrain
6068 from using the chunked encoding for responses of unknown length, while this
6069 is totally unrelated. The immediate effect is that this prevents haproxy from
6070 maintaining the client connection alive. A second effect is that a client or
6071 a cache could receive an incomplete response without being aware of it, and
6072 consider the response complete.
6073
6074 By setting "option http-pretend-keepalive", haproxy will make the server
6075 believe it will keep the connection alive. The server will then not fall back
6076 to the abnormal undesired above. When haproxy gets the whole response, it
6077 will close the connection with the server just as it would do with the
Christopher Faulet315b39c2018-09-21 16:26:19 +02006078 "option httpclose". That way the client gets a normal response and the
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006079 connection is correctly closed on the server side.
6080
6081 It is recommended not to enable this option by default, because most servers
6082 will more efficiently close the connection themselves after the last packet,
6083 and release its buffers slightly earlier. Also, the added packet on the
6084 network could slightly reduce the overall peak performance. However it is
6085 worth noting that when this option is enabled, haproxy will have slightly
6086 less work to do. So if haproxy is the bottleneck on the whole architecture,
6087 enabling this option might save a few CPU cycles.
6088
Christopher Faulet98db9762018-09-21 10:25:19 +02006089 This option may be set in backend and listen sections. Using it in a frontend
6090 section will be ignored and a warning will be reported during startup. It is
6091 a backend related option, so there is no real reason to set it on a
6092 frontend. This option may be combined with "option httpclose", which will
6093 cause keepalive to be announced to the server and close to be announced to
6094 the client. This practice is discouraged though.
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006095
6096 If this option has been enabled in a "defaults" section, it can be disabled
6097 in a specific instance by prepending the "no" keyword before it.
6098
Christopher Faulet315b39c2018-09-21 16:26:19 +02006099 See also : "option httpclose", "option http-server-close", and
Willy Tarreau16bfb022010-01-16 19:48:41 +01006100 "option http-keep-alive"
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006101
Willy Tarreauc27debf2008-01-06 08:57:02 +01006102
Willy Tarreaub608feb2010-01-02 22:47:18 +01006103option http-server-close
6104no option http-server-close
6105 Enable or disable HTTP connection closing on the server side
6106 May be used in sections : defaults | frontend | listen | backend
6107 yes | yes | yes | yes
6108 Arguments : none
6109
Willy Tarreau70dffda2014-01-30 03:07:23 +01006110 By default HAProxy operates in keep-alive mode with regards to persistent
6111 connections: for each connection it processes each request and response, and
6112 leaves the connection idle on both sides between the end of a response and
6113 the start of a new request. This mode may be changed by several options such
Christopher Faulet315b39c2018-09-21 16:26:19 +02006114 as "option http-server-close", "option httpclose" or "option http-tunnel".
6115 Setting "option http-server-close" enables HTTP connection-close mode on the
6116 server side while keeping the ability to support HTTP keep-alive and
6117 pipelining on the client side. This provides the lowest latency on the client
6118 side (slow network) and the fastest session reuse on the server side to save
6119 server resources, similarly to "option httpclose". It also permits
6120 non-keepalive capable servers to be served in keep-alive mode to the clients
6121 if they conform to the requirements of RFC7230. Please note that some servers
6122 do not always conform to those requirements when they see "Connection: close"
6123 in the request. The effect will be that keep-alive will never be used. A
6124 workaround consists in enabling "option http-pretend-keepalive".
Willy Tarreaub608feb2010-01-02 22:47:18 +01006125
6126 At the moment, logs will not indicate whether requests came from the same
6127 session or not. The accept date reported in the logs corresponds to the end
6128 of the previous request, and the request time corresponds to the time spent
6129 waiting for a new request. The keep-alive request time is still bound to the
Willy Tarreaub16a5742010-01-10 14:46:16 +01006130 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
6131 not set.
Willy Tarreaub608feb2010-01-02 22:47:18 +01006132
6133 This option may be set both in a frontend and in a backend. It is enabled if
6134 at least one of the frontend or backend holding a connection has it enabled.
Christopher Faulet315b39c2018-09-21 16:26:19 +02006135 It disables and replaces any previous "option httpclose", "option http-tunnel"
6136 or "option http-keep-alive". Please check section 4 ("Proxies") to see how
6137 this option combines with others when frontend and backend options differ.
Willy Tarreaub608feb2010-01-02 22:47:18 +01006138
6139 If this option has been enabled in a "defaults" section, it can be disabled
6140 in a specific instance by prepending the "no" keyword before it.
6141
Christopher Faulet315b39c2018-09-21 16:26:19 +02006142 See also : "option httpclose", "option http-pretend-keepalive",
6143 "option http-keep-alive", and "1.1. The HTTP transaction model".
Willy Tarreaub608feb2010-01-02 22:47:18 +01006144
6145
Willy Tarreau02bce8b2014-01-30 00:15:28 +01006146option http-tunnel
6147no option http-tunnel
6148 Disable or enable HTTP connection processing after first transaction
6149 May be used in sections : defaults | frontend | listen | backend
Christopher Faulet4212a302018-09-21 10:42:19 +02006150 yes | yes | yes | no
Willy Tarreau02bce8b2014-01-30 00:15:28 +01006151 Arguments : none
6152
Willy Tarreau70dffda2014-01-30 03:07:23 +01006153 By default HAProxy operates in keep-alive mode with regards to persistent
6154 connections: for each connection it processes each request and response, and
6155 leaves the connection idle on both sides between the end of a response and
6156 the start of a new request. This mode may be changed by several options such
Christopher Faulet315b39c2018-09-21 16:26:19 +02006157 as "option http-server-close", "option httpclose" or "option http-tunnel".
Willy Tarreau70dffda2014-01-30 03:07:23 +01006158
6159 Option "http-tunnel" disables any HTTP processing past the first request and
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03006160 the first response. This is the mode which was used by default in versions
Willy Tarreau70dffda2014-01-30 03:07:23 +01006161 1.0 to 1.5-dev21. It is the mode with the lowest processing overhead, which
6162 is normally not needed anymore unless in very specific cases such as when
6163 using an in-house protocol that looks like HTTP but is not compatible, or
6164 just to log one request per client in order to reduce log size. Note that
6165 everything which works at the HTTP level, including header parsing/addition,
6166 cookie processing or content switching will only work for the first request
6167 and will be ignored after the first response.
Willy Tarreau02bce8b2014-01-30 00:15:28 +01006168
Christopher Faulet4212a302018-09-21 10:42:19 +02006169 This option may be set on frontend and listen sections. Using it on a backend
6170 section will be ignored and a warning will be reported during the startup. It
6171 is a frontend related option, so there is no real reason to set it on a
6172 backend.
6173
Willy Tarreau02bce8b2014-01-30 00:15:28 +01006174 If this option has been enabled in a "defaults" section, it can be disabled
6175 in a specific instance by prepending the "no" keyword before it.
6176
Christopher Faulet315b39c2018-09-21 16:26:19 +02006177 See also : "option httpclose", "option http-server-close",
6178 "option http-keep-alive", and "1.1. The HTTP transaction model".
Willy Tarreau02bce8b2014-01-30 00:15:28 +01006179
6180
Willy Tarreau88d349d2010-01-25 12:15:43 +01006181option http-use-proxy-header
Cyril Bontéf0c60612010-02-06 14:44:47 +01006182no option http-use-proxy-header
Willy Tarreau88d349d2010-01-25 12:15:43 +01006183 Make use of non-standard Proxy-Connection header instead of Connection
6184 May be used in sections : defaults | frontend | listen | backend
6185 yes | yes | yes | no
6186 Arguments : none
6187
Lukas Tribus23953682017-04-28 13:24:30 +00006188 While RFC7230 explicitly states that HTTP/1.1 agents must use the
Willy Tarreau88d349d2010-01-25 12:15:43 +01006189 Connection header to indicate their wish of persistent or non-persistent
6190 connections, both browsers and proxies ignore this header for proxied
6191 connections and make use of the undocumented, non-standard Proxy-Connection
6192 header instead. The issue begins when trying to put a load balancer between
6193 browsers and such proxies, because there will be a difference between what
6194 haproxy understands and what the client and the proxy agree on.
6195
6196 By setting this option in a frontend, haproxy can automatically switch to use
6197 that non-standard header if it sees proxied requests. A proxied request is
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01006198 defined here as one where the URI begins with neither a '/' nor a '*'. This
6199 is incompatible with the HTTP tunnel mode. Note that this option can only be
6200 specified in a frontend and will affect the request along its whole life.
Willy Tarreau88d349d2010-01-25 12:15:43 +01006201
Willy Tarreau844a7e72010-01-31 21:46:18 +01006202 Also, when this option is set, a request which requires authentication will
6203 automatically switch to use proxy authentication headers if it is itself a
6204 proxied request. That makes it possible to check or enforce authentication in
6205 front of an existing proxy.
6206
Willy Tarreau88d349d2010-01-25 12:15:43 +01006207 This option should normally never be used, except in front of a proxy.
6208
Christopher Faulet315b39c2018-09-21 16:26:19 +02006209 See also : "option httpclose", and "option http-server-close".
Willy Tarreau88d349d2010-01-25 12:15:43 +01006210
6211
Willy Tarreau68ad3a42018-10-22 11:49:15 +02006212option http-use-htx
6213no option http-use-htx
6214 Switch to the new HTX internal representation for HTTP protocol elements
6215 May be used in sections : defaults | frontend | listen | backend
6216 yes | yes | yes | yes
6217 Arguments : none
6218
6219 By default, the HTTP protocol is processed as-is. Inserting, deleting, or
6220 modifying a header field requires to rewrite the affected part in the buffer
6221 and to move the buffer's tail accordingly. Since this principle has deep
6222 roots in haproxy, the HTTP/2 protocol is converted to HTTP/1.1 before being
6223 processed this way. It also results in the inability to establish HTTP/2
6224 connections to servers because of the loss of HTTP/2 semantics in the HTTP/1
6225 representation.
6226
6227 HTX is the name of a totally new native internal representation for the HTTP
6228 protocol, that is agnostic to the version and aims at preserving semantics
6229 all along the chain. It relies on a fast parsing, tokenizing and indexing of
6230 the protocol elements so that no more memory moves are necessary and that
6231 most elements are directly accessed. This mechanism is still limited to the
6232 most basic operations (no compression, filters, Lua, applets, cache, etc).
6233 But it supports using either HTTP/1 or HTTP/2 on any side regardless of the
6234 other side's version.
6235
6236 This option indicates that HTX needs to be used. It will cause errors to be
6237 emitted if incompatible features are used, but will allow H2 to be selected
6238 as a server protocol. It is recommended to use this option on new reasonably
6239 simple configurations, but since the feature still has incomplete functional
6240 coverage, it is not enabled by default.
6241
6242 See also : "mode http"
6243
6244
Willy Tarreaud63335a2010-02-26 12:56:52 +01006245option httpchk
6246option httpchk <uri>
6247option httpchk <method> <uri>
6248option httpchk <method> <uri> <version>
6249 Enable HTTP protocol to check on the servers health
6250 May be used in sections : defaults | frontend | listen | backend
6251 yes | no | yes | yes
6252 Arguments :
6253 <method> is the optional HTTP method used with the requests. When not set,
6254 the "OPTIONS" method is used, as it generally requires low server
6255 processing and is easy to filter out from the logs. Any method
6256 may be used, though it is not recommended to invent non-standard
6257 ones.
6258
6259 <uri> is the URI referenced in the HTTP requests. It defaults to " / "
6260 which is accessible by default on almost any server, but may be
6261 changed to any other URI. Query strings are permitted.
6262
6263 <version> is the optional HTTP version string. It defaults to "HTTP/1.0"
6264 but some servers might behave incorrectly in HTTP 1.0, so turning
6265 it to HTTP/1.1 may sometimes help. Note that the Host field is
6266 mandatory in HTTP/1.1, and as a trick, it is possible to pass it
6267 after "\r\n" following the version string.
6268
6269 By default, server health checks only consist in trying to establish a TCP
6270 connection. When "option httpchk" is specified, a complete HTTP request is
6271 sent once the TCP connection is established, and responses 2xx and 3xx are
6272 considered valid, while all other ones indicate a server failure, including
6273 the lack of any response.
6274
6275 The port and interval are specified in the server configuration.
6276
6277 This option does not necessarily require an HTTP backend, it also works with
6278 plain TCP backends. This is particularly useful to check simple scripts bound
6279 to some dedicated ports using the inetd daemon.
6280
6281 Examples :
6282 # Relay HTTPS traffic to Apache instance and check service availability
6283 # using HTTP request "OPTIONS * HTTP/1.1" on port 80.
6284 backend https_relay
6285 mode tcp
6286 option httpchk OPTIONS * HTTP/1.1\r\nHost:\ www
6287 server apache1 192.168.1.1:443 check port 80
6288
Simon Hormanafc47ee2013-11-25 10:46:35 +09006289 See also : "option ssl-hello-chk", "option smtpchk", "option mysql-check",
6290 "option pgsql-check", "http-check" and the "check", "port" and
6291 "inter" server options.
Willy Tarreaud63335a2010-02-26 12:56:52 +01006292
6293
Willy Tarreauc27debf2008-01-06 08:57:02 +01006294option httpclose
6295no option httpclose
Christopher Faulet315b39c2018-09-21 16:26:19 +02006296 Enable or disable HTTP connection closing
Willy Tarreauc27debf2008-01-06 08:57:02 +01006297 May be used in sections : defaults | frontend | listen | backend
6298 yes | yes | yes | yes
6299 Arguments : none
6300
Willy Tarreau70dffda2014-01-30 03:07:23 +01006301 By default HAProxy operates in keep-alive mode with regards to persistent
6302 connections: for each connection it processes each request and response, and
6303 leaves the connection idle on both sides between the end of a response and
6304 the start of a new request. This mode may be changed by several options such
Christopher Faulet315b39c2018-09-21 16:26:19 +02006305 as "option http-server-close", "option httpclose" or "option http-tunnel".
Willy Tarreau70dffda2014-01-30 03:07:23 +01006306
Christopher Faulet315b39c2018-09-21 16:26:19 +02006307 If "option httpclose" is set, HAProxy will close connections with the server
6308 and the client as soon as the request and the response are received. It will
6309 alos check if a "Connection: close" header is already set in each direction,
6310 and will add one if missing. Any "Connection" header different from "close"
6311 will also be removed.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006312
Christopher Faulet315b39c2018-09-21 16:26:19 +02006313 This option may also be combined with "option http-pretend-keepalive", which
6314 will disable sending of the "Connection: close" header, but will still cause
6315 the connection to be closed once the whole response is received.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006316
6317 This option may be set both in a frontend and in a backend. It is enabled if
6318 at least one of the frontend or backend holding a connection has it enabled.
Cyril Bonté653dcd62014-02-20 00:13:15 +01006319 It disables and replaces any previous "option http-server-close",
Christopher Faulet315b39c2018-09-21 16:26:19 +02006320 "option http-keep-alive" or "option http-tunnel". Please check section 4
6321 ("Proxies") to see how this option combines with others when frontend and
6322 backend options differ.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006323
6324 If this option has been enabled in a "defaults" section, it can be disabled
6325 in a specific instance by prepending the "no" keyword before it.
6326
Christopher Faulet315b39c2018-09-21 16:26:19 +02006327 See also : "option http-server-close" and "1.1. The HTTP transaction model".
Willy Tarreauc27debf2008-01-06 08:57:02 +01006328
6329
Emeric Brun3a058f32009-06-30 18:26:00 +02006330option httplog [ clf ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01006331 Enable logging of HTTP request, session state and timers
6332 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01006333 yes | yes | yes | no
Emeric Brun3a058f32009-06-30 18:26:00 +02006334 Arguments :
6335 clf if the "clf" argument is added, then the output format will be
6336 the CLF format instead of HAProxy's default HTTP format. You can
6337 use this when you need to feed HAProxy's logs through a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01006338 log analyzer which only support the CLF format and which is not
Emeric Brun3a058f32009-06-30 18:26:00 +02006339 extensible.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006340
6341 By default, the log output format is very poor, as it only contains the
6342 source and destination addresses, and the instance name. By specifying
6343 "option httplog", each log line turns into a much richer format including,
6344 but not limited to, the HTTP request, the connection timers, the session
6345 status, the connections numbers, the captured headers and cookies, the
6346 frontend, backend and server name, and of course the source address and
6347 ports.
6348
PiBa-NLbd556bf2014-12-11 21:31:54 +01006349 Specifying only "option httplog" will automatically clear the 'clf' mode
6350 if it was set by default.
Emeric Brun3a058f32009-06-30 18:26:00 +02006351
Guillaume de Lafond29f45602017-03-31 19:52:15 +02006352 "option httplog" overrides any previous "log-format" directive.
6353
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006354 See also : section 8 about logging.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006355
Willy Tarreau55165fe2009-05-10 12:02:55 +02006356
6357option http_proxy
6358no option http_proxy
6359 Enable or disable plain HTTP proxy mode
6360 May be used in sections : defaults | frontend | listen | backend
6361 yes | yes | yes | yes
6362 Arguments : none
6363
6364 It sometimes happens that people need a pure HTTP proxy which understands
6365 basic proxy requests without caching nor any fancy feature. In this case,
6366 it may be worth setting up an HAProxy instance with the "option http_proxy"
6367 set. In this mode, no server is declared, and the connection is forwarded to
6368 the IP address and port found in the URL after the "http://" scheme.
6369
6370 No host address resolution is performed, so this only works when pure IP
6371 addresses are passed. Since this option's usage perimeter is rather limited,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01006372 it will probably be used only by experts who know they need exactly it. This
6373 is incompatible with the HTTP tunnel mode.
Willy Tarreau55165fe2009-05-10 12:02:55 +02006374
6375 If this option has been enabled in a "defaults" section, it can be disabled
6376 in a specific instance by prepending the "no" keyword before it.
6377
6378 Example :
6379 # this backend understands HTTP proxy requests and forwards them directly.
6380 backend direct_forward
6381 option httpclose
6382 option http_proxy
6383
6384 See also : "option httpclose"
6385
Willy Tarreau211ad242009-10-03 21:45:07 +02006386
Jamie Gloudon801a0a32012-08-25 00:18:33 -04006387option independent-streams
6388no option independent-streams
6389 Enable or disable independent timeout processing for both directions
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006390 May be used in sections : defaults | frontend | listen | backend
6391 yes | yes | yes | yes
6392 Arguments : none
6393
6394 By default, when data is sent over a socket, both the write timeout and the
6395 read timeout for that socket are refreshed, because we consider that there is
6396 activity on that socket, and we have no other means of guessing if we should
6397 receive data or not.
6398
Davor Ocelice9ed2812017-12-25 17:49:28 +01006399 While this default behavior is desirable for almost all applications, there
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006400 exists a situation where it is desirable to disable it, and only refresh the
6401 read timeout if there are incoming data. This happens on sessions with large
6402 timeouts and low amounts of exchanged data such as telnet session. If the
6403 server suddenly disappears, the output data accumulates in the system's
6404 socket buffers, both timeouts are correctly refreshed, and there is no way
6405 to know the server does not receive them, so we don't timeout. However, when
6406 the underlying protocol always echoes sent data, it would be enough by itself
6407 to detect the issue using the read timeout. Note that this problem does not
6408 happen with more verbose protocols because data won't accumulate long in the
6409 socket buffers.
6410
6411 When this option is set on the frontend, it will disable read timeout updates
6412 on data sent to the client. There probably is little use of this case. When
6413 the option is set on the backend, it will disable read timeout updates on
6414 data sent to the server. Doing so will typically break large HTTP posts from
6415 slow lines, so use it with caution.
6416
Lukas Tribus745f15e2018-11-08 12:41:42 +01006417 Note: older versions used to call this setting "option independant-streams"
Jamie Gloudon801a0a32012-08-25 00:18:33 -04006418 with a spelling mistake. This spelling is still supported but
6419 deprecated.
6420
Willy Tarreauce887fd2012-05-12 12:50:00 +02006421 See also : "timeout client", "timeout server" and "timeout tunnel"
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006422
6423
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02006424option ldap-check
6425 Use LDAPv3 health checks for server testing
6426 May be used in sections : defaults | frontend | listen | backend
6427 yes | no | yes | yes
6428 Arguments : none
6429
6430 It is possible to test that the server correctly talks LDAPv3 instead of just
6431 testing that it accepts the TCP connection. When this option is set, an
6432 LDAPv3 anonymous simple bind message is sent to the server, and the response
6433 is analyzed to find an LDAPv3 bind response message.
6434
6435 The server is considered valid only when the LDAP response contains success
6436 resultCode (http://tools.ietf.org/html/rfc4511#section-4.1.9).
6437
6438 Logging of bind requests is server dependent see your documentation how to
6439 configure it.
6440
6441 Example :
6442 option ldap-check
6443
6444 See also : "option httpchk"
6445
6446
Simon Horman98637e52014-06-20 12:30:16 +09006447option external-check
6448 Use external processes for server health checks
6449 May be used in sections : defaults | frontend | listen | backend
6450 yes | no | yes | yes
6451
6452 It is possible to test the health of a server using an external command.
6453 This is achieved by running the executable set using "external-check
6454 command".
6455
6456 Requires the "external-check" global to be set.
6457
6458 See also : "external-check", "external-check command", "external-check path"
6459
6460
Willy Tarreau211ad242009-10-03 21:45:07 +02006461option log-health-checks
6462no option log-health-checks
Willy Tarreaubef1b322014-05-13 21:01:39 +02006463 Enable or disable logging of health checks status updates
Willy Tarreau211ad242009-10-03 21:45:07 +02006464 May be used in sections : defaults | frontend | listen | backend
6465 yes | no | yes | yes
6466 Arguments : none
6467
Willy Tarreaubef1b322014-05-13 21:01:39 +02006468 By default, failed health check are logged if server is UP and successful
6469 health checks are logged if server is DOWN, so the amount of additional
6470 information is limited.
Willy Tarreau211ad242009-10-03 21:45:07 +02006471
Willy Tarreaubef1b322014-05-13 21:01:39 +02006472 When this option is enabled, any change of the health check status or to
6473 the server's health will be logged, so that it becomes possible to know
6474 that a server was failing occasional checks before crashing, or exactly when
6475 it failed to respond a valid HTTP status, then when the port started to
6476 reject connections, then when the server stopped responding at all.
6477
Davor Ocelice9ed2812017-12-25 17:49:28 +01006478 Note that status changes not caused by health checks (e.g. enable/disable on
Willy Tarreaubef1b322014-05-13 21:01:39 +02006479 the CLI) are intentionally not logged by this option.
Willy Tarreau211ad242009-10-03 21:45:07 +02006480
Willy Tarreaubef1b322014-05-13 21:01:39 +02006481 See also: "option httpchk", "option ldap-check", "option mysql-check",
6482 "option pgsql-check", "option redis-check", "option smtpchk",
6483 "option tcp-check", "log" and section 8 about logging.
Willy Tarreau211ad242009-10-03 21:45:07 +02006484
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006485
6486option log-separate-errors
6487no option log-separate-errors
6488 Change log level for non-completely successful connections
6489 May be used in sections : defaults | frontend | listen | backend
6490 yes | yes | yes | no
6491 Arguments : none
6492
6493 Sometimes looking for errors in logs is not easy. This option makes haproxy
6494 raise the level of logs containing potentially interesting information such
6495 as errors, timeouts, retries, redispatches, or HTTP status codes 5xx. The
6496 level changes from "info" to "err". This makes it possible to log them
6497 separately to a different file with most syslog daemons. Be careful not to
6498 remove them from the original file, otherwise you would lose ordering which
6499 provides very important information.
6500
6501 Using this option, large sites dealing with several thousand connections per
6502 second may log normal traffic to a rotating buffer and only archive smaller
6503 error logs.
6504
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006505 See also : "log", "dontlognull", "dontlog-normal" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006506 logging.
6507
Willy Tarreauc27debf2008-01-06 08:57:02 +01006508
6509option logasap
6510no option logasap
6511 Enable or disable early logging of HTTP requests
6512 May be used in sections : defaults | frontend | listen | backend
6513 yes | yes | yes | no
6514 Arguments : none
6515
6516 By default, HTTP requests are logged upon termination so that the total
6517 transfer time and the number of bytes appear in the logs. When large objects
6518 are being transferred, it may take a while before the request appears in the
6519 logs. Using "option logasap", the request gets logged as soon as the server
6520 sends the complete headers. The only missing information in the logs will be
6521 the total number of bytes which will indicate everything except the amount
6522 of data transferred, and the total time which will not take the transfer
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01006523 time into account. In such a situation, it's a good practice to capture the
Willy Tarreauc27debf2008-01-06 08:57:02 +01006524 "Content-Length" response header so that the logs at least indicate how many
6525 bytes are expected to be transferred.
6526
Willy Tarreaucc6c8912009-02-22 10:53:55 +01006527 Examples :
6528 listen http_proxy 0.0.0.0:80
6529 mode http
6530 option httplog
6531 option logasap
6532 log 192.168.2.200 local3
6533
6534 >>> Feb 6 12:14:14 localhost \
6535 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
6536 static/srv1 9/10/7/14/+30 200 +243 - - ---- 3/1/1/1/0 1/0 \
6537 "GET /image.iso HTTP/1.0"
6538
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006539 See also : "option httplog", "capture response header", and section 8 about
Willy Tarreauc27debf2008-01-06 08:57:02 +01006540 logging.
6541
6542
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02006543option mysql-check [ user <username> [ post-41 ] ]
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006544 Use MySQL health checks for server testing
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006545 May be used in sections : defaults | frontend | listen | backend
6546 yes | no | yes | yes
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006547 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +02006548 <username> This is the username which will be used when connecting to MySQL
6549 server.
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02006550 post-41 Send post v4.1 client compatible checks
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006551
6552 If you specify a username, the check consists of sending two MySQL packet,
6553 one Client Authentication packet, and one QUIT packet, to correctly close
Davor Ocelice9ed2812017-12-25 17:49:28 +01006554 MySQL session. We then parse the MySQL Handshake Initialization packet and/or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006555 Error packet. It is a basic but useful test which does not produce error nor
6556 aborted connect on the server. However, it requires adding an authorization
6557 in the MySQL table, like this :
6558
6559 USE mysql;
6560 INSERT INTO user (Host,User) values ('<ip_of_haproxy>','<username>');
6561 FLUSH PRIVILEGES;
6562
6563 If you don't specify a username (it is deprecated and not recommended), the
Davor Ocelice9ed2812017-12-25 17:49:28 +01006564 check only consists in parsing the Mysql Handshake Initialization packet or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006565 Error packet, we don't send anything in this mode. It was reported that it
6566 can generate lockout if check is too frequent and/or if there is not enough
6567 traffic. In fact, you need in this case to check MySQL "max_connect_errors"
6568 value as if a connection is established successfully within fewer than MySQL
6569 "max_connect_errors" attempts after a previous connection was interrupted,
6570 the error count for the host is cleared to zero. If HAProxy's server get
6571 blocked, the "FLUSH HOSTS" statement is the only way to unblock it.
6572
6573 Remember that this does not check database presence nor database consistency.
6574 To do this, you can use an external check with xinetd for example.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006575
Hervé COMMOWICK212f7782011-06-10 14:05:59 +02006576 The check requires MySQL >=3.22, for older version, please use TCP check.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006577
6578 Most often, an incoming MySQL server needs to see the client's IP address for
6579 various purposes, including IP privilege matching and connection logging.
6580 When possible, it is often wise to masquerade the client's IP address when
6581 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02006582 which requires the transparent proxy feature to be compiled in, and the MySQL
6583 server to route the client via the machine hosting haproxy.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006584
6585 See also: "option httpchk"
6586
6587
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006588option nolinger
6589no option nolinger
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006590 Enable or disable immediate session resource cleaning after close
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006591 May be used in sections: defaults | frontend | listen | backend
6592 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006593 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006594
Davor Ocelice9ed2812017-12-25 17:49:28 +01006595 When clients or servers abort connections in a dirty way (e.g. they are
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006596 physically disconnected), the session timeouts triggers and the session is
6597 closed. But it will remain in FIN_WAIT1 state for some time in the system,
6598 using some resources and possibly limiting the ability to establish newer
6599 connections.
6600
6601 When this happens, it is possible to activate "option nolinger" which forces
6602 the system to immediately remove any socket's pending data on close. Thus,
6603 the session is instantly purged from the system's tables. This usually has
6604 side effects such as increased number of TCP resets due to old retransmits
6605 getting immediately rejected. Some firewalls may sometimes complain about
6606 this too.
6607
6608 For this reason, it is not recommended to use this option when not absolutely
6609 needed. You know that you need it when you have thousands of FIN_WAIT1
6610 sessions on your system (TIME_WAIT ones do not count).
6611
6612 This option may be used both on frontends and backends, depending on the side
6613 where it is required. Use it on the frontend for clients, and on the backend
6614 for servers.
6615
6616 If this option has been enabled in a "defaults" section, it can be disabled
6617 in a specific instance by prepending the "no" keyword before it.
6618
6619
Willy Tarreau55165fe2009-05-10 12:02:55 +02006620option originalto [ except <network> ] [ header <name> ]
6621 Enable insertion of the X-Original-To header to requests sent to servers
6622 May be used in sections : defaults | frontend | listen | backend
6623 yes | yes | yes | yes
6624 Arguments :
6625 <network> is an optional argument used to disable this option for sources
6626 matching <network>
6627 <name> an optional argument to specify a different "X-Original-To"
6628 header name.
6629
6630 Since HAProxy can work in transparent mode, every request from a client can
6631 be redirected to the proxy and HAProxy itself can proxy every request to a
6632 complex SQUID environment and the destination host from SO_ORIGINAL_DST will
6633 be lost. This is annoying when you want access rules based on destination ip
6634 addresses. To solve this problem, a new HTTP header "X-Original-To" may be
6635 added by HAProxy to all requests sent to the server. This header contains a
6636 value representing the original destination IP address. Since this must be
6637 configured to always use the last occurrence of this header only. Note that
6638 only the last occurrence of the header must be used, since it is really
6639 possible that the client has already brought one.
6640
6641 The keyword "header" may be used to supply a different header name to replace
6642 the default "X-Original-To". This can be useful where you might already
6643 have a "X-Original-To" header from a different application, and you need
6644 preserve it. Also if your backend server doesn't use the "X-Original-To"
6645 header and requires different one.
6646
6647 Sometimes, a same HAProxy instance may be shared between a direct client
6648 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
6649 used to decrypt HTTPS traffic). It is possible to disable the addition of the
6650 header for a known source address or network by adding the "except" keyword
6651 followed by the network address. In this case, any source IP matching the
6652 network will not cause an addition of this header. Most common uses are with
6653 private networks or 127.0.0.1.
6654
6655 This option may be specified either in the frontend or in the backend. If at
6656 least one of them uses it, the header will be added. Note that the backend's
6657 setting of the header subargument takes precedence over the frontend's if
6658 both are defined.
6659
Willy Tarreau55165fe2009-05-10 12:02:55 +02006660 Examples :
6661 # Original Destination address
6662 frontend www
6663 mode http
6664 option originalto except 127.0.0.1
6665
6666 # Those servers want the IP Address in X-Client-Dst
6667 backend www
6668 mode http
6669 option originalto header X-Client-Dst
6670
Christopher Faulet315b39c2018-09-21 16:26:19 +02006671 See also : "option httpclose", "option http-server-close".
Willy Tarreau55165fe2009-05-10 12:02:55 +02006672
6673
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006674option persist
6675no option persist
6676 Enable or disable forced persistence on down servers
6677 May be used in sections: defaults | frontend | listen | backend
6678 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006679 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006680
6681 When an HTTP request reaches a backend with a cookie which references a dead
6682 server, by default it is redispatched to another server. It is possible to
6683 force the request to be sent to the dead server first using "option persist"
6684 if absolutely needed. A common use case is when servers are under extreme
6685 load and spend their time flapping. In this case, the users would still be
6686 directed to the server they opened the session on, in the hope they would be
6687 correctly served. It is recommended to use "option redispatch" in conjunction
6688 with this option so that in the event it would not be possible to connect to
6689 the server at all (server definitely dead), the client would finally be
6690 redirected to another valid server.
6691
6692 If this option has been enabled in a "defaults" section, it can be disabled
6693 in a specific instance by prepending the "no" keyword before it.
6694
Willy Tarreau4de91492010-01-22 19:10:05 +01006695 See also : "option redispatch", "retries", "force-persist"
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006696
6697
Willy Tarreau0c122822013-12-15 18:49:01 +01006698option pgsql-check [ user <username> ]
6699 Use PostgreSQL health checks for server testing
6700 May be used in sections : defaults | frontend | listen | backend
6701 yes | no | yes | yes
6702 Arguments :
6703 <username> This is the username which will be used when connecting to
6704 PostgreSQL server.
6705
6706 The check sends a PostgreSQL StartupMessage and waits for either
6707 Authentication request or ErrorResponse message. It is a basic but useful
6708 test which does not produce error nor aborted connect on the server.
6709 This check is identical with the "mysql-check".
6710
6711 See also: "option httpchk"
6712
6713
Willy Tarreau9420b122013-12-15 18:58:25 +01006714option prefer-last-server
6715no option prefer-last-server
6716 Allow multiple load balanced requests to remain on the same server
6717 May be used in sections: defaults | frontend | listen | backend
6718 yes | no | yes | yes
6719 Arguments : none
6720
6721 When the load balancing algorithm in use is not deterministic, and a previous
6722 request was sent to a server to which haproxy still holds a connection, it is
6723 sometimes desirable that subsequent requests on a same session go to the same
6724 server as much as possible. Note that this is different from persistence, as
6725 we only indicate a preference which haproxy tries to apply without any form
6726 of warranty. The real use is for keep-alive connections sent to servers. When
6727 this option is used, haproxy will try to reuse the same connection that is
6728 attached to the server instead of rebalancing to another server, causing a
6729 close of the connection. This can make sense for static file servers. It does
Willy Tarreau068621e2013-12-23 15:11:25 +01006730 not make much sense to use this in combination with hashing algorithms. Note,
6731 haproxy already automatically tries to stick to a server which sends a 401 or
Lukas Tribus80512b12018-10-27 20:07:40 +02006732 to a proxy which sends a 407 (authentication required), when the load
6733 balancing algorithm is not deterministic. This is mandatory for use with the
6734 broken NTLM authentication challenge, and significantly helps in
Willy Tarreau068621e2013-12-23 15:11:25 +01006735 troubleshooting some faulty applications. Option prefer-last-server might be
6736 desirable in these environments as well, to avoid redistributing the traffic
6737 after every other response.
Willy Tarreau9420b122013-12-15 18:58:25 +01006738
6739 If this option has been enabled in a "defaults" section, it can be disabled
6740 in a specific instance by prepending the "no" keyword before it.
6741
6742 See also: "option http-keep-alive"
6743
6744
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006745option redispatch
Joseph Lynch726ab712015-05-11 23:25:34 -07006746option redispatch <interval>
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006747no option redispatch
6748 Enable or disable session redistribution in case of connection failure
6749 May be used in sections: defaults | frontend | listen | backend
6750 yes | no | yes | yes
Joseph Lynch726ab712015-05-11 23:25:34 -07006751 Arguments :
6752 <interval> The optional integer value that controls how often redispatches
6753 occur when retrying connections. Positive value P indicates a
6754 redispatch is desired on every Pth retry, and negative value
Davor Ocelice9ed2812017-12-25 17:49:28 +01006755 N indicate a redispatch is desired on the Nth retry prior to the
Joseph Lynch726ab712015-05-11 23:25:34 -07006756 last retry. For example, the default of -1 preserves the
Davor Ocelice9ed2812017-12-25 17:49:28 +01006757 historical behavior of redispatching on the last retry, a
Joseph Lynch726ab712015-05-11 23:25:34 -07006758 positive value of 1 would indicate a redispatch on every retry,
6759 and a positive value of 3 would indicate a redispatch on every
6760 third retry. You can disable redispatches with a value of 0.
6761
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006762
6763 In HTTP mode, if a server designated by a cookie is down, clients may
6764 definitely stick to it because they cannot flush the cookie, so they will not
6765 be able to access the service anymore.
6766
Willy Tarreau59884a62019-01-02 14:48:31 +01006767 Specifying "option redispatch" will allow the proxy to break cookie or
6768 consistent hash based persistence and redistribute them to a working server.
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006769
Joseph Lynch726ab712015-05-11 23:25:34 -07006770 It also allows to retry connections to another server in case of multiple
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006771 connection failures. Of course, it requires having "retries" set to a nonzero
6772 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01006773
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006774 This form is the preferred form, which replaces both the "redispatch" and
6775 "redisp" keywords.
6776
6777 If this option has been enabled in a "defaults" section, it can be disabled
6778 in a specific instance by prepending the "no" keyword before it.
6779
Willy Tarreau4de91492010-01-22 19:10:05 +01006780 See also : "redispatch", "retries", "force-persist"
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006781
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006782
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02006783option redis-check
6784 Use redis health checks for server testing
6785 May be used in sections : defaults | frontend | listen | backend
6786 yes | no | yes | yes
6787 Arguments : none
6788
6789 It is possible to test that the server correctly talks REDIS protocol instead
6790 of just testing that it accepts the TCP connection. When this option is set,
6791 a PING redis command is sent to the server, and the response is analyzed to
6792 find the "+PONG" response message.
6793
6794 Example :
6795 option redis-check
6796
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03006797 See also : "option httpchk", "option tcp-check", "tcp-check expect"
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02006798
6799
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006800option smtpchk
6801option smtpchk <hello> <domain>
6802 Use SMTP health checks for server testing
6803 May be used in sections : defaults | frontend | listen | backend
6804 yes | no | yes | yes
Willy Tarreaud72758d2010-01-12 10:42:19 +01006805 Arguments :
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006806 <hello> is an optional argument. It is the "hello" command to use. It can
Lukas Tribus27935782018-10-01 02:00:16 +02006807 be either "HELO" (for SMTP) or "EHLO" (for ESMTP). All other
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006808 values will be turned into the default command ("HELO").
6809
6810 <domain> is the domain name to present to the server. It may only be
6811 specified (and is mandatory) if the hello command has been
6812 specified. By default, "localhost" is used.
6813
6814 When "option smtpchk" is set, the health checks will consist in TCP
6815 connections followed by an SMTP command. By default, this command is
6816 "HELO localhost". The server's return code is analyzed and only return codes
6817 starting with a "2" will be considered as valid. All other responses,
6818 including a lack of response will constitute an error and will indicate a
6819 dead server.
6820
6821 This test is meant to be used with SMTP servers or relays. Depending on the
6822 request, it is possible that some servers do not log each connection attempt,
Davor Ocelice9ed2812017-12-25 17:49:28 +01006823 so you may want to experiment to improve the behavior. Using telnet on port
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006824 25 is often easier than adjusting the configuration.
6825
6826 Most often, an incoming SMTP server needs to see the client's IP address for
6827 various purposes, including spam filtering, anti-spoofing and logging. When
6828 possible, it is often wise to masquerade the client's IP address when
6829 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02006830 which requires the transparent proxy feature to be compiled in.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006831
6832 Example :
6833 option smtpchk HELO mydomain.org
6834
6835 See also : "option httpchk", "source"
6836
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006837
Krzysztof Piotr Oledzkiaeebf9b2009-10-04 15:43:17 +02006838option socket-stats
6839no option socket-stats
6840
6841 Enable or disable collecting & providing separate statistics for each socket.
6842 May be used in sections : defaults | frontend | listen | backend
6843 yes | yes | yes | no
6844
6845 Arguments : none
6846
6847
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006848option splice-auto
6849no option splice-auto
6850 Enable or disable automatic kernel acceleration on sockets in both directions
6851 May be used in sections : defaults | frontend | listen | backend
6852 yes | yes | yes | yes
6853 Arguments : none
6854
6855 When this option is enabled either on a frontend or on a backend, haproxy
6856 will automatically evaluate the opportunity to use kernel tcp splicing to
Davor Ocelice9ed2812017-12-25 17:49:28 +01006857 forward data between the client and the server, in either direction. HAProxy
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006858 uses heuristics to estimate if kernel splicing might improve performance or
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006859 not. Both directions are handled independently. Note that the heuristics used
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006860 are not much aggressive in order to limit excessive use of splicing. This
6861 option requires splicing to be enabled at compile time, and may be globally
6862 disabled with the global option "nosplice". Since splice uses pipes, using it
6863 requires that there are enough spare pipes.
6864
6865 Important note: kernel-based TCP splicing is a Linux-specific feature which
6866 first appeared in kernel 2.6.25. It offers kernel-based acceleration to
6867 transfer data between sockets without copying these data to user-space, thus
6868 providing noticeable performance gains and CPU cycles savings. Since many
6869 early implementations are buggy, corrupt data and/or are inefficient, this
6870 feature is not enabled by default, and it should be used with extreme care.
6871 While it is not possible to detect the correctness of an implementation,
6872 2.6.29 is the first version offering a properly working implementation. In
6873 case of doubt, splicing may be globally disabled using the global "nosplice"
6874 keyword.
6875
6876 Example :
6877 option splice-auto
6878
6879 If this option has been enabled in a "defaults" section, it can be disabled
6880 in a specific instance by prepending the "no" keyword before it.
6881
6882 See also : "option splice-request", "option splice-response", and global
6883 options "nosplice" and "maxpipes"
6884
6885
6886option splice-request
6887no option splice-request
6888 Enable or disable automatic kernel acceleration on sockets for requests
6889 May be used in sections : defaults | frontend | listen | backend
6890 yes | yes | yes | yes
6891 Arguments : none
6892
6893 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04006894 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006895 the client to the server. It might still use the recv/send scheme if there
6896 are no spare pipes left. This option requires splicing to be enabled at
6897 compile time, and may be globally disabled with the global option "nosplice".
6898 Since splice uses pipes, using it requires that there are enough spare pipes.
6899
6900 Important note: see "option splice-auto" for usage limitations.
6901
6902 Example :
6903 option splice-request
6904
6905 If this option has been enabled in a "defaults" section, it can be disabled
6906 in a specific instance by prepending the "no" keyword before it.
6907
6908 See also : "option splice-auto", "option splice-response", and global options
6909 "nosplice" and "maxpipes"
6910
6911
6912option splice-response
6913no option splice-response
6914 Enable or disable automatic kernel acceleration on sockets for responses
6915 May be used in sections : defaults | frontend | listen | backend
6916 yes | yes | yes | yes
6917 Arguments : none
6918
6919 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04006920 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006921 the server to the client. It might still use the recv/send scheme if there
6922 are no spare pipes left. This option requires splicing to be enabled at
6923 compile time, and may be globally disabled with the global option "nosplice".
6924 Since splice uses pipes, using it requires that there are enough spare pipes.
6925
6926 Important note: see "option splice-auto" for usage limitations.
6927
6928 Example :
6929 option splice-response
6930
6931 If this option has been enabled in a "defaults" section, it can be disabled
6932 in a specific instance by prepending the "no" keyword before it.
6933
6934 See also : "option splice-auto", "option splice-request", and global options
6935 "nosplice" and "maxpipes"
6936
6937
Christopher Fauletba7bc162016-11-07 21:07:38 +01006938option spop-check
6939 Use SPOP health checks for server testing
6940 May be used in sections : defaults | frontend | listen | backend
6941 no | no | no | yes
6942 Arguments : none
6943
6944 It is possible to test that the server correctly talks SPOP protocol instead
6945 of just testing that it accepts the TCP connection. When this option is set,
6946 a HELLO handshake is performed between HAProxy and the server, and the
6947 response is analyzed to check no error is reported.
6948
6949 Example :
6950 option spop-check
6951
6952 See also : "option httpchk"
6953
6954
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006955option srvtcpka
6956no option srvtcpka
6957 Enable or disable the sending of TCP keepalive packets on the server side
6958 May be used in sections : defaults | frontend | listen | backend
6959 yes | no | yes | yes
6960 Arguments : none
6961
6962 When there is a firewall or any session-aware component between a client and
6963 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01006964 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006965 components decides to expire a session which has remained idle for too long.
6966
6967 Enabling socket-level TCP keep-alives makes the system regularly send packets
6968 to the other end of the connection, leaving it active. The delay between
6969 keep-alive probes is controlled by the system only and depends both on the
6970 operating system and its tuning parameters.
6971
6972 It is important to understand that keep-alive packets are neither emitted nor
6973 received at the application level. It is only the network stacks which sees
6974 them. For this reason, even if one side of the proxy already uses keep-alives
6975 to maintain its connection alive, those keep-alive packets will not be
6976 forwarded to the other side of the proxy.
6977
6978 Please note that this has nothing to do with HTTP keep-alive.
6979
6980 Using option "srvtcpka" enables the emission of TCP keep-alive probes on the
6981 server side of a connection, which should help when session expirations are
6982 noticed between HAProxy and a server.
6983
6984 If this option has been enabled in a "defaults" section, it can be disabled
6985 in a specific instance by prepending the "no" keyword before it.
6986
6987 See also : "option clitcpka", "option tcpka"
6988
6989
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006990option ssl-hello-chk
6991 Use SSLv3 client hello health checks for server testing
6992 May be used in sections : defaults | frontend | listen | backend
6993 yes | no | yes | yes
6994 Arguments : none
6995
6996 When some SSL-based protocols are relayed in TCP mode through HAProxy, it is
6997 possible to test that the server correctly talks SSL instead of just testing
6998 that it accepts the TCP connection. When "option ssl-hello-chk" is set, pure
6999 SSLv3 client hello messages are sent once the connection is established to
7000 the server, and the response is analyzed to find an SSL server hello message.
7001 The server is considered valid only when the response contains this server
7002 hello message.
7003
7004 All servers tested till there correctly reply to SSLv3 client hello messages,
7005 and most servers tested do not even log the requests containing only hello
7006 messages, which is appreciable.
7007
Willy Tarreau763a95b2012-10-04 23:15:39 +02007008 Note that this check works even when SSL support was not built into haproxy
7009 because it forges the SSL message. When SSL support is available, it is best
7010 to use native SSL health checks instead of this one.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007011
Willy Tarreau763a95b2012-10-04 23:15:39 +02007012 See also: "option httpchk", "check-ssl"
7013
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007014
Willy Tarreaued179852013-12-16 01:07:00 +01007015option tcp-check
7016 Perform health checks using tcp-check send/expect sequences
7017 May be used in sections: defaults | frontend | listen | backend
7018 yes | no | yes | yes
7019
7020 This health check method is intended to be combined with "tcp-check" command
7021 lists in order to support send/expect types of health check sequences.
7022
7023 TCP checks currently support 4 modes of operations :
7024 - no "tcp-check" directive : the health check only consists in a connection
7025 attempt, which remains the default mode.
7026
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007027 - "tcp-check send" or "tcp-check send-binary" only is mentioned : this is
Willy Tarreaued179852013-12-16 01:07:00 +01007028 used to send a string along with a connection opening. With some
7029 protocols, it helps sending a "QUIT" message for example that prevents
7030 the server from logging a connection error for each health check. The
7031 check result will still be based on the ability to open the connection
7032 only.
7033
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007034 - "tcp-check expect" only is mentioned : this is used to test a banner.
Willy Tarreaued179852013-12-16 01:07:00 +01007035 The connection is opened and haproxy waits for the server to present some
7036 contents which must validate some rules. The check result will be based
7037 on the matching between the contents and the rules. This is suited for
7038 POP, IMAP, SMTP, FTP, SSH, TELNET.
7039
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007040 - both "tcp-check send" and "tcp-check expect" are mentioned : this is
Davor Ocelice9ed2812017-12-25 17:49:28 +01007041 used to test a hello-type protocol. HAProxy sends a message, the server
7042 responds and its response is analyzed. the check result will be based on
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007043 the matching between the response contents and the rules. This is often
Willy Tarreaued179852013-12-16 01:07:00 +01007044 suited for protocols which require a binding or a request/response model.
7045 LDAP, MySQL, Redis and SSL are example of such protocols, though they
7046 already all have their dedicated checks with a deeper understanding of
7047 the respective protocols.
7048 In this mode, many questions may be sent and many answers may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01007049 analyzed.
Willy Tarreaued179852013-12-16 01:07:00 +01007050
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007051 A fifth mode can be used to insert comments in different steps of the
7052 script.
7053
7054 For each tcp-check rule you create, you can add a "comment" directive,
7055 followed by a string. This string will be reported in the log and stderr
7056 in debug mode. It is useful to make user-friendly error reporting.
7057 The "comment" is of course optional.
7058
7059
Willy Tarreaued179852013-12-16 01:07:00 +01007060 Examples :
Davor Ocelice9ed2812017-12-25 17:49:28 +01007061 # perform a POP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01007062 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007063 tcp-check expect string +OK\ POP3\ ready comment POP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01007064
Davor Ocelice9ed2812017-12-25 17:49:28 +01007065 # perform an IMAP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01007066 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007067 tcp-check expect string *\ OK\ IMAP4\ ready comment IMAP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01007068
7069 # look for the redis master server after ensuring it speaks well
7070 # redis protocol, then it exits properly.
Davor Ocelice9ed2812017-12-25 17:49:28 +01007071 # (send a command then analyze the response 3 times)
Willy Tarreaued179852013-12-16 01:07:00 +01007072 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007073 tcp-check comment PING\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01007074 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02007075 tcp-check expect string +PONG
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007076 tcp-check comment role\ check
Willy Tarreaued179852013-12-16 01:07:00 +01007077 tcp-check send info\ replication\r\n
7078 tcp-check expect string role:master
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007079 tcp-check comment QUIT\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01007080 tcp-check send QUIT\r\n
7081 tcp-check expect string +OK
7082
Davor Ocelice9ed2812017-12-25 17:49:28 +01007083 forge a HTTP request, then analyze the response
Willy Tarreaued179852013-12-16 01:07:00 +01007084 (send many headers before analyzing)
7085 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007086 tcp-check comment forge\ and\ send\ HTTP\ request
Willy Tarreaued179852013-12-16 01:07:00 +01007087 tcp-check send HEAD\ /\ HTTP/1.1\r\n
7088 tcp-check send Host:\ www.mydomain.com\r\n
7089 tcp-check send User-Agent:\ HAProxy\ tcpcheck\r\n
7090 tcp-check send \r\n
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007091 tcp-check expect rstring HTTP/1\..\ (2..|3..) comment check\ HTTP\ response
Willy Tarreaued179852013-12-16 01:07:00 +01007092
7093
7094 See also : "tcp-check expect", "tcp-check send"
7095
7096
Willy Tarreau9ea05a72009-06-14 12:07:01 +02007097option tcp-smart-accept
7098no option tcp-smart-accept
7099 Enable or disable the saving of one ACK packet during the accept sequence
7100 May be used in sections : defaults | frontend | listen | backend
7101 yes | yes | yes | no
7102 Arguments : none
7103
7104 When an HTTP connection request comes in, the system acknowledges it on
7105 behalf of HAProxy, then the client immediately sends its request, and the
7106 system acknowledges it too while it is notifying HAProxy about the new
7107 connection. HAProxy then reads the request and responds. This means that we
7108 have one TCP ACK sent by the system for nothing, because the request could
7109 very well be acknowledged by HAProxy when it sends its response.
7110
7111 For this reason, in HTTP mode, HAProxy automatically asks the system to avoid
7112 sending this useless ACK on platforms which support it (currently at least
7113 Linux). It must not cause any problem, because the system will send it anyway
7114 after 40 ms if the response takes more time than expected to come.
7115
7116 During complex network debugging sessions, it may be desirable to disable
7117 this optimization because delayed ACKs can make troubleshooting more complex
7118 when trying to identify where packets are delayed. It is then possible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01007119 fall back to normal behavior by specifying "no option tcp-smart-accept".
Willy Tarreau9ea05a72009-06-14 12:07:01 +02007120
7121 It is also possible to force it for non-HTTP proxies by simply specifying
7122 "option tcp-smart-accept". For instance, it can make sense with some services
7123 such as SMTP where the server speaks first.
7124
7125 It is recommended to avoid forcing this option in a defaults section. In case
7126 of doubt, consider setting it back to automatic values by prepending the
7127 "default" keyword before it, or disabling it using the "no" keyword.
7128
Willy Tarreaud88edf22009-06-14 15:48:17 +02007129 See also : "option tcp-smart-connect"
7130
7131
7132option tcp-smart-connect
7133no option tcp-smart-connect
7134 Enable or disable the saving of one ACK packet during the connect sequence
7135 May be used in sections : defaults | frontend | listen | backend
7136 yes | no | yes | yes
7137 Arguments : none
7138
7139 On certain systems (at least Linux), HAProxy can ask the kernel not to
7140 immediately send an empty ACK upon a connection request, but to directly
7141 send the buffer request instead. This saves one packet on the network and
7142 thus boosts performance. It can also be useful for some servers, because they
7143 immediately get the request along with the incoming connection.
7144
7145 This feature is enabled when "option tcp-smart-connect" is set in a backend.
7146 It is not enabled by default because it makes network troubleshooting more
7147 complex.
7148
7149 It only makes sense to enable it with protocols where the client speaks first
7150 such as HTTP. In other situations, if there is no data to send in place of
7151 the ACK, a normal ACK is sent.
7152
7153 If this option has been enabled in a "defaults" section, it can be disabled
7154 in a specific instance by prepending the "no" keyword before it.
7155
7156 See also : "option tcp-smart-accept"
7157
Willy Tarreau9ea05a72009-06-14 12:07:01 +02007158
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007159option tcpka
7160 Enable or disable the sending of TCP keepalive packets on both sides
7161 May be used in sections : defaults | frontend | listen | backend
7162 yes | yes | yes | yes
7163 Arguments : none
7164
7165 When there is a firewall or any session-aware component between a client and
7166 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01007167 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007168 components decides to expire a session which has remained idle for too long.
7169
7170 Enabling socket-level TCP keep-alives makes the system regularly send packets
7171 to the other end of the connection, leaving it active. The delay between
7172 keep-alive probes is controlled by the system only and depends both on the
7173 operating system and its tuning parameters.
7174
7175 It is important to understand that keep-alive packets are neither emitted nor
7176 received at the application level. It is only the network stacks which sees
7177 them. For this reason, even if one side of the proxy already uses keep-alives
7178 to maintain its connection alive, those keep-alive packets will not be
7179 forwarded to the other side of the proxy.
7180
7181 Please note that this has nothing to do with HTTP keep-alive.
7182
7183 Using option "tcpka" enables the emission of TCP keep-alive probes on both
7184 the client and server sides of a connection. Note that this is meaningful
7185 only in "defaults" or "listen" sections. If this option is used in a
7186 frontend, only the client side will get keep-alives, and if this option is
7187 used in a backend, only the server side will get keep-alives. For this
7188 reason, it is strongly recommended to explicitly use "option clitcpka" and
7189 "option srvtcpka" when the configuration is split between frontends and
7190 backends.
7191
7192 See also : "option clitcpka", "option srvtcpka"
7193
Willy Tarreau844e3c52008-01-11 16:28:18 +01007194
7195option tcplog
7196 Enable advanced logging of TCP connections with session state and timers
7197 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01007198 yes | yes | yes | no
Willy Tarreau844e3c52008-01-11 16:28:18 +01007199 Arguments : none
7200
7201 By default, the log output format is very poor, as it only contains the
7202 source and destination addresses, and the instance name. By specifying
7203 "option tcplog", each log line turns into a much richer format including, but
7204 not limited to, the connection timers, the session status, the connections
7205 numbers, the frontend, backend and server name, and of course the source
7206 address and ports. This option is useful for pure TCP proxies in order to
7207 find which of the client or server disconnects or times out. For normal HTTP
7208 proxies, it's better to use "option httplog" which is even more complete.
7209
Guillaume de Lafond29f45602017-03-31 19:52:15 +02007210 "option tcplog" overrides any previous "log-format" directive.
7211
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007212 See also : "option httplog", and section 8 about logging.
Willy Tarreau844e3c52008-01-11 16:28:18 +01007213
7214
Willy Tarreau844e3c52008-01-11 16:28:18 +01007215option transparent
7216no option transparent
7217 Enable client-side transparent proxying
7218 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +01007219 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +01007220 Arguments : none
7221
7222 This option was introduced in order to provide layer 7 persistence to layer 3
7223 load balancers. The idea is to use the OS's ability to redirect an incoming
7224 connection for a remote address to a local process (here HAProxy), and let
7225 this process know what address was initially requested. When this option is
7226 used, sessions without cookies will be forwarded to the original destination
7227 IP address of the incoming request (which should match that of another
7228 equipment), while requests with cookies will still be forwarded to the
7229 appropriate server.
7230
7231 Note that contrary to a common belief, this option does NOT make HAProxy
7232 present the client's IP to the server when establishing the connection.
7233
Willy Tarreaua1146052011-03-01 09:51:54 +01007234 See also: the "usesrc" argument of the "source" keyword, and the
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007235 "transparent" option of the "bind" keyword.
Willy Tarreau844e3c52008-01-11 16:28:18 +01007236
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007237
Simon Horman98637e52014-06-20 12:30:16 +09007238external-check command <command>
7239 Executable to run when performing an external-check
7240 May be used in sections : defaults | frontend | listen | backend
7241 yes | no | yes | yes
7242
7243 Arguments :
7244 <command> is the external command to run
7245
Simon Horman98637e52014-06-20 12:30:16 +09007246 The arguments passed to the to the command are:
7247
Cyril Bonté777be862014-12-02 21:21:35 +01007248 <proxy_address> <proxy_port> <server_address> <server_port>
Simon Horman98637e52014-06-20 12:30:16 +09007249
Cyril Bonté777be862014-12-02 21:21:35 +01007250 The <proxy_address> and <proxy_port> are derived from the first listener
7251 that is either IPv4, IPv6 or a UNIX socket. In the case of a UNIX socket
7252 listener the proxy_address will be the path of the socket and the
7253 <proxy_port> will be the string "NOT_USED". In a backend section, it's not
7254 possible to determine a listener, and both <proxy_address> and <proxy_port>
7255 will have the string value "NOT_USED".
Simon Horman98637e52014-06-20 12:30:16 +09007256
Cyril Bonté72cda2a2014-12-27 22:28:39 +01007257 Some values are also provided through environment variables.
7258
7259 Environment variables :
7260 HAPROXY_PROXY_ADDR The first bind address if available (or empty if not
7261 applicable, for example in a "backend" section).
7262
7263 HAPROXY_PROXY_ID The backend id.
7264
7265 HAPROXY_PROXY_NAME The backend name.
7266
7267 HAPROXY_PROXY_PORT The first bind port if available (or empty if not
7268 applicable, for example in a "backend" section or
7269 for a UNIX socket).
7270
7271 HAPROXY_SERVER_ADDR The server address.
7272
7273 HAPROXY_SERVER_CURCONN The current number of connections on the server.
7274
7275 HAPROXY_SERVER_ID The server id.
7276
7277 HAPROXY_SERVER_MAXCONN The server max connections.
7278
7279 HAPROXY_SERVER_NAME The server name.
7280
7281 HAPROXY_SERVER_PORT The server port if available (or empty for a UNIX
7282 socket).
7283
7284 PATH The PATH environment variable used when executing
7285 the command may be set using "external-check path".
7286
Simon Horman98637e52014-06-20 12:30:16 +09007287 If the command executed and exits with a zero status then the check is
7288 considered to have passed, otherwise the check is considered to have
7289 failed.
7290
7291 Example :
7292 external-check command /bin/true
7293
7294 See also : "external-check", "option external-check", "external-check path"
7295
7296
7297external-check path <path>
7298 The value of the PATH environment variable used when running an external-check
7299 May be used in sections : defaults | frontend | listen | backend
7300 yes | no | yes | yes
7301
7302 Arguments :
7303 <path> is the path used when executing external command to run
7304
7305 The default path is "".
7306
7307 Example :
7308 external-check path "/usr/bin:/bin"
7309
7310 See also : "external-check", "option external-check",
7311 "external-check command"
7312
7313
Emeric Brun647caf12009-06-30 17:57:00 +02007314persist rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02007315persist rdp-cookie(<name>)
Emeric Brun647caf12009-06-30 17:57:00 +02007316 Enable RDP cookie-based persistence
7317 May be used in sections : defaults | frontend | listen | backend
7318 yes | no | yes | yes
7319 Arguments :
7320 <name> is the optional name of the RDP cookie to check. If omitted, the
Willy Tarreau61e28f22010-05-16 22:31:05 +02007321 default cookie name "msts" will be used. There currently is no
7322 valid reason to change this name.
Emeric Brun647caf12009-06-30 17:57:00 +02007323
7324 This statement enables persistence based on an RDP cookie. The RDP cookie
7325 contains all information required to find the server in the list of known
Davor Ocelice9ed2812017-12-25 17:49:28 +01007326 servers. So when this option is set in the backend, the request is analyzed
Emeric Brun647caf12009-06-30 17:57:00 +02007327 and if an RDP cookie is found, it is decoded. If it matches a known server
7328 which is still UP (or if "option persist" is set), then the connection is
7329 forwarded to this server.
7330
7331 Note that this only makes sense in a TCP backend, but for this to work, the
7332 frontend must have waited long enough to ensure that an RDP cookie is present
7333 in the request buffer. This is the same requirement as with the "rdp-cookie"
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007334 load-balancing method. Thus it is highly recommended to put all statements in
Emeric Brun647caf12009-06-30 17:57:00 +02007335 a single "listen" section.
7336
Willy Tarreau61e28f22010-05-16 22:31:05 +02007337 Also, it is important to understand that the terminal server will emit this
7338 RDP cookie only if it is configured for "token redirection mode", which means
7339 that the "IP address redirection" option is disabled.
7340
Emeric Brun647caf12009-06-30 17:57:00 +02007341 Example :
7342 listen tse-farm
7343 bind :3389
7344 # wait up to 5s for an RDP cookie in the request
7345 tcp-request inspect-delay 5s
7346 tcp-request content accept if RDP_COOKIE
7347 # apply RDP cookie persistence
7348 persist rdp-cookie
7349 # if server is unknown, let's balance on the same cookie.
Cyril Bontédc4d9032012-04-08 21:57:39 +02007350 # alternatively, "balance leastconn" may be useful too.
Emeric Brun647caf12009-06-30 17:57:00 +02007351 balance rdp-cookie
7352 server srv1 1.1.1.1:3389
7353 server srv2 1.1.1.2:3389
7354
Simon Hormanab814e02011-06-24 14:50:20 +09007355 See also : "balance rdp-cookie", "tcp-request", the "req_rdp_cookie" ACL and
7356 the rdp_cookie pattern fetch function.
Emeric Brun647caf12009-06-30 17:57:00 +02007357
7358
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007359rate-limit sessions <rate>
7360 Set a limit on the number of new sessions accepted per second on a frontend
7361 May be used in sections : defaults | frontend | listen | backend
7362 yes | yes | yes | no
7363 Arguments :
7364 <rate> The <rate> parameter is an integer designating the maximum number
7365 of new sessions per second to accept on the frontend.
7366
7367 When the frontend reaches the specified number of new sessions per second, it
7368 stops accepting new connections until the rate drops below the limit again.
7369 During this time, the pending sessions will be kept in the socket's backlog
7370 (in system buffers) and haproxy will not even be aware that sessions are
7371 pending. When applying very low limit on a highly loaded service, it may make
7372 sense to increase the socket's backlog using the "backlog" keyword.
7373
7374 This feature is particularly efficient at blocking connection-based attacks
7375 or service abuse on fragile servers. Since the session rate is measured every
7376 millisecond, it is extremely accurate. Also, the limit applies immediately,
7377 no delay is needed at all to detect the threshold.
7378
7379 Example : limit the connection rate on SMTP to 10 per second max
7380 listen smtp
7381 mode tcp
7382 bind :25
7383 rate-limit sessions 10
Panagiotis Panagiotopoulos7282d8e2016-02-11 16:37:15 +02007384 server smtp1 127.0.0.1:1025
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007385
Willy Tarreaua17c2d92011-07-25 08:16:20 +02007386 Note : when the maximum rate is reached, the frontend's status is not changed
7387 but its sockets appear as "WAITING" in the statistics if the
7388 "socket-stats" option is enabled.
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007389
7390 See also : the "backlog" keyword and the "fe_sess_rate" ACL criterion.
7391
7392
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007393redirect location <loc> [code <code>] <option> [{if | unless} <condition>]
7394redirect prefix <pfx> [code <code>] <option> [{if | unless} <condition>]
7395redirect scheme <sch> [code <code>] <option> [{if | unless} <condition>]
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007396 Return an HTTP redirection if/unless a condition is matched
7397 May be used in sections : defaults | frontend | listen | backend
7398 no | yes | yes | yes
7399
7400 If/unless the condition is matched, the HTTP request will lead to a redirect
Willy Tarreauf285f542010-01-03 20:03:03 +01007401 response. If no condition is specified, the redirect applies unconditionally.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007402
Willy Tarreau0140f252008-11-19 21:07:09 +01007403 Arguments :
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007404 <loc> With "redirect location", the exact value in <loc> is placed into
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007405 the HTTP "Location" header. When used in an "http-request" rule,
7406 <loc> value follows the log-format rules and can include some
7407 dynamic values (see Custom Log Format in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007408
7409 <pfx> With "redirect prefix", the "Location" header is built from the
7410 concatenation of <pfx> and the complete URI path, including the
7411 query string, unless the "drop-query" option is specified (see
7412 below). As a special case, if <pfx> equals exactly "/", then
7413 nothing is inserted before the original URI. It allows one to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007414 redirect to the same URL (for instance, to insert a cookie). When
7415 used in an "http-request" rule, <pfx> value follows the log-format
7416 rules and can include some dynamic values (see Custom Log Format
7417 in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007418
7419 <sch> With "redirect scheme", then the "Location" header is built by
7420 concatenating <sch> with "://" then the first occurrence of the
7421 "Host" header, and then the URI path, including the query string
7422 unless the "drop-query" option is specified (see below). If no
7423 path is found or if the path is "*", then "/" is used instead. If
7424 no "Host" header is found, then an empty host component will be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007425 returned, which most recent browsers interpret as redirecting to
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007426 the same host. This directive is mostly used to redirect HTTP to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007427 HTTPS. When used in an "http-request" rule, <sch> value follows
7428 the log-format rules and can include some dynamic values (see
7429 Custom Log Format in section 8.2.4).
Willy Tarreau0140f252008-11-19 21:07:09 +01007430
7431 <code> The code is optional. It indicates which type of HTTP redirection
Willy Tarreaub67fdc42013-03-29 19:28:11 +01007432 is desired. Only codes 301, 302, 303, 307 and 308 are supported,
7433 with 302 used by default if no code is specified. 301 means
7434 "Moved permanently", and a browser may cache the Location. 302
Baptiste Assmannea849c02015-08-03 11:42:50 +02007435 means "Moved temporarily" and means that the browser should not
Willy Tarreaub67fdc42013-03-29 19:28:11 +01007436 cache the redirection. 303 is equivalent to 302 except that the
7437 browser will fetch the location with a GET method. 307 is just
7438 like 302 but makes it clear that the same method must be reused.
7439 Likewise, 308 replaces 301 if the same method must be used.
Willy Tarreau0140f252008-11-19 21:07:09 +01007440
7441 <option> There are several options which can be specified to adjust the
Davor Ocelice9ed2812017-12-25 17:49:28 +01007442 expected behavior of a redirection :
Willy Tarreau0140f252008-11-19 21:07:09 +01007443
7444 - "drop-query"
7445 When this keyword is used in a prefix-based redirection, then the
7446 location will be set without any possible query-string, which is useful
7447 for directing users to a non-secure page for instance. It has no effect
7448 with a location-type redirect.
7449
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01007450 - "append-slash"
7451 This keyword may be used in conjunction with "drop-query" to redirect
7452 users who use a URL not ending with a '/' to the same one with the '/'.
7453 It can be useful to ensure that search engines will only see one URL.
7454 For this, a return code 301 is preferred.
7455
Willy Tarreau0140f252008-11-19 21:07:09 +01007456 - "set-cookie NAME[=value]"
7457 A "Set-Cookie" header will be added with NAME (and optionally "=value")
7458 to the response. This is sometimes used to indicate that a user has
7459 been seen, for instance to protect against some types of DoS. No other
7460 cookie option is added, so the cookie will be a session cookie. Note
7461 that for a browser, a sole cookie name without an equal sign is
7462 different from a cookie with an equal sign.
7463
7464 - "clear-cookie NAME[=]"
7465 A "Set-Cookie" header will be added with NAME (and optionally "="), but
7466 with the "Max-Age" attribute set to zero. This will tell the browser to
7467 delete this cookie. It is useful for instance on logout pages. It is
7468 important to note that clearing the cookie "NAME" will not remove a
7469 cookie set with "NAME=value". You have to clear the cookie "NAME=" for
7470 that, because the browser makes the difference.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007471
7472 Example: move the login URL only to HTTPS.
7473 acl clear dst_port 80
7474 acl secure dst_port 8080
7475 acl login_page url_beg /login
Willy Tarreau0140f252008-11-19 21:07:09 +01007476 acl logout url_beg /logout
Willy Tarreau79da4692008-11-19 20:03:04 +01007477 acl uid_given url_reg /login?userid=[^&]+
Willy Tarreau0140f252008-11-19 21:07:09 +01007478 acl cookie_set hdr_sub(cookie) SEEN=1
7479
7480 redirect prefix https://mysite.com set-cookie SEEN=1 if !cookie_set
Willy Tarreau79da4692008-11-19 20:03:04 +01007481 redirect prefix https://mysite.com if login_page !secure
7482 redirect prefix http://mysite.com drop-query if login_page !uid_given
7483 redirect location http://mysite.com/ if !login_page secure
Willy Tarreau0140f252008-11-19 21:07:09 +01007484 redirect location / clear-cookie USERID= if logout
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007485
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01007486 Example: send redirects for request for articles without a '/'.
7487 acl missing_slash path_reg ^/article/[^/]*$
7488 redirect code 301 prefix / drop-query append-slash if missing_slash
7489
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007490 Example: redirect all HTTP traffic to HTTPS when SSL is handled by haproxy.
David BERARDe7153042012-11-03 00:11:31 +01007491 redirect scheme https if !{ ssl_fc }
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007492
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007493 Example: append 'www.' prefix in front of all hosts not having it
Coen Rosdorff596659b2016-04-11 11:33:49 +02007494 http-request redirect code 301 location \
7495 http://www.%[hdr(host)]%[capture.req.uri] \
7496 unless { hdr_beg(host) -i www }
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007497
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007498 See section 7 about ACL usage.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007499
7500
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007501redisp (deprecated)
7502redispatch (deprecated)
7503 Enable or disable session redistribution in case of connection failure
7504 May be used in sections: defaults | frontend | listen | backend
7505 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007506 Arguments : none
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007507
7508 In HTTP mode, if a server designated by a cookie is down, clients may
7509 definitely stick to it because they cannot flush the cookie, so they will not
7510 be able to access the service anymore.
7511
7512 Specifying "redispatch" will allow the proxy to break their persistence and
7513 redistribute them to a working server.
7514
7515 It also allows to retry last connection to another server in case of multiple
7516 connection failures. Of course, it requires having "retries" set to a nonzero
7517 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01007518
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007519 This form is deprecated, do not use it in any new configuration, use the new
7520 "option redispatch" instead.
7521
7522 See also : "option redispatch"
7523
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007524
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01007525reqadd <string> [{if | unless} <cond>]
Willy Tarreau303c0352008-01-17 19:01:39 +01007526 Add a header at the end of the HTTP request
7527 May be used in sections : defaults | frontend | listen | backend
7528 no | yes | yes | yes
7529 Arguments :
7530 <string> is the complete line to be added. Any space or known delimiter
7531 must be escaped using a backslash ('\'). Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007532 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01007533
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01007534 <cond> is an optional matching condition built from ACLs. It makes it
7535 possible to ignore this rule when other conditions are not met.
7536
Willy Tarreau303c0352008-01-17 19:01:39 +01007537 A new line consisting in <string> followed by a line feed will be added after
7538 the last header of an HTTP request.
7539
7540 Header transformations only apply to traffic which passes through HAProxy,
7541 and not to traffic generated by HAProxy, such as health-checks or error
7542 responses.
7543
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01007544 Example : add "X-Proto: SSL" to requests coming via port 81
7545 acl is-ssl dst_port 81
7546 reqadd X-Proto:\ SSL if is-ssl
7547
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007548 See also: "rspadd", "http-request", section 6 about HTTP header manipulation,
7549 and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007550
7551
Willy Tarreau5321c422010-01-28 20:35:13 +01007552reqallow <search> [{if | unless} <cond>]
7553reqiallow <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007554 Definitely allow an HTTP request if a line matches a regular expression
7555 May be used in sections : defaults | frontend | listen | backend
7556 no | yes | yes | yes
7557 Arguments :
7558 <search> is the regular expression applied to HTTP headers and to the
7559 request line. This is an extended regular expression. Parenthesis
7560 grouping is supported and no preliminary backslash is required.
7561 Any space or known delimiter must be escaped using a backslash
7562 ('\'). The pattern applies to a full line at a time. The
7563 "reqallow" keyword strictly matches case while "reqiallow"
7564 ignores case.
7565
Willy Tarreau5321c422010-01-28 20:35:13 +01007566 <cond> is an optional matching condition built from ACLs. It makes it
7567 possible to ignore this rule when other conditions are not met.
7568
Willy Tarreau303c0352008-01-17 19:01:39 +01007569 A request containing any line which matches extended regular expression
7570 <search> will mark the request as allowed, even if any later test would
7571 result in a deny. The test applies both to the request line and to request
7572 headers. Keep in mind that URLs in request line are case-sensitive while
Willy Tarreaud72758d2010-01-12 10:42:19 +01007573 header names are not.
Willy Tarreau303c0352008-01-17 19:01:39 +01007574
7575 It is easier, faster and more powerful to use ACLs to write access policies.
7576 Reqdeny, reqallow and reqpass should be avoided in new designs.
7577
7578 Example :
7579 # allow www.* but refuse *.local
7580 reqiallow ^Host:\ www\.
7581 reqideny ^Host:\ .*\.local
7582
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007583 See also: "reqdeny", "block", "http-request", section 6 about HTTP header
7584 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007585
7586
Willy Tarreau5321c422010-01-28 20:35:13 +01007587reqdel <search> [{if | unless} <cond>]
7588reqidel <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007589 Delete all headers matching a regular expression in an HTTP request
7590 May be used in sections : defaults | frontend | listen | backend
7591 no | yes | yes | yes
7592 Arguments :
7593 <search> is the regular expression applied to HTTP headers and to the
7594 request line. This is an extended regular expression. Parenthesis
7595 grouping is supported and no preliminary backslash is required.
7596 Any space or known delimiter must be escaped using a backslash
7597 ('\'). The pattern applies to a full line at a time. The "reqdel"
7598 keyword strictly matches case while "reqidel" ignores case.
7599
Willy Tarreau5321c422010-01-28 20:35:13 +01007600 <cond> is an optional matching condition built from ACLs. It makes it
7601 possible to ignore this rule when other conditions are not met.
7602
Willy Tarreau303c0352008-01-17 19:01:39 +01007603 Any header line matching extended regular expression <search> in the request
7604 will be completely deleted. Most common use of this is to remove unwanted
7605 and/or dangerous headers or cookies from a request before passing it to the
7606 next servers.
7607
7608 Header transformations only apply to traffic which passes through HAProxy,
7609 and not to traffic generated by HAProxy, such as health-checks or error
7610 responses. Keep in mind that header names are not case-sensitive.
7611
7612 Example :
7613 # remove X-Forwarded-For header and SERVER cookie
7614 reqidel ^X-Forwarded-For:.*
7615 reqidel ^Cookie:.*SERVER=
7616
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007617 See also: "reqadd", "reqrep", "rspdel", "http-request", section 6 about
7618 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007619
7620
Willy Tarreau5321c422010-01-28 20:35:13 +01007621reqdeny <search> [{if | unless} <cond>]
7622reqideny <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007623 Deny an HTTP request if a line matches a regular expression
7624 May be used in sections : defaults | frontend | listen | backend
7625 no | yes | yes | yes
7626 Arguments :
7627 <search> is the regular expression applied to HTTP headers and to the
7628 request line. This is an extended regular expression. Parenthesis
7629 grouping is supported and no preliminary backslash is required.
7630 Any space or known delimiter must be escaped using a backslash
7631 ('\'). The pattern applies to a full line at a time. The
7632 "reqdeny" keyword strictly matches case while "reqideny" ignores
7633 case.
7634
Willy Tarreau5321c422010-01-28 20:35:13 +01007635 <cond> is an optional matching condition built from ACLs. It makes it
7636 possible to ignore this rule when other conditions are not met.
7637
Willy Tarreau303c0352008-01-17 19:01:39 +01007638 A request containing any line which matches extended regular expression
7639 <search> will mark the request as denied, even if any later test would
7640 result in an allow. The test applies both to the request line and to request
7641 headers. Keep in mind that URLs in request line are case-sensitive while
Willy Tarreaud72758d2010-01-12 10:42:19 +01007642 header names are not.
Willy Tarreau303c0352008-01-17 19:01:39 +01007643
Willy Tarreauced27012008-01-17 20:35:34 +01007644 A denied request will generate an "HTTP 403 forbidden" response once the
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01007645 complete request has been parsed. This is consistent with what is practiced
Willy Tarreaud72758d2010-01-12 10:42:19 +01007646 using ACLs.
Willy Tarreauced27012008-01-17 20:35:34 +01007647
Willy Tarreau303c0352008-01-17 19:01:39 +01007648 It is easier, faster and more powerful to use ACLs to write access policies.
7649 Reqdeny, reqallow and reqpass should be avoided in new designs.
7650
7651 Example :
7652 # refuse *.local, then allow www.*
7653 reqideny ^Host:\ .*\.local
7654 reqiallow ^Host:\ www\.
7655
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007656 See also: "reqallow", "rspdeny", "block", "http-request", section 6 about
7657 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007658
7659
Willy Tarreau5321c422010-01-28 20:35:13 +01007660reqpass <search> [{if | unless} <cond>]
7661reqipass <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007662 Ignore any HTTP request line matching a regular expression in next rules
7663 May be used in sections : defaults | frontend | listen | backend
7664 no | yes | yes | yes
7665 Arguments :
7666 <search> is the regular expression applied to HTTP headers and to the
7667 request line. This is an extended regular expression. Parenthesis
7668 grouping is supported and no preliminary backslash is required.
7669 Any space or known delimiter must be escaped using a backslash
7670 ('\'). The pattern applies to a full line at a time. The
7671 "reqpass" keyword strictly matches case while "reqipass" ignores
7672 case.
7673
Willy Tarreau5321c422010-01-28 20:35:13 +01007674 <cond> is an optional matching condition built from ACLs. It makes it
7675 possible to ignore this rule when other conditions are not met.
7676
Willy Tarreau303c0352008-01-17 19:01:39 +01007677 A request containing any line which matches extended regular expression
7678 <search> will skip next rules, without assigning any deny or allow verdict.
7679 The test applies both to the request line and to request headers. Keep in
7680 mind that URLs in request line are case-sensitive while header names are not.
7681
7682 It is easier, faster and more powerful to use ACLs to write access policies.
7683 Reqdeny, reqallow and reqpass should be avoided in new designs.
7684
7685 Example :
7686 # refuse *.local, then allow www.*, but ignore "www.private.local"
7687 reqipass ^Host:\ www.private\.local
7688 reqideny ^Host:\ .*\.local
7689 reqiallow ^Host:\ www\.
7690
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007691 See also: "reqallow", "reqdeny", "block", "http-request", section 6 about
7692 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007693
7694
Willy Tarreau5321c422010-01-28 20:35:13 +01007695reqrep <search> <string> [{if | unless} <cond>]
7696reqirep <search> <string> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007697 Replace a regular expression with a string in an HTTP request line
7698 May be used in sections : defaults | frontend | listen | backend
7699 no | yes | yes | yes
7700 Arguments :
7701 <search> is the regular expression applied to HTTP headers and to the
7702 request line. This is an extended regular expression. Parenthesis
7703 grouping is supported and no preliminary backslash is required.
7704 Any space or known delimiter must be escaped using a backslash
7705 ('\'). The pattern applies to a full line at a time. The "reqrep"
7706 keyword strictly matches case while "reqirep" ignores case.
7707
7708 <string> is the complete line to be added. Any space or known delimiter
7709 must be escaped using a backslash ('\'). References to matched
7710 pattern groups are possible using the common \N form, with N
7711 being a single digit between 0 and 9. Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007712 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01007713
Willy Tarreau5321c422010-01-28 20:35:13 +01007714 <cond> is an optional matching condition built from ACLs. It makes it
7715 possible to ignore this rule when other conditions are not met.
7716
Willy Tarreau303c0352008-01-17 19:01:39 +01007717 Any line matching extended regular expression <search> in the request (both
7718 the request line and header lines) will be completely replaced with <string>.
7719 Most common use of this is to rewrite URLs or domain names in "Host" headers.
7720
7721 Header transformations only apply to traffic which passes through HAProxy,
7722 and not to traffic generated by HAProxy, such as health-checks or error
7723 responses. Note that for increased readability, it is suggested to add enough
7724 spaces between the request and the response. Keep in mind that URLs in
7725 request line are case-sensitive while header names are not.
7726
7727 Example :
7728 # replace "/static/" with "/" at the beginning of any request path.
Dmitry Sivachenko7823de32012-05-16 14:00:26 +04007729 reqrep ^([^\ :]*)\ /static/(.*) \1\ /\2
Willy Tarreau303c0352008-01-17 19:01:39 +01007730 # replace "www.mydomain.com" with "www" in the host name.
7731 reqirep ^Host:\ www.mydomain.com Host:\ www
7732
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007733 See also: "reqadd", "reqdel", "rsprep", "tune.bufsize", "http-request",
7734 section 6 about HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007735
7736
Willy Tarreau5321c422010-01-28 20:35:13 +01007737reqtarpit <search> [{if | unless} <cond>]
7738reqitarpit <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007739 Tarpit an HTTP request containing a line matching a regular expression
7740 May be used in sections : defaults | frontend | listen | backend
7741 no | yes | yes | yes
7742 Arguments :
7743 <search> is the regular expression applied to HTTP headers and to the
7744 request line. This is an extended regular expression. Parenthesis
7745 grouping is supported and no preliminary backslash is required.
7746 Any space or known delimiter must be escaped using a backslash
7747 ('\'). The pattern applies to a full line at a time. The
7748 "reqtarpit" keyword strictly matches case while "reqitarpit"
7749 ignores case.
7750
Willy Tarreau5321c422010-01-28 20:35:13 +01007751 <cond> is an optional matching condition built from ACLs. It makes it
7752 possible to ignore this rule when other conditions are not met.
7753
Willy Tarreau303c0352008-01-17 19:01:39 +01007754 A request containing any line which matches extended regular expression
7755 <search> will be tarpitted, which means that it will connect to nowhere, will
Willy Tarreauced27012008-01-17 20:35:34 +01007756 be kept open for a pre-defined time, then will return an HTTP error 500 so
7757 that the attacker does not suspect it has been tarpitted. The status 500 will
7758 be reported in the logs, but the completion flags will indicate "PT". The
Willy Tarreau303c0352008-01-17 19:01:39 +01007759 delay is defined by "timeout tarpit", or "timeout connect" if the former is
7760 not set.
7761
7762 The goal of the tarpit is to slow down robots attacking servers with
7763 identifiable requests. Many robots limit their outgoing number of connections
7764 and stay connected waiting for a reply which can take several minutes to
7765 come. Depending on the environment and attack, it may be particularly
7766 efficient at reducing the load on the network and firewalls.
7767
Willy Tarreau5321c422010-01-28 20:35:13 +01007768 Examples :
Davor Ocelice9ed2812017-12-25 17:49:28 +01007769 # ignore user-agents reporting any flavor of "Mozilla" or "MSIE", but
Willy Tarreau303c0352008-01-17 19:01:39 +01007770 # block all others.
7771 reqipass ^User-Agent:\.*(Mozilla|MSIE)
7772 reqitarpit ^User-Agent:
7773
Willy Tarreau5321c422010-01-28 20:35:13 +01007774 # block bad guys
7775 acl badguys src 10.1.0.3 172.16.13.20/28
7776 reqitarpit . if badguys
7777
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007778 See also: "reqallow", "reqdeny", "reqpass", "http-request", section 6
7779 about HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007780
7781
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02007782retries <value>
7783 Set the number of retries to perform on a server after a connection failure
7784 May be used in sections: defaults | frontend | listen | backend
7785 yes | no | yes | yes
7786 Arguments :
7787 <value> is the number of times a connection attempt should be retried on
7788 a server when a connection either is refused or times out. The
7789 default value is 3.
7790
7791 It is important to understand that this value applies to the number of
7792 connection attempts, not full requests. When a connection has effectively
7793 been established to a server, there will be no more retry.
7794
7795 In order to avoid immediate reconnections to a server which is restarting,
Joseph Lynch726ab712015-05-11 23:25:34 -07007796 a turn-around timer of min("timeout connect", one second) is applied before
7797 a retry occurs.
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02007798
7799 When "option redispatch" is set, the last retry may be performed on another
7800 server even if a cookie references a different server.
7801
7802 See also : "option redispatch"
7803
7804
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007805rspadd <string> [{if | unless} <cond>]
Willy Tarreau303c0352008-01-17 19:01:39 +01007806 Add a header at the end of the HTTP response
7807 May be used in sections : defaults | frontend | listen | backend
7808 no | yes | yes | yes
7809 Arguments :
7810 <string> is the complete line to be added. Any space or known delimiter
7811 must be escaped using a backslash ('\'). Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007812 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01007813
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007814 <cond> is an optional matching condition built from ACLs. It makes it
7815 possible to ignore this rule when other conditions are not met.
7816
Willy Tarreau303c0352008-01-17 19:01:39 +01007817 A new line consisting in <string> followed by a line feed will be added after
7818 the last header of an HTTP response.
7819
7820 Header transformations only apply to traffic which passes through HAProxy,
7821 and not to traffic generated by HAProxy, such as health-checks or error
7822 responses.
7823
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007824 See also: "rspdel" "reqadd", "http-response", section 6 about HTTP header
7825 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007826
7827
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007828rspdel <search> [{if | unless} <cond>]
7829rspidel <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007830 Delete all headers matching a regular expression in an HTTP response
7831 May be used in sections : defaults | frontend | listen | backend
7832 no | yes | yes | yes
7833 Arguments :
7834 <search> is the regular expression applied to HTTP headers and to the
7835 response line. This is an extended regular expression, so
7836 parenthesis grouping is supported and no preliminary backslash
7837 is required. Any space or known delimiter must be escaped using
7838 a backslash ('\'). The pattern applies to a full line at a time.
7839 The "rspdel" keyword strictly matches case while "rspidel"
7840 ignores case.
7841
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007842 <cond> is an optional matching condition built from ACLs. It makes it
7843 possible to ignore this rule when other conditions are not met.
7844
Willy Tarreau303c0352008-01-17 19:01:39 +01007845 Any header line matching extended regular expression <search> in the response
7846 will be completely deleted. Most common use of this is to remove unwanted
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02007847 and/or sensitive headers or cookies from a response before passing it to the
Willy Tarreau303c0352008-01-17 19:01:39 +01007848 client.
7849
7850 Header transformations only apply to traffic which passes through HAProxy,
7851 and not to traffic generated by HAProxy, such as health-checks or error
7852 responses. Keep in mind that header names are not case-sensitive.
7853
7854 Example :
7855 # remove the Server header from responses
Willy Tarreau5e80e022013-05-25 08:31:25 +02007856 rspidel ^Server:.*
Willy Tarreau303c0352008-01-17 19:01:39 +01007857
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007858 See also: "rspadd", "rsprep", "reqdel", "http-response", section 6 about
7859 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007860
7861
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007862rspdeny <search> [{if | unless} <cond>]
7863rspideny <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007864 Block an HTTP response if a line matches a regular expression
7865 May be used in sections : defaults | frontend | listen | backend
7866 no | yes | yes | yes
7867 Arguments :
7868 <search> is the regular expression applied to HTTP headers and to the
7869 response line. This is an extended regular expression, so
7870 parenthesis grouping is supported and no preliminary backslash
7871 is required. Any space or known delimiter must be escaped using
7872 a backslash ('\'). The pattern applies to a full line at a time.
7873 The "rspdeny" keyword strictly matches case while "rspideny"
7874 ignores case.
7875
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007876 <cond> is an optional matching condition built from ACLs. It makes it
7877 possible to ignore this rule when other conditions are not met.
7878
Willy Tarreau303c0352008-01-17 19:01:39 +01007879 A response containing any line which matches extended regular expression
7880 <search> will mark the request as denied. The test applies both to the
7881 response line and to response headers. Keep in mind that header names are not
7882 case-sensitive.
7883
7884 Main use of this keyword is to prevent sensitive information leak and to
Willy Tarreauced27012008-01-17 20:35:34 +01007885 block the response before it reaches the client. If a response is denied, it
7886 will be replaced with an HTTP 502 error so that the client never retrieves
7887 any sensitive data.
Willy Tarreau303c0352008-01-17 19:01:39 +01007888
7889 It is easier, faster and more powerful to use ACLs to write access policies.
7890 Rspdeny should be avoided in new designs.
7891
7892 Example :
7893 # Ensure that no content type matching ms-word will leak
7894 rspideny ^Content-type:\.*/ms-word
7895
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007896 See also: "reqdeny", "acl", "block", "http-response", section 6 about
7897 HTTP header manipulation and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007898
7899
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007900rsprep <search> <string> [{if | unless} <cond>]
7901rspirep <search> <string> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007902 Replace a regular expression with a string in an HTTP response line
7903 May be used in sections : defaults | frontend | listen | backend
7904 no | yes | yes | yes
7905 Arguments :
7906 <search> is the regular expression applied to HTTP headers and to the
7907 response line. This is an extended regular expression, so
7908 parenthesis grouping is supported and no preliminary backslash
7909 is required. Any space or known delimiter must be escaped using
7910 a backslash ('\'). The pattern applies to a full line at a time.
7911 The "rsprep" keyword strictly matches case while "rspirep"
7912 ignores case.
7913
7914 <string> is the complete line to be added. Any space or known delimiter
7915 must be escaped using a backslash ('\'). References to matched
7916 pattern groups are possible using the common \N form, with N
7917 being a single digit between 0 and 9. Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007918 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01007919
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007920 <cond> is an optional matching condition built from ACLs. It makes it
7921 possible to ignore this rule when other conditions are not met.
7922
Willy Tarreau303c0352008-01-17 19:01:39 +01007923 Any line matching extended regular expression <search> in the response (both
7924 the response line and header lines) will be completely replaced with
7925 <string>. Most common use of this is to rewrite Location headers.
7926
7927 Header transformations only apply to traffic which passes through HAProxy,
7928 and not to traffic generated by HAProxy, such as health-checks or error
7929 responses. Note that for increased readability, it is suggested to add enough
7930 spaces between the request and the response. Keep in mind that header names
7931 are not case-sensitive.
7932
7933 Example :
7934 # replace "Location: 127.0.0.1:8080" with "Location: www.mydomain.com"
7935 rspirep ^Location:\ 127.0.0.1:8080 Location:\ www.mydomain.com
7936
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007937 See also: "rspadd", "rspdel", "reqrep", "http-response", section 6 about
7938 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007939
7940
David du Colombier486df472011-03-17 10:40:26 +01007941server <name> <address>[:[port]] [param*]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007942 Declare a server in a backend
7943 May be used in sections : defaults | frontend | listen | backend
7944 no | no | yes | yes
7945 Arguments :
7946 <name> is the internal name assigned to this server. This name will
Davor Ocelice9ed2812017-12-25 17:49:28 +01007947 appear in logs and alerts. If "http-send-name-header" is
Mark Lamourinec2247f02012-01-04 13:02:01 -05007948 set, it will be added to the request header sent to the server.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007949
David du Colombier486df472011-03-17 10:40:26 +01007950 <address> is the IPv4 or IPv6 address of the server. Alternatively, a
7951 resolvable hostname is supported, but this name will be resolved
7952 during start-up. Address "0.0.0.0" or "*" has a special meaning.
7953 It indicates that the connection will be forwarded to the same IP
Willy Tarreaud669a4f2010-07-13 14:49:50 +02007954 address as the one from the client connection. This is useful in
7955 transparent proxy architectures where the client's connection is
7956 intercepted and haproxy must forward to the original destination
7957 address. This is more or less what the "transparent" keyword does
7958 except that with a server it's possible to limit concurrency and
Willy Tarreau24709282013-03-10 21:32:12 +01007959 to report statistics. Optionally, an address family prefix may be
7960 used before the address to force the family regardless of the
7961 address format, which can be useful to specify a path to a unix
7962 socket with no slash ('/'). Currently supported prefixes are :
7963 - 'ipv4@' -> address is always IPv4
7964 - 'ipv6@' -> address is always IPv6
7965 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02007966 - 'abns@' -> address is in abstract namespace (Linux only)
William Lallemand2fe7dd02018-09-11 16:51:29 +02007967 - 'sockpair@' -> address is the FD of a connected unix
7968 socket or of a socketpair. During a connection, the
7969 backend creates a pair of connected sockets, and passes
7970 one of them over the FD. The bind part will use the
7971 received socket as the client FD. Should be used
7972 carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02007973 You may want to reference some environment variables in the
7974 address parameter, see section 2.3 about environment
Willy Tarreau6a031d12016-11-07 19:42:35 +01007975 variables. The "init-addr" setting can be used to modify the way
7976 IP addresses should be resolved upon startup.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007977
Willy Tarreaub6205fd2012-09-24 12:27:33 +02007978 <port> is an optional port specification. If set, all connections will
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007979 be sent to this port. If unset, the same port the client
7980 connected to will be used. The port may also be prefixed by a "+"
7981 or a "-". In this case, the server's port will be determined by
7982 adding this value to the client's port.
7983
7984 <param*> is a list of parameters for this server. The "server" keywords
7985 accepts an important number of options and has a complete section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007986 dedicated to it. Please refer to section 5 for more details.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007987
7988 Examples :
7989 server first 10.1.1.1:1080 cookie first check inter 1000
7990 server second 10.1.1.2:1080 cookie second check inter 1000
Willy Tarreau24709282013-03-10 21:32:12 +01007991 server transp ipv4@
William Lallemandb2f07452015-05-12 14:27:13 +02007992 server backup "${SRV_BACKUP}:1080" backup
7993 server www1_dc1 "${LAN_DC1}.101:80"
7994 server www1_dc2 "${LAN_DC2}.101:80"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007995
Willy Tarreau55dcaf62015-09-27 15:03:15 +02007996 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
7997 sun_path length is used for the address length. Some other programs
7998 such as socat use the string length only by default. Pass the option
7999 ",unix-tightsocklen=0" to any abstract socket definition in socat to
8000 make it compatible with HAProxy's.
8001
Mark Lamourinec2247f02012-01-04 13:02:01 -05008002 See also: "default-server", "http-send-name-header" and section 5 about
8003 server options
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008004
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02008005server-state-file-name [<file>]
8006 Set the server state file to read, load and apply to servers available in
8007 this backend. It only applies when the directive "load-server-state-from-file"
8008 is set to "local". When <file> is not provided or if this directive is not
8009 set, then backend name is used. If <file> starts with a slash '/', then it is
8010 considered as an absolute path. Otherwise, <file> is concatenated to the
8011 global directive "server-state-file-base".
8012
8013 Example: the minimal configuration below would make HAProxy look for the
8014 state server file '/etc/haproxy/states/bk':
8015
8016 global
8017 server-state-file-base /etc/haproxy/states
8018
8019 backend bk
8020 load-server-state-from-file
8021
8022 See also: "server-state-file-base", "load-server-state-from-file", and
8023 "show servers state"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008024
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02008025server-template <prefix> <num | range> <fqdn>[:<port>] [params*]
8026 Set a template to initialize servers with shared parameters.
8027 The names of these servers are built from <prefix> and <num | range> parameters.
8028 May be used in sections : defaults | frontend | listen | backend
8029 no | no | yes | yes
8030
8031 Arguments:
8032 <prefix> A prefix for the server names to be built.
8033
8034 <num | range>
8035 If <num> is provided, this template initializes <num> servers
8036 with 1 up to <num> as server name suffixes. A range of numbers
8037 <num_low>-<num_high> may also be used to use <num_low> up to
8038 <num_high> as server name suffixes.
8039
8040 <fqdn> A FQDN for all the servers this template initializes.
8041
8042 <port> Same meaning as "server" <port> argument (see "server" keyword).
8043
8044 <params*>
8045 Remaining server parameters among all those supported by "server"
8046 keyword.
8047
8048 Examples:
8049 # Initializes 3 servers with srv1, srv2 and srv3 as names,
8050 # google.com as FQDN, and health-check enabled.
8051 server-template srv 1-3 google.com:80 check
8052
8053 # or
8054 server-template srv 3 google.com:80 check
8055
8056 # would be equivalent to:
8057 server srv1 google.com:80 check
8058 server srv2 google.com:80 check
8059 server srv3 google.com:80 check
8060
8061
8062
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008063source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +02008064source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreaud53f96b2009-02-04 18:46:54 +01008065source <addr>[:<port>] [interface <name>]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008066 Set the source address for outgoing connections
8067 May be used in sections : defaults | frontend | listen | backend
8068 yes | no | yes | yes
8069 Arguments :
8070 <addr> is the IPv4 address HAProxy will bind to before connecting to a
8071 server. This address is also used as a source for health checks.
Willy Tarreau24709282013-03-10 21:32:12 +01008072
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008073 The default value of 0.0.0.0 means that the system will select
Willy Tarreau24709282013-03-10 21:32:12 +01008074 the most appropriate address to reach its destination. Optionally
8075 an address family prefix may be used before the address to force
8076 the family regardless of the address format, which can be useful
8077 to specify a path to a unix socket with no slash ('/'). Currently
8078 supported prefixes are :
8079 - 'ipv4@' -> address is always IPv4
8080 - 'ipv6@' -> address is always IPv6
8081 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02008082 - 'abns@' -> address is in abstract namespace (Linux only)
Cyril Bonté307ee1e2015-09-28 23:16:06 +02008083 You may want to reference some environment variables in the
8084 address parameter, see section 2.3 about environment variables.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008085
8086 <port> is an optional port. It is normally not needed but may be useful
8087 in some very specific contexts. The default value of zero means
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02008088 the system will select a free port. Note that port ranges are not
8089 supported in the backend. If you want to force port ranges, you
8090 have to specify them on each "server" line.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008091
8092 <addr2> is the IP address to present to the server when connections are
8093 forwarded in full transparent proxy mode. This is currently only
8094 supported on some patched Linux kernels. When this address is
8095 specified, clients connecting to the server will be presented
8096 with this address, while health checks will still use the address
8097 <addr>.
8098
8099 <port2> is the optional port to present to the server when connections
8100 are forwarded in full transparent proxy mode (see <addr2> above).
8101 The default value of zero means the system will select a free
8102 port.
8103
Willy Tarreaubce70882009-09-07 11:51:47 +02008104 <hdr> is the name of a HTTP header in which to fetch the IP to bind to.
8105 This is the name of a comma-separated header list which can
8106 contain multiple IP addresses. By default, the last occurrence is
8107 used. This is designed to work with the X-Forwarded-For header
Baptiste Assmannea3e73b2013-02-02 23:47:49 +01008108 and to automatically bind to the client's IP address as seen
Willy Tarreaubce70882009-09-07 11:51:47 +02008109 by previous proxy, typically Stunnel. In order to use another
8110 occurrence from the last one, please see the <occ> parameter
8111 below. When the header (or occurrence) is not found, no binding
8112 is performed so that the proxy's default IP address is used. Also
8113 keep in mind that the header name is case insensitive, as for any
8114 HTTP header.
8115
8116 <occ> is the occurrence number of a value to be used in a multi-value
8117 header. This is to be used in conjunction with "hdr_ip(<hdr>)",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04008118 in order to specify which occurrence to use for the source IP
Willy Tarreaubce70882009-09-07 11:51:47 +02008119 address. Positive values indicate a position from the first
8120 occurrence, 1 being the first one. Negative values indicate
8121 positions relative to the last one, -1 being the last one. This
8122 is helpful for situations where an X-Forwarded-For header is set
8123 at the entry point of an infrastructure and must be used several
8124 proxy layers away. When this value is not specified, -1 is
8125 assumed. Passing a zero here disables the feature.
8126
Willy Tarreaud53f96b2009-02-04 18:46:54 +01008127 <name> is an optional interface name to which to bind to for outgoing
8128 traffic. On systems supporting this features (currently, only
8129 Linux), this allows one to bind all traffic to the server to
8130 this interface even if it is not the one the system would select
8131 based on routing tables. This should be used with extreme care.
8132 Note that using this option requires root privileges.
8133
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008134 The "source" keyword is useful in complex environments where a specific
8135 address only is allowed to connect to the servers. It may be needed when a
8136 private address must be used through a public gateway for instance, and it is
8137 known that the system cannot determine the adequate source address by itself.
8138
8139 An extension which is available on certain patched Linux kernels may be used
8140 through the "usesrc" optional keyword. It makes it possible to connect to the
8141 servers with an IP address which does not belong to the system itself. This
8142 is called "full transparent proxy mode". For this to work, the destination
8143 servers have to route their traffic back to this address through the machine
8144 running HAProxy, and IP forwarding must generally be enabled on this machine.
8145
8146 In this "full transparent proxy" mode, it is possible to force a specific IP
8147 address to be presented to the servers. This is not much used in fact. A more
8148 common use is to tell HAProxy to present the client's IP address. For this,
8149 there are two methods :
8150
8151 - present the client's IP and port addresses. This is the most transparent
8152 mode, but it can cause problems when IP connection tracking is enabled on
8153 the machine, because a same connection may be seen twice with different
8154 states. However, this solution presents the huge advantage of not
8155 limiting the system to the 64k outgoing address+port couples, because all
8156 of the client ranges may be used.
8157
8158 - present only the client's IP address and select a spare port. This
8159 solution is still quite elegant but slightly less transparent (downstream
8160 firewalls logs will not match upstream's). It also presents the downside
8161 of limiting the number of concurrent connections to the usual 64k ports.
8162 However, since the upstream and downstream ports are different, local IP
8163 connection tracking on the machine will not be upset by the reuse of the
8164 same session.
8165
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008166 This option sets the default source for all servers in the backend. It may
8167 also be specified in a "defaults" section. Finer source address specification
8168 is possible at the server level using the "source" server option. Refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008169 section 5 for more information.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008170
Baptiste Assmann91bd3372015-07-17 21:59:42 +02008171 In order to work, "usesrc" requires root privileges.
8172
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008173 Examples :
8174 backend private
8175 # Connect to the servers using our 192.168.1.200 source address
8176 source 192.168.1.200
8177
8178 backend transparent_ssl1
8179 # Connect to the SSL farm from the client's source address
8180 source 192.168.1.200 usesrc clientip
8181
8182 backend transparent_ssl2
8183 # Connect to the SSL farm from the client's source address and port
8184 # not recommended if IP conntrack is present on the local machine.
8185 source 192.168.1.200 usesrc client
8186
8187 backend transparent_ssl3
8188 # Connect to the SSL farm from the client's source address. It
8189 # is more conntrack-friendly.
8190 source 192.168.1.200 usesrc clientip
8191
8192 backend transparent_smtp
8193 # Connect to the SMTP farm from the client's source address/port
8194 # with Tproxy version 4.
8195 source 0.0.0.0 usesrc clientip
8196
Willy Tarreaubce70882009-09-07 11:51:47 +02008197 backend transparent_http
8198 # Connect to the servers using the client's IP as seen by previous
8199 # proxy.
8200 source 0.0.0.0 usesrc hdr_ip(x-forwarded-for,-1)
8201
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008202 See also : the "source" server option in section 5, the Tproxy patches for
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008203 the Linux kernel on www.balabit.com, the "bind" keyword.
8204
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008205
Willy Tarreau844e3c52008-01-11 16:28:18 +01008206srvtimeout <timeout> (deprecated)
8207 Set the maximum inactivity time on the server side.
8208 May be used in sections : defaults | frontend | listen | backend
8209 yes | no | yes | yes
8210 Arguments :
8211 <timeout> is the timeout value specified in milliseconds by default, but
8212 can be in any other unit if the number is suffixed by the unit,
8213 as explained at the top of this document.
8214
8215 The inactivity timeout applies when the server is expected to acknowledge or
8216 send data. In HTTP mode, this timeout is particularly important to consider
8217 during the first phase of the server's response, when it has to send the
8218 headers, as it directly represents the server's processing time for the
8219 request. To find out what value to put there, it's often good to start with
8220 what would be considered as unacceptable response times, then check the logs
8221 to observe the response time distribution, and adjust the value accordingly.
8222
8223 The value is specified in milliseconds by default, but can be in any other
8224 unit if the number is suffixed by the unit, as specified at the top of this
8225 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
8226 recommended that the client timeout remains equal to the server timeout in
8227 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01008228 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +01008229 packet losses by specifying timeouts that are slightly above multiples of 3
Davor Ocelice9ed2812017-12-25 17:49:28 +01008230 seconds (e.g. 4 or 5 seconds minimum).
Willy Tarreau844e3c52008-01-11 16:28:18 +01008231
8232 This parameter is specific to backends, but can be specified once for all in
8233 "defaults" sections. This is in fact one of the easiest solutions not to
8234 forget about it. An unspecified timeout results in an infinite timeout, which
8235 is not recommended. Such a usage is accepted and works but reports a warning
8236 during startup because it may results in accumulation of expired sessions in
8237 the system if the system's timeouts are not configured either.
8238
8239 This parameter is provided for compatibility but is currently deprecated.
8240 Please use "timeout server" instead.
8241
Willy Tarreauce887fd2012-05-12 12:50:00 +02008242 See also : "timeout server", "timeout tunnel", "timeout client" and
8243 "clitimeout".
Willy Tarreau844e3c52008-01-11 16:28:18 +01008244
8245
Cyril Bonté66c327d2010-10-12 00:14:37 +02008246stats admin { if | unless } <cond>
8247 Enable statistics admin level if/unless a condition is matched
8248 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008249 no | yes | yes | yes
Cyril Bonté66c327d2010-10-12 00:14:37 +02008250
8251 This statement enables the statistics admin level if/unless a condition is
8252 matched.
8253
8254 The admin level allows to enable/disable servers from the web interface. By
8255 default, statistics page is read-only for security reasons.
8256
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008257 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8258 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008259 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008260
Cyril Bonté23b39d92011-02-10 22:54:44 +01008261 Currently, the POST request is limited to the buffer size minus the reserved
8262 buffer space, which means that if the list of servers is too long, the
8263 request won't be processed. It is recommended to alter few servers at a
8264 time.
Cyril Bonté66c327d2010-10-12 00:14:37 +02008265
8266 Example :
8267 # statistics admin level only for localhost
8268 backend stats_localhost
8269 stats enable
8270 stats admin if LOCALHOST
8271
8272 Example :
8273 # statistics admin level always enabled because of the authentication
8274 backend stats_auth
8275 stats enable
8276 stats auth admin:AdMiN123
8277 stats admin if TRUE
8278
8279 Example :
8280 # statistics admin level depends on the authenticated user
8281 userlist stats-auth
8282 group admin users admin
8283 user admin insecure-password AdMiN123
8284 group readonly users haproxy
8285 user haproxy insecure-password haproxy
8286
8287 backend stats_auth
8288 stats enable
8289 acl AUTH http_auth(stats-auth)
8290 acl AUTH_ADMIN http_auth_group(stats-auth) admin
8291 stats http-request auth unless AUTH
8292 stats admin if AUTH_ADMIN
8293
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008294 See also : "stats enable", "stats auth", "stats http-request", "nbproc",
8295 "bind-process", section 3.4 about userlists and section 7 about
8296 ACL usage.
Cyril Bonté66c327d2010-10-12 00:14:37 +02008297
8298
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008299stats auth <user>:<passwd>
8300 Enable statistics with authentication and grant access to an account
8301 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008302 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008303 Arguments :
8304 <user> is a user name to grant access to
8305
8306 <passwd> is the cleartext password associated to this user
8307
8308 This statement enables statistics with default settings, and restricts access
8309 to declared users only. It may be repeated as many times as necessary to
8310 allow as many users as desired. When a user tries to access the statistics
8311 without a valid account, a "401 Forbidden" response will be returned so that
8312 the browser asks the user to provide a valid user and password. The real
8313 which will be returned to the browser is configurable using "stats realm".
8314
8315 Since the authentication method is HTTP Basic Authentication, the passwords
8316 circulate in cleartext on the network. Thus, it was decided that the
8317 configuration file would also use cleartext passwords to remind the users
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02008318 that those ones should not be sensitive and not shared with any other account.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008319
8320 It is also possible to reduce the scope of the proxies which appear in the
8321 report using "stats scope".
8322
8323 Though this statement alone is enough to enable statistics reporting, it is
8324 recommended to set all other settings in order to avoid relying on default
8325 unobvious parameters.
8326
8327 Example :
8328 # public access (limited to this backend only)
8329 backend public_www
8330 server srv1 192.168.0.1:80
8331 stats enable
8332 stats hide-version
8333 stats scope .
8334 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008335 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008336 stats auth admin1:AdMiN123
8337 stats auth admin2:AdMiN321
8338
8339 # internal monitoring access (unlimited)
8340 backend private_monitoring
8341 stats enable
8342 stats uri /admin?stats
8343 stats refresh 5s
8344
8345 See also : "stats enable", "stats realm", "stats scope", "stats uri"
8346
8347
8348stats enable
8349 Enable statistics reporting with default settings
8350 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008351 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008352 Arguments : none
8353
8354 This statement enables statistics reporting with default settings defined
8355 at build time. Unless stated otherwise, these settings are used :
8356 - stats uri : /haproxy?stats
8357 - stats realm : "HAProxy Statistics"
8358 - stats auth : no authentication
8359 - stats scope : no restriction
8360
8361 Though this statement alone is enough to enable statistics reporting, it is
8362 recommended to set all other settings in order to avoid relying on default
8363 unobvious parameters.
8364
8365 Example :
8366 # public access (limited to this backend only)
8367 backend public_www
8368 server srv1 192.168.0.1:80
8369 stats enable
8370 stats hide-version
8371 stats scope .
8372 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008373 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008374 stats auth admin1:AdMiN123
8375 stats auth admin2:AdMiN321
8376
8377 # internal monitoring access (unlimited)
8378 backend private_monitoring
8379 stats enable
8380 stats uri /admin?stats
8381 stats refresh 5s
8382
8383 See also : "stats auth", "stats realm", "stats uri"
8384
8385
Willy Tarreaud63335a2010-02-26 12:56:52 +01008386stats hide-version
8387 Enable statistics and hide HAProxy version reporting
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008388 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008389 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008390 Arguments : none
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008391
Willy Tarreaud63335a2010-02-26 12:56:52 +01008392 By default, the stats page reports some useful status information along with
8393 the statistics. Among them is HAProxy's version. However, it is generally
8394 considered dangerous to report precise version to anyone, as it can help them
8395 target known weaknesses with specific attacks. The "stats hide-version"
8396 statement removes the version from the statistics report. This is recommended
8397 for public sites or any site with a weak login/password.
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008398
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02008399 Though this statement alone is enough to enable statistics reporting, it is
8400 recommended to set all other settings in order to avoid relying on default
8401 unobvious parameters.
8402
Willy Tarreaud63335a2010-02-26 12:56:52 +01008403 Example :
8404 # public access (limited to this backend only)
8405 backend public_www
8406 server srv1 192.168.0.1:80
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02008407 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01008408 stats hide-version
8409 stats scope .
8410 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008411 stats realm HAProxy\ Statistics
Willy Tarreaud63335a2010-02-26 12:56:52 +01008412 stats auth admin1:AdMiN123
8413 stats auth admin2:AdMiN321
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008414
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008415 # internal monitoring access (unlimited)
8416 backend private_monitoring
8417 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01008418 stats uri /admin?stats
8419 stats refresh 5s
Krzysztof Piotr Oledzki15514c22010-01-04 16:03:09 +01008420
Willy Tarreaud63335a2010-02-26 12:56:52 +01008421 See also : "stats auth", "stats enable", "stats realm", "stats uri"
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008422
Willy Tarreau983e01e2010-01-11 18:42:06 +01008423
Cyril Bonté2be1b3f2010-09-30 23:46:30 +02008424stats http-request { allow | deny | auth [realm <realm>] }
8425 [ { if | unless } <condition> ]
8426 Access control for statistics
8427
8428 May be used in sections: defaults | frontend | listen | backend
8429 no | no | yes | yes
8430
8431 As "http-request", these set of options allow to fine control access to
8432 statistics. Each option may be followed by if/unless and acl.
8433 First option with matched condition (or option without condition) is final.
8434 For "deny" a 403 error will be returned, for "allow" normal processing is
8435 performed, for "auth" a 401/407 error code is returned so the client
8436 should be asked to enter a username and password.
8437
8438 There is no fixed limit to the number of http-request statements per
8439 instance.
8440
8441 See also : "http-request", section 3.4 about userlists and section 7
8442 about ACL usage.
8443
8444
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008445stats realm <realm>
8446 Enable statistics and set authentication realm
8447 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008448 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008449 Arguments :
8450 <realm> is the name of the HTTP Basic Authentication realm reported to
8451 the browser. The browser uses it to display it in the pop-up
8452 inviting the user to enter a valid username and password.
8453
8454 The realm is read as a single word, so any spaces in it should be escaped
8455 using a backslash ('\').
8456
8457 This statement is useful only in conjunction with "stats auth" since it is
8458 only related to authentication.
8459
8460 Though this statement alone is enough to enable statistics reporting, it is
8461 recommended to set all other settings in order to avoid relying on default
8462 unobvious parameters.
8463
8464 Example :
8465 # public access (limited to this backend only)
8466 backend public_www
8467 server srv1 192.168.0.1:80
8468 stats enable
8469 stats hide-version
8470 stats scope .
8471 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008472 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008473 stats auth admin1:AdMiN123
8474 stats auth admin2:AdMiN321
8475
8476 # internal monitoring access (unlimited)
8477 backend private_monitoring
8478 stats enable
8479 stats uri /admin?stats
8480 stats refresh 5s
8481
8482 See also : "stats auth", "stats enable", "stats uri"
8483
8484
8485stats refresh <delay>
8486 Enable statistics with automatic refresh
8487 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008488 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008489 Arguments :
8490 <delay> is the suggested refresh delay, specified in seconds, which will
8491 be returned to the browser consulting the report page. While the
8492 browser is free to apply any delay, it will generally respect it
8493 and refresh the page this every seconds. The refresh interval may
8494 be specified in any other non-default time unit, by suffixing the
8495 unit after the value, as explained at the top of this document.
8496
8497 This statement is useful on monitoring displays with a permanent page
8498 reporting the load balancer's activity. When set, the HTML report page will
8499 include a link "refresh"/"stop refresh" so that the user can select whether
8500 he wants automatic refresh of the page or not.
8501
8502 Though this statement alone is enough to enable statistics reporting, it is
8503 recommended to set all other settings in order to avoid relying on default
8504 unobvious parameters.
8505
8506 Example :
8507 # public access (limited to this backend only)
8508 backend public_www
8509 server srv1 192.168.0.1:80
8510 stats enable
8511 stats hide-version
8512 stats scope .
8513 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008514 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008515 stats auth admin1:AdMiN123
8516 stats auth admin2:AdMiN321
8517
8518 # internal monitoring access (unlimited)
8519 backend private_monitoring
8520 stats enable
8521 stats uri /admin?stats
8522 stats refresh 5s
8523
8524 See also : "stats auth", "stats enable", "stats realm", "stats uri"
8525
8526
8527stats scope { <name> | "." }
8528 Enable statistics and limit access scope
8529 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008530 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008531 Arguments :
8532 <name> is the name of a listen, frontend or backend section to be
8533 reported. The special name "." (a single dot) designates the
8534 section in which the statement appears.
8535
8536 When this statement is specified, only the sections enumerated with this
8537 statement will appear in the report. All other ones will be hidden. This
8538 statement may appear as many times as needed if multiple sections need to be
8539 reported. Please note that the name checking is performed as simple string
8540 comparisons, and that it is never checked that a give section name really
8541 exists.
8542
8543 Though this statement alone is enough to enable statistics reporting, it is
8544 recommended to set all other settings in order to avoid relying on default
8545 unobvious parameters.
8546
8547 Example :
8548 # public access (limited to this backend only)
8549 backend public_www
8550 server srv1 192.168.0.1:80
8551 stats enable
8552 stats hide-version
8553 stats scope .
8554 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008555 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008556 stats auth admin1:AdMiN123
8557 stats auth admin2:AdMiN321
8558
8559 # internal monitoring access (unlimited)
8560 backend private_monitoring
8561 stats enable
8562 stats uri /admin?stats
8563 stats refresh 5s
8564
8565 See also : "stats auth", "stats enable", "stats realm", "stats uri"
8566
Willy Tarreaud63335a2010-02-26 12:56:52 +01008567
Willy Tarreauc9705a12010-07-27 20:05:50 +02008568stats show-desc [ <desc> ]
Willy Tarreaud63335a2010-02-26 12:56:52 +01008569 Enable reporting of a description on the statistics page.
8570 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008571 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008572
Willy Tarreauc9705a12010-07-27 20:05:50 +02008573 <desc> is an optional description to be reported. If unspecified, the
Willy Tarreaud63335a2010-02-26 12:56:52 +01008574 description from global section is automatically used instead.
8575
8576 This statement is useful for users that offer shared services to their
8577 customers, where node or description should be different for each customer.
8578
8579 Though this statement alone is enough to enable statistics reporting, it is
8580 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +01008581 unobvious parameters. By default description is not shown.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008582
8583 Example :
8584 # internal monitoring access (unlimited)
8585 backend private_monitoring
8586 stats enable
8587 stats show-desc Master node for Europe, Asia, Africa
8588 stats uri /admin?stats
8589 stats refresh 5s
8590
8591 See also: "show-node", "stats enable", "stats uri" and "description" in
8592 global section.
8593
8594
8595stats show-legends
Willy Tarreaued2119c2014-04-24 22:10:39 +02008596 Enable reporting additional information on the statistics page
8597 May be used in sections : defaults | frontend | listen | backend
8598 yes | yes | yes | yes
8599 Arguments : none
8600
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008601 Enable reporting additional information on the statistics page :
Willy Tarreaud63335a2010-02-26 12:56:52 +01008602 - cap: capabilities (proxy)
8603 - mode: one of tcp, http or health (proxy)
8604 - id: SNMP ID (proxy, socket, server)
8605 - IP (socket, server)
8606 - cookie (backend, server)
8607
8608 Though this statement alone is enough to enable statistics reporting, it is
8609 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +01008610 unobvious parameters. Default behavior is not to show this information.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008611
8612 See also: "stats enable", "stats uri".
8613
8614
8615stats show-node [ <name> ]
8616 Enable reporting of a host name on the statistics page.
8617 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008618 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008619 Arguments:
8620 <name> is an optional name to be reported. If unspecified, the
8621 node name from global section is automatically used instead.
8622
8623 This statement is useful for users that offer shared services to their
8624 customers, where node or description might be different on a stats page
Davor Ocelice9ed2812017-12-25 17:49:28 +01008625 provided for each customer. Default behavior is not to show host name.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008626
8627 Though this statement alone is enough to enable statistics reporting, it is
8628 recommended to set all other settings in order to avoid relying on default
8629 unobvious parameters.
8630
8631 Example:
8632 # internal monitoring access (unlimited)
8633 backend private_monitoring
8634 stats enable
8635 stats show-node Europe-1
8636 stats uri /admin?stats
8637 stats refresh 5s
8638
8639 See also: "show-desc", "stats enable", "stats uri", and "node" in global
8640 section.
8641
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008642
8643stats uri <prefix>
8644 Enable statistics and define the URI prefix to access them
8645 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008646 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008647 Arguments :
8648 <prefix> is the prefix of any URI which will be redirected to stats. This
8649 prefix may contain a question mark ('?') to indicate part of a
8650 query string.
8651
8652 The statistics URI is intercepted on the relayed traffic, so it appears as a
8653 page within the normal application. It is strongly advised to ensure that the
8654 selected URI will never appear in the application, otherwise it will never be
8655 possible to reach it in the application.
8656
8657 The default URI compiled in haproxy is "/haproxy?stats", but this may be
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01008658 changed at build time, so it's better to always explicitly specify it here.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008659 It is generally a good idea to include a question mark in the URI so that
8660 intermediate proxies refrain from caching the results. Also, since any string
8661 beginning with the prefix will be accepted as a stats request, the question
8662 mark helps ensuring that no valid URI will begin with the same words.
8663
8664 It is sometimes very convenient to use "/" as the URI prefix, and put that
8665 statement in a "listen" instance of its own. That makes it easy to dedicate
8666 an address or a port to statistics only.
8667
8668 Though this statement alone is enough to enable statistics reporting, it is
8669 recommended to set all other settings in order to avoid relying on default
8670 unobvious parameters.
8671
8672 Example :
8673 # public access (limited to this backend only)
8674 backend public_www
8675 server srv1 192.168.0.1:80
8676 stats enable
8677 stats hide-version
8678 stats scope .
8679 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008680 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008681 stats auth admin1:AdMiN123
8682 stats auth admin2:AdMiN321
8683
8684 # internal monitoring access (unlimited)
8685 backend private_monitoring
8686 stats enable
8687 stats uri /admin?stats
8688 stats refresh 5s
8689
8690 See also : "stats auth", "stats enable", "stats realm"
8691
8692
Willy Tarreaud63335a2010-02-26 12:56:52 +01008693stick match <pattern> [table <table>] [{if | unless} <cond>]
8694 Define a request pattern matching condition to stick a user to a server
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008695 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaud63335a2010-02-26 12:56:52 +01008696 no | no | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008697
8698 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02008699 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008700 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01008701 will be analyzed in the hope to find a matching entry in a
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008702 stickiness table. This rule is mandatory.
8703
8704 <table> is an optional stickiness table name. If unspecified, the same
8705 backend's table is used. A stickiness table is declared using
8706 the "stick-table" statement.
8707
8708 <cond> is an optional matching condition. It makes it possible to match
8709 on a certain criterion only when other conditions are met (or
8710 not met). For instance, it could be used to match on a source IP
8711 address except when a request passes through a known proxy, in
8712 which case we'd match on a header containing that IP address.
8713
8714 Some protocols or applications require complex stickiness rules and cannot
8715 always simply rely on cookies nor hashing. The "stick match" statement
8716 describes a rule to extract the stickiness criterion from an incoming request
8717 or connection. See section 7 for a complete list of possible patterns and
8718 transformation rules.
8719
8720 The table has to be declared using the "stick-table" statement. It must be of
8721 a type compatible with the pattern. By default it is the one which is present
8722 in the same backend. It is possible to share a table with other backends by
8723 referencing it using the "table" keyword. If another table is referenced,
8724 the server's ID inside the backends are used. By default, all server IDs
8725 start at 1 in each backend, so the server ordering is enough. But in case of
8726 doubt, it is highly recommended to force server IDs using their "id" setting.
8727
8728 It is possible to restrict the conditions where a "stick match" statement
8729 will apply, using "if" or "unless" followed by a condition. See section 7 for
8730 ACL based conditions.
8731
8732 There is no limit on the number of "stick match" statements. The first that
8733 applies and matches will cause the request to be directed to the same server
8734 as was used for the request which created the entry. That way, multiple
8735 matches can be used as fallbacks.
8736
8737 The stick rules are checked after the persistence cookies, so they will not
8738 affect stickiness if a cookie has already been used to select a server. That
8739 way, it becomes very easy to insert cookies and match on IP addresses in
8740 order to maintain stickiness between HTTP and HTTPS.
8741
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008742 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8743 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008744 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008745
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008746 Example :
8747 # forward SMTP users to the same server they just used for POP in the
8748 # last 30 minutes
8749 backend pop
8750 mode tcp
8751 balance roundrobin
8752 stick store-request src
8753 stick-table type ip size 200k expire 30m
8754 server s1 192.168.1.1:110
8755 server s2 192.168.1.1:110
8756
8757 backend smtp
8758 mode tcp
8759 balance roundrobin
8760 stick match src table pop
8761 server s1 192.168.1.1:25
8762 server s2 192.168.1.1:25
8763
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008764 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02008765 about ACLs and samples fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008766
8767
8768stick on <pattern> [table <table>] [{if | unless} <condition>]
8769 Define a request pattern to associate a user to a server
8770 May be used in sections : defaults | frontend | listen | backend
8771 no | no | yes | yes
8772
8773 Note : This form is exactly equivalent to "stick match" followed by
8774 "stick store-request", all with the same arguments. Please refer
8775 to both keywords for details. It is only provided as a convenience
8776 for writing more maintainable configurations.
8777
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008778 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8779 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008780 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008781
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008782 Examples :
8783 # The following form ...
Willy Tarreauec579d82010-02-26 19:15:04 +01008784 stick on src table pop if !localhost
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008785
8786 # ...is strictly equivalent to this one :
8787 stick match src table pop if !localhost
8788 stick store-request src table pop if !localhost
8789
8790
8791 # Use cookie persistence for HTTP, and stick on source address for HTTPS as
8792 # well as HTTP without cookie. Share the same table between both accesses.
8793 backend http
8794 mode http
8795 balance roundrobin
8796 stick on src table https
8797 cookie SRV insert indirect nocache
8798 server s1 192.168.1.1:80 cookie s1
8799 server s2 192.168.1.1:80 cookie s2
8800
8801 backend https
8802 mode tcp
8803 balance roundrobin
8804 stick-table type ip size 200k expire 30m
8805 stick on src
8806 server s1 192.168.1.1:443
8807 server s2 192.168.1.1:443
8808
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008809 See also : "stick match", "stick store-request", "nbproc" and "bind-process".
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008810
8811
8812stick store-request <pattern> [table <table>] [{if | unless} <condition>]
8813 Define a request pattern used to create an entry in a stickiness table
8814 May be used in sections : defaults | frontend | listen | backend
8815 no | no | yes | yes
8816
8817 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02008818 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008819 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01008820 will be analyzed, extracted and stored in the table once a
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008821 server is selected.
8822
8823 <table> is an optional stickiness table name. If unspecified, the same
8824 backend's table is used. A stickiness table is declared using
8825 the "stick-table" statement.
8826
8827 <cond> is an optional storage condition. It makes it possible to store
8828 certain criteria only when some conditions are met (or not met).
8829 For instance, it could be used to store the source IP address
8830 except when the request passes through a known proxy, in which
8831 case we'd store a converted form of a header containing that IP
8832 address.
8833
8834 Some protocols or applications require complex stickiness rules and cannot
8835 always simply rely on cookies nor hashing. The "stick store-request" statement
8836 describes a rule to decide what to extract from the request and when to do
8837 it, in order to store it into a stickiness table for further requests to
8838 match it using the "stick match" statement. Obviously the extracted part must
8839 make sense and have a chance to be matched in a further request. Storing a
8840 client's IP address for instance often makes sense. Storing an ID found in a
8841 URL parameter also makes sense. Storing a source port will almost never make
8842 any sense because it will be randomly matched. See section 7 for a complete
8843 list of possible patterns and transformation rules.
8844
8845 The table has to be declared using the "stick-table" statement. It must be of
8846 a type compatible with the pattern. By default it is the one which is present
8847 in the same backend. It is possible to share a table with other backends by
8848 referencing it using the "table" keyword. If another table is referenced,
8849 the server's ID inside the backends are used. By default, all server IDs
8850 start at 1 in each backend, so the server ordering is enough. But in case of
8851 doubt, it is highly recommended to force server IDs using their "id" setting.
8852
8853 It is possible to restrict the conditions where a "stick store-request"
8854 statement will apply, using "if" or "unless" followed by a condition. This
8855 condition will be evaluated while parsing the request, so any criteria can be
8856 used. See section 7 for ACL based conditions.
8857
8858 There is no limit on the number of "stick store-request" statements, but
8859 there is a limit of 8 simultaneous stores per request or response. This
8860 makes it possible to store up to 8 criteria, all extracted from either the
8861 request or the response, regardless of the number of rules. Only the 8 first
8862 ones which match will be kept. Using this, it is possible to feed multiple
8863 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01008864 another protocol or access method. Using multiple store-request rules with
8865 the same table is possible and may be used to find the best criterion to rely
8866 on, by arranging the rules by decreasing preference order. Only the first
8867 extracted criterion for a given table will be stored. All subsequent store-
8868 request rules referencing the same table will be skipped and their ACLs will
8869 not be evaluated.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008870
8871 The "store-request" rules are evaluated once the server connection has been
8872 established, so that the table will contain the real server that processed
8873 the request.
8874
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008875 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8876 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008877 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008878
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008879 Example :
8880 # forward SMTP users to the same server they just used for POP in the
8881 # last 30 minutes
8882 backend pop
8883 mode tcp
8884 balance roundrobin
8885 stick store-request src
8886 stick-table type ip size 200k expire 30m
8887 server s1 192.168.1.1:110
8888 server s2 192.168.1.1:110
8889
8890 backend smtp
8891 mode tcp
8892 balance roundrobin
8893 stick match src table pop
8894 server s1 192.168.1.1:25
8895 server s2 192.168.1.1:25
8896
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008897 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02008898 about ACLs and sample fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008899
8900
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008901stick-table type {ip | integer | string [len <length>] | binary [len <length>]}
Emeric Brunf099e792010-09-27 12:05:28 +02008902 size <size> [expire <expire>] [nopurge] [peers <peersect>]
8903 [store <data_type>]*
Godbach64cef792013-12-04 16:08:22 +08008904 Configure the stickiness table for the current section
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008905 May be used in sections : defaults | frontend | listen | backend
Willy Tarreauc00cdc22010-06-06 16:48:26 +02008906 no | yes | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008907
8908 Arguments :
8909 ip a table declared with "type ip" will only store IPv4 addresses.
8910 This form is very compact (about 50 bytes per entry) and allows
8911 very fast entry lookup and stores with almost no overhead. This
8912 is mainly used to store client source IP addresses.
8913
David du Colombier9a6d3c92011-03-17 10:40:24 +01008914 ipv6 a table declared with "type ipv6" will only store IPv6 addresses.
8915 This form is very compact (about 60 bytes per entry) and allows
8916 very fast entry lookup and stores with almost no overhead. This
8917 is mainly used to store client source IP addresses.
8918
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008919 integer a table declared with "type integer" will store 32bit integers
8920 which can represent a client identifier found in a request for
8921 instance.
8922
8923 string a table declared with "type string" will store substrings of up
8924 to <len> characters. If the string provided by the pattern
8925 extractor is larger than <len>, it will be truncated before
8926 being stored. During matching, at most <len> characters will be
8927 compared between the string in the table and the extracted
8928 pattern. When not specified, the string is automatically limited
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008929 to 32 characters.
8930
8931 binary a table declared with "type binary" will store binary blocks
8932 of <len> bytes. If the block provided by the pattern
8933 extractor is larger than <len>, it will be truncated before
Willy Tarreaube722a22014-06-13 16:31:59 +02008934 being stored. If the block provided by the sample expression
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008935 is shorter than <len>, it will be padded by 0. When not
8936 specified, the block is automatically limited to 32 bytes.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008937
8938 <length> is the maximum number of characters that will be stored in a
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008939 "string" type table (See type "string" above). Or the number
8940 of bytes of the block in "binary" type table. Be careful when
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008941 changing this parameter as memory usage will proportionally
8942 increase.
8943
8944 <size> is the maximum number of entries that can fit in the table. This
Cyril Bonté78caf842010-03-10 22:41:43 +01008945 value directly impacts memory usage. Count approximately
8946 50 bytes per entry, plus the size of a string if any. The size
8947 supports suffixes "k", "m", "g" for 2^10, 2^20 and 2^30 factors.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008948
8949 [nopurge] indicates that we refuse to purge older entries when the table
8950 is full. When not specified and the table is full when haproxy
8951 wants to store an entry in it, it will flush a few of the oldest
8952 entries in order to release some space for the new ones. This is
Davor Ocelice9ed2812017-12-25 17:49:28 +01008953 most often the desired behavior. In some specific cases, it
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008954 be desirable to refuse new entries instead of purging the older
8955 ones. That may be the case when the amount of data to store is
8956 far above the hardware limits and we prefer not to offer access
8957 to new clients than to reject the ones already connected. When
8958 using this parameter, be sure to properly set the "expire"
8959 parameter (see below).
8960
Emeric Brunf099e792010-09-27 12:05:28 +02008961 <peersect> is the name of the peers section to use for replication. Entries
8962 which associate keys to server IDs are kept synchronized with
8963 the remote peers declared in this section. All entries are also
8964 automatically learned from the local peer (old process) during a
8965 soft restart.
8966
Willy Tarreau1abc6732015-05-01 19:21:02 +02008967 NOTE : each peers section may be referenced only by tables
8968 belonging to the same unique process.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008969
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008970 <expire> defines the maximum duration of an entry in the table since it
8971 was last created, refreshed or matched. The expiration delay is
8972 defined using the standard time format, similarly as the various
8973 timeouts. The maximum duration is slightly above 24 days. See
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +03008974 section 2.4 for more information. If this delay is not specified,
Cyril Bontédc4d9032012-04-08 21:57:39 +02008975 the session won't automatically expire, but older entries will
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008976 be removed once full. Be sure not to use the "nopurge" parameter
8977 if not expiration delay is specified.
8978
Willy Tarreau08d5f982010-06-06 13:34:54 +02008979 <data_type> is used to store additional information in the stick-table. This
8980 may be used by ACLs in order to control various criteria related
8981 to the activity of the client matching the stick-table. For each
8982 item specified here, the size of each entry will be inflated so
Willy Tarreauc9705a12010-07-27 20:05:50 +02008983 that the additional data can fit. Several data types may be
8984 stored with an entry. Multiple data types may be specified after
8985 the "store" keyword, as a comma-separated list. Alternatively,
8986 it is possible to repeat the "store" keyword followed by one or
8987 several data types. Except for the "server_id" type which is
8988 automatically detected and enabled, all data types must be
8989 explicitly declared to be stored. If an ACL references a data
8990 type which is not stored, the ACL will simply not match. Some
8991 data types require an argument which must be passed just after
8992 the type between parenthesis. See below for the supported data
8993 types and their arguments.
8994
8995 The data types that can be stored with an entry are the following :
8996 - server_id : this is an integer which holds the numeric ID of the server a
8997 request was assigned to. It is used by the "stick match", "stick store",
8998 and "stick on" rules. It is automatically enabled when referenced.
8999
9000 - gpc0 : first General Purpose Counter. It is a positive 32-bit integer
9001 integer which may be used for anything. Most of the time it will be used
9002 to put a special tag on some entries, for instance to note that a
Davor Ocelice9ed2812017-12-25 17:49:28 +01009003 specific behavior was detected and must be known for future matches.
Willy Tarreauc9705a12010-07-27 20:05:50 +02009004
Willy Tarreauba2ffd12013-05-29 15:54:14 +02009005 - gpc0_rate(<period>) : increment rate of the first General Purpose Counter
9006 over a period. It is a positive 32-bit integer integer which may be used
9007 for anything. Just like <gpc0>, it counts events, but instead of keeping
Davor Ocelice9ed2812017-12-25 17:49:28 +01009008 a cumulative number, it maintains the rate at which the counter is
Willy Tarreauba2ffd12013-05-29 15:54:14 +02009009 incremented. Most of the time it will be used to measure the frequency of
Davor Ocelice9ed2812017-12-25 17:49:28 +01009010 occurrence of certain events (e.g. requests to a specific URL).
Willy Tarreauba2ffd12013-05-29 15:54:14 +02009011
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009012 - gpc1 : second General Purpose Counter. It is a positive 32-bit integer
9013 integer which may be used for anything. Most of the time it will be used
9014 to put a special tag on some entries, for instance to note that a
9015 specific behavior was detected and must be known for future matches.
9016
9017 - gpc1_rate(<period>) : increment rate of the second General Purpose Counter
9018 over a period. It is a positive 32-bit integer integer which may be used
9019 for anything. Just like <gpc1>, it counts events, but instead of keeping
9020 a cumulative number, it maintains the rate at which the counter is
9021 incremented. Most of the time it will be used to measure the frequency of
9022 occurrence of certain events (e.g. requests to a specific URL).
9023
Willy Tarreauc9705a12010-07-27 20:05:50 +02009024 - conn_cnt : Connection Count. It is a positive 32-bit integer which counts
9025 the absolute number of connections received from clients which matched
9026 this entry. It does not mean the connections were accepted, just that
9027 they were received.
9028
9029 - conn_cur : Current Connections. It is a positive 32-bit integer which
9030 stores the concurrent connection counts for the entry. It is incremented
9031 once an incoming connection matches the entry, and decremented once the
9032 connection leaves. That way it is possible to know at any time the exact
9033 number of concurrent connections for an entry.
9034
9035 - conn_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9036 integer parameter <period> which indicates in milliseconds the length
9037 of the period over which the average is measured. It reports the average
9038 incoming connection rate over that period, in connections per period. The
9039 result is an integer which can be matched using ACLs.
9040
9041 - sess_cnt : Session Count. It is a positive 32-bit integer which counts
9042 the absolute number of sessions received from clients which matched this
9043 entry. A session is a connection that was accepted by the layer 4 rules.
9044
9045 - sess_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9046 integer parameter <period> which indicates in milliseconds the length
9047 of the period over which the average is measured. It reports the average
9048 incoming session rate over that period, in sessions per period. The
9049 result is an integer which can be matched using ACLs.
9050
9051 - http_req_cnt : HTTP request Count. It is a positive 32-bit integer which
9052 counts the absolute number of HTTP requests received from clients which
9053 matched this entry. It does not matter whether they are valid requests or
9054 not. Note that this is different from sessions when keep-alive is used on
9055 the client side.
9056
9057 - http_req_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9058 integer parameter <period> which indicates in milliseconds the length
9059 of the period over which the average is measured. It reports the average
9060 HTTP request rate over that period, in requests per period. The result is
9061 an integer which can be matched using ACLs. It does not matter whether
9062 they are valid requests or not. Note that this is different from sessions
9063 when keep-alive is used on the client side.
9064
9065 - http_err_cnt : HTTP Error Count. It is a positive 32-bit integer which
9066 counts the absolute number of HTTP requests errors induced by clients
9067 which matched this entry. Errors are counted on invalid and truncated
9068 requests, as well as on denied or tarpitted requests, and on failed
9069 authentications. If the server responds with 4xx, then the request is
9070 also counted as an error since it's an error triggered by the client
Davor Ocelice9ed2812017-12-25 17:49:28 +01009071 (e.g. vulnerability scan).
Willy Tarreauc9705a12010-07-27 20:05:50 +02009072
9073 - http_err_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9074 integer parameter <period> which indicates in milliseconds the length
9075 of the period over which the average is measured. It reports the average
9076 HTTP request error rate over that period, in requests per period (see
9077 http_err_cnt above for what is accounted as an error). The result is an
9078 integer which can be matched using ACLs.
9079
9080 - bytes_in_cnt : client to server byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +01009081 integer which counts the cumulative number of bytes received from clients
Willy Tarreauc9705a12010-07-27 20:05:50 +02009082 which matched this entry. Headers are included in the count. This may be
9083 used to limit abuse of upload features on photo or video servers.
9084
9085 - bytes_in_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9086 integer parameter <period> which indicates in milliseconds the length
9087 of the period over which the average is measured. It reports the average
9088 incoming bytes rate over that period, in bytes per period. It may be used
9089 to detect users which upload too much and too fast. Warning: with large
9090 uploads, it is possible that the amount of uploaded data will be counted
9091 once upon termination, thus causing spikes in the average transfer speed
9092 instead of having a smooth one. This may partially be smoothed with
9093 "option contstats" though this is not perfect yet. Use of byte_in_cnt is
9094 recommended for better fairness.
9095
9096 - bytes_out_cnt : server to client byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +01009097 integer which counts the cumulative number of bytes sent to clients which
Willy Tarreauc9705a12010-07-27 20:05:50 +02009098 matched this entry. Headers are included in the count. This may be used
9099 to limit abuse of bots sucking the whole site.
9100
9101 - bytes_out_rate(<period>) : frequency counter (takes 12 bytes). It takes
9102 an integer parameter <period> which indicates in milliseconds the length
9103 of the period over which the average is measured. It reports the average
9104 outgoing bytes rate over that period, in bytes per period. It may be used
9105 to detect users which download too much and too fast. Warning: with large
9106 transfers, it is possible that the amount of transferred data will be
9107 counted once upon termination, thus causing spikes in the average
9108 transfer speed instead of having a smooth one. This may partially be
9109 smoothed with "option contstats" though this is not perfect yet. Use of
9110 byte_out_cnt is recommended for better fairness.
Willy Tarreau08d5f982010-06-06 13:34:54 +02009111
Willy Tarreauc00cdc22010-06-06 16:48:26 +02009112 There is only one stick-table per proxy. At the moment of writing this doc,
9113 it does not seem useful to have multiple tables per proxy. If this happens
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009114 to be required, simply create a dummy backend with a stick-table in it and
9115 reference it.
9116
9117 It is important to understand that stickiness based on learning information
9118 has some limitations, including the fact that all learned associations are
Baptiste Assmann123ff042016-03-06 23:29:28 +01009119 lost upon restart unless peers are properly configured to transfer such
9120 information upon restart (recommended). In general it can be good as a
9121 complement but not always as an exclusive stickiness.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009122
Willy Tarreauc9705a12010-07-27 20:05:50 +02009123 Last, memory requirements may be important when storing many data types.
9124 Indeed, storing all indicators above at once in each entry requires 116 bytes
9125 per entry, or 116 MB for a 1-million entries table. This is definitely not
9126 something that can be ignored.
9127
9128 Example:
9129 # Keep track of counters of up to 1 million IP addresses over 5 minutes
9130 # and store a general purpose counter and the average connection rate
9131 # computed over a sliding window of 30 seconds.
9132 stick-table type ip size 1m expire 5m store gpc0,conn_rate(30s)
9133
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +03009134 See also : "stick match", "stick on", "stick store-request", section 2.4
David du Colombiera13d1b92011-03-17 10:40:22 +01009135 about time format and section 7 about ACLs.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009136
9137
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009138stick store-response <pattern> [table <table>] [{if | unless} <condition>]
Baptiste Assmann2f2d2ec2016-03-06 23:27:24 +01009139 Define a response pattern used to create an entry in a stickiness table
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009140 May be used in sections : defaults | frontend | listen | backend
9141 no | no | yes | yes
9142
9143 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02009144 <pattern> is a sample expression rule as described in section 7.3. It
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009145 describes what elements of the response or connection will
Davor Ocelice9ed2812017-12-25 17:49:28 +01009146 be analyzed, extracted and stored in the table once a
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009147 server is selected.
9148
9149 <table> is an optional stickiness table name. If unspecified, the same
9150 backend's table is used. A stickiness table is declared using
9151 the "stick-table" statement.
9152
9153 <cond> is an optional storage condition. It makes it possible to store
9154 certain criteria only when some conditions are met (or not met).
9155 For instance, it could be used to store the SSL session ID only
9156 when the response is a SSL server hello.
9157
9158 Some protocols or applications require complex stickiness rules and cannot
9159 always simply rely on cookies nor hashing. The "stick store-response"
9160 statement describes a rule to decide what to extract from the response and
9161 when to do it, in order to store it into a stickiness table for further
9162 requests to match it using the "stick match" statement. Obviously the
9163 extracted part must make sense and have a chance to be matched in a further
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009164 request. Storing an ID found in a header of a response makes sense.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009165 See section 7 for a complete list of possible patterns and transformation
9166 rules.
9167
9168 The table has to be declared using the "stick-table" statement. It must be of
9169 a type compatible with the pattern. By default it is the one which is present
9170 in the same backend. It is possible to share a table with other backends by
9171 referencing it using the "table" keyword. If another table is referenced,
9172 the server's ID inside the backends are used. By default, all server IDs
9173 start at 1 in each backend, so the server ordering is enough. But in case of
9174 doubt, it is highly recommended to force server IDs using their "id" setting.
9175
9176 It is possible to restrict the conditions where a "stick store-response"
9177 statement will apply, using "if" or "unless" followed by a condition. This
9178 condition will be evaluated while parsing the response, so any criteria can
9179 be used. See section 7 for ACL based conditions.
9180
9181 There is no limit on the number of "stick store-response" statements, but
9182 there is a limit of 8 simultaneous stores per request or response. This
9183 makes it possible to store up to 8 criteria, all extracted from either the
9184 request or the response, regardless of the number of rules. Only the 8 first
9185 ones which match will be kept. Using this, it is possible to feed multiple
9186 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01009187 another protocol or access method. Using multiple store-response rules with
9188 the same table is possible and may be used to find the best criterion to rely
9189 on, by arranging the rules by decreasing preference order. Only the first
9190 extracted criterion for a given table will be stored. All subsequent store-
9191 response rules referencing the same table will be skipped and their ACLs will
9192 not be evaluated. However, even if a store-request rule references a table, a
9193 store-response rule may also use the same table. This means that each table
9194 may learn exactly one element from the request and one element from the
9195 response at once.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009196
9197 The table will contain the real server that processed the request.
9198
9199 Example :
9200 # Learn SSL session ID from both request and response and create affinity.
9201 backend https
9202 mode tcp
9203 balance roundrobin
Cyril Bontédc4d9032012-04-08 21:57:39 +02009204 # maximum SSL session ID length is 32 bytes.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009205 stick-table type binary len 32 size 30k expire 30m
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009206
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009207 acl clienthello req_ssl_hello_type 1
9208 acl serverhello rep_ssl_hello_type 2
9209
9210 # use tcp content accepts to detects ssl client and server hello.
9211 tcp-request inspect-delay 5s
9212 tcp-request content accept if clienthello
9213
9214 # no timeout on response inspect delay by default.
9215 tcp-response content accept if serverhello
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009216
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009217 # SSL session ID (SSLID) may be present on a client or server hello.
9218 # Its length is coded on 1 byte at offset 43 and its value starts
9219 # at offset 44.
9220
9221 # Match and learn on request if client hello.
9222 stick on payload_lv(43,1) if clienthello
9223
9224 # Learn on response if server hello.
9225 stick store-response payload_lv(43,1) if serverhello
Cyril Bontédc4d9032012-04-08 21:57:39 +02009226
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009227 server s1 192.168.1.1:443
9228 server s2 192.168.1.1:443
9229
9230 See also : "stick-table", "stick on", and section 7 about ACLs and pattern
9231 extraction.
9232
9233
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009234tcp-check connect [params*]
9235 Opens a new connection
9236 May be used in sections: defaults | frontend | listen | backend
9237 no | no | yes | yes
9238
9239 When an application lies on more than a single TCP port or when HAProxy
9240 load-balance many services in a single backend, it makes sense to probe all
9241 the services individually before considering a server as operational.
9242
9243 When there are no TCP port configured on the server line neither server port
9244 directive, then the 'tcp-check connect port <port>' must be the first step
9245 of the sequence.
9246
9247 In a tcp-check ruleset a 'connect' is required, it is also mandatory to start
9248 the ruleset with a 'connect' rule. Purpose is to ensure admin know what they
9249 do.
9250
9251 Parameters :
9252 They are optional and can be used to describe how HAProxy should open and
9253 use the TCP connection.
9254
9255 port if not set, check port or server port is used.
9256 It tells HAProxy where to open the connection to.
9257 <port> must be a valid TCP port source integer, from 1 to 65535.
9258
9259 send-proxy send a PROXY protocol string
9260
9261 ssl opens a ciphered connection
9262
9263 Examples:
9264 # check HTTP and HTTPs services on a server.
9265 # first open port 80 thanks to server line port directive, then
9266 # tcp-check opens port 443, ciphered and run a request on it:
9267 option tcp-check
9268 tcp-check connect
9269 tcp-check send GET\ /\ HTTP/1.0\r\n
9270 tcp-check send Host:\ haproxy.1wt.eu\r\n
9271 tcp-check send \r\n
9272 tcp-check expect rstring (2..|3..)
9273 tcp-check connect port 443 ssl
9274 tcp-check send GET\ /\ HTTP/1.0\r\n
9275 tcp-check send Host:\ haproxy.1wt.eu\r\n
9276 tcp-check send \r\n
9277 tcp-check expect rstring (2..|3..)
9278 server www 10.0.0.1 check port 80
9279
9280 # check both POP and IMAP from a single server:
9281 option tcp-check
9282 tcp-check connect port 110
9283 tcp-check expect string +OK\ POP3\ ready
9284 tcp-check connect port 143
9285 tcp-check expect string *\ OK\ IMAP4\ ready
9286 server mail 10.0.0.1 check
9287
9288 See also : "option tcp-check", "tcp-check send", "tcp-check expect"
9289
9290
9291tcp-check expect [!] <match> <pattern>
Davor Ocelice9ed2812017-12-25 17:49:28 +01009292 Specify data to be collected and analyzed during a generic health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009293 May be used in sections: defaults | frontend | listen | backend
9294 no | no | yes | yes
9295
9296 Arguments :
9297 <match> is a keyword indicating how to look for a specific pattern in the
9298 response. The keyword may be one of "string", "rstring" or
9299 binary.
9300 The keyword may be preceded by an exclamation mark ("!") to negate
9301 the match. Spaces are allowed between the exclamation mark and the
9302 keyword. See below for more details on the supported keywords.
9303
9304 <pattern> is the pattern to look for. It may be a string or a regular
9305 expression. If the pattern contains spaces, they must be escaped
9306 with the usual backslash ('\').
9307 If the match is set to binary, then the pattern must be passed as
Davor Ocelice9ed2812017-12-25 17:49:28 +01009308 a series of hexadecimal digits in an even number. Each sequence of
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009309 two digits will represent a byte. The hexadecimal digits may be
9310 used upper or lower case.
9311
9312
9313 The available matches are intentionally similar to their http-check cousins :
9314
9315 string <string> : test the exact string matches in the response buffer.
9316 A health check response will be considered valid if the
9317 response's buffer contains this exact string. If the
9318 "string" keyword is prefixed with "!", then the response
9319 will be considered invalid if the body contains this
9320 string. This can be used to look for a mandatory pattern
9321 in a protocol response, or to detect a failure when a
9322 specific error appears in a protocol banner.
9323
9324 rstring <regex> : test a regular expression on the response buffer.
9325 A health check response will be considered valid if the
9326 response's buffer matches this expression. If the
9327 "rstring" keyword is prefixed with "!", then the response
9328 will be considered invalid if the body matches the
9329 expression.
9330
9331 binary <hexstring> : test the exact string in its hexadecimal form matches
9332 in the response buffer. A health check response will
9333 be considered valid if the response's buffer contains
9334 this exact hexadecimal string.
9335 Purpose is to match data on binary protocols.
9336
9337 It is important to note that the responses will be limited to a certain size
9338 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
9339 Thus, too large responses may not contain the mandatory pattern when using
9340 "string", "rstring" or binary. If a large response is absolutely required, it
9341 is possible to change the default max size by setting the global variable.
9342 However, it is worth keeping in mind that parsing very large responses can
9343 waste some CPU cycles, especially when regular expressions are used, and that
9344 it is always better to focus the checks on smaller resources. Also, in its
9345 current state, the check will not find any string nor regex past a null
9346 character in the response. Similarly it is not possible to request matching
9347 the null character.
9348
9349 Examples :
9350 # perform a POP check
9351 option tcp-check
9352 tcp-check expect string +OK\ POP3\ ready
9353
9354 # perform an IMAP check
9355 option tcp-check
9356 tcp-check expect string *\ OK\ IMAP4\ ready
9357
9358 # look for the redis master server
9359 option tcp-check
9360 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02009361 tcp-check expect string +PONG
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009362 tcp-check send info\ replication\r\n
9363 tcp-check expect string role:master
9364 tcp-check send QUIT\r\n
9365 tcp-check expect string +OK
9366
9367
9368 See also : "option tcp-check", "tcp-check connect", "tcp-check send",
9369 "tcp-check send-binary", "http-check expect", tune.chksize
9370
9371
9372tcp-check send <data>
9373 Specify a string to be sent as a question during a generic health check
9374 May be used in sections: defaults | frontend | listen | backend
9375 no | no | yes | yes
9376
9377 <data> : the data to be sent as a question during a generic health check
9378 session. For now, <data> must be a string.
9379
9380 Examples :
9381 # look for the redis master server
9382 option tcp-check
9383 tcp-check send info\ replication\r\n
9384 tcp-check expect string role:master
9385
9386 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
9387 "tcp-check send-binary", tune.chksize
9388
9389
Davor Ocelice9ed2812017-12-25 17:49:28 +01009390tcp-check send-binary <hexstring>
9391 Specify a hex digits string to be sent as a binary question during a raw
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009392 tcp health check
9393 May be used in sections: defaults | frontend | listen | backend
9394 no | no | yes | yes
9395
9396 <data> : the data to be sent as a question during a generic health check
9397 session. For now, <data> must be a string.
Davor Ocelice9ed2812017-12-25 17:49:28 +01009398 <hexstring> : test the exact string in its hexadecimal form matches in the
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009399 response buffer. A health check response will be considered
9400 valid if the response's buffer contains this exact
9401 hexadecimal string.
9402 Purpose is to send binary data to ask on binary protocols.
9403
9404 Examples :
9405 # redis check in binary
9406 option tcp-check
9407 tcp-check send-binary 50494e470d0a # PING\r\n
9408 tcp-check expect binary 2b504F4e47 # +PONG
9409
9410
9411 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
9412 "tcp-check send", tune.chksize
9413
9414
Willy Tarreaue9656522010-08-17 15:40:09 +02009415tcp-request connection <action> [{if | unless} <condition>]
9416 Perform an action on an incoming connection depending on a layer 4 condition
Willy Tarreau1a687942010-05-23 22:40:30 +02009417 May be used in sections : defaults | frontend | listen | backend
9418 no | yes | yes | no
Willy Tarreaue9656522010-08-17 15:40:09 +02009419 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02009420 <action> defines the action to perform if the condition applies. See
9421 below.
Willy Tarreau1a687942010-05-23 22:40:30 +02009422
Willy Tarreaue9656522010-08-17 15:40:09 +02009423 <condition> is a standard layer4-only ACL-based condition (see section 7).
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009424
9425 Immediately after acceptance of a new incoming connection, it is possible to
9426 evaluate some conditions to decide whether this connection must be accepted
Willy Tarreaue9656522010-08-17 15:40:09 +02009427 or dropped or have its counters tracked. Those conditions cannot make use of
9428 any data contents because the connection has not been read from yet, and the
9429 buffers are not yet allocated. This is used to selectively and very quickly
9430 accept or drop connections from various sources with a very low overhead. If
9431 some contents need to be inspected in order to take the decision, the
9432 "tcp-request content" statements must be used instead.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009433
Willy Tarreaue9656522010-08-17 15:40:09 +02009434 The "tcp-request connection" rules are evaluated in their exact declaration
9435 order. If no rule matches or if there is no rule, the default action is to
9436 accept the incoming connection. There is no specific limit to the number of
9437 rules which may be inserted.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009438
Willy Tarreaua9083d02015-05-08 15:27:59 +02009439 Four types of actions are supported :
Willy Tarreaue9656522010-08-17 15:40:09 +02009440 - accept :
9441 accepts the connection if the condition is true (when used with "if")
9442 or false (when used with "unless"). The first such rule executed ends
9443 the rules evaluation.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009444
Willy Tarreaue9656522010-08-17 15:40:09 +02009445 - reject :
9446 rejects the connection if the condition is true (when used with "if")
9447 or false (when used with "unless"). The first such rule executed ends
9448 the rules evaluation. Rejected connections do not even become a
9449 session, which is why they are accounted separately for in the stats,
9450 as "denied connections". They are not considered for the session
9451 rate-limit and are not logged either. The reason is that these rules
9452 should only be used to filter extremely high connection rates such as
9453 the ones encountered during a massive DDoS attack. Under these extreme
9454 conditions, the simple action of logging each event would make the
9455 system collapse and would considerably lower the filtering capacity. If
9456 logging is absolutely desired, then "tcp-request content" rules should
Willy Tarreau4f614292016-10-21 17:49:36 +02009457 be used instead, as "tcp-request session" rules will not log either.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009458
Willy Tarreau4f0d9192013-06-11 20:40:55 +02009459 - expect-proxy layer4 :
9460 configures the client-facing connection to receive a PROXY protocol
9461 header before any byte is read from the socket. This is equivalent to
9462 having the "accept-proxy" keyword on the "bind" line, except that using
9463 the TCP rule allows the PROXY protocol to be accepted only for certain
9464 IP address ranges using an ACL. This is convenient when multiple layers
9465 of load balancers are passed through by traffic coming from public
9466 hosts.
9467
Bertrand Jacquin90759682016-06-06 15:35:39 +01009468 - expect-netscaler-cip layer4 :
9469 configures the client-facing connection to receive a NetScaler Client
9470 IP insertion protocol header before any byte is read from the socket.
9471 This is equivalent to having the "accept-netscaler-cip" keyword on the
9472 "bind" line, except that using the TCP rule allows the PROXY protocol
9473 to be accepted only for certain IP address ranges using an ACL. This
9474 is convenient when multiple layers of load balancers are passed
9475 through by traffic coming from public hosts.
9476
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009477 - capture <sample> len <length> :
9478 This only applies to "tcp-request content" rules. It captures sample
9479 expression <sample> from the request buffer, and converts it to a
9480 string of at most <len> characters. The resulting string is stored into
9481 the next request "capture" slot, so it will possibly appear next to
9482 some captured HTTP headers. It will then automatically appear in the
9483 logs, and it will be possible to extract it using sample fetch rules to
9484 feed it into headers or anything. The length should be limited given
9485 that this size will be allocated for each capture during the whole
Willy Tarreaua9083d02015-05-08 15:27:59 +02009486 session life. Please check section 7.3 (Fetching samples) and "capture
9487 request header" for more information.
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009488
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009489 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] :
Willy Tarreaue9656522010-08-17 15:40:09 +02009490 enables tracking of sticky counters from current connection. These
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +02009491 rules do not stop evaluation and do not change default action. The
9492 number of counters that may be simultaneously tracked by the same
9493 connection is set in MAX_SESS_STKCTR at build time (reported in
9494 haproxy -vv) whichs defaults to 3, so the track-sc number is between 0
9495 and (MAX_SESS_STCKTR-1). The first "track-sc0" rule executed enables
9496 tracking of the counters of the specified table as the first set. The
9497 first "track-sc1" rule executed enables tracking of the counters of the
9498 specified table as the second set. The first "track-sc2" rule executed
9499 enables tracking of the counters of the specified table as the third
9500 set. It is a recommended practice to use the first set of counters for
9501 the per-frontend counters and the second set for the per-backend ones.
9502 But this is just a guideline, all may be used everywhere.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009503
Willy Tarreaue9656522010-08-17 15:40:09 +02009504 These actions take one or two arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02009505 <key> is mandatory, and is a sample expression rule as described
Willy Tarreau74ca5042013-06-11 23:12:07 +02009506 in section 7.3. It describes what elements of the incoming
Davor Ocelice9ed2812017-12-25 17:49:28 +01009507 request or connection will be analyzed, extracted, combined,
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009508 and used to select which table entry to update the counters.
9509 Note that "tcp-request connection" cannot use content-based
9510 fetches.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009511
Willy Tarreaue9656522010-08-17 15:40:09 +02009512 <table> is an optional table to be used instead of the default one,
9513 which is the stick-table declared in the current proxy. All
9514 the counters for the matches and updates for the key will
9515 then be performed in that table until the session ends.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009516
Willy Tarreaue9656522010-08-17 15:40:09 +02009517 Once a "track-sc*" rule is executed, the key is looked up in the table
9518 and if it is not found, an entry is allocated for it. Then a pointer to
9519 that entry is kept during all the session's life, and this entry's
9520 counters are updated as often as possible, every time the session's
9521 counters are updated, and also systematically when the session ends.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009522 Counters are only updated for events that happen after the tracking has
9523 been started. For example, connection counters will not be updated when
9524 tracking layer 7 information, since the connection event happens before
9525 layer7 information is extracted.
9526
Willy Tarreaue9656522010-08-17 15:40:09 +02009527 If the entry tracks concurrent connection counters, one connection is
9528 counted for as long as the entry is tracked, and the entry will not
9529 expire during that time. Tracking counters also provides a performance
9530 advantage over just checking the keys, because only one table lookup is
9531 performed for all ACL checks that make use of it.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009532
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02009533 - sc-inc-gpc0(<sc-id>):
9534 The "sc-inc-gpc0" increments the GPC0 counter according to the sticky
9535 counter designated by <sc-id>. If an error occurs, this action silently
9536 fails and the actions evaluation continues.
9537
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009538 - sc-inc-gpc1(<sc-id>):
9539 The "sc-inc-gpc1" increments the GPC1 counter according to the sticky
9540 counter designated by <sc-id>. If an error occurs, this action silently
9541 fails and the actions evaluation continues.
9542
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009543 - sc-set-gpt0(<sc-id>) <int>:
9544 This action sets the GPT0 tag according to the sticky counter designated
9545 by <sc-id> and the value of <int>. The expected result is a boolean. If
9546 an error occurs, this action silently fails and the actions evaluation
9547 continues.
9548
William Lallemand2e785f22016-05-25 01:48:42 +02009549 - set-src <expr> :
9550 Is used to set the source IP address to the value of specified
9551 expression. Useful if you want to mask source IP for privacy.
9552 If you want to provide an IP from a HTTP header use "http-request
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02009553 set-src".
William Lallemand2e785f22016-05-25 01:48:42 +02009554
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02009555 Arguments:
9556 <expr> Is a standard HAProxy expression formed by a sample-fetch
9557 followed by some converters.
William Lallemand2e785f22016-05-25 01:48:42 +02009558
9559 Example:
William Lallemand2e785f22016-05-25 01:48:42 +02009560 tcp-request connection set-src src,ipmask(24)
9561
Willy Tarreau0c630532016-10-21 17:52:58 +02009562 When possible, set-src preserves the original source port as long as the
9563 address family allows it, otherwise the source port is set to 0.
William Lallemand2e785f22016-05-25 01:48:42 +02009564
William Lallemand44be6402016-05-25 01:51:35 +02009565 - set-src-port <expr> :
9566 Is used to set the source port address to the value of specified
9567 expression.
9568
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02009569 Arguments:
9570 <expr> Is a standard HAProxy expression formed by a sample-fetch
9571 followed by some converters.
William Lallemand44be6402016-05-25 01:51:35 +02009572
9573 Example:
William Lallemand44be6402016-05-25 01:51:35 +02009574 tcp-request connection set-src-port int(4000)
9575
Willy Tarreau0c630532016-10-21 17:52:58 +02009576 When possible, set-src-port preserves the original source address as long
9577 as the address family supports a port, otherwise it forces the source
9578 address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +02009579
William Lallemand13e9b0c2016-05-25 02:34:07 +02009580 - set-dst <expr> :
9581 Is used to set the destination IP address to the value of specified
9582 expression. Useful if you want to mask IP for privacy in log.
9583 If you want to provide an IP from a HTTP header use "http-request
9584 set-dst". If you want to connect to the new address/port, use
9585 '0.0.0.0:0' as a server address in the backend.
9586
9587 <expr> Is a standard HAProxy expression formed by a sample-fetch
9588 followed by some converters.
9589
9590 Example:
9591
9592 tcp-request connection set-dst dst,ipmask(24)
9593 tcp-request connection set-dst ipv4(10.0.0.1)
9594
Willy Tarreau0c630532016-10-21 17:52:58 +02009595 When possible, set-dst preserves the original destination port as long as
9596 the address family allows it, otherwise the destination port is set to 0.
9597
William Lallemand13e9b0c2016-05-25 02:34:07 +02009598 - set-dst-port <expr> :
9599 Is used to set the destination port address to the value of specified
9600 expression. If you want to connect to the new address/port, use
9601 '0.0.0.0:0' as a server address in the backend.
9602
9603
9604 <expr> Is a standard HAProxy expression formed by a sample-fetch
9605 followed by some converters.
9606
9607 Example:
9608
9609 tcp-request connection set-dst-port int(4000)
9610
Willy Tarreau0c630532016-10-21 17:52:58 +02009611 When possible, set-dst-port preserves the original destination address as
9612 long as the address family supports a port, otherwise it forces the
9613 destination address to IPv4 "0.0.0.0" before rewriting the port.
9614
Willy Tarreau2d392c22015-08-24 01:43:45 +02009615 - "silent-drop" :
9616 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +01009617 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +02009618 to prevent the client from being notified. The effect it then that the
9619 client still sees an established connection while there's none on
9620 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
9621 except that it doesn't use any local resource at all on the machine
9622 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +01009623 slow down stronger attackers. It is important to understand the impact
9624 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009625 client and HAProxy (firewalls, proxies, load balancers) will also keep
9626 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +01009627 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009628 TCP_REPAIR socket option is used to block the emission of a TCP
9629 reset. On other systems, the socket's TTL is reduced to 1 so that the
9630 TCP reset doesn't pass the first router, though it's still delivered to
9631 local networks. Do not use it unless you fully understand how it works.
9632
Willy Tarreaue9656522010-08-17 15:40:09 +02009633 Note that the "if/unless" condition is optional. If no condition is set on
9634 the action, it is simply performed unconditionally. That can be useful for
9635 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009636
Willy Tarreaue9656522010-08-17 15:40:09 +02009637 Example: accept all connections from white-listed hosts, reject too fast
9638 connection without counting them, and track accepted connections.
9639 This results in connection rate being capped from abusive sources.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009640
Willy Tarreaue9656522010-08-17 15:40:09 +02009641 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009642 tcp-request connection reject if { src_conn_rate gt 10 }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009643 tcp-request connection track-sc0 src
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009644
Willy Tarreaue9656522010-08-17 15:40:09 +02009645 Example: accept all connections from white-listed hosts, count all other
9646 connections and reject too fast ones. This results in abusive ones
9647 being blocked as long as they don't slow down.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009648
Willy Tarreaue9656522010-08-17 15:40:09 +02009649 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009650 tcp-request connection track-sc0 src
9651 tcp-request connection reject if { sc0_conn_rate gt 10 }
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009652
Willy Tarreau4f0d9192013-06-11 20:40:55 +02009653 Example: enable the PROXY protocol for traffic coming from all known proxies.
9654
9655 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
9656
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009657 See section 7 about ACL usage.
9658
Willy Tarreau4f614292016-10-21 17:49:36 +02009659 See also : "tcp-request session", "tcp-request content", "stick-table"
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009660
9661
Willy Tarreaue9656522010-08-17 15:40:09 +02009662tcp-request content <action> [{if | unless} <condition>]
9663 Perform an action on a new session depending on a layer 4-7 condition
Willy Tarreau62644772008-07-16 18:36:06 +02009664 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +02009665 no | yes | yes | yes
Willy Tarreaue9656522010-08-17 15:40:09 +02009666 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02009667 <action> defines the action to perform if the condition applies. See
9668 below.
Willy Tarreau62644772008-07-16 18:36:06 +02009669
Willy Tarreaue9656522010-08-17 15:40:09 +02009670 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
Willy Tarreau62644772008-07-16 18:36:06 +02009671
Davor Ocelice9ed2812017-12-25 17:49:28 +01009672 A request's contents can be analyzed at an early stage of request processing
Willy Tarreaue9656522010-08-17 15:40:09 +02009673 called "TCP content inspection". During this stage, ACL-based rules are
9674 evaluated every time the request contents are updated, until either an
9675 "accept" or a "reject" rule matches, or the TCP request inspection delay
9676 expires with no matching rule.
Willy Tarreau62644772008-07-16 18:36:06 +02009677
Willy Tarreaue9656522010-08-17 15:40:09 +02009678 The first difference between these rules and "tcp-request connection" rules
9679 is that "tcp-request content" rules can make use of contents to take a
9680 decision. Most often, these decisions will consider a protocol recognition or
9681 validity. The second difference is that content-based rules can be used in
Willy Tarreauf3338342014-01-28 21:40:28 +01009682 both frontends and backends. In case of HTTP keep-alive with the client, all
9683 tcp-request content rules are evaluated again, so haproxy keeps a record of
9684 what sticky counters were assigned by a "tcp-request connection" versus a
9685 "tcp-request content" rule, and flushes all the content-related ones after
9686 processing an HTTP request, so that they may be evaluated again by the rules
9687 being evaluated again for the next request. This is of particular importance
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009688 when the rule tracks some L7 information or when it is conditioned by an
Willy Tarreauf3338342014-01-28 21:40:28 +01009689 L7-based ACL, since tracking may change between requests.
Willy Tarreau62644772008-07-16 18:36:06 +02009690
Willy Tarreaue9656522010-08-17 15:40:09 +02009691 Content-based rules are evaluated in their exact declaration order. If no
9692 rule matches or if there is no rule, the default action is to accept the
9693 contents. There is no specific limit to the number of rules which may be
9694 inserted.
Willy Tarreau62644772008-07-16 18:36:06 +02009695
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009696 Several types of actions are supported :
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009697 - accept : the request is accepted
9698 - reject : the request is rejected and the connection is closed
9699 - capture : the specified sample expression is captured
Patrick Hemmer268a7072018-05-11 12:52:31 -04009700 - set-priority-class <expr> | set-priority-offset <expr>
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009701 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02009702 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009703 - sc-inc-gpc1(<sc-id>)
Thierry Fournierb9125672016-03-29 19:34:37 +02009704 - sc-set-gpt0(<sc-id>) <int>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009705 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +01009706 - unset-var(<var-name>)
Willy Tarreau2d392c22015-08-24 01:43:45 +02009707 - silent-drop
Davor Ocelice9ed2812017-12-25 17:49:28 +01009708 - send-spoe-group <engine-name> <group-name>
Willy Tarreau62644772008-07-16 18:36:06 +02009709
Willy Tarreaue9656522010-08-17 15:40:09 +02009710 They have the same meaning as their counter-parts in "tcp-request connection"
9711 so please refer to that section for a complete description.
Willy Tarreau62644772008-07-16 18:36:06 +02009712
Willy Tarreauf3338342014-01-28 21:40:28 +01009713 While there is nothing mandatory about it, it is recommended to use the
9714 track-sc0 in "tcp-request connection" rules, track-sc1 for "tcp-request
9715 content" rules in the frontend, and track-sc2 for "tcp-request content"
9716 rules in the backend, because that makes the configuration more readable
9717 and easier to troubleshoot, but this is just a guideline and all counters
9718 may be used everywhere.
Willy Tarreau62644772008-07-16 18:36:06 +02009719
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009720 Note that the "if/unless" condition is optional. If no condition is set on
Willy Tarreaue9656522010-08-17 15:40:09 +02009721 the action, it is simply performed unconditionally. That can be useful for
9722 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau62644772008-07-16 18:36:06 +02009723
Willy Tarreaue9656522010-08-17 15:40:09 +02009724 It is perfectly possible to match layer 7 contents with "tcp-request content"
Willy Tarreauc0239e02012-04-16 14:42:55 +02009725 rules, since HTTP-specific ACL matches are able to preliminarily parse the
9726 contents of a buffer before extracting the required data. If the buffered
9727 contents do not parse as a valid HTTP message, then the ACL does not match.
9728 The parser which is involved there is exactly the same as for all other HTTP
Willy Tarreauf3338342014-01-28 21:40:28 +01009729 processing, so there is no risk of parsing something differently. In an HTTP
9730 backend connected to from an HTTP frontend, it is guaranteed that HTTP
9731 contents will always be immediately present when the rule is evaluated first.
Willy Tarreau62644772008-07-16 18:36:06 +02009732
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009733 Tracking layer7 information is also possible provided that the information
Willy Tarreau4d54c7c2014-09-16 15:48:15 +02009734 are present when the rule is processed. The rule processing engine is able to
9735 wait until the inspect delay expires when the data to be tracked is not yet
9736 available.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009737
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009738 The "set-var" is used to set the content of a variable. The variable is
Willy Tarreau4f614292016-10-21 17:49:36 +02009739 declared inline. For "tcp-request session" rules, only session-level
9740 variables can be used, without any layer7 contents.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009741
Daniel Schneller0b547052016-03-21 20:46:57 +01009742 <var-name> The name of the variable starts with an indication about
9743 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +01009744 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +01009745 "sess" : the variable is shared with the whole session
9746 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009747 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +01009748 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009749 processing
Daniel Schneller0b547052016-03-21 20:46:57 +01009750 "res" : the variable is shared only during response
9751 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009752 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +01009753 The name may only contain characters 'a-z', 'A-Z', '0-9',
9754 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009755
9756 <expr> Is a standard HAProxy expression formed by a sample-fetch
9757 followed by some converters.
9758
Christopher Faulet85d79c92016-11-09 16:54:56 +01009759 The "unset-var" is used to unset a variable. See above for details about
9760 <var-name>.
9761
Patrick Hemmer268a7072018-05-11 12:52:31 -04009762 The "set-priority-class" is used to set the queue priority class of the
9763 current request. The value must be a sample expression which converts to an
9764 integer in the range -2047..2047. Results outside this range will be
9765 truncated. The priority class determines the order in which queued requests
9766 are processed. Lower values have higher priority.
9767
9768 The "set-priority-offset" is used to set the queue priority timestamp offset
9769 of the current request. The value must be a sample expression which converts
9770 to an integer in the range -524287..524287. Results outside this range will be
9771 truncated. When a request is queued, it is ordered first by the priority
9772 class, then by the current timestamp adjusted by the given offset in
9773 milliseconds. Lower values have higher priority.
9774 Note that the resulting timestamp is is only tracked with enough precision for
9775 524,287ms (8m44s287ms). If the request is queued long enough to where the
9776 adjusted timestamp exceeds this value, it will be misidentified as highest
9777 priority. Thus it is important to set "timeout queue" to a value, where when
9778 combined with the offset, does not exceed this limit.
9779
Christopher Faulet76c09ef2017-09-21 11:03:52 +02009780 The "send-spoe-group" is used to trigger sending of a group of SPOE
9781 messages. To do so, the SPOE engine used to send messages must be defined, as
9782 well as the SPOE group to send. Of course, the SPOE engine must refer to an
9783 existing SPOE filter. If not engine name is provided on the SPOE filter line,
9784 the SPOE agent name must be used.
9785
9786 <engine-name> The SPOE engine name.
9787
9788 <group-name> The SPOE group name as specified in the engine configuration.
9789
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009790 Example:
9791
9792 tcp-request content set-var(sess.my_var) src
Christopher Faulet85d79c92016-11-09 16:54:56 +01009793 tcp-request content unset-var(sess.my_var2)
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009794
Willy Tarreau62644772008-07-16 18:36:06 +02009795 Example:
Willy Tarreaue9656522010-08-17 15:40:09 +02009796 # Accept HTTP requests containing a Host header saying "example.com"
9797 # and reject everything else.
9798 acl is_host_com hdr(Host) -i example.com
9799 tcp-request inspect-delay 30s
Willy Tarreauc0239e02012-04-16 14:42:55 +02009800 tcp-request content accept if is_host_com
Willy Tarreaue9656522010-08-17 15:40:09 +02009801 tcp-request content reject
9802
9803 Example:
Willy Tarreau62644772008-07-16 18:36:06 +02009804 # reject SMTP connection if client speaks first
9805 tcp-request inspect-delay 30s
9806 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009807 tcp-request content reject if content_present
Willy Tarreau62644772008-07-16 18:36:06 +02009808
9809 # Forward HTTPS connection only if client speaks
9810 tcp-request inspect-delay 30s
9811 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009812 tcp-request content accept if content_present
Willy Tarreaue9656522010-08-17 15:40:09 +02009813 tcp-request content reject
9814
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009815 Example:
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009816 # Track the last IP(stick-table type string) from X-Forwarded-For
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009817 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +02009818 tcp-request content track-sc0 hdr(x-forwarded-for,-1)
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009819 # Or track the last IP(stick-table type ip|ipv6) from X-Forwarded-For
9820 tcp-request content track-sc0 req.hdr_ip(x-forwarded-for,-1)
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009821
9822 Example:
9823 # track request counts per "base" (concatenation of Host+URL)
9824 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +02009825 tcp-request content track-sc0 base table req-rate
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009826
Willy Tarreaue9656522010-08-17 15:40:09 +02009827 Example: track per-frontend and per-backend counters, block abusers at the
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009828 frontend when the backend detects abuse(and marks gpc0).
Willy Tarreaue9656522010-08-17 15:40:09 +02009829
9830 frontend http
Davor Ocelice9ed2812017-12-25 17:49:28 +01009831 # Use General Purpose Counter 0 in SC0 as a global abuse counter
Willy Tarreaue9656522010-08-17 15:40:09 +02009832 # protecting all our sites
9833 stick-table type ip size 1m expire 5m store gpc0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009834 tcp-request connection track-sc0 src
9835 tcp-request connection reject if { sc0_get_gpc0 gt 0 }
Willy Tarreaue9656522010-08-17 15:40:09 +02009836 ...
9837 use_backend http_dynamic if { path_end .php }
9838
9839 backend http_dynamic
9840 # if a source makes too fast requests to this dynamic site (tracked
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009841 # by SC1), block it globally in the frontend.
Willy Tarreaue9656522010-08-17 15:40:09 +02009842 stick-table type ip size 1m expire 5m store http_req_rate(10s)
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009843 acl click_too_fast sc1_http_req_rate gt 10
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009844 acl mark_as_abuser sc0_inc_gpc0(http) gt 0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009845 tcp-request content track-sc1 src
Willy Tarreaue9656522010-08-17 15:40:09 +02009846 tcp-request content reject if click_too_fast mark_as_abuser
Willy Tarreau62644772008-07-16 18:36:06 +02009847
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009848 See section 7 about ACL usage.
Willy Tarreau62644772008-07-16 18:36:06 +02009849
Jarno Huuskonen95b012b2017-04-06 13:59:14 +03009850 See also : "tcp-request connection", "tcp-request session",
9851 "tcp-request inspect-delay", and "http-request".
Willy Tarreau62644772008-07-16 18:36:06 +02009852
9853
9854tcp-request inspect-delay <timeout>
9855 Set the maximum allowed time to wait for data during content inspection
9856 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +02009857 no | yes | yes | yes
Willy Tarreau62644772008-07-16 18:36:06 +02009858 Arguments :
9859 <timeout> is the timeout value specified in milliseconds by default, but
9860 can be in any other unit if the number is suffixed by the unit,
9861 as explained at the top of this document.
9862
9863 People using haproxy primarily as a TCP relay are often worried about the
9864 risk of passing any type of protocol to a server without any analysis. In
9865 order to be able to analyze the request contents, we must first withhold
9866 the data then analyze them. This statement simply enables withholding of
9867 data for at most the specified amount of time.
9868
Willy Tarreaufb356202010-08-03 14:02:05 +02009869 TCP content inspection applies very early when a connection reaches a
9870 frontend, then very early when the connection is forwarded to a backend. This
9871 means that a connection may experience a first delay in the frontend and a
9872 second delay in the backend if both have tcp-request rules.
9873
Willy Tarreau62644772008-07-16 18:36:06 +02009874 Note that when performing content inspection, haproxy will evaluate the whole
9875 rules for every new chunk which gets in, taking into account the fact that
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009876 those data are partial. If no rule matches before the aforementioned delay,
Willy Tarreau62644772008-07-16 18:36:06 +02009877 a last check is performed upon expiration, this time considering that the
Willy Tarreaud869b242009-03-15 14:43:58 +01009878 contents are definitive. If no delay is set, haproxy will not wait at all
9879 and will immediately apply a verdict based on the available information.
9880 Obviously this is unlikely to be very useful and might even be racy, so such
9881 setups are not recommended.
Willy Tarreau62644772008-07-16 18:36:06 +02009882
9883 As soon as a rule matches, the request is released and continues as usual. If
9884 the timeout is reached and no rule matches, the default policy will be to let
9885 it pass through unaffected.
9886
9887 For most protocols, it is enough to set it to a few seconds, as most clients
9888 send the full request immediately upon connection. Add 3 or more seconds to
9889 cover TCP retransmits but that's all. For some protocols, it may make sense
Willy Tarreaud72758d2010-01-12 10:42:19 +01009890 to use large values, for instance to ensure that the client never talks
Davor Ocelice9ed2812017-12-25 17:49:28 +01009891 before the server (e.g. SMTP), or to wait for a client to talk before passing
9892 data to the server (e.g. SSL). Note that the client timeout must cover at
Willy Tarreaub824b002010-09-29 16:36:16 +02009893 least the inspection delay, otherwise it will expire first. If the client
9894 closes the connection or if the buffer is full, the delay immediately expires
9895 since the contents will not be able to change anymore.
Willy Tarreau62644772008-07-16 18:36:06 +02009896
Willy Tarreau55165fe2009-05-10 12:02:55 +02009897 See also : "tcp-request content accept", "tcp-request content reject",
Willy Tarreau62644772008-07-16 18:36:06 +02009898 "timeout client".
9899
9900
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009901tcp-response content <action> [{if | unless} <condition>]
9902 Perform an action on a session response depending on a layer 4-7 condition
9903 May be used in sections : defaults | frontend | listen | backend
9904 no | no | yes | yes
9905 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02009906 <action> defines the action to perform if the condition applies. See
9907 below.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009908
9909 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
9910
Davor Ocelice9ed2812017-12-25 17:49:28 +01009911 Response contents can be analyzed at an early stage of response processing
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009912 called "TCP content inspection". During this stage, ACL-based rules are
9913 evaluated every time the response contents are updated, until either an
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02009914 "accept", "close" or a "reject" rule matches, or a TCP response inspection
9915 delay is set and expires with no matching rule.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009916
9917 Most often, these decisions will consider a protocol recognition or validity.
9918
9919 Content-based rules are evaluated in their exact declaration order. If no
9920 rule matches or if there is no rule, the default action is to accept the
9921 contents. There is no specific limit to the number of rules which may be
9922 inserted.
9923
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009924 Several types of actions are supported :
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009925 - accept :
9926 accepts the response if the condition is true (when used with "if")
9927 or false (when used with "unless"). The first such rule executed ends
9928 the rules evaluation.
9929
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02009930 - close :
9931 immediately closes the connection with the server if the condition is
9932 true (when used with "if"), or false (when used with "unless"). The
9933 first such rule executed ends the rules evaluation. The main purpose of
9934 this action is to force a connection to be finished between a client
9935 and a server after an exchange when the application protocol expects
9936 some long time outs to elapse first. The goal is to eliminate idle
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009937 connections which take significant resources on servers with certain
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02009938 protocols.
9939
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009940 - reject :
9941 rejects the response if the condition is true (when used with "if")
9942 or false (when used with "unless"). The first such rule executed ends
Jamie Gloudonaaa21002012-08-25 00:18:33 -04009943 the rules evaluation. Rejected session are immediately closed.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009944
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009945 - set-var(<var-name>) <expr>
9946 Sets a variable.
9947
Christopher Faulet85d79c92016-11-09 16:54:56 +01009948 - unset-var(<var-name>)
9949 Unsets a variable.
9950
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02009951 - sc-inc-gpc0(<sc-id>):
9952 This action increments the GPC0 counter according to the sticky
9953 counter designated by <sc-id>. If an error occurs, this action fails
9954 silently and the actions evaluation continues.
9955
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009956 - sc-inc-gpc1(<sc-id>):
9957 This action increments the GPC1 counter according to the sticky
9958 counter designated by <sc-id>. If an error occurs, this action fails
9959 silently and the actions evaluation continues.
9960
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009961 - sc-set-gpt0(<sc-id>) <int> :
9962 This action sets the GPT0 tag according to the sticky counter designated
9963 by <sc-id> and the value of <int>. The expected result is a boolean. If
9964 an error occurs, this action silently fails and the actions evaluation
9965 continues.
9966
Willy Tarreau2d392c22015-08-24 01:43:45 +02009967 - "silent-drop" :
9968 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +01009969 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +02009970 to prevent the client from being notified. The effect it then that the
9971 client still sees an established connection while there's none on
9972 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
9973 except that it doesn't use any local resource at all on the machine
9974 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +01009975 slow down stronger attackers. It is important to understand the impact
9976 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009977 client and HAProxy (firewalls, proxies, load balancers) will also keep
9978 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +01009979 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009980 TCP_REPAIR socket option is used to block the emission of a TCP
9981 reset. On other systems, the socket's TTL is reduced to 1 so that the
9982 TCP reset doesn't pass the first router, though it's still delivered to
9983 local networks. Do not use it unless you fully understand how it works.
9984
Christopher Faulet76c09ef2017-09-21 11:03:52 +02009985 - send-spoe-group <engine-name> <group-name>
9986 Send a group of SPOE messages.
9987
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009988 Note that the "if/unless" condition is optional. If no condition is set on
9989 the action, it is simply performed unconditionally. That can be useful for
9990 for changing the default action to a reject.
9991
Jamie Gloudonaaa21002012-08-25 00:18:33 -04009992 It is perfectly possible to match layer 7 contents with "tcp-response
9993 content" rules, but then it is important to ensure that a full response has
9994 been buffered, otherwise no contents will match. In order to achieve this,
9995 the best solution involves detecting the HTTP protocol during the inspection
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009996 period.
9997
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009998 The "set-var" is used to set the content of a variable. The variable is
9999 declared inline.
10000
Daniel Schneller0b547052016-03-21 20:46:57 +010010001 <var-name> The name of the variable starts with an indication about
10002 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010010003 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010010004 "sess" : the variable is shared with the whole session
10005 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010006 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +010010007 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010008 processing
Daniel Schneller0b547052016-03-21 20:46:57 +010010009 "res" : the variable is shared only during response
10010 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010011 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +010010012 The name may only contain characters 'a-z', 'A-Z', '0-9',
10013 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010014
10015 <expr> Is a standard HAProxy expression formed by a sample-fetch
10016 followed by some converters.
10017
10018 Example:
10019
10020 tcp-request content set-var(sess.my_var) src
10021
Christopher Faulet85d79c92016-11-09 16:54:56 +010010022 The "unset-var" is used to unset a variable. See above for details about
10023 <var-name>.
10024
10025 Example:
10026
10027 tcp-request content unset-var(sess.my_var)
10028
Christopher Faulet76c09ef2017-09-21 11:03:52 +020010029 The "send-spoe-group" is used to trigger sending of a group of SPOE
10030 messages. To do so, the SPOE engine used to send messages must be defined, as
10031 well as the SPOE group to send. Of course, the SPOE engine must refer to an
10032 existing SPOE filter. If not engine name is provided on the SPOE filter line,
10033 the SPOE agent name must be used.
10034
10035 <engine-name> The SPOE engine name.
10036
10037 <group-name> The SPOE group name as specified in the engine configuration.
10038
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010039 See section 7 about ACL usage.
10040
10041 See also : "tcp-request content", "tcp-response inspect-delay"
10042
10043
Willy Tarreau4f614292016-10-21 17:49:36 +020010044tcp-request session <action> [{if | unless} <condition>]
10045 Perform an action on a validated session depending on a layer 5 condition
10046 May be used in sections : defaults | frontend | listen | backend
10047 no | yes | yes | no
10048 Arguments :
10049 <action> defines the action to perform if the condition applies. See
10050 below.
10051
10052 <condition> is a standard layer5-only ACL-based condition (see section 7).
10053
Davor Ocelice9ed2812017-12-25 17:49:28 +010010054 Once a session is validated, (i.e. after all handshakes have been completed),
Willy Tarreau4f614292016-10-21 17:49:36 +020010055 it is possible to evaluate some conditions to decide whether this session
10056 must be accepted or dropped or have its counters tracked. Those conditions
10057 cannot make use of any data contents because no buffers are allocated yet and
10058 the processing cannot wait at this stage. The main use case it to copy some
10059 early information into variables (since variables are accessible in the
10060 session), or to keep track of some information collected after the handshake,
10061 such as SSL-level elements (SNI, ciphers, client cert's CN) or information
Davor Ocelice9ed2812017-12-25 17:49:28 +010010062 from the PROXY protocol header (e.g. track a source forwarded this way). The
Willy Tarreau4f614292016-10-21 17:49:36 +020010063 extracted information can thus be copied to a variable or tracked using
10064 "track-sc" rules. Of course it is also possible to decide to accept/reject as
10065 with other rulesets. Most operations performed here could also be performed
10066 in "tcp-request content" rules, except that in HTTP these rules are evaluated
10067 for each new request, and that might not always be acceptable. For example a
10068 rule might increment a counter on each evaluation. It would also be possible
10069 that a country is resolved by geolocation from the source IP address,
10070 assigned to a session-wide variable, then the source address rewritten from
10071 an HTTP header for all requests. If some contents need to be inspected in
10072 order to take the decision, the "tcp-request content" statements must be used
10073 instead.
10074
10075 The "tcp-request session" rules are evaluated in their exact declaration
10076 order. If no rule matches or if there is no rule, the default action is to
10077 accept the incoming session. There is no specific limit to the number of
10078 rules which may be inserted.
10079
10080 Several types of actions are supported :
10081 - accept : the request is accepted
10082 - reject : the request is rejected and the connection is closed
10083 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
10084 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +010010085 - sc-inc-gpc1(<sc-id>)
Willy Tarreau4f614292016-10-21 17:49:36 +020010086 - sc-set-gpt0(<sc-id>) <int>
10087 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +010010088 - unset-var(<var-name>)
Willy Tarreau4f614292016-10-21 17:49:36 +020010089 - silent-drop
10090
10091 These actions have the same meaning as their respective counter-parts in
10092 "tcp-request connection" and "tcp-request content", so please refer to these
10093 sections for a complete description.
10094
10095 Note that the "if/unless" condition is optional. If no condition is set on
10096 the action, it is simply performed unconditionally. That can be useful for
10097 "track-sc*" actions as well as for changing the default action to a reject.
10098
10099 Example: track the original source address by default, or the one advertised
10100 in the PROXY protocol header for connection coming from the local
10101 proxies. The first connection-level rule enables receipt of the
10102 PROXY protocol for these ones, the second rule tracks whatever
10103 address we decide to keep after optional decoding.
10104
10105 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
10106 tcp-request session track-sc0 src
10107
10108 Example: accept all sessions from white-listed hosts, reject too fast
10109 sessions without counting them, and track accepted sessions.
10110 This results in session rate being capped from abusive sources.
10111
10112 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
10113 tcp-request session reject if { src_sess_rate gt 10 }
10114 tcp-request session track-sc0 src
10115
10116 Example: accept all sessions from white-listed hosts, count all other
10117 sessions and reject too fast ones. This results in abusive ones
10118 being blocked as long as they don't slow down.
10119
10120 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
10121 tcp-request session track-sc0 src
10122 tcp-request session reject if { sc0_sess_rate gt 10 }
10123
10124 See section 7 about ACL usage.
10125
10126 See also : "tcp-request connection", "tcp-request content", "stick-table"
10127
10128
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010129tcp-response inspect-delay <timeout>
10130 Set the maximum allowed time to wait for a response during content inspection
10131 May be used in sections : defaults | frontend | listen | backend
10132 no | no | yes | yes
10133 Arguments :
10134 <timeout> is the timeout value specified in milliseconds by default, but
10135 can be in any other unit if the number is suffixed by the unit,
10136 as explained at the top of this document.
10137
10138 See also : "tcp-response content", "tcp-request inspect-delay".
10139
10140
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010141timeout check <timeout>
10142 Set additional check timeout, but only after a connection has been already
10143 established.
10144
10145 May be used in sections: defaults | frontend | listen | backend
10146 yes | no | yes | yes
10147 Arguments:
10148 <timeout> is the timeout value specified in milliseconds by default, but
10149 can be in any other unit if the number is suffixed by the unit,
10150 as explained at the top of this document.
10151
10152 If set, haproxy uses min("timeout connect", "inter") as a connect timeout
10153 for check and "timeout check" as an additional read timeout. The "min" is
Davor Ocelice9ed2812017-12-25 17:49:28 +010010154 used so that people running with *very* long "timeout connect" (e.g. those
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010155 who needed this due to the queue or tarpit) do not slow down their checks.
Willy Tarreaud7550a22010-02-10 05:10:19 +010010156 (Please also note that there is no valid reason to have such long connect
10157 timeouts, because "timeout queue" and "timeout tarpit" can always be used to
10158 avoid that).
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010159
10160 If "timeout check" is not set haproxy uses "inter" for complete check
10161 timeout (connect + read) exactly like all <1.3.15 version.
10162
10163 In most cases check request is much simpler and faster to handle than normal
10164 requests and people may want to kick out laggy servers so this timeout should
Willy Tarreau41a340d2008-01-22 12:25:31 +010010165 be smaller than "timeout server".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010166
10167 This parameter is specific to backends, but can be specified once for all in
10168 "defaults" sections. This is in fact one of the easiest solutions not to
10169 forget about it.
10170
Willy Tarreau41a340d2008-01-22 12:25:31 +010010171 See also: "timeout connect", "timeout queue", "timeout server",
10172 "timeout tarpit".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010173
10174
Willy Tarreau0ba27502007-12-24 16:55:16 +010010175timeout client <timeout>
10176timeout clitimeout <timeout> (deprecated)
10177 Set the maximum inactivity time on the client side.
10178 May be used in sections : defaults | frontend | listen | backend
10179 yes | yes | yes | no
10180 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010181 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010010182 can be in any other unit if the number is suffixed by the unit,
10183 as explained at the top of this document.
10184
10185 The inactivity timeout applies when the client is expected to acknowledge or
10186 send data. In HTTP mode, this timeout is particularly important to consider
10187 during the first phase, when the client sends the request, and during the
Baptiste Assmann2e1941e2016-03-06 23:24:12 +010010188 response while it is reading data sent by the server. That said, for the
10189 first phase, it is preferable to set the "timeout http-request" to better
10190 protect HAProxy from Slowloris like attacks. The value is specified in
10191 milliseconds by default, but can be in any other unit if the number is
Willy Tarreau0ba27502007-12-24 16:55:16 +010010192 suffixed by the unit, as specified at the top of this document. In TCP mode
10193 (and to a lesser extent, in HTTP mode), it is highly recommended that the
10194 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010195 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +010010196 losses by specifying timeouts that are slightly above multiples of 3 seconds
Davor Ocelice9ed2812017-12-25 17:49:28 +010010197 (e.g. 4 or 5 seconds). If some long-lived sessions are mixed with short-lived
10198 sessions (e.g. WebSocket and HTTP), it's worth considering "timeout tunnel",
Willy Tarreau05cdd962014-05-10 14:30:07 +020010199 which overrides "timeout client" and "timeout server" for tunnels, as well as
10200 "timeout client-fin" for half-closed connections.
Willy Tarreau0ba27502007-12-24 16:55:16 +010010201
10202 This parameter is specific to frontends, but can be specified once for all in
10203 "defaults" sections. This is in fact one of the easiest solutions not to
10204 forget about it. An unspecified timeout results in an infinite timeout, which
10205 is not recommended. Such a usage is accepted and works but reports a warning
10206 during startup because it may results in accumulation of expired sessions in
10207 the system if the system's timeouts are not configured either.
10208
Willy Tarreau95c4e142017-11-26 12:18:55 +010010209 This also applies to HTTP/2 connections, which will be closed with GOAWAY.
Lukas Tribus75df9d72017-11-24 19:05:12 +010010210
Willy Tarreau0ba27502007-12-24 16:55:16 +010010211 This parameter replaces the old, deprecated "clitimeout". It is recommended
10212 to use it to write new configurations. The form "timeout clitimeout" is
10213 provided only by backwards compatibility but its use is strongly discouraged.
10214
Baptiste Assmann2e1941e2016-03-06 23:24:12 +010010215 See also : "clitimeout", "timeout server", "timeout tunnel",
10216 "timeout http-request".
Willy Tarreau0ba27502007-12-24 16:55:16 +010010217
10218
Willy Tarreau05cdd962014-05-10 14:30:07 +020010219timeout client-fin <timeout>
10220 Set the inactivity timeout on the client side for half-closed connections.
10221 May be used in sections : defaults | frontend | listen | backend
10222 yes | yes | yes | no
10223 Arguments :
10224 <timeout> is the timeout value specified in milliseconds by default, but
10225 can be in any other unit if the number is suffixed by the unit,
10226 as explained at the top of this document.
10227
10228 The inactivity timeout applies when the client is expected to acknowledge or
10229 send data while one direction is already shut down. This timeout is different
10230 from "timeout client" in that it only applies to connections which are closed
10231 in one direction. This is particularly useful to avoid keeping connections in
10232 FIN_WAIT state for too long when clients do not disconnect cleanly. This
10233 problem is particularly common long connections such as RDP or WebSocket.
10234 Note that this timeout can override "timeout tunnel" when a connection shuts
Willy Tarreau599391a2017-11-24 10:16:00 +010010235 down in one direction. It is applied to idle HTTP/2 connections once a GOAWAY
10236 frame was sent, often indicating an expectation that the connection quickly
10237 ends.
Willy Tarreau05cdd962014-05-10 14:30:07 +020010238
10239 This parameter is specific to frontends, but can be specified once for all in
10240 "defaults" sections. By default it is not set, so half-closed connections
10241 will use the other timeouts (timeout.client or timeout.tunnel).
10242
10243 See also : "timeout client", "timeout server-fin", and "timeout tunnel".
10244
10245
Willy Tarreau0ba27502007-12-24 16:55:16 +010010246timeout connect <timeout>
10247timeout contimeout <timeout> (deprecated)
10248 Set the maximum time to wait for a connection attempt to a server to succeed.
10249 May be used in sections : defaults | frontend | listen | backend
10250 yes | no | yes | yes
10251 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010252 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010010253 can be in any other unit if the number is suffixed by the unit,
10254 as explained at the top of this document.
10255
10256 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010257 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +010010258 cover one or several TCP packet losses by specifying timeouts that are
Davor Ocelice9ed2812017-12-25 17:49:28 +010010259 slightly above multiples of 3 seconds (e.g. 4 or 5 seconds). By default, the
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010260 connect timeout also presets both queue and tarpit timeouts to the same value
10261 if these have not been specified.
Willy Tarreau0ba27502007-12-24 16:55:16 +010010262
10263 This parameter is specific to backends, but can be specified once for all in
10264 "defaults" sections. This is in fact one of the easiest solutions not to
10265 forget about it. An unspecified timeout results in an infinite timeout, which
10266 is not recommended. Such a usage is accepted and works but reports a warning
10267 during startup because it may results in accumulation of failed sessions in
10268 the system if the system's timeouts are not configured either.
10269
10270 This parameter replaces the old, deprecated "contimeout". It is recommended
10271 to use it to write new configurations. The form "timeout contimeout" is
10272 provided only by backwards compatibility but its use is strongly discouraged.
10273
Willy Tarreau41a340d2008-01-22 12:25:31 +010010274 See also: "timeout check", "timeout queue", "timeout server", "contimeout",
10275 "timeout tarpit".
Willy Tarreau0ba27502007-12-24 16:55:16 +010010276
10277
Willy Tarreaub16a5742010-01-10 14:46:16 +010010278timeout http-keep-alive <timeout>
10279 Set the maximum allowed time to wait for a new HTTP request to appear
10280 May be used in sections : defaults | frontend | listen | backend
10281 yes | yes | yes | yes
10282 Arguments :
10283 <timeout> is the timeout value specified in milliseconds by default, but
10284 can be in any other unit if the number is suffixed by the unit,
10285 as explained at the top of this document.
10286
10287 By default, the time to wait for a new request in case of keep-alive is set
10288 by "timeout http-request". However this is not always convenient because some
10289 people want very short keep-alive timeouts in order to release connections
10290 faster, and others prefer to have larger ones but still have short timeouts
10291 once the request has started to present itself.
10292
10293 The "http-keep-alive" timeout covers these needs. It will define how long to
10294 wait for a new HTTP request to start coming after a response was sent. Once
10295 the first byte of request has been seen, the "http-request" timeout is used
10296 to wait for the complete request to come. Note that empty lines prior to a
10297 new request do not refresh the timeout and are not counted as a new request.
10298
10299 There is also another difference between the two timeouts : when a connection
10300 expires during timeout http-keep-alive, no error is returned, the connection
10301 just closes. If the connection expires in "http-request" while waiting for a
10302 connection to complete, a HTTP 408 error is returned.
10303
10304 In general it is optimal to set this value to a few tens to hundreds of
10305 milliseconds, to allow users to fetch all objects of a page at once but
Davor Ocelice9ed2812017-12-25 17:49:28 +010010306 without waiting for further clicks. Also, if set to a very small value (e.g.
Willy Tarreaub16a5742010-01-10 14:46:16 +010010307 1 millisecond) it will probably only accept pipelined requests but not the
10308 non-pipelined ones. It may be a nice trade-off for very large sites running
Patrick Mézard2382ad62010-05-09 10:43:32 +020010309 with tens to hundreds of thousands of clients.
Willy Tarreaub16a5742010-01-10 14:46:16 +010010310
10311 If this parameter is not set, the "http-request" timeout applies, and if both
10312 are not set, "timeout client" still applies at the lower level. It should be
10313 set in the frontend to take effect, unless the frontend is in TCP mode, in
10314 which case the HTTP backend's timeout will be used.
10315
Willy Tarreau95c4e142017-11-26 12:18:55 +010010316 When using HTTP/2 "timeout client" is applied instead. This is so we can keep
10317 using short keep-alive timeouts in HTTP/1.1 while using longer ones in HTTP/2
Lukas Tribus75df9d72017-11-24 19:05:12 +010010318 (where we only have one connection per client and a connection setup).
10319
Willy Tarreaub16a5742010-01-10 14:46:16 +010010320 See also : "timeout http-request", "timeout client".
10321
10322
Willy Tarreau036fae02008-01-06 13:24:40 +010010323timeout http-request <timeout>
10324 Set the maximum allowed time to wait for a complete HTTP request
10325 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaucd7afc02009-07-12 10:03:17 +020010326 yes | yes | yes | yes
Willy Tarreau036fae02008-01-06 13:24:40 +010010327 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010328 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau036fae02008-01-06 13:24:40 +010010329 can be in any other unit if the number is suffixed by the unit,
10330 as explained at the top of this document.
10331
10332 In order to offer DoS protection, it may be required to lower the maximum
10333 accepted time to receive a complete HTTP request without affecting the client
10334 timeout. This helps protecting against established connections on which
10335 nothing is sent. The client timeout cannot offer a good protection against
10336 this abuse because it is an inactivity timeout, which means that if the
10337 attacker sends one character every now and then, the timeout will not
10338 trigger. With the HTTP request timeout, no matter what speed the client
Willy Tarreau2705a612014-05-23 17:38:34 +020010339 types, the request will be aborted if it does not complete in time. When the
10340 timeout expires, an HTTP 408 response is sent to the client to inform it
10341 about the problem, and the connection is closed. The logs will report
10342 termination codes "cR". Some recent browsers are having problems with this
Davor Ocelice9ed2812017-12-25 17:49:28 +010010343 standard, well-documented behavior, so it might be needed to hide the 408
Willy Tarreau0f228a02015-05-01 15:37:53 +020010344 code using "option http-ignore-probes" or "errorfile 408 /dev/null". See
10345 more details in the explanations of the "cR" termination code in section 8.5.
Willy Tarreau036fae02008-01-06 13:24:40 +010010346
Baptiste Assmanneccdf432015-10-28 13:49:01 +010010347 By default, this timeout only applies to the header part of the request,
10348 and not to any data. As soon as the empty line is received, this timeout is
10349 not used anymore. When combined with "option http-buffer-request", this
10350 timeout also applies to the body of the request..
10351 It is used again on keep-alive connections to wait for a second
Willy Tarreaub16a5742010-01-10 14:46:16 +010010352 request if "timeout http-keep-alive" is not set.
Willy Tarreau036fae02008-01-06 13:24:40 +010010353
10354 Generally it is enough to set it to a few seconds, as most clients send the
10355 full request immediately upon connection. Add 3 or more seconds to cover TCP
Davor Ocelice9ed2812017-12-25 17:49:28 +010010356 retransmits but that's all. Setting it to very low values (e.g. 50 ms) will
Willy Tarreau036fae02008-01-06 13:24:40 +010010357 generally work on local networks as long as there are no packet losses. This
10358 will prevent people from sending bare HTTP requests using telnet.
10359
10360 If this parameter is not set, the client timeout still applies between each
Willy Tarreaucd7afc02009-07-12 10:03:17 +020010361 chunk of the incoming request. It should be set in the frontend to take
10362 effect, unless the frontend is in TCP mode, in which case the HTTP backend's
10363 timeout will be used.
Willy Tarreau036fae02008-01-06 13:24:40 +010010364
Willy Tarreau0f228a02015-05-01 15:37:53 +020010365 See also : "errorfile", "http-ignore-probes", "timeout http-keep-alive", and
Baptiste Assmanneccdf432015-10-28 13:49:01 +010010366 "timeout client", "option http-buffer-request".
Willy Tarreau036fae02008-01-06 13:24:40 +010010367
Willy Tarreau844e3c52008-01-11 16:28:18 +010010368
10369timeout queue <timeout>
10370 Set the maximum time to wait in the queue for a connection slot to be free
10371 May be used in sections : defaults | frontend | listen | backend
10372 yes | no | yes | yes
10373 Arguments :
10374 <timeout> is the timeout value specified in milliseconds by default, but
10375 can be in any other unit if the number is suffixed by the unit,
10376 as explained at the top of this document.
10377
10378 When a server's maxconn is reached, connections are left pending in a queue
10379 which may be server-specific or global to the backend. In order not to wait
10380 indefinitely, a timeout is applied to requests pending in the queue. If the
10381 timeout is reached, it is considered that the request will almost never be
10382 served, so it is dropped and a 503 error is returned to the client.
10383
10384 The "timeout queue" statement allows to fix the maximum time for a request to
10385 be left pending in a queue. If unspecified, the same value as the backend's
10386 connection timeout ("timeout connect") is used, for backwards compatibility
10387 with older versions with no "timeout queue" parameter.
10388
10389 See also : "timeout connect", "contimeout".
10390
10391
10392timeout server <timeout>
10393timeout srvtimeout <timeout> (deprecated)
10394 Set the maximum inactivity time on the server side.
10395 May be used in sections : defaults | frontend | listen | backend
10396 yes | no | yes | yes
10397 Arguments :
10398 <timeout> is the timeout value specified in milliseconds by default, but
10399 can be in any other unit if the number is suffixed by the unit,
10400 as explained at the top of this document.
10401
10402 The inactivity timeout applies when the server is expected to acknowledge or
10403 send data. In HTTP mode, this timeout is particularly important to consider
10404 during the first phase of the server's response, when it has to send the
10405 headers, as it directly represents the server's processing time for the
10406 request. To find out what value to put there, it's often good to start with
10407 what would be considered as unacceptable response times, then check the logs
10408 to observe the response time distribution, and adjust the value accordingly.
10409
10410 The value is specified in milliseconds by default, but can be in any other
10411 unit if the number is suffixed by the unit, as specified at the top of this
10412 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
10413 recommended that the client timeout remains equal to the server timeout in
10414 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010415 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +010010416 packet losses by specifying timeouts that are slightly above multiples of 3
Davor Ocelice9ed2812017-12-25 17:49:28 +010010417 seconds (e.g. 4 or 5 seconds minimum). If some long-lived sessions are mixed
10418 with short-lived sessions (e.g. WebSocket and HTTP), it's worth considering
Willy Tarreauce887fd2012-05-12 12:50:00 +020010419 "timeout tunnel", which overrides "timeout client" and "timeout server" for
10420 tunnels.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010421
10422 This parameter is specific to backends, but can be specified once for all in
10423 "defaults" sections. This is in fact one of the easiest solutions not to
10424 forget about it. An unspecified timeout results in an infinite timeout, which
10425 is not recommended. Such a usage is accepted and works but reports a warning
10426 during startup because it may results in accumulation of expired sessions in
10427 the system if the system's timeouts are not configured either.
10428
10429 This parameter replaces the old, deprecated "srvtimeout". It is recommended
10430 to use it to write new configurations. The form "timeout srvtimeout" is
10431 provided only by backwards compatibility but its use is strongly discouraged.
10432
Willy Tarreauce887fd2012-05-12 12:50:00 +020010433 See also : "srvtimeout", "timeout client" and "timeout tunnel".
Willy Tarreau844e3c52008-01-11 16:28:18 +010010434
Willy Tarreau05cdd962014-05-10 14:30:07 +020010435
10436timeout server-fin <timeout>
10437 Set the inactivity timeout on the server side for half-closed connections.
10438 May be used in sections : defaults | frontend | listen | backend
10439 yes | no | yes | yes
10440 Arguments :
10441 <timeout> is the timeout value specified in milliseconds by default, but
10442 can be in any other unit if the number is suffixed by the unit,
10443 as explained at the top of this document.
10444
10445 The inactivity timeout applies when the server is expected to acknowledge or
10446 send data while one direction is already shut down. This timeout is different
10447 from "timeout server" in that it only applies to connections which are closed
10448 in one direction. This is particularly useful to avoid keeping connections in
10449 FIN_WAIT state for too long when a remote server does not disconnect cleanly.
10450 This problem is particularly common long connections such as RDP or WebSocket.
10451 Note that this timeout can override "timeout tunnel" when a connection shuts
10452 down in one direction. This setting was provided for completeness, but in most
10453 situations, it should not be needed.
10454
10455 This parameter is specific to backends, but can be specified once for all in
10456 "defaults" sections. By default it is not set, so half-closed connections
10457 will use the other timeouts (timeout.server or timeout.tunnel).
10458
10459 See also : "timeout client-fin", "timeout server", and "timeout tunnel".
10460
Willy Tarreau844e3c52008-01-11 16:28:18 +010010461
10462timeout tarpit <timeout>
Cyril Bonté78caf842010-03-10 22:41:43 +010010463 Set the duration for which tarpitted connections will be maintained
Willy Tarreau844e3c52008-01-11 16:28:18 +010010464 May be used in sections : defaults | frontend | listen | backend
10465 yes | yes | yes | yes
10466 Arguments :
10467 <timeout> is the tarpit duration specified in milliseconds by default, but
10468 can be in any other unit if the number is suffixed by the unit,
10469 as explained at the top of this document.
10470
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030010471 When a connection is tarpitted using "http-request tarpit" or
10472 "reqtarpit", it is maintained open with no activity for a certain
10473 amount of time, then closed. "timeout tarpit" defines how long it will
10474 be maintained open.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010475
10476 The value is specified in milliseconds by default, but can be in any other
10477 unit if the number is suffixed by the unit, as specified at the top of this
10478 document. If unspecified, the same value as the backend's connection timeout
10479 ("timeout connect") is used, for backwards compatibility with older versions
Cyril Bonté78caf842010-03-10 22:41:43 +010010480 with no "timeout tarpit" parameter.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010481
10482 See also : "timeout connect", "contimeout".
10483
10484
Willy Tarreauce887fd2012-05-12 12:50:00 +020010485timeout tunnel <timeout>
10486 Set the maximum inactivity time on the client and server side for tunnels.
10487 May be used in sections : defaults | frontend | listen | backend
10488 yes | no | yes | yes
10489 Arguments :
10490 <timeout> is the timeout value specified in milliseconds by default, but
10491 can be in any other unit if the number is suffixed by the unit,
10492 as explained at the top of this document.
10493
Jamie Gloudonaaa21002012-08-25 00:18:33 -040010494 The tunnel timeout applies when a bidirectional connection is established
Willy Tarreauce887fd2012-05-12 12:50:00 +020010495 between a client and a server, and the connection remains inactive in both
10496 directions. This timeout supersedes both the client and server timeouts once
10497 the connection becomes a tunnel. In TCP, this timeout is used as soon as no
Davor Ocelice9ed2812017-12-25 17:49:28 +010010498 analyzer remains attached to either connection (e.g. tcp content rules are
10499 accepted). In HTTP, this timeout is used when a connection is upgraded (e.g.
Willy Tarreauce887fd2012-05-12 12:50:00 +020010500 when switching to the WebSocket protocol, or forwarding a CONNECT request
10501 to a proxy), or after the first response when no keepalive/close option is
10502 specified.
10503
Willy Tarreau05cdd962014-05-10 14:30:07 +020010504 Since this timeout is usually used in conjunction with long-lived connections,
10505 it usually is a good idea to also set "timeout client-fin" to handle the
10506 situation where a client suddenly disappears from the net and does not
10507 acknowledge a close, or sends a shutdown and does not acknowledge pending
10508 data anymore. This can happen in lossy networks where firewalls are present,
10509 and is detected by the presence of large amounts of sessions in a FIN_WAIT
10510 state.
10511
Willy Tarreauce887fd2012-05-12 12:50:00 +020010512 The value is specified in milliseconds by default, but can be in any other
10513 unit if the number is suffixed by the unit, as specified at the top of this
10514 document. Whatever the expected normal idle time, it is a good practice to
10515 cover at least one or several TCP packet losses by specifying timeouts that
Davor Ocelice9ed2812017-12-25 17:49:28 +010010516 are slightly above multiples of 3 seconds (e.g. 4 or 5 seconds minimum).
Willy Tarreauce887fd2012-05-12 12:50:00 +020010517
10518 This parameter is specific to backends, but can be specified once for all in
10519 "defaults" sections. This is in fact one of the easiest solutions not to
10520 forget about it.
10521
10522 Example :
10523 defaults http
10524 option http-server-close
10525 timeout connect 5s
10526 timeout client 30s
Willy Tarreau05cdd962014-05-10 14:30:07 +020010527 timeout client-fin 30s
Willy Tarreauce887fd2012-05-12 12:50:00 +020010528 timeout server 30s
10529 timeout tunnel 1h # timeout to use with WebSocket and CONNECT
10530
Willy Tarreau05cdd962014-05-10 14:30:07 +020010531 See also : "timeout client", "timeout client-fin", "timeout server".
Willy Tarreauce887fd2012-05-12 12:50:00 +020010532
10533
Willy Tarreau844e3c52008-01-11 16:28:18 +010010534transparent (deprecated)
10535 Enable client-side transparent proxying
10536 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +010010537 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +010010538 Arguments : none
10539
10540 This keyword was introduced in order to provide layer 7 persistence to layer
10541 3 load balancers. The idea is to use the OS's ability to redirect an incoming
10542 connection for a remote address to a local process (here HAProxy), and let
10543 this process know what address was initially requested. When this option is
10544 used, sessions without cookies will be forwarded to the original destination
10545 IP address of the incoming request (which should match that of another
10546 equipment), while requests with cookies will still be forwarded to the
10547 appropriate server.
10548
10549 The "transparent" keyword is deprecated, use "option transparent" instead.
10550
10551 Note that contrary to a common belief, this option does NOT make HAProxy
10552 present the client's IP to the server when establishing the connection.
10553
Willy Tarreau844e3c52008-01-11 16:28:18 +010010554 See also: "option transparent"
10555
William Lallemanda73203e2012-03-12 12:48:57 +010010556unique-id-format <string>
10557 Generate a unique ID for each request.
10558 May be used in sections : defaults | frontend | listen | backend
10559 yes | yes | yes | no
10560 Arguments :
10561 <string> is a log-format string.
10562
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010563 This keyword creates a ID for each request using the custom log format. A
10564 unique ID is useful to trace a request passing through many components of
10565 a complex infrastructure. The newly created ID may also be logged using the
10566 %ID tag the log-format string.
William Lallemanda73203e2012-03-12 12:48:57 +010010567
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010568 The format should be composed from elements that are guaranteed to be
10569 unique when combined together. For instance, if multiple haproxy instances
10570 are involved, it might be important to include the node name. It is often
10571 needed to log the incoming connection's source and destination addresses
10572 and ports. Note that since multiple requests may be performed over the same
10573 connection, including a request counter may help differentiate them.
10574 Similarly, a timestamp may protect against a rollover of the counter.
10575 Logging the process ID will avoid collisions after a service restart.
William Lallemanda73203e2012-03-12 12:48:57 +010010576
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010577 It is recommended to use hexadecimal notation for many fields since it
10578 makes them more compact and saves space in logs.
William Lallemanda73203e2012-03-12 12:48:57 +010010579
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010580 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010010581
Julien Vehentf21be322014-03-07 08:27:34 -050010582 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010010583
10584 will generate:
10585
10586 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
10587
10588 See also: "unique-id-header"
10589
10590unique-id-header <name>
10591 Add a unique ID header in the HTTP request.
10592 May be used in sections : defaults | frontend | listen | backend
10593 yes | yes | yes | no
10594 Arguments :
10595 <name> is the name of the header.
10596
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010597 Add a unique-id header in the HTTP request sent to the server, using the
10598 unique-id-format. It can't work if the unique-id-format doesn't exist.
William Lallemanda73203e2012-03-12 12:48:57 +010010599
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010600 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010010601
Julien Vehentf21be322014-03-07 08:27:34 -050010602 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010010603 unique-id-header X-Unique-ID
10604
10605 will generate:
10606
10607 X-Unique-ID: 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
10608
10609 See also: "unique-id-format"
Willy Tarreau844e3c52008-01-11 16:28:18 +010010610
Willy Tarreauf51658d2014-04-23 01:21:56 +020010611use_backend <backend> [{if | unless} <condition>]
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020010612 Switch to a specific backend if/unless an ACL-based condition is matched.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010613 May be used in sections : defaults | frontend | listen | backend
10614 no | yes | yes | no
10615 Arguments :
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010010616 <backend> is the name of a valid backend or "listen" section, or a
10617 "log-format" string resolving to a backend name.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010618
Willy Tarreauf51658d2014-04-23 01:21:56 +020010619 <condition> is a condition composed of ACLs, as described in section 7. If
10620 it is omitted, the rule is unconditionally applied.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010621
10622 When doing content-switching, connections arrive on a frontend and are then
10623 dispatched to various backends depending on a number of conditions. The
10624 relation between the conditions and the backends is described with the
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020010625 "use_backend" keyword. While it is normally used with HTTP processing, it can
Davor Ocelice9ed2812017-12-25 17:49:28 +010010626 also be used in pure TCP, either without content using stateless ACLs (e.g.
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020010627 source address validation) or combined with a "tcp-request" rule to wait for
10628 some payload.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010629
10630 There may be as many "use_backend" rules as desired. All of these rules are
10631 evaluated in their declaration order, and the first one which matches will
10632 assign the backend.
10633
10634 In the first form, the backend will be used if the condition is met. In the
10635 second form, the backend will be used if the condition is not met. If no
10636 condition is valid, the backend defined with "default_backend" will be used.
10637 If no default backend is defined, either the servers in the same section are
10638 used (in case of a "listen" section) or, in case of a frontend, no server is
10639 used and a 503 service unavailable response is returned.
10640
Willy Tarreau51aecc72009-07-12 09:47:04 +020010641 Note that it is possible to switch from a TCP frontend to an HTTP backend. In
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010010642 this case, either the frontend has already checked that the protocol is HTTP,
Willy Tarreau51aecc72009-07-12 09:47:04 +020010643 and backend processing will immediately follow, or the backend will wait for
10644 a complete HTTP request to get in. This feature is useful when a frontend
10645 must decode several protocols on a unique port, one of them being HTTP.
10646
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010010647 When <backend> is a simple name, it is resolved at configuration time, and an
10648 error is reported if the specified backend does not exist. If <backend> is
10649 a log-format string instead, no check may be done at configuration time, so
10650 the backend name is resolved dynamically at run time. If the resulting
10651 backend name does not correspond to any valid backend, no other rule is
10652 evaluated, and the default_backend directive is applied instead. Note that
10653 when using dynamic backend names, it is highly recommended to use a prefix
10654 that no other backend uses in order to ensure that an unauthorized backend
10655 cannot be forced from the request.
10656
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010657 It is worth mentioning that "use_backend" rules with an explicit name are
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010010658 used to detect the association between frontends and backends to compute the
10659 backend's "fullconn" setting. This cannot be done for dynamic names.
10660
10661 See also: "default_backend", "tcp-request", "fullconn", "log-format", and
10662 section 7 about ACLs.
Willy Tarreaud72758d2010-01-12 10:42:19 +010010663
Willy Tarreau036fae02008-01-06 13:24:40 +010010664
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010665use-server <server> if <condition>
10666use-server <server> unless <condition>
10667 Only use a specific server if/unless an ACL-based condition is matched.
10668 May be used in sections : defaults | frontend | listen | backend
10669 no | no | yes | yes
10670 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010671 <server> is the name of a valid server in the same backend section.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010672
10673 <condition> is a condition composed of ACLs, as described in section 7.
10674
10675 By default, connections which arrive to a backend are load-balanced across
10676 the available servers according to the configured algorithm, unless a
10677 persistence mechanism such as a cookie is used and found in the request.
10678
10679 Sometimes it is desirable to forward a particular request to a specific
10680 server without having to declare a dedicated backend for this server. This
10681 can be achieved using the "use-server" rules. These rules are evaluated after
10682 the "redirect" rules and before evaluating cookies, and they have precedence
10683 on them. There may be as many "use-server" rules as desired. All of these
10684 rules are evaluated in their declaration order, and the first one which
10685 matches will assign the server.
10686
10687 If a rule designates a server which is down, and "option persist" is not used
10688 and no force-persist rule was validated, it is ignored and evaluation goes on
10689 with the next rules until one matches.
10690
10691 In the first form, the server will be used if the condition is met. In the
10692 second form, the server will be used if the condition is not met. If no
10693 condition is valid, the processing continues and the server will be assigned
10694 according to other persistence mechanisms.
10695
10696 Note that even if a rule is matched, cookie processing is still performed but
10697 does not assign the server. This allows prefixed cookies to have their prefix
10698 stripped.
10699
10700 The "use-server" statement works both in HTTP and TCP mode. This makes it
10701 suitable for use with content-based inspection. For instance, a server could
10702 be selected in a farm according to the TLS SNI field. And if these servers
10703 have their weight set to zero, they will not be used for other traffic.
10704
10705 Example :
10706 # intercept incoming TLS requests based on the SNI field
10707 use-server www if { req_ssl_sni -i www.example.com }
10708 server www 192.168.0.1:443 weight 0
10709 use-server mail if { req_ssl_sni -i mail.example.com }
10710 server mail 192.168.0.1:587 weight 0
10711 use-server imap if { req_ssl_sni -i imap.example.com }
Lukas Tribus98a3e3f2017-03-26 12:55:35 +000010712 server imap 192.168.0.1:993 weight 0
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010713 # all the rest is forwarded to this server
10714 server default 192.168.0.2:443 check
10715
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010716 See also: "use_backend", section 5 about server and section 7 about ACLs.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010717
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010718
Davor Ocelice9ed2812017-12-25 17:49:28 +0100107195. Bind and server options
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010720--------------------------
10721
10722The "bind", "server" and "default-server" keywords support a number of settings
10723depending on some build options and on the system HAProxy was built on. These
10724settings generally each consist in one word sometimes followed by a value,
10725written on the same line as the "bind" or "server" line. All these options are
10726described in this section.
10727
10728
107295.1. Bind options
10730-----------------
10731
10732The "bind" keyword supports a certain number of settings which are all passed
10733as arguments on the same line. The order in which those arguments appear makes
10734no importance, provided that they appear after the bind address. All of these
10735parameters are optional. Some of them consist in a single words (booleans),
10736while other ones expect a value after them. In this case, the value must be
10737provided immediately after the setting name.
10738
10739The currently supported settings are the following ones.
10740
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010010741accept-netscaler-cip <magic number>
10742 Enforces the use of the NetScaler Client IP insertion protocol over any
10743 connection accepted by any of the TCP sockets declared on the same line. The
10744 NetScaler Client IP insertion protocol dictates the layer 3/4 addresses of
10745 the incoming connection to be used everywhere an address is used, with the
10746 only exception of "tcp-request connection" rules which will only see the
10747 real connection address. Logs will reflect the addresses indicated in the
10748 protocol, unless it is violated, in which case the real address will still
10749 be used. This keyword combined with support from external components can be
10750 used as an efficient and reliable alternative to the X-Forwarded-For
Bertrand Jacquin90759682016-06-06 15:35:39 +010010751 mechanism which is not always reliable and not even always usable. See also
10752 "tcp-request connection expect-netscaler-cip" for a finer-grained setting of
10753 which client is allowed to use the protocol.
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010010754
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010755accept-proxy
10756 Enforces the use of the PROXY protocol over any connection accepted by any of
Willy Tarreau77992672014-06-14 11:06:17 +020010757 the sockets declared on the same line. Versions 1 and 2 of the PROXY protocol
10758 are supported and correctly detected. The PROXY protocol dictates the layer
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010759 3/4 addresses of the incoming connection to be used everywhere an address is
10760 used, with the only exception of "tcp-request connection" rules which will
10761 only see the real connection address. Logs will reflect the addresses
10762 indicated in the protocol, unless it is violated, in which case the real
Davor Ocelice9ed2812017-12-25 17:49:28 +010010763 address will still be used. This keyword combined with support from external
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010764 components can be used as an efficient and reliable alternative to the
10765 X-Forwarded-For mechanism which is not always reliable and not even always
Willy Tarreau4f0d9192013-06-11 20:40:55 +020010766 usable. See also "tcp-request connection expect-proxy" for a finer-grained
10767 setting of which client is allowed to use the protocol.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010768
Olivier Houchardc2aae742017-09-22 18:26:28 +020010769allow-0rtt
Bertrand Jacquina25282b2018-08-14 00:56:13 +010010770 Allow receiving early data when using TLSv1.3. This is disabled by default,
Olivier Houchardc2aae742017-09-22 18:26:28 +020010771 due to security considerations.
10772
Willy Tarreauab861d32013-04-02 02:30:41 +020010773alpn <protocols>
10774 This enables the TLS ALPN extension and advertises the specified protocol
10775 list as supported on top of ALPN. The protocol list consists in a comma-
10776 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
10777 quotes). This requires that the SSL library is build with support for TLS
10778 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
Willy Tarreau95c4e142017-11-26 12:18:55 +010010779 initial NPN extension. ALPN is required to enable HTTP/2 on an HTTP frontend.
10780 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
10781 now obsolete NPN extension. At the time of writing this, most browsers still
10782 support both ALPN and NPN for HTTP/2 so a fallback to NPN may still work for
10783 a while. But ALPN must be used whenever possible. If both HTTP/2 and HTTP/1.1
10784 are expected to be supported, both versions can be advertised, in order of
10785 preference, like below :
10786
10787 bind :443 ssl crt pub.pem alpn h2,http/1.1
Willy Tarreauab861d32013-04-02 02:30:41 +020010788
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010789backlog <backlog>
10790 Sets the socket's backlog to this value. If unspecified, the frontend's
10791 backlog is used instead, which generally defaults to the maxconn value.
10792
Emmanuel Hocdete7f2b732017-01-09 16:15:54 +010010793curves <curves>
10794 This setting is only available when support for OpenSSL was built in. It sets
10795 the string describing the list of elliptic curves algorithms ("curve suite")
10796 that are negotiated during the SSL/TLS handshake with ECDHE. The format of the
10797 string is a colon-delimited list of curve name.
10798 Example: "X25519:P-256" (without quote)
10799 When "curves" is set, "ecdhe" parameter is ignored.
10800
Emeric Brun7fb34422012-09-28 15:26:15 +020010801ecdhe <named curve>
10802 This setting is only available when support for OpenSSL was built in. It sets
Emeric Brun6924ef82013-03-06 14:08:53 +010010803 the named curve (RFC 4492) used to generate ECDH ephemeral keys. By default,
10804 used named curve is prime256v1.
Emeric Brun7fb34422012-09-28 15:26:15 +020010805
Emeric Brunfd33a262012-10-11 16:28:27 +020010806ca-file <cafile>
Emeric Brun1a073b42012-09-28 17:07:34 +020010807 This setting is only available when support for OpenSSL was built in. It
10808 designates a PEM file from which to load CA certificates used to verify
10809 client's certificate.
10810
Emeric Brunb6dc9342012-09-28 17:55:37 +020010811ca-ignore-err [all|<errorID>,...]
10812 This setting is only available when support for OpenSSL was built in.
10813 Sets a comma separated list of errorIDs to ignore during verify at depth > 0.
10814 If set to 'all', all errors are ignored. SSL handshake is not aborted if an
10815 error is ignored.
10816
Christopher Faulet31af49d2015-06-09 17:29:50 +020010817ca-sign-file <cafile>
10818 This setting is only available when support for OpenSSL was built in. It
10819 designates a PEM file containing both the CA certificate and the CA private
10820 key used to create and sign server's certificates. This is a mandatory
10821 setting when the dynamic generation of certificates is enabled. See
10822 'generate-certificates' for details.
10823
Bertrand Jacquind4d0a232016-11-13 16:37:12 +000010824ca-sign-pass <passphrase>
Christopher Faulet31af49d2015-06-09 17:29:50 +020010825 This setting is only available when support for OpenSSL was built in. It is
10826 the CA private key passphrase. This setting is optional and used only when
10827 the dynamic generation of certificates is enabled. See
10828 'generate-certificates' for details.
10829
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010830ciphers <ciphers>
10831 This setting is only available when support for OpenSSL was built in. It sets
10832 the string describing the list of cipher algorithms ("cipher suite") that are
Dirkjan Bussink415150f2018-09-14 11:14:21 +020010833 negotiated during the SSL/TLS handshake except for TLSv1.3. The format of the
10834 string is defined in "man 1 ciphers" from OpenSSL man pages, and can be for
10835 instance a string such as "AES:ALL:!aNULL:!eNULL:+RC4:@STRENGTH" (without
10836 quotes). Depending on the compatibility and security requirements, the list
10837 of suitable ciphers depends on a variety of variables. For background
10838 information and recommendations see e.g.
10839 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
10840 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
10841 cipher configuration, please check the "ciphersuites" keyword.
10842
10843ciphersuites <ciphersuites>
10844 This setting is only available when support for OpenSSL was built in and
10845 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the string describing
10846 the list of cipher algorithms ("cipher suite") that are negotiated during the
10847 TLSv1.3 handshake. The format of the string is defined in "man 1 ciphers" from
10848 OpenSSL man pages under the "ciphersuites" section, and can be for instance a
10849 string such as
10850 "TLS_AES_256_GCM_SHA384:TLS_CHACHA20_POLY1305_SHA256:TLS_AES_128_GCM_SHA256"
10851 (without quotes). For cipher configuration for TLSv1.2 and earlier, please check
10852 the "ciphers" keyword.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010853
Emeric Brunfd33a262012-10-11 16:28:27 +020010854crl-file <crlfile>
Emeric Brun1a073b42012-09-28 17:07:34 +020010855 This setting is only available when support for OpenSSL was built in. It
10856 designates a PEM file from which to load certificate revocation list used
10857 to verify client's certificate.
10858
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010859crt <cert>
Alex Davies0fbf0162013-03-02 16:04:50 +000010860 This setting is only available when support for OpenSSL was built in. It
10861 designates a PEM file containing both the required certificates and any
10862 associated private keys. This file can be built by concatenating multiple
10863 PEM files into one (e.g. cat cert.pem key.pem > combined.pem). If your CA
10864 requires an intermediate certificate, this can also be concatenated into this
10865 file.
10866
10867 If the OpenSSL used supports Diffie-Hellman, parameters present in this file
10868 are loaded.
10869
10870 If a directory name is used instead of a PEM file, then all files found in
Cyril Bonté3180f7b2015-01-25 00:16:08 +010010871 that directory will be loaded in alphabetic order unless their name ends with
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010010872 '.issuer', '.ocsp' or '.sctl' (reserved extensions). This directive may be
10873 specified multiple times in order to load certificates from multiple files or
10874 directories. The certificates will be presented to clients who provide a
10875 valid TLS Server Name Indication field matching one of their CN or alt
Davor Ocelice9ed2812017-12-25 17:49:28 +010010876 subjects. Wildcards are supported, where a wildcard character '*' is used
10877 instead of the first hostname component (e.g. *.example.org matches
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010010878 www.example.org but not www.sub.example.org).
Alex Davies0fbf0162013-03-02 16:04:50 +000010879
10880 If no SNI is provided by the client or if the SSL library does not support
10881 TLS extensions, or if the client provides an SNI hostname which does not
10882 match any certificate, then the first loaded certificate will be presented.
10883 This means that when loading certificates from a directory, it is highly
Cyril Bonté3180f7b2015-01-25 00:16:08 +010010884 recommended to load the default one first as a file or to ensure that it will
10885 always be the first one in the directory.
Alex Davies0fbf0162013-03-02 16:04:50 +000010886
Emeric Brune032bfa2012-09-28 13:01:45 +020010887 Note that the same cert may be loaded multiple times without side effects.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010888
Davor Ocelice9ed2812017-12-25 17:49:28 +010010889 Some CAs (such as GoDaddy) offer a drop down list of server types that do not
Alex Davies0fbf0162013-03-02 16:04:50 +000010890 include HAProxy when obtaining a certificate. If this happens be sure to
Davor Ocelice9ed2812017-12-25 17:49:28 +010010891 choose a web server that the CA believes requires an intermediate CA (for
10892 GoDaddy, selection Apache Tomcat will get the correct bundle, but many
Alex Davies0fbf0162013-03-02 16:04:50 +000010893 others, e.g. nginx, result in a wrong bundle that will not work for some
10894 clients).
10895
Emeric Brun4147b2e2014-06-16 18:36:30 +020010896 For each PEM file, haproxy checks for the presence of file at the same path
10897 suffixed by ".ocsp". If such file is found, support for the TLS Certificate
10898 Status Request extension (also known as "OCSP stapling") is automatically
10899 enabled. The content of this file is optional. If not empty, it must contain
10900 a valid OCSP Response in DER format. In order to be valid an OCSP Response
10901 must comply with the following rules: it has to indicate a good status,
10902 it has to be a single response for the certificate of the PEM file, and it
10903 has to be valid at the moment of addition. If these rules are not respected
10904 the OCSP Response is ignored and a warning is emitted. In order to identify
10905 which certificate an OCSP Response applies to, the issuer's certificate is
10906 necessary. If the issuer's certificate is not found in the PEM file, it will
10907 be loaded from a file at the same path as the PEM file suffixed by ".issuer"
10908 if it exists otherwise it will fail with an error.
10909
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010010910 For each PEM file, haproxy also checks for the presence of file at the same
10911 path suffixed by ".sctl". If such file is found, support for Certificate
10912 Transparency (RFC6962) TLS extension is enabled. The file must contain a
10913 valid Signed Certificate Timestamp List, as described in RFC. File is parsed
10914 to check basic syntax, but no signatures are verified.
10915
yanbzhu6c25e9e2016-01-05 12:52:02 -050010916 There are cases where it is desirable to support multiple key types, e.g. RSA
10917 and ECDSA in the cipher suites offered to the clients. This allows clients
10918 that support EC certificates to be able to use EC ciphers, while
10919 simultaneously supporting older, RSA only clients.
yanbzhud19630c2015-12-14 15:10:25 -050010920
10921 In order to provide this functionality, multiple PEM files, each with a
10922 different key type, are required. To associate these PEM files into a
10923 "cert bundle" that is recognized by haproxy, they must be named in the
10924 following way: All PEM files that are to be bundled must have the same base
10925 name, with a suffix indicating the key type. Currently, three suffixes are
10926 supported: rsa, dsa and ecdsa. For example, if www.example.com has two PEM
10927 files, an RSA file and an ECDSA file, they must be named: "example.pem.rsa"
10928 and "example.pem.ecdsa". The first part of the filename is arbitrary; only the
10929 suffix matters. To load this bundle into haproxy, specify the base name only:
10930
10931 Example : bind :8443 ssl crt example.pem
10932
yanbzhu6c25e9e2016-01-05 12:52:02 -050010933 Note that the suffix is not given to haproxy; this tells haproxy to look for
yanbzhud19630c2015-12-14 15:10:25 -050010934 a cert bundle.
10935
Davor Ocelice9ed2812017-12-25 17:49:28 +010010936 HAProxy will load all PEM files in the bundle at the same time to try to
yanbzhud19630c2015-12-14 15:10:25 -050010937 support multiple key types. PEM files are combined based on Common Name
10938 (CN) and Subject Alternative Name (SAN) to support SNI lookups. This means
10939 that even if you give haproxy a cert bundle, if there are no shared CN/SAN
10940 entries in the certificates in that bundle, haproxy will not be able to
10941 provide multi-cert support.
10942
10943 Assuming bundle in the example above contained the following:
10944
10945 Filename | CN | SAN
10946 -------------------+-----------------+-------------------
10947 example.pem.rsa | www.example.com | rsa.example.com
yanbzhu6c25e9e2016-01-05 12:52:02 -050010948 -------------------+-----------------+-------------------
yanbzhud19630c2015-12-14 15:10:25 -050010949 example.pem.ecdsa | www.example.com | ecdsa.example.com
10950 -------------------+-----------------+-------------------
10951
10952 Users connecting with an SNI of "www.example.com" will be able
10953 to use both RSA and ECDSA cipher suites. Users connecting with an SNI of
10954 "rsa.example.com" will only be able to use RSA cipher suites, and users
10955 connecting with "ecdsa.example.com" will only be able to use ECDSA cipher
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020010956 suites. With BoringSSL and Openssl >= 1.1.1 multi-cert is natively supported,
10957 no need to bundle certificates. ECDSA certificate will be preferred if client
10958 support it.
yanbzhud19630c2015-12-14 15:10:25 -050010959
10960 If a directory name is given as the <cert> argument, haproxy will
10961 automatically search and load bundled files in that directory.
10962
10963 OSCP files (.ocsp) and issuer files (.issuer) are supported with multi-cert
10964 bundling. Each certificate can have its own .ocsp and .issuer file. At this
10965 time, sctl is not supported in multi-certificate bundling.
10966
Emeric Brunb6dc9342012-09-28 17:55:37 +020010967crt-ignore-err <errors>
Alex Davies0fbf0162013-03-02 16:04:50 +000010968 This setting is only available when support for OpenSSL was built in. Sets a
Davor Ocelice9ed2812017-12-25 17:49:28 +010010969 comma separated list of errorIDs to ignore during verify at depth == 0. If
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010970 set to 'all', all errors are ignored. SSL handshake is not aborted if an error
Alex Davies0fbf0162013-03-02 16:04:50 +000010971 is ignored.
Emeric Brunb6dc9342012-09-28 17:55:37 +020010972
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010973crt-list <file>
10974 This setting is only available when support for OpenSSL was built in. It
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010975 designates a list of PEM file with an optional ssl configuration and a SNI
10976 filter per certificate, with the following format for each line :
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010977
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010978 <crtfile> [\[<sslbindconf> ...\]] [[!]<snifilter> ...]
10979
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020010980 sslbindconf support "npn", "alpn", "verify", "ca-file", "no-ca-names",
10981 crl-file", "ecdhe", "curves", "ciphers" configuration. With BoringSSL
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020010982 and Openssl >= 1.1.1 "ssl-min-ver" and "ssl-max-ver" are also supported.
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010983 It override the configuration set in bind line for the certificate.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010984
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +020010985 Wildcards are supported in the SNI filter. Negative filter are also supported,
10986 only useful in combination with a wildcard filter to exclude a particular SNI.
10987 The certificates will be presented to clients who provide a valid TLS Server
10988 Name Indication field matching one of the SNI filters. If no SNI filter is
10989 specified, the CN and alt subjects are used. This directive may be specified
10990 multiple times. See the "crt" option for more information. The default
10991 certificate is still needed to meet OpenSSL expectations. If it is not used,
10992 the 'strict-sni' option may be used.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010993
yanbzhu6c25e9e2016-01-05 12:52:02 -050010994 Multi-cert bundling (see "crt") is supported with crt-list, as long as only
Emmanuel Hocdetd294aea2016-05-13 11:14:06 +020010995 the base name is given in the crt-list. SNI filter will do the same work on
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020010996 all bundled certificates. With BoringSSL and Openssl >= 1.1.1 multi-cert is
10997 natively supported, avoid multi-cert bundling. RSA and ECDSA certificates can
10998 be declared in a row, and set different ssl and filter parameter.
yanbzhud19630c2015-12-14 15:10:25 -050010999
Emmanuel Hocdet98263292016-12-29 18:26:15 +010011000 crt-list file example:
11001 cert1.pem
Emmanuel Hocdet05942112017-02-20 16:11:50 +010011002 cert2.pem [alpn h2,http/1.1]
Emmanuel Hocdet98263292016-12-29 18:26:15 +010011003 certW.pem *.domain.tld !secure.domain.tld
Emmanuel Hocdet05942112017-02-20 16:11:50 +010011004 certS.pem [curves X25519:P-256 ciphers ECDHE-ECDSA-AES256-GCM-SHA384] secure.domain.tld
Emmanuel Hocdet98263292016-12-29 18:26:15 +010011005
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011006defer-accept
11007 Is an optional keyword which is supported only on certain Linux kernels. It
11008 states that a connection will only be accepted once some data arrive on it,
11009 or at worst after the first retransmit. This should be used only on protocols
Davor Ocelice9ed2812017-12-25 17:49:28 +010011010 for which the client talks first (e.g. HTTP). It can slightly improve
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011011 performance by ensuring that most of the request is already available when
11012 the connection is accepted. On the other hand, it will not be able to detect
11013 connections which don't talk. It is important to note that this option is
11014 broken in all kernels up to 2.6.31, as the connection is never accepted until
11015 the client talks. This can cause issues with front firewalls which would see
11016 an established connection while the proxy will only see it in SYN_RECV. This
11017 option is only supported on TCPv4/TCPv6 sockets and ignored by other ones.
11018
William Lallemandf6975e92017-05-26 17:42:10 +020011019expose-fd listeners
11020 This option is only usable with the stats socket. It gives your stats socket
11021 the capability to pass listeners FD to another HAProxy process.
William Lallemande202b1e2017-06-01 17:38:56 +020011022 During a reload with the master-worker mode, the process is automatically
11023 reexecuted adding -x and one of the stats socket with this option.
Davor Ocelice9ed2812017-12-25 17:49:28 +010011024 See also "-x" in the management guide.
William Lallemandf6975e92017-05-26 17:42:10 +020011025
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011026force-sslv3
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011027 This option enforces use of SSLv3 only on SSL connections instantiated from
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011028 this listener. SSLv3 is generally less expensive than the TLS counterparts
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011029 for high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011030 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011031
11032force-tlsv10
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011033 This option enforces use of TLSv1.0 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011034 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011035 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011036
11037force-tlsv11
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011038 This option enforces use of TLSv1.1 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011039 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011040 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011041
11042force-tlsv12
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011043 This option enforces use of TLSv1.2 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011044 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011045 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011046
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011047force-tlsv13
11048 This option enforces use of TLSv1.3 only on SSL connections instantiated from
11049 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011050 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011051
Christopher Faulet31af49d2015-06-09 17:29:50 +020011052generate-certificates
11053 This setting is only available when support for OpenSSL was built in. It
11054 enables the dynamic SSL certificates generation. A CA certificate and its
11055 private key are necessary (see 'ca-sign-file'). When HAProxy is configured as
11056 a transparent forward proxy, SSL requests generate errors because of a common
11057 name mismatch on the certificate presented to the client. With this option
11058 enabled, HAProxy will try to forge a certificate using the SNI hostname
11059 indicated by the client. This is done only if no certificate matches the SNI
11060 hostname (see 'crt-list'). If an error occurs, the default certificate is
11061 used, else the 'strict-sni' option is set.
11062 It can also be used when HAProxy is configured as a reverse proxy to ease the
11063 deployment of an architecture with many backends.
11064
11065 Creating a SSL certificate is an expensive operation, so a LRU cache is used
11066 to store forged certificates (see 'tune.ssl.ssl-ctx-cache-size'). It
Davor Ocelice9ed2812017-12-25 17:49:28 +010011067 increases the HAProxy's memory footprint to reduce latency when the same
Christopher Faulet31af49d2015-06-09 17:29:50 +020011068 certificate is used many times.
11069
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011070gid <gid>
11071 Sets the group of the UNIX sockets to the designated system gid. It can also
11072 be set by default in the global section's "unix-bind" statement. Note that
11073 some platforms simply ignore this. This setting is equivalent to the "group"
11074 setting except that the group ID is used instead of its name. This setting is
11075 ignored by non UNIX sockets.
11076
11077group <group>
11078 Sets the group of the UNIX sockets to the designated system group. It can
11079 also be set by default in the global section's "unix-bind" statement. Note
11080 that some platforms simply ignore this. This setting is equivalent to the
11081 "gid" setting except that the group name is used instead of its gid. This
11082 setting is ignored by non UNIX sockets.
11083
11084id <id>
11085 Fixes the socket ID. By default, socket IDs are automatically assigned, but
11086 sometimes it is more convenient to fix them to ease monitoring. This value
11087 must be strictly positive and unique within the listener/frontend. This
11088 option can only be used when defining only a single socket.
11089
11090interface <interface>
Lukas Tribusfce2e962013-02-12 22:13:19 +010011091 Restricts the socket to a specific interface. When specified, only packets
11092 received from that particular interface are processed by the socket. This is
11093 currently only supported on Linux. The interface must be a primary system
11094 interface, not an aliased interface. It is also possible to bind multiple
11095 frontends to the same address if they are bound to different interfaces. Note
11096 that binding to a network interface requires root privileges. This parameter
Jérôme Magnin61275192018-02-07 11:39:58 +010011097 is only compatible with TCPv4/TCPv6 sockets. When specified, return traffic
11098 uses the same interface as inbound traffic, and its associated routing table,
11099 even if there are explicit routes through different interfaces configured.
11100 This can prove useful to address asymmetric routing issues when the same
11101 client IP addresses need to be able to reach frontends hosted on different
11102 interfaces.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011103
Willy Tarreauabb175f2012-09-24 12:43:26 +020011104level <level>
11105 This setting is used with the stats sockets only to restrict the nature of
11106 the commands that can be issued on the socket. It is ignored by other
11107 sockets. <level> can be one of :
Davor Ocelice9ed2812017-12-25 17:49:28 +010011108 - "user" is the least privileged level; only non-sensitive stats can be
Willy Tarreauabb175f2012-09-24 12:43:26 +020011109 read, and no change is allowed. It would make sense on systems where it
11110 is not easy to restrict access to the socket.
11111 - "operator" is the default level and fits most common uses. All data can
Davor Ocelice9ed2812017-12-25 17:49:28 +010011112 be read, and only non-sensitive changes are permitted (e.g. clear max
Willy Tarreauabb175f2012-09-24 12:43:26 +020011113 counters).
Davor Ocelice9ed2812017-12-25 17:49:28 +010011114 - "admin" should be used with care, as everything is permitted (e.g. clear
Willy Tarreauabb175f2012-09-24 12:43:26 +020011115 all counters).
11116
Andjelko Iharosc4df59e2017-07-20 11:59:48 +020011117severity-output <format>
11118 This setting is used with the stats sockets only to configure severity
11119 level output prepended to informational feedback messages. Severity
11120 level of messages can range between 0 and 7, conforming to syslog
11121 rfc5424. Valid and successful socket commands requesting data
11122 (i.e. "show map", "get acl foo" etc.) will never have a severity level
11123 prepended. It is ignored by other sockets. <format> can be one of :
11124 - "none" (default) no severity level is prepended to feedback messages.
11125 - "number" severity level is prepended as a number.
11126 - "string" severity level is prepended as a string following the
11127 rfc5424 convention.
11128
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011129maxconn <maxconn>
11130 Limits the sockets to this number of concurrent connections. Extraneous
11131 connections will remain in the system's backlog until a connection is
11132 released. If unspecified, the limit will be the same as the frontend's
11133 maxconn. Note that in case of port ranges or multiple addresses, the same
11134 value will be applied to each socket. This setting enables different
11135 limitations on expensive sockets, for instance SSL entries which may easily
11136 eat all memory.
11137
11138mode <mode>
11139 Sets the octal mode used to define access permissions on the UNIX socket. It
11140 can also be set by default in the global section's "unix-bind" statement.
11141 Note that some platforms simply ignore this. This setting is ignored by non
11142 UNIX sockets.
11143
11144mss <maxseg>
11145 Sets the TCP Maximum Segment Size (MSS) value to be advertised on incoming
11146 connections. This can be used to force a lower MSS for certain specific
11147 ports, for instance for connections passing through a VPN. Note that this
11148 relies on a kernel feature which is theoretically supported under Linux but
11149 was buggy in all versions prior to 2.6.28. It may or may not work on other
11150 operating systems. It may also not change the advertised value but change the
11151 effective size of outgoing segments. The commonly advertised value for TCPv4
11152 over Ethernet networks is 1460 = 1500(MTU) - 40(IP+TCP). If this value is
11153 positive, it will be used as the advertised MSS. If it is negative, it will
11154 indicate by how much to reduce the incoming connection's advertised MSS for
11155 outgoing segments. This parameter is only compatible with TCP v4/v6 sockets.
11156
11157name <name>
11158 Sets an optional name for these sockets, which will be reported on the stats
11159 page.
11160
Willy Tarreaud72f0f32015-10-13 14:50:22 +020011161namespace <name>
11162 On Linux, it is possible to specify which network namespace a socket will
11163 belong to. This directive makes it possible to explicitly bind a listener to
11164 a namespace different from the default one. Please refer to your operating
11165 system's documentation to find more details about network namespaces.
11166
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011167nice <nice>
11168 Sets the 'niceness' of connections initiated from the socket. Value must be
11169 in the range -1024..1024 inclusive, and defaults to zero. Positive values
11170 means that such connections are more friendly to others and easily offer
11171 their place in the scheduler. On the opposite, negative values mean that
11172 connections want to run with a higher priority than others. The difference
11173 only happens under high loads when the system is close to saturation.
11174 Negative values are appropriate for low-latency or administration services,
11175 and high values are generally recommended for CPU intensive tasks such as SSL
11176 processing or bulk transfers which are less sensible to latency. For example,
11177 it may make sense to use a positive value for an SMTP socket and a negative
11178 one for an RDP socket.
11179
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020011180no-ca-names
11181 This setting is only available when support for OpenSSL was built in. It
11182 prevents from send CA names in server hello message when ca-file is used.
11183
Emeric Brun9b3009b2012-10-05 11:55:06 +020011184no-sslv3
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011185 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011186 disables support for SSLv3 on any sockets instantiated from the listener when
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011187 SSL is supported. Note that SSLv2 is forced disabled in the code and cannot
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011188 be enabled using any configuration option. This option is also available on
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011189 global statement "ssl-default-bind-options". Use "ssl-min-ver" and
11190 "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011191
Emeric Brun90ad8722012-10-02 14:00:59 +020011192no-tls-tickets
11193 This setting is only available when support for OpenSSL was built in. It
11194 disables the stateless session resumption (RFC 5077 TLS Ticket
11195 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011196 session resumption is more expensive in CPU usage. This option is also
11197 available on global statement "ssl-default-bind-options".
Emeric Brun90ad8722012-10-02 14:00:59 +020011198
Emeric Brun9b3009b2012-10-05 11:55:06 +020011199no-tlsv10
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011200 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011201 disables support for TLSv1.0 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011202 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011203 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011204 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11205 and "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011206
Emeric Brun9b3009b2012-10-05 11:55:06 +020011207no-tlsv11
Emeric Brunf5da4932012-09-28 19:42:54 +020011208 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011209 disables support for TLSv1.1 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011210 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011211 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011212 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11213 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020011214
Emeric Brun9b3009b2012-10-05 11:55:06 +020011215no-tlsv12
Emeric Brunf5da4932012-09-28 19:42:54 +020011216 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011217 disables support for TLSv1.2 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011218 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011219 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011220 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11221 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020011222
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011223no-tlsv13
11224 This setting is only available when support for OpenSSL was built in. It
11225 disables support for TLSv1.3 on any sockets instantiated from the listener
11226 when SSL is supported. Note that SSLv2 is forced disabled in the code and
11227 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011228 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11229 and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011230
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020011231npn <protocols>
11232 This enables the NPN TLS extension and advertises the specified protocol list
11233 as supported on top of NPN. The protocol list consists in a comma-delimited
11234 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
11235 This requires that the SSL library is build with support for TLS extensions
Willy Tarreauab861d32013-04-02 02:30:41 +020011236 enabled (check with haproxy -vv). Note that the NPN extension has been
Willy Tarreau95c4e142017-11-26 12:18:55 +010011237 replaced with the ALPN extension (see the "alpn" keyword), though this one is
11238 only available starting with OpenSSL 1.0.2. If HTTP/2 is desired on an older
11239 version of OpenSSL, NPN might still be used as most clients still support it
11240 at the time of writing this. It is possible to enable both NPN and ALPN
11241 though it probably doesn't make any sense out of testing.
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020011242
Lukas Tribus53ae85c2017-05-04 15:45:40 +000011243prefer-client-ciphers
11244 Use the client's preference when selecting the cipher suite, by default
11245 the server's preference is enforced. This option is also available on
11246 global statement "ssl-default-bind-options".
Lukas Tribus926594f2018-05-18 17:55:57 +020011247 Note that with OpenSSL >= 1.1.1 ChaCha20-Poly1305 is reprioritized anyway
11248 (without setting this option), if a ChaCha20-Poly1305 cipher is at the top of
11249 the client cipher list.
Lukas Tribus53ae85c2017-05-04 15:45:40 +000011250
Christopher Fauletc644fa92017-11-23 22:44:11 +010011251process <process-set>[/<thread-set>]
11252 This restricts the list of processes and/or threads on which this listener is
11253 allowed to run. It does not enforce any process but eliminates those which do
Davor Ocelice9ed2812017-12-25 17:49:28 +010011254 not match. If the frontend uses a "bind-process" setting, the intersection
Christopher Fauletc644fa92017-11-23 22:44:11 +010011255 between the two is applied. If in the end the listener is not allowed to run
11256 on any remaining process, a warning is emitted, and the listener will either
11257 run on the first process of the listener if a single process was specified,
11258 or on all of its processes if multiple processes were specified. If a thread
Davor Ocelice9ed2812017-12-25 17:49:28 +010011259 set is specified, it limits the threads allowed to process incoming
Christopher Fauletc644fa92017-11-23 22:44:11 +010011260 connections for this listener, for the corresponding process set. For the
11261 unlikely case where several ranges are needed, this directive may be
11262 repeated. <process-set> and <thread-set> must use the format
11263
11264 all | odd | even | number[-[number]]
11265
11266 Ranges can be partially defined. The higher bound can be omitted. In such
11267 case, it is replaced by the corresponding maximum value. The main purpose of
11268 this directive is to be used with the stats sockets and have one different
11269 socket per process. The second purpose is to have multiple bind lines sharing
11270 the same IP:port but not the same process in a listener, so that the system
11271 can distribute the incoming connections into multiple queues and allow a
11272 smoother inter-process load balancing. Currently Linux 3.9 and above is known
11273 for supporting this. See also "bind-process" and "nbproc".
Willy Tarreau6ae1ba62014-05-07 19:01:58 +020011274
Christopher Fauleta717b992018-04-10 14:43:00 +020011275proto <name>
11276 Forces the multiplexer's protocol to use for the incoming connections. It
11277 must be compatible with the mode of the frontend (TCP or HTTP). It must also
11278 be usable on the frontend side. The list of available protocols is reported
11279 in haproxy -vv.
11280 Idea behind this optipon is to bypass the selection of the best multiplexer's
11281 protocol for all connections instantiated from this listening socket. For
Joseph Herlant71b4b152018-11-13 16:55:16 -080011282 instance, it is possible to force the http/2 on clear TCP by specifying "proto
Christopher Fauleta717b992018-04-10 14:43:00 +020011283 h2" on the bind line.
11284
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011285ssl
11286 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011287 enables SSL deciphering on connections instantiated from this listener. A
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011288 certificate is necessary (see "crt" above). All contents in the buffers will
11289 appear in clear text, so that ACLs and HTTP processing will only have access
Emmanuel Hocdetbd695fe2017-05-15 15:53:41 +020011290 to deciphered contents. SSLv3 is disabled per default, use "ssl-min-ver SSLv3"
11291 to enable it.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011292
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011293ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
11294 This option enforces use of <version> or lower on SSL connections instantiated
11295 from this listener. This option is also available on global statement
11296 "ssl-default-bind-options". See also "ssl-min-ver".
11297
11298ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
11299 This option enforces use of <version> or upper on SSL connections instantiated
11300 from this listener. This option is also available on global statement
11301 "ssl-default-bind-options". See also "ssl-max-ver".
11302
Emmanuel Hocdet65623372013-01-24 17:17:15 +010011303strict-sni
11304 This setting is only available when support for OpenSSL was built in. The
11305 SSL/TLS negotiation is allow only if the client provided an SNI which match
11306 a certificate. The default certificate is not used.
11307 See the "crt" option for more information.
11308
Willy Tarreau2af207a2015-02-04 00:45:58 +010011309tcp-ut <delay>
Tim Düsterhus4896c442016-11-29 02:15:19 +010011310 Sets the TCP User Timeout for all incoming connections instantiated from this
Willy Tarreau2af207a2015-02-04 00:45:58 +010011311 listening socket. This option is available on Linux since version 2.6.37. It
11312 allows haproxy to configure a timeout for sockets which contain data not
Davor Ocelice9ed2812017-12-25 17:49:28 +010011313 receiving an acknowledgment for the configured delay. This is especially
Willy Tarreau2af207a2015-02-04 00:45:58 +010011314 useful on long-lived connections experiencing long idle periods such as
11315 remote terminals or database connection pools, where the client and server
11316 timeouts must remain high to allow a long period of idle, but where it is
11317 important to detect that the client has disappeared in order to release all
11318 resources associated with its connection (and the server's session). The
11319 argument is a delay expressed in milliseconds by default. This only works
11320 for regular TCP connections, and is ignored for other protocols.
11321
Willy Tarreau1c862c52012-10-05 16:21:00 +020011322tfo
Lukas Tribus0defb902013-02-13 23:35:39 +010011323 Is an optional keyword which is supported only on Linux kernels >= 3.7. It
Willy Tarreau1c862c52012-10-05 16:21:00 +020011324 enables TCP Fast Open on the listening socket, which means that clients which
11325 support this feature will be able to send a request and receive a response
11326 during the 3-way handshake starting from second connection, thus saving one
11327 round-trip after the first connection. This only makes sense with protocols
11328 that use high connection rates and where each round trip matters. This can
11329 possibly cause issues with many firewalls which do not accept data on SYN
11330 packets, so this option should only be enabled once well tested. This option
Lukas Tribus0999f762013-04-02 16:43:24 +020011331 is only supported on TCPv4/TCPv6 sockets and ignored by other ones. You may
11332 need to build HAProxy with USE_TFO=1 if your libc doesn't define
11333 TCP_FASTOPEN.
Willy Tarreau1c862c52012-10-05 16:21:00 +020011334
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010011335tls-ticket-keys <keyfile>
11336 Sets the TLS ticket keys file to load the keys from. The keys need to be 48
11337 bytes long, encoded with base64 (ex. openssl rand -base64 48). Number of keys
11338 is specified by the TLS_TICKETS_NO build option (default 3) and at least as
11339 many keys need to be present in the file. Last TLS_TICKETS_NO keys will be
11340 used for decryption and the penultimate one for encryption. This enables easy
11341 key rotation by just appending new key to the file and reloading the process.
11342 Keys must be periodically rotated (ex. every 12h) or Perfect Forward Secrecy
11343 is compromised. It is also a good idea to keep the keys off any permanent
11344 storage such as hard drives (hint: use tmpfs and don't swap those files).
11345 Lifetime hint can be changed using tune.ssl.timeout.
11346
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011347transparent
11348 Is an optional keyword which is supported only on certain Linux kernels. It
11349 indicates that the addresses will be bound even if they do not belong to the
11350 local machine, and that packets targeting any of these addresses will be
11351 intercepted just as if the addresses were locally configured. This normally
11352 requires that IP forwarding is enabled. Caution! do not use this with the
11353 default address '*', as it would redirect any traffic for the specified port.
11354 This keyword is available only when HAProxy is built with USE_LINUX_TPROXY=1.
11355 This parameter is only compatible with TCPv4 and TCPv6 sockets, depending on
11356 kernel version. Some distribution kernels include backports of the feature,
11357 so check for support with your vendor.
11358
Willy Tarreau77e3af92012-11-24 15:07:23 +010011359v4v6
11360 Is an optional keyword which is supported only on most recent systems
11361 including Linux kernels >= 2.4.21. It is used to bind a socket to both IPv4
11362 and IPv6 when it uses the default address. Doing so is sometimes necessary
11363 on systems which bind to IPv6 only by default. It has no effect on non-IPv6
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011364 sockets, and is overridden by the "v6only" option.
Willy Tarreau77e3af92012-11-24 15:07:23 +010011365
Willy Tarreau9b6700f2012-11-24 11:55:28 +010011366v6only
11367 Is an optional keyword which is supported only on most recent systems
11368 including Linux kernels >= 2.4.21. It is used to bind a socket to IPv6 only
11369 when it uses the default address. Doing so is sometimes preferred to doing it
Willy Tarreau77e3af92012-11-24 15:07:23 +010011370 system-wide as it is per-listener. It has no effect on non-IPv6 sockets and
11371 has precedence over the "v4v6" option.
Willy Tarreau9b6700f2012-11-24 11:55:28 +010011372
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011373uid <uid>
11374 Sets the owner of the UNIX sockets to the designated system uid. It can also
11375 be set by default in the global section's "unix-bind" statement. Note that
11376 some platforms simply ignore this. This setting is equivalent to the "user"
11377 setting except that the user numeric ID is used instead of its name. This
11378 setting is ignored by non UNIX sockets.
11379
11380user <user>
11381 Sets the owner of the UNIX sockets to the designated system user. It can also
11382 be set by default in the global section's "unix-bind" statement. Note that
11383 some platforms simply ignore this. This setting is equivalent to the "uid"
11384 setting except that the user name is used instead of its uid. This setting is
11385 ignored by non UNIX sockets.
11386
Emeric Brun1a073b42012-09-28 17:07:34 +020011387verify [none|optional|required]
11388 This setting is only available when support for OpenSSL was built in. If set
11389 to 'none', client certificate is not requested. This is the default. In other
11390 cases, a client certificate is requested. If the client does not provide a
11391 certificate after the request and if 'verify' is set to 'required', then the
11392 handshake is aborted, while it would have succeeded if set to 'optional'. The
Emeric Brunfd33a262012-10-11 16:28:27 +020011393 certificate provided by the client is always verified using CAs from
11394 'ca-file' and optional CRLs from 'crl-file'. On verify failure the handshake
11395 is aborted, regardless of the 'verify' option, unless the error code exactly
11396 matches one of those listed with 'ca-ignore-err' or 'crt-ignore-err'.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020011397
Willy Tarreaub6205fd2012-09-24 12:27:33 +0200113985.2. Server and default-server options
Cyril Bontéf0c60612010-02-06 14:44:47 +010011399------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020011400
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010011401The "server" and "default-server" keywords support a certain number of settings
11402which are all passed as arguments on the server line. The order in which those
11403arguments appear does not count, and they are all optional. Some of those
11404settings are single words (booleans) while others expect one or several values
11405after them. In this case, the values must immediately follow the setting name.
11406Except default-server, all those settings must be specified after the server's
11407address if they are used:
Willy Tarreau6a06a402007-07-15 20:15:28 +020011408
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011409 server <name> <address>[:port] [settings ...]
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010011410 default-server [settings ...]
Willy Tarreau6a06a402007-07-15 20:15:28 +020011411
Frédéric Lécailled2376272017-03-21 18:52:12 +010011412Note that all these settings are supported both by "server" and "default-server"
11413keywords, except "id" which is only supported by "server".
11414
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011415The currently supported settings are the following ones.
Willy Tarreau0ba27502007-12-24 16:55:16 +010011416
Willy Tarreauceb4ac92012-04-28 00:41:46 +020011417addr <ipv4|ipv6>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011418 Using the "addr" parameter, it becomes possible to use a different IP address
Baptiste Assmann13f83532016-03-06 23:14:36 +010011419 to send health-checks or to probe the agent-check. On some servers, it may be
11420 desirable to dedicate an IP address to specific component able to perform
11421 complex tests which are more suitable to health-checks than the application.
11422 This parameter is ignored if the "check" parameter is not set. See also the
11423 "port" parameter.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011424
Simon Hormand60d6912013-11-25 10:46:36 +090011425agent-check
11426 Enable an auxiliary agent check which is run independently of a regular
Willy Tarreau81f5d942013-12-09 20:51:51 +010011427 health check. An agent health check is performed by making a TCP connection
Willy Tarreau7a0139e2018-12-16 08:42:56 +010011428 to the port set by the "agent-port" parameter and reading an ASCII string
11429 terminated by the first '\r' or '\n' met. The string is made of a series of
11430 words delimited by spaces, tabs or commas in any order, each consisting of :
Simon Hormand60d6912013-11-25 10:46:36 +090011431
Willy Tarreau81f5d942013-12-09 20:51:51 +010011432 - An ASCII representation of a positive integer percentage, e.g. "75%".
Simon Hormand60d6912013-11-25 10:46:36 +090011433 Values in this format will set the weight proportional to the initial
Willy Tarreauc5af3a62014-10-07 15:27:33 +020011434 weight of a server as configured when haproxy starts. Note that a zero
11435 weight is reported on the stats page as "DRAIN" since it has the same
11436 effect on the server (it's removed from the LB farm).
Simon Hormand60d6912013-11-25 10:46:36 +090011437
Davor Ocelice9ed2812017-12-25 17:49:28 +010011438 - The string "maxconn:" followed by an integer (no space between). Values
11439 in this format will set the maxconn of a server. The maximum number of
11440 connections advertised needs to be multiplied by the number of load
11441 balancers and different backends that use this health check to get the
11442 total number of connections the server might receive. Example: maxconn:30
Nenad Merdanovic174dd372016-04-24 23:10:06 +020011443
Willy Tarreau81f5d942013-12-09 20:51:51 +010011444 - The word "ready". This will turn the server's administrative state to the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011445 READY mode, thus canceling any DRAIN or MAINT state
Simon Hormand60d6912013-11-25 10:46:36 +090011446
Willy Tarreau81f5d942013-12-09 20:51:51 +010011447 - The word "drain". This will turn the server's administrative state to the
11448 DRAIN mode, thus it will not accept any new connections other than those
11449 that are accepted via persistence.
Simon Hormand60d6912013-11-25 10:46:36 +090011450
Willy Tarreau81f5d942013-12-09 20:51:51 +010011451 - The word "maint". This will turn the server's administrative state to the
11452 MAINT mode, thus it will not accept any new connections at all, and health
11453 checks will be stopped.
Simon Hormand60d6912013-11-25 10:46:36 +090011454
Willy Tarreau81f5d942013-12-09 20:51:51 +010011455 - The words "down", "failed", or "stopped", optionally followed by a
11456 description string after a sharp ('#'). All of these mark the server's
11457 operating state as DOWN, but since the word itself is reported on the stats
11458 page, the difference allows an administrator to know if the situation was
11459 expected or not : the service may intentionally be stopped, may appear up
Davor Ocelice9ed2812017-12-25 17:49:28 +010011460 but fail some validity tests, or may be seen as down (e.g. missing process,
Willy Tarreau81f5d942013-12-09 20:51:51 +010011461 or port not responding).
Simon Hormand60d6912013-11-25 10:46:36 +090011462
Willy Tarreau81f5d942013-12-09 20:51:51 +010011463 - The word "up" sets back the server's operating state as UP if health checks
11464 also report that the service is accessible.
Simon Hormand60d6912013-11-25 10:46:36 +090011465
Willy Tarreau81f5d942013-12-09 20:51:51 +010011466 Parameters which are not advertised by the agent are not changed. For
11467 example, an agent might be designed to monitor CPU usage and only report a
11468 relative weight and never interact with the operating status. Similarly, an
11469 agent could be designed as an end-user interface with 3 radio buttons
11470 allowing an administrator to change only the administrative state. However,
11471 it is important to consider that only the agent may revert its own actions,
11472 so if a server is set to DRAIN mode or to DOWN state using the agent, the
11473 agent must implement the other equivalent actions to bring the service into
11474 operations again.
Simon Hormand60d6912013-11-25 10:46:36 +090011475
Simon Horman2f1f9552013-11-25 10:46:37 +090011476 Failure to connect to the agent is not considered an error as connectivity
11477 is tested by the regular health check which is enabled by the "check"
Willy Tarreau81f5d942013-12-09 20:51:51 +010011478 parameter. Warning though, it is not a good idea to stop an agent after it
11479 reports "down", since only an agent reporting "up" will be able to turn the
11480 server up again. Note that the CLI on the Unix stats socket is also able to
Willy Tarreau989222a2016-01-15 10:26:26 +010011481 force an agent's result in order to work around a bogus agent if needed.
Simon Horman2f1f9552013-11-25 10:46:37 +090011482
Willy Tarreau81f5d942013-12-09 20:51:51 +010011483 Requires the "agent-port" parameter to be set. See also the "agent-inter"
Frédéric Lécailled2376272017-03-21 18:52:12 +010011484 and "no-agent-check" parameters.
Simon Hormand60d6912013-11-25 10:46:36 +090011485
James Brown55f9ff12015-10-21 18:19:05 -070011486agent-send <string>
11487 If this option is specified, haproxy will send the given string (verbatim)
11488 to the agent server upon connection. You could, for example, encode
11489 the backend name into this string, which would enable your agent to send
11490 different responses based on the backend. Make sure to include a '\n' if
11491 you want to terminate your request with a newline.
11492
Simon Hormand60d6912013-11-25 10:46:36 +090011493agent-inter <delay>
11494 The "agent-inter" parameter sets the interval between two agent checks
11495 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
11496
11497 Just as with every other time-based parameter, it may be entered in any
11498 other explicit unit among { us, ms, s, m, h, d }. The "agent-inter"
11499 parameter also serves as a timeout for agent checks "timeout check" is
11500 not set. In order to reduce "resonance" effects when multiple servers are
11501 hosted on the same hardware, the agent and health checks of all servers
11502 are started with a small time offset between them. It is also possible to
11503 add some random noise in the agent and health checks interval using the
11504 global "spread-checks" keyword. This makes sense for instance when a lot
11505 of backends use the same servers.
11506
11507 See also the "agent-check" and "agent-port" parameters.
11508
Misiek768d8602017-01-09 09:52:43 +010011509agent-addr <addr>
11510 The "agent-addr" parameter sets address for agent check.
11511
11512 You can offload agent-check to another target, so you can make single place
11513 managing status and weights of servers defined in haproxy in case you can't
11514 make self-aware and self-managing services. You can specify both IP or
11515 hostname, it will be resolved.
11516
Simon Hormand60d6912013-11-25 10:46:36 +090011517agent-port <port>
11518 The "agent-port" parameter sets the TCP port used for agent checks.
11519
11520 See also the "agent-check" and "agent-inter" parameters.
11521
Olivier Houchardc7566002018-11-20 23:33:50 +010011522alpn <protocols>
11523 This enables the TLS ALPN extension and advertises the specified protocol
11524 list as supported on top of ALPN. The protocol list consists in a comma-
11525 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
11526 quotes). This requires that the SSL library is build with support for TLS
11527 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
11528 initial NPN extension. ALPN is required to connect to HTTP/2 servers.
11529 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
11530 now obsolete NPN extension.
11531 If both HTTP/2 and HTTP/1.1 are expected to be supported, both versions can
11532 be advertised, in order of preference, like below :
11533
11534 server 127.0.0.1:443 ssl crt pub.pem alpn h2,http/1.1
11535
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011536backup
11537 When "backup" is present on a server line, the server is only used in load
11538 balancing when all other non-backup servers are unavailable. Requests coming
11539 with a persistence cookie referencing the server will always be served
11540 though. By default, only the first operational backup server is used, unless
Frédéric Lécailled2376272017-03-21 18:52:12 +010011541 the "allbackups" option is set in the backend. See also the "no-backup" and
11542 "allbackups" options.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011543
Emeric Brunef42d922012-10-11 16:11:36 +020011544ca-file <cafile>
11545 This setting is only available when support for OpenSSL was built in. It
11546 designates a PEM file from which to load CA certificates used to verify
11547 server's certificate.
11548
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011549check
11550 This option enables health checks on the server. By default, a server is
Patrick Mézardb7aeec62012-01-22 16:01:22 +010011551 always considered available. If "check" is set, the server is available when
11552 accepting periodic TCP connections, to ensure that it is really able to serve
11553 requests. The default address and port to send the tests to are those of the
11554 server, and the default source is the same as the one defined in the
11555 backend. It is possible to change the address using the "addr" parameter, the
11556 port using the "port" parameter, the source address using the "source"
11557 address, and the interval and timers using the "inter", "rise" and "fall"
Simon Hormanafc47ee2013-11-25 10:46:35 +090011558 parameters. The request method is define in the backend using the "httpchk",
11559 "smtpchk", "mysql-check", "pgsql-check" and "ssl-hello-chk" options. Please
Frédéric Lécailled2376272017-03-21 18:52:12 +010011560 refer to those options and parameters for more information. See also
11561 "no-check" option.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011562
Willy Tarreau6c16adc2012-10-05 00:04:16 +020011563check-send-proxy
11564 This option forces emission of a PROXY protocol line with outgoing health
11565 checks, regardless of whether the server uses send-proxy or not for the
11566 normal traffic. By default, the PROXY protocol is enabled for health checks
11567 if it is already enabled for normal traffic and if no "port" nor "addr"
11568 directive is present. However, if such a directive is present, the
11569 "check-send-proxy" option needs to be used to force the use of the
11570 protocol. See also the "send-proxy" option for more information.
11571
Olivier Houchard92150142018-12-21 19:47:01 +010011572check-alpn <protocols>
11573 Defines which protocols to advertise with ALPN. The protocol list consists in
11574 a comma-delimited list of protocol names, for instance: "http/1.1,http/1.0"
11575 (without quotes). If it is not set, the server ALPN is used.
11576
Jérôme Magninae9bb762018-12-09 16:08:26 +010011577check-sni <sni>
Olivier Houchard9130a962017-10-17 17:33:43 +020011578 This option allows you to specify the SNI to be used when doing health checks
Jérôme Magninae9bb762018-12-09 16:08:26 +010011579 over SSL. It is only possible to use a string to set <sni>. If you want to
11580 set a SNI for proxied traffic, see "sni".
Olivier Houchard9130a962017-10-17 17:33:43 +020011581
Willy Tarreau763a95b2012-10-04 23:15:39 +020011582check-ssl
11583 This option forces encryption of all health checks over SSL, regardless of
11584 whether the server uses SSL or not for the normal traffic. This is generally
11585 used when an explicit "port" or "addr" directive is specified and SSL health
11586 checks are not inherited. It is important to understand that this option
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011587 inserts an SSL transport layer below the checks, so that a simple TCP connect
Willy Tarreau763a95b2012-10-04 23:15:39 +020011588 check becomes an SSL connect, which replaces the old ssl-hello-chk. The most
11589 common use is to send HTTPS checks by combining "httpchk" with SSL checks.
Davor Ocelice9ed2812017-12-25 17:49:28 +010011590 All SSL settings are common to health checks and traffic (e.g. ciphers).
Frédéric Lécailled2376272017-03-21 18:52:12 +010011591 See the "ssl" option for more information and "no-check-ssl" to disable
11592 this option.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011593
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011594ciphers <ciphers>
Dirkjan Bussink415150f2018-09-14 11:14:21 +020011595 This setting is only available when support for OpenSSL was built in. This
11596 option sets the string describing the list of cipher algorithms that is
11597 negotiated during the SSL/TLS handshake with the server. The format of the
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011598 string is defined in "man 1 ciphers". When SSL is used to communicate with
11599 servers on the local network, it is common to see a weaker set of algorithms
11600 than what is used over the internet. Doing so reduces CPU usage on both the
11601 server and haproxy while still keeping it compatible with deployed software.
11602 Some algorithms such as RC4-SHA1 are reasonably cheap. If no security at all
11603 is needed and just connectivity, using DES can be appropriate.
11604
Dirkjan Bussink415150f2018-09-14 11:14:21 +020011605ciphersuites <ciphersuites>
11606 This setting is only available when support for OpenSSL was built in and
11607 OpenSSL 1.1.1 or later was used to build HAProxy. This option sets the string
11608 describing the list of cipher algorithms that is negotiated during the TLS
11609 1.3 handshake with the server. The format of the string is defined in
11610 "man 1 ciphers" under the "ciphersuites" section.
11611
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011612cookie <value>
11613 The "cookie" parameter sets the cookie value assigned to the server to
11614 <value>. This value will be checked in incoming requests, and the first
11615 operational server possessing the same value will be selected. In return, in
11616 cookie insertion or rewrite modes, this value will be assigned to the cookie
11617 sent to the client. There is nothing wrong in having several servers sharing
11618 the same cookie value, and it is in fact somewhat common between normal and
11619 backup servers. See also the "cookie" keyword in backend section.
11620
Emeric Brunef42d922012-10-11 16:11:36 +020011621crl-file <crlfile>
11622 This setting is only available when support for OpenSSL was built in. It
11623 designates a PEM file from which to load certificate revocation list used
11624 to verify server's certificate.
11625
Emeric Bruna7aa3092012-10-26 12:58:00 +020011626crt <cert>
11627 This setting is only available when support for OpenSSL was built in.
11628 It designates a PEM file from which to load both a certificate and the
11629 associated private key. This file can be built by concatenating both PEM
11630 files into one. This certificate will be sent if the server send a client
11631 certificate request.
11632
Willy Tarreau96839092010-03-29 10:02:24 +020011633disabled
11634 The "disabled" keyword starts the server in the "disabled" state. That means
11635 that it is marked down in maintenance mode, and no connection other than the
11636 ones allowed by persist mode will reach it. It is very well suited to setup
11637 new servers, because normal traffic will never reach them, while it is still
11638 possible to test the service by making use of the force-persist mechanism.
Frédéric Lécailled2376272017-03-21 18:52:12 +010011639 See also "enabled" setting.
Willy Tarreau96839092010-03-29 10:02:24 +020011640
Frédéric Lécailled2376272017-03-21 18:52:12 +010011641enabled
11642 This option may be used as 'server' setting to reset any 'disabled'
11643 setting which would have been inherited from 'default-server' directive as
11644 default value.
11645 It may also be used as 'default-server' setting to reset any previous
11646 'default-server' 'disabled' setting.
Willy Tarreau96839092010-03-29 10:02:24 +020011647
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011648error-limit <count>
Willy Tarreau983e01e2010-01-11 18:42:06 +010011649 If health observing is enabled, the "error-limit" parameter specifies the
11650 number of consecutive errors that triggers event selected by the "on-error"
11651 option. By default it is set to 10 consecutive errors.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011652
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011653 See also the "check", "error-limit" and "on-error".
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011654
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011655fall <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011656 The "fall" parameter states that a server will be considered as dead after
11657 <count> consecutive unsuccessful health checks. This value defaults to 3 if
11658 unspecified. See also the "check", "inter" and "rise" parameters.
11659
Emeric Brun8694b9a2012-10-05 14:39:07 +020011660force-sslv3
11661 This option enforces use of SSLv3 only when SSL is used to communicate with
11662 the server. SSLv3 is generally less expensive than the TLS counterparts for
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011663 high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011664 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011665
11666force-tlsv10
11667 This option enforces use of TLSv1.0 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011668 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011669 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011670
11671force-tlsv11
11672 This option enforces use of TLSv1.1 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011673 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011674 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011675
11676force-tlsv12
11677 This option enforces use of TLSv1.2 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011678 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011679 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011680
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011681force-tlsv13
11682 This option enforces use of TLSv1.3 only when SSL is used to communicate with
11683 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011684 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011685
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011686id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +020011687 Set a persistent ID for the server. This ID must be positive and unique for
11688 the proxy. An unused ID will automatically be assigned if unset. The first
11689 assigned value will be 1. This ID is currently only returned in statistics.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011690
Willy Tarreau6a031d12016-11-07 19:42:35 +010011691init-addr {last | libc | none | <ip>},[...]*
11692 Indicate in what order the server's address should be resolved upon startup
11693 if it uses an FQDN. Attempts are made to resolve the address by applying in
Davor Ocelice9ed2812017-12-25 17:49:28 +010011694 turn each of the methods mentioned in the comma-delimited list. The first
Willy Tarreau6a031d12016-11-07 19:42:35 +010011695 method which succeeds is used. If the end of the list is reached without
11696 finding a working method, an error is thrown. Method "last" suggests to pick
11697 the address which appears in the state file (see "server-state-file"). Method
11698 "libc" uses the libc's internal resolver (gethostbyname() or getaddrinfo()
11699 depending on the operating system and build options). Method "none"
11700 specifically indicates that the server should start without any valid IP
11701 address in a down state. It can be useful to ignore some DNS issues upon
11702 startup, waiting for the situation to get fixed later. Finally, an IP address
11703 (IPv4 or IPv6) may be provided. It can be the currently known address of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011704 server (e.g. filled by a configuration generator), or the address of a dummy
Willy Tarreau6a031d12016-11-07 19:42:35 +010011705 server used to catch old sessions and present them with a decent error
11706 message for example. When the "first" load balancing algorithm is used, this
11707 IP address could point to a fake server used to trigger the creation of new
11708 instances on the fly. This option defaults to "last,libc" indicating that the
11709 previous address found in the state file (if any) is used first, otherwise
11710 the libc's resolver is used. This ensures continued compatibility with the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011711 historic behavior.
Willy Tarreau6a031d12016-11-07 19:42:35 +010011712
11713 Example:
11714 defaults
11715 # never fail on address resolution
11716 default-server init-addr last,libc,none
11717
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011718inter <delay>
11719fastinter <delay>
11720downinter <delay>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011721 The "inter" parameter sets the interval between two consecutive health checks
11722 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
11723 It is also possible to use "fastinter" and "downinter" to optimize delays
11724 between checks depending on the server state :
11725
Pieter Baauw44fc9df2015-09-17 21:30:46 +020011726 Server state | Interval used
11727 ----------------------------------------+----------------------------------
11728 UP 100% (non-transitional) | "inter"
11729 ----------------------------------------+----------------------------------
11730 Transitionally UP (going down "fall"), | "fastinter" if set,
11731 Transitionally DOWN (going up "rise"), | "inter" otherwise.
11732 or yet unchecked. |
11733 ----------------------------------------+----------------------------------
11734 DOWN 100% (non-transitional) | "downinter" if set,
11735 | "inter" otherwise.
11736 ----------------------------------------+----------------------------------
Willy Tarreaud72758d2010-01-12 10:42:19 +010011737
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011738 Just as with every other time-based parameter, they can be entered in any
11739 other explicit unit among { us, ms, s, m, h, d }. The "inter" parameter also
11740 serves as a timeout for health checks sent to servers if "timeout check" is
11741 not set. In order to reduce "resonance" effects when multiple servers are
Simon Hormand60d6912013-11-25 10:46:36 +090011742 hosted on the same hardware, the agent and health checks of all servers
11743 are started with a small time offset between them. It is also possible to
11744 add some random noise in the agent and health checks interval using the
11745 global "spread-checks" keyword. This makes sense for instance when a lot
11746 of backends use the same servers.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011747
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011748maxconn <maxconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011749 The "maxconn" parameter specifies the maximal number of concurrent
11750 connections that will be sent to this server. If the number of incoming
11751 concurrent requests goes higher than this value, they will be queued, waiting
11752 for a connection to be released. This parameter is very important as it can
11753 save fragile servers from going down under extreme loads. If a "minconn"
11754 parameter is specified, the limit becomes dynamic. The default value is "0"
11755 which means unlimited. See also the "minconn" and "maxqueue" parameters, and
11756 the backend's "fullconn" keyword.
11757
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011758maxqueue <maxqueue>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011759 The "maxqueue" parameter specifies the maximal number of connections which
11760 will wait in the queue for this server. If this limit is reached, next
11761 requests will be redispatched to other servers instead of indefinitely
11762 waiting to be served. This will break persistence but may allow people to
11763 quickly re-log in when the server they try to connect to is dying. The
11764 default value is "0" which means the queue is unlimited. See also the
11765 "maxconn" and "minconn" parameters.
11766
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011767minconn <minconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011768 When the "minconn" parameter is set, the maxconn limit becomes a dynamic
11769 limit following the backend's load. The server will always accept at least
11770 <minconn> connections, never more than <maxconn>, and the limit will be on
11771 the ramp between both values when the backend has less than <fullconn>
11772 concurrent connections. This makes it possible to limit the load on the
11773 server during normal loads, but push it further for important loads without
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011774 overloading the server during exceptional loads. See also the "maxconn"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011775 and "maxqueue" parameters, as well as the "fullconn" backend keyword.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011776
Willy Tarreaud72f0f32015-10-13 14:50:22 +020011777namespace <name>
11778 On Linux, it is possible to specify which network namespace a socket will
11779 belong to. This directive makes it possible to explicitly bind a server to
11780 a namespace different from the default one. Please refer to your operating
11781 system's documentation to find more details about network namespaces.
11782
Frédéric Lécailled2376272017-03-21 18:52:12 +010011783no-agent-check
11784 This option may be used as "server" setting to reset any "agent-check"
11785 setting which would have been inherited from "default-server" directive as
11786 default value.
11787 It may also be used as "default-server" setting to reset any previous
11788 "default-server" "agent-check" setting.
11789
11790no-backup
11791 This option may be used as "server" setting to reset any "backup"
11792 setting which would have been inherited from "default-server" directive as
11793 default value.
11794 It may also be used as "default-server" setting to reset any previous
11795 "default-server" "backup" setting.
11796
11797no-check
11798 This option may be used as "server" setting to reset any "check"
11799 setting which would have been inherited from "default-server" directive as
11800 default value.
11801 It may also be used as "default-server" setting to reset any previous
11802 "default-server" "check" setting.
11803
11804no-check-ssl
11805 This option may be used as "server" setting to reset any "check-ssl"
11806 setting which would have been inherited from "default-server" directive as
11807 default value.
11808 It may also be used as "default-server" setting to reset any previous
11809 "default-server" "check-ssl" setting.
11810
Frédéric Lécailled2376272017-03-21 18:52:12 +010011811no-send-proxy
11812 This option may be used as "server" setting to reset any "send-proxy"
11813 setting which would have been inherited from "default-server" directive as
11814 default value.
11815 It may also be used as "default-server" setting to reset any previous
11816 "default-server" "send-proxy" setting.
11817
11818no-send-proxy-v2
11819 This option may be used as "server" setting to reset any "send-proxy-v2"
11820 setting which would have been inherited from "default-server" directive as
11821 default value.
11822 It may also be used as "default-server" setting to reset any previous
11823 "default-server" "send-proxy-v2" setting.
11824
11825no-send-proxy-v2-ssl
11826 This option may be used as "server" setting to reset any "send-proxy-v2-ssl"
11827 setting which would have been inherited from "default-server" directive as
11828 default value.
11829 It may also be used as "default-server" setting to reset any previous
11830 "default-server" "send-proxy-v2-ssl" setting.
11831
11832no-send-proxy-v2-ssl-cn
11833 This option may be used as "server" setting to reset any "send-proxy-v2-ssl-cn"
11834 setting which would have been inherited from "default-server" directive as
11835 default value.
11836 It may also be used as "default-server" setting to reset any previous
11837 "default-server" "send-proxy-v2-ssl-cn" setting.
11838
11839no-ssl
11840 This option may be used as "server" setting to reset any "ssl"
11841 setting which would have been inherited from "default-server" directive as
11842 default value.
11843 It may also be used as "default-server" setting to reset any previous
11844 "default-server" "ssl" setting.
11845
Willy Tarreau2a3fb1c2015-02-05 16:47:07 +010011846no-ssl-reuse
11847 This option disables SSL session reuse when SSL is used to communicate with
11848 the server. It will force the server to perform a full handshake for every
11849 new connection. It's probably only useful for benchmarking, troubleshooting,
11850 and for paranoid users.
11851
Emeric Brun9b3009b2012-10-05 11:55:06 +020011852no-sslv3
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011853 This option disables support for SSLv3 when SSL is used to communicate with
11854 the server. Note that SSLv2 is disabled in the code and cannot be enabled
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011855 using any configuration option. Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011856
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011857 Supported in default-server: No
11858
Emeric Brunf9c5c472012-10-11 15:28:34 +020011859no-tls-tickets
11860 This setting is only available when support for OpenSSL was built in. It
11861 disables the stateless session resumption (RFC 5077 TLS Ticket
11862 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011863 session resumption is more expensive in CPU usage for servers. This option
11864 is also available on global statement "ssl-default-server-options".
Frédéric Lécailled2376272017-03-21 18:52:12 +010011865 See also "tls-tickets".
Emeric Brunf9c5c472012-10-11 15:28:34 +020011866
Emeric Brun9b3009b2012-10-05 11:55:06 +020011867no-tlsv10
Emeric Brun8694b9a2012-10-05 14:39:07 +020011868 This option disables support for TLSv1.0 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020011869 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11870 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011871 often makes sense to disable it when communicating with local servers. This
11872 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011873 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011874
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011875 Supported in default-server: No
11876
Emeric Brun9b3009b2012-10-05 11:55:06 +020011877no-tlsv11
Emeric Brun8694b9a2012-10-05 14:39:07 +020011878 This option disables support for TLSv1.1 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020011879 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11880 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011881 often makes sense to disable it when communicating with local servers. This
11882 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011883 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011884
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011885 Supported in default-server: No
11886
Emeric Brun9b3009b2012-10-05 11:55:06 +020011887no-tlsv12
Emeric Brun8694b9a2012-10-05 14:39:07 +020011888 This option disables support for TLSv1.2 when SSL is used to communicate with
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011889 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11890 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011891 often makes sense to disable it when communicating with local servers. This
11892 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011893 Use "ssl-min-ver" and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011894
11895 Supported in default-server: No
11896
11897no-tlsv13
11898 This option disables support for TLSv1.3 when SSL is used to communicate with
11899 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11900 using any configuration option. TLSv1 is more expensive than SSLv3 so it
11901 often makes sense to disable it when communicating with local servers. This
11902 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011903 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011904
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011905 Supported in default-server: No
11906
Frédéric Lécailled2376272017-03-21 18:52:12 +010011907no-verifyhost
11908 This option may be used as "server" setting to reset any "verifyhost"
11909 setting which would have been inherited from "default-server" directive as
11910 default value.
11911 It may also be used as "default-server" setting to reset any previous
11912 "default-server" "verifyhost" setting.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011913
Simon Hormanfa461682011-06-25 09:39:49 +090011914non-stick
11915 Never add connections allocated to this sever to a stick-table.
11916 This may be used in conjunction with backup to ensure that
11917 stick-table persistence is disabled for backup servers.
11918
Olivier Houchardc7566002018-11-20 23:33:50 +010011919npn <protocols>
11920 This enables the NPN TLS extension and advertises the specified protocol list
11921 as supported on top of NPN. The protocol list consists in a comma-delimited
11922 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
11923 This requires that the SSL library is build with support for TLS extensions
11924 enabled (check with haproxy -vv). Note that the NPN extension has been
11925 replaced with the ALPN extension (see the "alpn" keyword), though this one is
11926 only available starting with OpenSSL 1.0.2.
11927
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011928observe <mode>
11929 This option enables health adjusting based on observing communication with
11930 the server. By default this functionality is disabled and enabling it also
11931 requires to enable health checks. There are two supported modes: "layer4" and
11932 "layer7". In layer4 mode, only successful/unsuccessful tcp connections are
11933 significant. In layer7, which is only allowed for http proxies, responses
11934 received from server are verified, like valid/wrong http code, unparsable
Willy Tarreau150d1462012-03-10 08:19:02 +010011935 headers, a timeout, etc. Valid status codes include 100 to 499, 501 and 505.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011936
11937 See also the "check", "on-error" and "error-limit".
11938
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011939on-error <mode>
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011940 Select what should happen when enough consecutive errors are detected.
11941 Currently, four modes are available:
11942 - fastinter: force fastinter
11943 - fail-check: simulate a failed check, also forces fastinter (default)
11944 - sudden-death: simulate a pre-fatal failed health check, one more failed
11945 check will mark a server down, forces fastinter
11946 - mark-down: mark the server immediately down and force fastinter
11947
11948 See also the "check", "observe" and "error-limit".
11949
Simon Hormane0d1bfb2011-06-21 14:34:58 +090011950on-marked-down <action>
11951 Modify what occurs when a server is marked down.
11952 Currently one action is available:
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070011953 - shutdown-sessions: Shutdown peer sessions. When this setting is enabled,
11954 all connections to the server are immediately terminated when the server
11955 goes down. It might be used if the health check detects more complex cases
11956 than a simple connection status, and long timeouts would cause the service
11957 to remain unresponsive for too long a time. For instance, a health check
11958 might detect that a database is stuck and that there's no chance to reuse
11959 existing connections anymore. Connections killed this way are logged with
11960 a 'D' termination code (for "Down").
Simon Hormane0d1bfb2011-06-21 14:34:58 +090011961
11962 Actions are disabled by default
11963
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070011964on-marked-up <action>
11965 Modify what occurs when a server is marked up.
11966 Currently one action is available:
11967 - shutdown-backup-sessions: Shutdown sessions on all backup servers. This is
11968 done only if the server is not in backup state and if it is not disabled
11969 (it must have an effective weight > 0). This can be used sometimes to force
11970 an active server to take all the traffic back after recovery when dealing
Davor Ocelice9ed2812017-12-25 17:49:28 +010011971 with long sessions (e.g. LDAP, SQL, ...). Doing this can cause more trouble
11972 than it tries to solve (e.g. incomplete transactions), so use this feature
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070011973 with extreme care. Sessions killed because a server comes up are logged
11974 with an 'U' termination code (for "Up").
11975
11976 Actions are disabled by default
11977
Olivier Houchard006e3102018-12-10 18:30:32 +010011978pool-max-conn <max>
11979 Set the maximum number of idling connections for a server. -1 means unlimited
11980 connections, 0 means no idle connections. The default is -1. When idle
11981 connections are enabled, orphaned idle connections which do not belong to any
11982 client session anymore are moved to a dedicated pool so that they remain
11983 usable by future clients. This only applies to connections that can be shared
11984 according to the same principles as those applying to "http-reuse".
11985
Olivier Houchardb7b3faa2018-12-14 18:15:36 +010011986pool-purge-delay <delay>
11987 Sets the delay to start purging idle connections. Each <delay> interval, half
11988 of the idle connections are closed. 0 means it's never purged. The default is
11989 1s.
11990
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011991port <port>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011992 Using the "port" parameter, it becomes possible to use a different port to
11993 send health-checks. On some servers, it may be desirable to dedicate a port
11994 to a specific component able to perform complex tests which are more suitable
11995 to health-checks than the application. It is common to run a simple script in
11996 inetd for instance. This parameter is ignored if the "check" parameter is not
11997 set. See also the "addr" parameter.
11998
Christopher Faulet8ed0a3e2018-04-10 14:45:45 +020011999proto <name>
12000
12001 Forces the multiplexer's protocol to use for the outgoing connections to this
12002 server. It must be compatible with the mode of the backend (TCP or HTTP). It
12003 must also be usable on the backend side. The list of available protocols is
12004 reported in haproxy -vv.
12005 Idea behind this optipon is to bypass the selection of the best multiplexer's
12006 protocol for all connections established to this server.
12007
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012008redir <prefix>
12009 The "redir" parameter enables the redirection mode for all GET and HEAD
12010 requests addressing this server. This means that instead of having HAProxy
12011 forward the request to the server, it will send an "HTTP 302" response with
12012 the "Location" header composed of this prefix immediately followed by the
12013 requested URI beginning at the leading '/' of the path component. That means
12014 that no trailing slash should be used after <prefix>. All invalid requests
12015 will be rejected, and all non-GET or HEAD requests will be normally served by
12016 the server. Note that since the response is completely forged, no header
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012017 mangling nor cookie insertion is possible in the response. However, cookies in
Davor Ocelice9ed2812017-12-25 17:49:28 +010012018 requests are still analyzed, making this solution completely usable to direct
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012019 users to a remote location in case of local disaster. Main use consists in
12020 increasing bandwidth for static servers by having the clients directly
12021 connect to them. Note: never use a relative location here, it would cause a
12022 loop between the client and HAProxy!
12023
12024 Example : server srv1 192.168.1.1:80 redir http://image1.mydomain.com check
12025
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012026rise <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012027 The "rise" parameter states that a server will be considered as operational
12028 after <count> consecutive successful health checks. This value defaults to 2
12029 if unspecified. See also the "check", "inter" and "fall" parameters.
12030
Baptiste Assmann8e2d9432018-06-22 15:04:43 +020012031resolve-opts <option>,<option>,...
12032 Comma separated list of options to apply to DNS resolution linked to this
12033 server.
12034
12035 Available options:
12036
12037 * allow-dup-ip
12038 By default, HAProxy prevents IP address duplication in a backend when DNS
12039 resolution at runtime is in operation.
12040 That said, for some cases, it makes sense that two servers (in the same
12041 backend, being resolved by the same FQDN) have the same IP address.
12042 For such case, simply enable this option.
12043 This is the opposite of prevent-dup-ip.
12044
12045 * prevent-dup-ip
12046 Ensure HAProxy's default behavior is enforced on a server: prevent re-using
12047 an IP address already set to a server in the same backend and sharing the
12048 same fqdn.
12049 This is the opposite of allow-dup-ip.
12050
12051 Example:
12052 backend b_myapp
12053 default-server init-addr none resolvers dns
12054 server s1 myapp.example.com:80 check resolve-opts allow-dup-ip
12055 server s2 myapp.example.com:81 check resolve-opts allow-dup-ip
12056
12057 With the option allow-dup-ip set:
12058 * if the nameserver returns a single IP address, then both servers will use
12059 it
12060 * If the nameserver returns 2 IP addresses, then each server will pick up a
12061 different address
12062
12063 Default value: not set
12064
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012065resolve-prefer <family>
12066 When DNS resolution is enabled for a server and multiple IP addresses from
12067 different families are returned, HAProxy will prefer using an IP address
12068 from the family mentioned in the "resolve-prefer" parameter.
12069 Available families: "ipv4" and "ipv6"
12070
Baptiste Assmannc4aabae2015-08-04 22:43:06 +020012071 Default value: ipv6
12072
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012073 Example:
12074
12075 server s1 app1.domain.com:80 resolvers mydns resolve-prefer ipv6
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012076
Thierry Fournierac88cfe2016-02-17 22:05:30 +010012077resolve-net <network>[,<network[,...]]
12078 This options prioritize th choice of an ip address matching a network. This is
12079 useful with clouds to prefer a local ip. In some cases, a cloud high
Tim Düsterhus4896c442016-11-29 02:15:19 +010012080 availability service can be announced with many ip addresses on many
Davor Ocelice9ed2812017-12-25 17:49:28 +010012081 different datacenters. The latency between datacenter is not negligible, so
12082 this patch permits to prefer a local datacenter. If no address matches the
Thierry Fournierac88cfe2016-02-17 22:05:30 +010012083 configured network, another address is selected.
12084
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012085 Example:
12086
12087 server s1 app1.domain.com:80 resolvers mydns resolve-net 10.0.0.0/8
Thierry Fournierac88cfe2016-02-17 22:05:30 +010012088
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012089resolvers <id>
12090 Points to an existing "resolvers" section to resolve current server's
12091 hostname.
12092
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012093 Example:
12094
12095 server s1 app1.domain.com:80 check resolvers mydns
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012096
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012097 See also section 5.3
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012098
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010012099send-proxy
12100 The "send-proxy" parameter enforces use of the PROXY protocol over any
12101 connection established to this server. The PROXY protocol informs the other
12102 end about the layer 3/4 addresses of the incoming connection, so that it can
12103 know the client's address or the public address it accessed to, whatever the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010012104 upper layer protocol. For connections accepted by an "accept-proxy" or
12105 "accept-netscaler-cip" listener, the advertised address will be used. Only
12106 TCPv4 and TCPv6 address families are supported. Other families such as
12107 Unix sockets, will report an UNKNOWN family. Servers using this option can
12108 fully be chained to another instance of haproxy listening with an
12109 "accept-proxy" setting. This setting must not be used if the server isn't
12110 aware of the protocol. When health checks are sent to the server, the PROXY
12111 protocol is automatically used when this option is set, unless there is an
12112 explicit "port" or "addr" directive, in which case an explicit
12113 "check-send-proxy" directive would also be needed to use the PROXY protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010012114 See also the "no-send-proxy" option of this section and "accept-proxy" and
12115 "accept-netscaler-cip" option of the "bind" keyword.
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010012116
David Safb76832014-05-08 23:42:08 -040012117send-proxy-v2
12118 The "send-proxy-v2" parameter enforces use of the PROXY protocol version 2
12119 over any connection established to this server. The PROXY protocol informs
12120 the other end about the layer 3/4 addresses of the incoming connection, so
12121 that it can know the client's address or the public address it accessed to,
Emmanuel Hocdet404d9782017-10-24 10:55:14 +020012122 whatever the upper layer protocol. It also send ALPN information if an alpn
12123 have been negotiated. This setting must not be used if the server isn't aware
12124 of this version of the protocol. See also the "no-send-proxy-v2" option of
12125 this section and send-proxy" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040012126
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010012127proxy-v2-options <option>[,<option>]*
12128 The "proxy-v2-options" parameter add option to send in PROXY protocol version
12129 2 when "send-proxy-v2" is used. Options available are "ssl" (see also
Emmanuel Hocdetfa8d0f12018-02-01 15:53:52 +010012130 send-proxy-v2-ssl), "cert-cn" (see also "send-proxy-v2-ssl-cn"), "ssl-cipher":
12131 name of the used cipher, "cert-sig": signature algorithm of the used
Emmanuel Hocdet253c3b72018-02-01 18:29:59 +010012132 certificate, "cert-key": key algorithm of the used certificate), "authority":
12133 host name value passed by the client (only sni from a tls connection is
Emmanuel Hocdet4399c752018-02-05 15:26:43 +010012134 supported), "crc32c": checksum of the proxy protocol v2 header.
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010012135
David Safb76832014-05-08 23:42:08 -040012136send-proxy-v2-ssl
12137 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
12138 2 over any connection established to this server. The PROXY protocol informs
12139 the other end about the layer 3/4 addresses of the incoming connection, so
12140 that it can know the client's address or the public address it accessed to,
12141 whatever the upper layer protocol. In addition, the SSL information extension
12142 of the PROXY protocol is added to the PROXY protocol header. This setting
12143 must not be used if the server isn't aware of this version of the protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010012144 See also the "no-send-proxy-v2-ssl" option of this section and the
12145 "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040012146
12147send-proxy-v2-ssl-cn
12148 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
12149 2 over any connection established to this server. The PROXY protocol informs
12150 the other end about the layer 3/4 addresses of the incoming connection, so
12151 that it can know the client's address or the public address it accessed to,
12152 whatever the upper layer protocol. In addition, the SSL information extension
12153 of the PROXY protocol, along along with the Common Name from the subject of
12154 the client certificate (if any), is added to the PROXY protocol header. This
12155 setting must not be used if the server isn't aware of this version of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012156 protocol. See also the "no-send-proxy-v2-ssl-cn" option of this section and
12157 the "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040012158
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012159slowstart <start_time_in_ms>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012160 The "slowstart" parameter for a server accepts a value in milliseconds which
12161 indicates after how long a server which has just come back up will run at
12162 full speed. Just as with every other time-based parameter, it can be entered
12163 in any other explicit unit among { us, ms, s, m, h, d }. The speed grows
12164 linearly from 0 to 100% during this time. The limitation applies to two
12165 parameters :
12166
12167 - maxconn: the number of connections accepted by the server will grow from 1
12168 to 100% of the usual dynamic limit defined by (minconn,maxconn,fullconn).
12169
12170 - weight: when the backend uses a dynamic weighted algorithm, the weight
12171 grows linearly from 1 to 100%. In this case, the weight is updated at every
12172 health-check. For this reason, it is important that the "inter" parameter
12173 is smaller than the "slowstart", in order to maximize the number of steps.
12174
12175 The slowstart never applies when haproxy starts, otherwise it would cause
12176 trouble to running servers. It only applies when a server has been previously
12177 seen as failed.
12178
Willy Tarreau732eac42015-07-09 11:40:25 +020012179sni <expression>
12180 The "sni" parameter evaluates the sample fetch expression, converts it to a
12181 string and uses the result as the host name sent in the SNI TLS extension to
12182 the server. A typical use case is to send the SNI received from the client in
12183 a bridged HTTPS scenario, using the "ssl_fc_sni" sample fetch for the
Willy Tarreau2ab88672017-07-05 18:23:03 +020012184 expression, though alternatives such as req.hdr(host) can also make sense. If
12185 "verify required" is set (which is the recommended setting), the resulting
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012186 name will also be matched against the server certificate's names. See the
Jérôme Magninb36a6d22018-12-09 16:03:40 +010012187 "verify" directive for more details. If you want to set a SNI for health
12188 checks, see the "check-sni" directive for more details.
Willy Tarreau732eac42015-07-09 11:40:25 +020012189
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020012190source <addr>[:<pl>[-<ph>]] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +020012191source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020012192source <addr>[:<pl>[-<ph>]] [interface <name>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012193 The "source" parameter sets the source address which will be used when
12194 connecting to the server. It follows the exact same parameters and principle
12195 as the backend "source" keyword, except that it only applies to the server
12196 referencing it. Please consult the "source" keyword for details.
12197
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020012198 Additionally, the "source" statement on a server line allows one to specify a
12199 source port range by indicating the lower and higher bounds delimited by a
12200 dash ('-'). Some operating systems might require a valid IP address when a
12201 source port range is specified. It is permitted to have the same IP/range for
12202 several servers. Doing so makes it possible to bypass the maximum of 64k
12203 total concurrent connections. The limit will then reach 64k connections per
12204 server.
12205
Lukas Tribus7d56c6d2016-09-13 09:51:15 +000012206 Since Linux 4.2/libc 2.23 IP_BIND_ADDRESS_NO_PORT is set for connections
12207 specifying the source address without port(s).
12208
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012209ssl
Willy Tarreau44f65392013-06-25 07:56:20 +020012210 This option enables SSL ciphering on outgoing connections to the server. It
12211 is critical to verify server certificates using "verify" when using SSL to
12212 connect to servers, otherwise the communication is prone to trivial man in
12213 the-middle attacks rendering SSL useless. When this option is used, health
12214 checks are automatically sent in SSL too unless there is a "port" or an
12215 "addr" directive indicating the check should be sent to a different location.
Frédéric Lécailled2376272017-03-21 18:52:12 +010012216 See the "no-ssl" to disable "ssl" option and "check-ssl" option to force
12217 SSL health checks.
Willy Tarreau763a95b2012-10-04 23:15:39 +020012218
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012219ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
12220 This option enforces use of <version> or lower when SSL is used to communicate
12221 with the server. This option is also available on global statement
12222 "ssl-default-server-options". See also "ssl-min-ver".
12223
12224ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
12225 This option enforces use of <version> or upper when SSL is used to communicate
12226 with the server. This option is also available on global statement
12227 "ssl-default-server-options". See also "ssl-max-ver".
12228
Frédéric Lécailled2376272017-03-21 18:52:12 +010012229ssl-reuse
12230 This option may be used as "server" setting to reset any "no-ssl-reuse"
12231 setting which would have been inherited from "default-server" directive as
12232 default value.
12233 It may also be used as "default-server" setting to reset any previous
12234 "default-server" "no-ssl-reuse" setting.
12235
12236stick
12237 This option may be used as "server" setting to reset any "non-stick"
12238 setting which would have been inherited from "default-server" directive as
12239 default value.
12240 It may also be used as "default-server" setting to reset any previous
12241 "default-server" "non-stick" setting.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012242
Willy Tarreau163d4622015-10-13 16:16:41 +020012243tcp-ut <delay>
12244 Sets the TCP User Timeout for all outgoing connections to this server. This
12245 option is available on Linux since version 2.6.37. It allows haproxy to
12246 configure a timeout for sockets which contain data not receiving an
Davor Ocelice9ed2812017-12-25 17:49:28 +010012247 acknowledgment for the configured delay. This is especially useful on
Willy Tarreau163d4622015-10-13 16:16:41 +020012248 long-lived connections experiencing long idle periods such as remote
12249 terminals or database connection pools, where the client and server timeouts
12250 must remain high to allow a long period of idle, but where it is important to
12251 detect that the server has disappeared in order to release all resources
12252 associated with its connection (and the client's session). One typical use
12253 case is also to force dead server connections to die when health checks are
12254 too slow or during a soft reload since health checks are then disabled. The
12255 argument is a delay expressed in milliseconds by default. This only works for
12256 regular TCP connections, and is ignored for other protocols.
12257
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012258track [<proxy>/]<server>
Willy Tarreau32091232014-05-16 13:52:00 +020012259 This option enables ability to set the current state of the server by tracking
12260 another one. It is possible to track a server which itself tracks another
12261 server, provided that at the end of the chain, a server has health checks
12262 enabled. If <proxy> is omitted the current one is used. If disable-on-404 is
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012263 used, it has to be enabled on both proxies.
12264
Frédéric Lécailled2376272017-03-21 18:52:12 +010012265tls-tickets
12266 This option may be used as "server" setting to reset any "no-tls-tickets"
12267 setting which would have been inherited from "default-server" directive as
12268 default value.
12269 It may also be used as "default-server" setting to reset any previous
12270 "default-server" "no-tlsv-tickets" setting.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012271
Emeric Brunef42d922012-10-11 16:11:36 +020012272verify [none|required]
12273 This setting is only available when support for OpenSSL was built in. If set
Emeric Brun850efd52014-01-29 12:24:34 +010012274 to 'none', server certificate is not verified. In the other case, The
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012275 certificate provided by the server is verified using CAs from 'ca-file' and
12276 optional CRLs from 'crl-file' after having checked that the names provided in
Davor Ocelice9ed2812017-12-25 17:49:28 +010012277 the certificate's subject and subjectAlternateNames attributes match either
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012278 the name passed using the "sni" directive, or if not provided, the static
12279 host name passed using the "verifyhost" directive. When no name is found, the
12280 certificate's names are ignored. For this reason, without SNI it's important
12281 to use "verifyhost". On verification failure the handshake is aborted. It is
12282 critically important to verify server certificates when using SSL to connect
12283 to servers, otherwise the communication is prone to trivial man-in-the-middle
12284 attacks rendering SSL totally useless. Unless "ssl_server_verify" appears in
12285 the global section, "verify" is set to "required" by default.
Emeric Brunef42d922012-10-11 16:11:36 +020012286
Evan Broderbe554312013-06-27 00:05:25 -070012287verifyhost <hostname>
12288 This setting is only available when support for OpenSSL was built in, and
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012289 only takes effect if 'verify required' is also specified. This directive sets
12290 a default static hostname to check the server's certificate against when no
12291 SNI was used to connect to the server. If SNI is not used, this is the only
12292 way to enable hostname verification. This static hostname, when set, will
12293 also be used for health checks (which cannot provide an SNI value). If none
12294 of the hostnames in the certificate match the specified hostname, the
12295 handshake is aborted. The hostnames in the server-provided certificate may
12296 include wildcards. See also "verify", "sni" and "no-verifyhost" options.
Evan Broderbe554312013-06-27 00:05:25 -070012297
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012298weight <weight>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012299 The "weight" parameter is used to adjust the server's weight relative to
12300 other servers. All servers will receive a load proportional to their weight
12301 relative to the sum of all weights, so the higher the weight, the higher the
Willy Tarreau6704d672009-06-15 10:56:05 +020012302 load. The default weight is 1, and the maximal value is 256. A value of 0
12303 means the server will not participate in load-balancing but will still accept
12304 persistent connections. If this parameter is used to distribute the load
12305 according to server's capacity, it is recommended to start with values which
12306 can both grow and shrink, for instance between 10 and 100 to leave enough
12307 room above and below for later adjustments.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012308
12309
Cyril Bonté46175dd2015-07-02 22:45:32 +0200123105.3. Server IP address resolution using DNS
12311-------------------------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012312
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012313HAProxy allows using a host name on the server line to retrieve its IP address
12314using name servers. By default, HAProxy resolves the name when parsing the
12315configuration file, at startup and cache the result for the process' life.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012316This is not sufficient in some cases, such as in Amazon where a server's IP
12317can change after a reboot or an ELB Virtual IP can change based on current
12318workload.
12319This chapter describes how HAProxy can be configured to process server's name
12320resolution at run time.
12321Whether run time server name resolution has been enable or not, HAProxy will
12322carry on doing the first resolution when parsing the configuration.
12323
12324
Cyril Bonté46175dd2015-07-02 22:45:32 +0200123255.3.1. Global overview
12326----------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012327
12328As we've seen in introduction, name resolution in HAProxy occurs at two
12329different steps of the process life:
12330
12331 1. when starting up, HAProxy parses the server line definition and matches a
12332 host name. It uses libc functions to get the host name resolved. This
12333 resolution relies on /etc/resolv.conf file.
12334
Christopher Faulet67957bd2017-09-27 11:00:59 +020012335 2. at run time, HAProxy performs periodically name resolutions for servers
12336 requiring DNS resolutions.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012337
12338A few other events can trigger a name resolution at run time:
12339 - when a server's health check ends up in a connection timeout: this may be
12340 because the server has a new IP address. So we need to trigger a name
12341 resolution to know this new IP.
12342
Christopher Faulet67957bd2017-09-27 11:00:59 +020012343When using resolvers, the server name can either be a hostname, or a SRV label.
Davor Ocelice9ed2812017-12-25 17:49:28 +010012344HAProxy considers anything that starts with an underscore as a SRV label. If a
Christopher Faulet67957bd2017-09-27 11:00:59 +020012345SRV label is specified, then the corresponding SRV records will be retrieved
12346from the DNS server, and the provided hostnames will be used. The SRV label
12347will be checked periodically, and if any server are added or removed, haproxy
12348will automatically do the same.
Olivier Houchardecfa18d2017-08-07 17:30:03 +020012349
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012350A few things important to notice:
12351 - all the name servers are queried in the mean time. HAProxy will process the
12352 first valid response.
12353
12354 - a resolution is considered as invalid (NX, timeout, refused), when all the
12355 servers return an error.
12356
12357
Cyril Bonté46175dd2015-07-02 22:45:32 +0200123585.3.2. The resolvers section
12359----------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012360
12361This section is dedicated to host information related to name resolution in
Christopher Faulet67957bd2017-09-27 11:00:59 +020012362HAProxy. There can be as many as resolvers section as needed. Each section can
12363contain many name servers.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012364
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012365When multiple name servers are configured in a resolvers section, then HAProxy
12366uses the first valid response. In case of invalid responses, only the last one
12367is treated. Purpose is to give the chance to a slow server to deliver a valid
12368answer after a fast faulty or outdated server.
12369
12370When each server returns a different error type, then only the last error is
Christopher Faulet67957bd2017-09-27 11:00:59 +020012371used by HAProxy. The following processing is applied on this error:
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012372
Christopher Faulet67957bd2017-09-27 11:00:59 +020012373 1. HAProxy retries the same DNS query with a new query type. The A queries are
12374 switch to AAAA or the opposite. SRV queries are not concerned here. Timeout
12375 errors are also excluded.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012376
Christopher Faulet67957bd2017-09-27 11:00:59 +020012377 2. When the fallback on the query type was done (or not applicable), HAProxy
12378 retries the original DNS query, with the preferred query type.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012379
Christopher Faulet67957bd2017-09-27 11:00:59 +020012380 3. HAProxy retries previous steps <resolve_retires> times. If no valid
12381 response is received after that, it stops the DNS resolution and reports
12382 the error.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012383
Christopher Faulet67957bd2017-09-27 11:00:59 +020012384For example, with 2 name servers configured in a resolvers section, the
12385following scenarios are possible:
12386
12387 - First response is valid and is applied directly, second response is
12388 ignored
12389
12390 - First response is invalid and second one is valid, then second response is
12391 applied
12392
12393 - First response is a NX domain and second one a truncated response, then
12394 HAProxy retries the query with a new type
12395
12396 - First response is a NX domain and second one is a timeout, then HAProxy
12397 retries the query with a new type
12398
12399 - Query timed out for both name servers, then HAProxy retries it with the
12400 same query type
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012401
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012402As a DNS server may not answer all the IPs in one DNS request, haproxy keeps
12403a cache of previous answers, an answer will be considered obsolete after
Christopher Faulet67957bd2017-09-27 11:00:59 +020012404<hold obsolete> seconds without the IP returned.
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012405
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012406
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012407resolvers <resolvers id>
Davor Ocelice9ed2812017-12-25 17:49:28 +010012408 Creates a new name server list labeled <resolvers id>
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012409
12410A resolvers section accept the following parameters:
12411
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020012412accepted_payload_size <nb>
Davor Ocelice9ed2812017-12-25 17:49:28 +010012413 Defines the maximum payload size accepted by HAProxy and announced to all the
Christopher Faulet67957bd2017-09-27 11:00:59 +020012414 name servers configured in this resolvers section.
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020012415 <nb> is in bytes. If not set, HAProxy announces 512. (minimal value defined
12416 by RFC 6891)
12417
Baptiste Assmann9d8dbbc2017-08-18 23:35:08 +020012418 Note: the maximum allowed value is 8192.
12419
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012420nameserver <id> <ip>:<port>
12421 DNS server description:
12422 <id> : label of the server, should be unique
12423 <ip> : IP address of the server
12424 <port> : port where the DNS service actually runs
12425
Ben Draut44e609b2018-05-29 15:40:08 -060012426parse-resolv-conf
12427 Adds all nameservers found in /etc/resolv.conf to this resolvers nameservers
12428 list. Ordered as if each nameserver in /etc/resolv.conf was individually
12429 placed in the resolvers section in place of this directive.
12430
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012431hold <status> <period>
12432 Defines <period> during which the last name resolution should be kept based
12433 on last resolution <status>
Baptiste Assmann987e16d2016-11-02 22:23:31 +010012434 <status> : last name resolution status. Acceptable values are "nx",
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012435 "other", "refused", "timeout", "valid", "obsolete".
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012436 <period> : interval between two successive name resolution when the last
12437 answer was in <status>. It follows the HAProxy time format.
12438 <period> is in milliseconds by default.
12439
Baptiste Assmann686408b2017-08-18 10:15:42 +020012440 Default value is 10s for "valid", 0s for "obsolete" and 30s for others.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012441
Christopher Faulet67957bd2017-09-27 11:00:59 +020012442resolution_pool_size <nb> (deprecated)
Baptiste Assmann201c07f2017-05-22 15:17:15 +020012443 Defines the number of resolutions available in the pool for this resolvers.
12444 If not defines, it defaults to 64. If your configuration requires more than
12445 <nb>, then HAProxy will return an error when parsing the configuration.
12446
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012447resolve_retries <nb>
12448 Defines the number <nb> of queries to send to resolve a server name before
12449 giving up.
12450 Default value: 3
12451
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012452 A retry occurs on name server timeout or when the full sequence of DNS query
12453 type failover is over and we need to start up from the default ANY query
12454 type.
12455
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012456timeout <event> <time>
12457 Defines timeouts related to name resolution
12458 <event> : the event on which the <time> timeout period applies to.
12459 events available are:
Christopher Faulet67957bd2017-09-27 11:00:59 +020012460 - resolve : default time to trigger name resolutions when no
12461 other time applied.
12462 Default value: 1s
12463 - retry : time between two DNS queries, when no valid response
12464 have been received.
12465 Default value: 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012466 <time> : time related to the event. It follows the HAProxy time format.
12467 <time> is expressed in milliseconds.
12468
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012469 Example:
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012470
12471 resolvers mydns
12472 nameserver dns1 10.0.0.1:53
12473 nameserver dns2 10.0.0.2:53
Ben Draut44e609b2018-05-29 15:40:08 -060012474 parse-resolv-conf
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012475 resolve_retries 3
Christopher Faulet67957bd2017-09-27 11:00:59 +020012476 timeout resolve 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012477 timeout retry 1s
Baptiste Assmann987e16d2016-11-02 22:23:31 +010012478 hold other 30s
12479 hold refused 30s
12480 hold nx 30s
12481 hold timeout 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012482 hold valid 10s
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012483 hold obsolete 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012484
12485
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200124866. HTTP header manipulation
12487---------------------------
12488
12489In HTTP mode, it is possible to rewrite, add or delete some of the request and
12490response headers based on regular expressions. It is also possible to block a
12491request or a response if a particular header matches a regular expression,
12492which is enough to stop most elementary protocol attacks, and to protect
Willy Tarreau70dffda2014-01-30 03:07:23 +010012493against information leak from the internal network.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012494
Willy Tarreau70dffda2014-01-30 03:07:23 +010012495If HAProxy encounters an "Informational Response" (status code 1xx), it is able
12496to process all rsp* rules which can allow, deny, rewrite or delete a header,
12497but it will refuse to add a header to any such messages as this is not
12498HTTP-compliant. The reason for still processing headers in such responses is to
12499stop and/or fix any possible information leak which may happen, for instance
12500because another downstream equipment would unconditionally add a header, or if
12501a server name appears there. When such messages are seen, normal processing
12502still occurs on the next non-informational messages.
Willy Tarreau816b9792009-09-15 21:25:21 +020012503
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012504This section covers common usage of the following keywords, described in detail
12505in section 4.2 :
12506
12507 - reqadd <string>
12508 - reqallow <search>
12509 - reqiallow <search>
12510 - reqdel <search>
12511 - reqidel <search>
12512 - reqdeny <search>
12513 - reqideny <search>
12514 - reqpass <search>
12515 - reqipass <search>
12516 - reqrep <search> <replace>
12517 - reqirep <search> <replace>
12518 - reqtarpit <search>
12519 - reqitarpit <search>
12520 - rspadd <string>
12521 - rspdel <search>
12522 - rspidel <search>
12523 - rspdeny <search>
12524 - rspideny <search>
12525 - rsprep <search> <replace>
12526 - rspirep <search> <replace>
12527
12528With all these keywords, the same conventions are used. The <search> parameter
12529is a POSIX extended regular expression (regex) which supports grouping through
12530parenthesis (without the backslash). Spaces and other delimiters must be
12531prefixed with a backslash ('\') to avoid confusion with a field delimiter.
12532Other characters may be prefixed with a backslash to change their meaning :
12533
12534 \t for a tab
12535 \r for a carriage return (CR)
12536 \n for a new line (LF)
12537 \ to mark a space and differentiate it from a delimiter
12538 \# to mark a sharp and differentiate it from a comment
12539 \\ to use a backslash in a regex
12540 \\\\ to use a backslash in the text (*2 for regex, *2 for haproxy)
12541 \xXX to write the ASCII hex code XX as in the C language
12542
12543The <replace> parameter contains the string to be used to replace the largest
12544portion of text matching the regex. It can make use of the special characters
12545above, and can reference a substring which is delimited by parenthesis in the
12546regex, by writing a backslash ('\') immediately followed by one digit from 0 to
125479 indicating the group position (0 designating the entire line). This practice
12548is very common to users of the "sed" program.
12549
12550The <string> parameter represents the string which will systematically be added
12551after the last header line. It can also use special character sequences above.
12552
12553Notes related to these keywords :
12554---------------------------------
12555 - these keywords are not always convenient to allow/deny based on header
12556 contents. It is strongly recommended to use ACLs with the "block" keyword
12557 instead, resulting in far more flexible and manageable rules.
12558
12559 - lines are always considered as a whole. It is not possible to reference
12560 a header name only or a value only. This is important because of the way
12561 headers are written (notably the number of spaces after the colon).
12562
12563 - the first line is always considered as a header, which makes it possible to
12564 rewrite or filter HTTP requests URIs or response codes, but in turn makes
12565 it harder to distinguish between headers and request line. The regex prefix
12566 ^[^\ \t]*[\ \t] matches any HTTP method followed by a space, and the prefix
12567 ^[^ \t:]*: matches any header name followed by a colon.
12568
12569 - for performances reasons, the number of characters added to a request or to
12570 a response is limited at build time to values between 1 and 4 kB. This
12571 should normally be far more than enough for most usages. If it is too short
12572 on occasional usages, it is possible to gain some space by removing some
12573 useless headers before adding new ones.
12574
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012575 - keywords beginning with "reqi" and "rspi" are the same as their counterpart
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012576 without the 'i' letter except that they ignore case when matching patterns.
12577
12578 - when a request passes through a frontend then a backend, all req* rules
12579 from the frontend will be evaluated, then all req* rules from the backend
12580 will be evaluated. The reverse path is applied to responses.
12581
12582 - req* statements are applied after "block" statements, so that "block" is
12583 always the first one, but before "use_backend" in order to permit rewriting
Willy Tarreaud72758d2010-01-12 10:42:19 +010012584 before switching.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012585
12586
Willy Tarreau74ca5042013-06-11 23:12:07 +0200125877. Using ACLs and fetching samples
12588----------------------------------
12589
Davor Ocelice9ed2812017-12-25 17:49:28 +010012590HAProxy is capable of extracting data from request or response streams, from
Willy Tarreau74ca5042013-06-11 23:12:07 +020012591client or server information, from tables, environmental information etc...
12592The action of extracting such data is called fetching a sample. Once retrieved,
12593these samples may be used for various purposes such as a key to a stick-table,
12594but most common usages consist in matching them against predefined constant
12595data called patterns.
12596
12597
125987.1. ACL basics
12599---------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012600
12601The use of Access Control Lists (ACL) provides a flexible solution to perform
12602content switching and generally to take decisions based on content extracted
12603from the request, the response or any environmental status. The principle is
12604simple :
12605
Willy Tarreau74ca5042013-06-11 23:12:07 +020012606 - extract a data sample from a stream, table or the environment
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012607 - optionally apply some format conversion to the extracted sample
Willy Tarreau74ca5042013-06-11 23:12:07 +020012608 - apply one or multiple pattern matching methods on this sample
12609 - perform actions only when a pattern matches the sample
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012610
Willy Tarreau74ca5042013-06-11 23:12:07 +020012611The actions generally consist in blocking a request, selecting a backend, or
12612adding a header.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012613
12614In order to define a test, the "acl" keyword is used. The syntax is :
12615
Willy Tarreau74ca5042013-06-11 23:12:07 +020012616 acl <aclname> <criterion> [flags] [operator] [<value>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012617
12618This creates a new ACL <aclname> or completes an existing one with new tests.
12619Those tests apply to the portion of request/response specified in <criterion>
12620and may be adjusted with optional flags [flags]. Some criteria also support
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012621an operator which may be specified before the set of values. Optionally some
12622conversion operators may be applied to the sample, and they will be specified
12623as a comma-delimited list of keywords just after the first keyword. The values
12624are of the type supported by the criterion, and are separated by spaces.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012625
12626ACL names must be formed from upper and lower case letters, digits, '-' (dash),
12627'_' (underscore) , '.' (dot) and ':' (colon). ACL names are case-sensitive,
12628which means that "my_acl" and "My_Acl" are two different ACLs.
12629
12630There is no enforced limit to the number of ACLs. The unused ones do not affect
12631performance, they just consume a small amount of memory.
12632
Willy Tarreau74ca5042013-06-11 23:12:07 +020012633The criterion generally is the name of a sample fetch method, or one of its ACL
12634specific declinations. The default test method is implied by the output type of
12635this sample fetch method. The ACL declinations can describe alternate matching
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012636methods of a same sample fetch method. The sample fetch methods are the only
12637ones supporting a conversion.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012638
12639Sample fetch methods return data which can be of the following types :
12640 - boolean
12641 - integer (signed or unsigned)
12642 - IPv4 or IPv6 address
12643 - string
12644 - data block
12645
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012646Converters transform any of these data into any of these. For example, some
12647converters might convert a string to a lower-case string while other ones
12648would turn a string to an IPv4 address, or apply a netmask to an IP address.
12649The resulting sample is of the type of the last converter applied to the list,
12650which defaults to the type of the sample fetch method.
12651
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012652Each sample or converter returns data of a specific type, specified with its
12653keyword in this documentation. When an ACL is declared using a standard sample
12654fetch method, certain types automatically involved a default matching method
12655which are summarized in the table below :
12656
12657 +---------------------+-----------------+
12658 | Sample or converter | Default |
12659 | output type | matching method |
12660 +---------------------+-----------------+
12661 | boolean | bool |
12662 +---------------------+-----------------+
12663 | integer | int |
12664 +---------------------+-----------------+
12665 | ip | ip |
12666 +---------------------+-----------------+
12667 | string | str |
12668 +---------------------+-----------------+
12669 | binary | none, use "-m" |
12670 +---------------------+-----------------+
12671
12672Note that in order to match a binary samples, it is mandatory to specify a
12673matching method, see below.
12674
Willy Tarreau74ca5042013-06-11 23:12:07 +020012675The ACL engine can match these types against patterns of the following types :
12676 - boolean
12677 - integer or integer range
12678 - IP address / network
12679 - string (exact, substring, suffix, prefix, subdir, domain)
12680 - regular expression
12681 - hex block
12682
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012683The following ACL flags are currently supported :
12684
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012685 -i : ignore case during matching of all subsequent patterns.
12686 -f : load patterns from a file.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012687 -m : use a specific pattern matching method
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010012688 -n : forbid the DNS resolutions
Thierry FOURNIER9860c412014-01-29 14:23:29 +010012689 -M : load the file pointed by -f like a map file.
Thierry FOURNIER3534d882014-01-20 17:01:44 +010012690 -u : force the unique id of the ACL
Willy Tarreau6a06a402007-07-15 20:15:28 +020012691 -- : force end of flags. Useful when a string looks like one of the flags.
12692
Willy Tarreau74ca5042013-06-11 23:12:07 +020012693The "-f" flag is followed by the name of a file from which all lines will be
12694read as individual values. It is even possible to pass multiple "-f" arguments
12695if the patterns are to be loaded from multiple files. Empty lines as well as
12696lines beginning with a sharp ('#') will be ignored. All leading spaces and tabs
12697will be stripped. If it is absolutely necessary to insert a valid pattern
12698beginning with a sharp, just prefix it with a space so that it is not taken for
12699a comment. Depending on the data type and match method, haproxy may load the
12700lines into a binary tree, allowing very fast lookups. This is true for IPv4 and
12701exact string matching. In this case, duplicates will automatically be removed.
12702
Thierry FOURNIER9860c412014-01-29 14:23:29 +010012703The "-M" flag allows an ACL to use a map file. If this flag is set, the file is
12704parsed as two column file. The first column contains the patterns used by the
12705ACL, and the second column contain the samples. The sample can be used later by
12706a map. This can be useful in some rare cases where an ACL would just be used to
12707check for the existence of a pattern in a map before a mapping is applied.
12708
Thierry FOURNIER3534d882014-01-20 17:01:44 +010012709The "-u" flag forces the unique id of the ACL. This unique id is used with the
12710socket interface to identify ACL and dynamically change its values. Note that a
12711file is always identified by its name even if an id is set.
12712
Willy Tarreau74ca5042013-06-11 23:12:07 +020012713Also, note that the "-i" flag applies to subsequent entries and not to entries
12714loaded from files preceding it. For instance :
12715
12716 acl valid-ua hdr(user-agent) -f exact-ua.lst -i -f generic-ua.lst test
12717
12718In this example, each line of "exact-ua.lst" will be exactly matched against
12719the "user-agent" header of the request. Then each line of "generic-ua" will be
12720case-insensitively matched. Then the word "test" will be insensitively matched
12721as well.
12722
12723The "-m" flag is used to select a specific pattern matching method on the input
12724sample. All ACL-specific criteria imply a pattern matching method and generally
12725do not need this flag. However, this flag is useful with generic sample fetch
12726methods to describe how they're going to be matched against the patterns. This
12727is required for sample fetches which return data type for which there is no
Davor Ocelice9ed2812017-12-25 17:49:28 +010012728obvious matching method (e.g. string or binary). When "-m" is specified and
Willy Tarreau74ca5042013-06-11 23:12:07 +020012729followed by a pattern matching method name, this method is used instead of the
12730default one for the criterion. This makes it possible to match contents in ways
12731that were not initially planned, or with sample fetch methods which return a
12732string. The matching method also affects the way the patterns are parsed.
12733
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010012734The "-n" flag forbids the dns resolutions. It is used with the load of ip files.
12735By default, if the parser cannot parse ip address it considers that the parsed
12736string is maybe a domain name and try dns resolution. The flag "-n" disable this
12737resolution. It is useful for detecting malformed ip lists. Note that if the DNS
12738server is not reachable, the haproxy configuration parsing may last many minutes
12739waiting fir the timeout. During this time no error messages are displayed. The
12740flag "-n" disable this behavior. Note also that during the runtime, this
12741function is disabled for the dynamic acl modifications.
12742
Willy Tarreau74ca5042013-06-11 23:12:07 +020012743There are some restrictions however. Not all methods can be used with all
12744sample fetch methods. Also, if "-m" is used in conjunction with "-f", it must
12745be placed first. The pattern matching method must be one of the following :
Willy Tarreau5adeda12013-03-31 22:13:34 +020012746
12747 - "found" : only check if the requested sample could be found in the stream,
12748 but do not compare it against any pattern. It is recommended not
Willy Tarreau74ca5042013-06-11 23:12:07 +020012749 to pass any pattern to avoid confusion. This matching method is
12750 particularly useful to detect presence of certain contents such
12751 as headers, cookies, etc... even if they are empty and without
12752 comparing them to anything nor counting them.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012753
12754 - "bool" : check the value as a boolean. It can only be applied to fetches
12755 which return a boolean or integer value, and takes no pattern.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012756 Value zero or false does not match, all other values do match.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012757
12758 - "int" : match the value as an integer. It can be used with integer and
Willy Tarreau74ca5042013-06-11 23:12:07 +020012759 boolean samples. Boolean false is integer 0, true is integer 1.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012760
12761 - "ip" : match the value as an IPv4 or IPv6 address. It is compatible
Willy Tarreau74ca5042013-06-11 23:12:07 +020012762 with IP address samples only, so it is implied and never needed.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012763
Davor Ocelice9ed2812017-12-25 17:49:28 +010012764 - "bin" : match the contents against a hexadecimal string representing a
Willy Tarreau5adeda12013-03-31 22:13:34 +020012765 binary sequence. This may be used with binary or string samples.
12766
12767 - "len" : match the sample's length as an integer. This may be used with
12768 binary or string samples.
12769
Willy Tarreau74ca5042013-06-11 23:12:07 +020012770 - "str" : exact match : match the contents against a string. This may be
12771 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012772
Willy Tarreau74ca5042013-06-11 23:12:07 +020012773 - "sub" : substring match : check that the contents contain at least one of
12774 the provided string patterns. This may be used with binary or
12775 string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012776
Willy Tarreau74ca5042013-06-11 23:12:07 +020012777 - "reg" : regex match : match the contents against a list of regular
12778 expressions. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012779
Willy Tarreau74ca5042013-06-11 23:12:07 +020012780 - "beg" : prefix match : check that the contents begin like the provided
12781 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012782
Willy Tarreau74ca5042013-06-11 23:12:07 +020012783 - "end" : suffix match : check that the contents end like the provided
12784 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012785
Willy Tarreau74ca5042013-06-11 23:12:07 +020012786 - "dir" : subdir match : check that a slash-delimited portion of the
12787 contents exactly matches one of the provided string patterns.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012788 This may be used with binary or string samples.
12789
Willy Tarreau74ca5042013-06-11 23:12:07 +020012790 - "dom" : domain match : check that a dot-delimited portion of the contents
12791 exactly match one of the provided string patterns. This may be
12792 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012793
12794For example, to quickly detect the presence of cookie "JSESSIONID" in an HTTP
12795request, it is possible to do :
12796
12797 acl jsess_present cook(JSESSIONID) -m found
12798
12799In order to apply a regular expression on the 500 first bytes of data in the
12800buffer, one would use the following acl :
12801
12802 acl script_tag payload(0,500) -m reg -i <script>
12803
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012804On systems where the regex library is much slower when using "-i", it is
12805possible to convert the sample to lowercase before matching, like this :
12806
12807 acl script_tag payload(0,500),lower -m reg <script>
12808
Willy Tarreau74ca5042013-06-11 23:12:07 +020012809All ACL-specific criteria imply a default matching method. Most often, these
12810criteria are composed by concatenating the name of the original sample fetch
12811method and the matching method. For example, "hdr_beg" applies the "beg" match
12812to samples retrieved using the "hdr" fetch method. Since all ACL-specific
12813criteria rely on a sample fetch method, it is always possible instead to use
12814the original sample fetch method and the explicit matching method using "-m".
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012815
Willy Tarreau74ca5042013-06-11 23:12:07 +020012816If an alternate match is specified using "-m" on an ACL-specific criterion,
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012817the matching method is simply applied to the underlying sample fetch method.
12818For example, all ACLs below are exact equivalent :
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012819
Willy Tarreau74ca5042013-06-11 23:12:07 +020012820 acl short_form hdr_beg(host) www.
12821 acl alternate1 hdr_beg(host) -m beg www.
12822 acl alternate2 hdr_dom(host) -m beg www.
12823 acl alternate3 hdr(host) -m beg www.
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012824
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012825
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012826The table below summarizes the compatibility matrix between sample or converter
12827types and the pattern types to fetch against. It indicates for each compatible
12828combination the name of the matching method to be used, surrounded with angle
12829brackets ">" and "<" when the method is the default one and will work by
12830default without "-m".
Willy Tarreau0ba27502007-12-24 16:55:16 +010012831
Willy Tarreau74ca5042013-06-11 23:12:07 +020012832 +-------------------------------------------------+
12833 | Input sample type |
12834 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012835 | pattern type | boolean | integer | ip | string | binary |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012836 +----------------------+---------+---------+---------+---------+---------+
12837 | none (presence only) | found | found | found | found | found |
12838 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012839 | none (boolean value) |> bool <| bool | | bool | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012840 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012841 | integer (value) | int |> int <| int | int | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012842 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012843 | integer (length) | len | len | len | len | len |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012844 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012845 | IP address | | |> ip <| ip | ip |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012846 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012847 | exact string | str | str | str |> str <| str |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012848 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012849 | prefix | beg | beg | beg | beg | beg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012850 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012851 | suffix | end | end | end | end | end |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012852 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012853 | substring | sub | sub | sub | sub | sub |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012854 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012855 | subdir | dir | dir | dir | dir | dir |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012856 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012857 | domain | dom | dom | dom | dom | dom |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012858 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012859 | regex | reg | reg | reg | reg | reg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012860 +----------------------+---------+---------+---------+---------+---------+
12861 | hex block | | | | bin | bin |
12862 +----------------------+---------+---------+---------+---------+---------+
Willy Tarreau6a06a402007-07-15 20:15:28 +020012863
12864
Willy Tarreau74ca5042013-06-11 23:12:07 +0200128657.1.1. Matching booleans
12866------------------------
12867
12868In order to match a boolean, no value is needed and all values are ignored.
12869Boolean matching is used by default for all fetch methods of type "boolean".
12870When boolean matching is used, the fetched value is returned as-is, which means
12871that a boolean "true" will always match and a boolean "false" will never match.
12872
12873Boolean matching may also be enforced using "-m bool" on fetch methods which
12874return an integer value. Then, integer value 0 is converted to the boolean
12875"false" and all other values are converted to "true".
12876
Willy Tarreau6a06a402007-07-15 20:15:28 +020012877
Willy Tarreau74ca5042013-06-11 23:12:07 +0200128787.1.2. Matching integers
12879------------------------
12880
12881Integer matching applies by default to integer fetch methods. It can also be
12882enforced on boolean fetches using "-m int". In this case, "false" is converted
12883to the integer 0, and "true" is converted to the integer 1.
12884
12885Integer matching also supports integer ranges and operators. Note that integer
12886matching only applies to positive values. A range is a value expressed with a
12887lower and an upper bound separated with a colon, both of which may be omitted.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012888
12889For instance, "1024:65535" is a valid range to represent a range of
12890unprivileged ports, and "1024:" would also work. "0:1023" is a valid
12891representation of privileged ports, and ":1023" would also work.
12892
Willy Tarreau62644772008-07-16 18:36:06 +020012893As a special case, some ACL functions support decimal numbers which are in fact
12894two integers separated by a dot. This is used with some version checks for
12895instance. All integer properties apply to those decimal numbers, including
12896ranges and operators.
12897
Willy Tarreau6a06a402007-07-15 20:15:28 +020012898For an easier usage, comparison operators are also supported. Note that using
Willy Tarreau0ba27502007-12-24 16:55:16 +010012899operators with ranges does not make much sense and is strongly discouraged.
12900Similarly, it does not make much sense to perform order comparisons with a set
12901of values.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012902
Willy Tarreau0ba27502007-12-24 16:55:16 +010012903Available operators for integer matching are :
Willy Tarreau6a06a402007-07-15 20:15:28 +020012904
12905 eq : true if the tested value equals at least one value
12906 ge : true if the tested value is greater than or equal to at least one value
12907 gt : true if the tested value is greater than at least one value
12908 le : true if the tested value is less than or equal to at least one value
12909 lt : true if the tested value is less than at least one value
12910
Willy Tarreau0ba27502007-12-24 16:55:16 +010012911For instance, the following ACL matches any negative Content-Length header :
Willy Tarreau6a06a402007-07-15 20:15:28 +020012912
12913 acl negative-length hdr_val(content-length) lt 0
12914
Willy Tarreau62644772008-07-16 18:36:06 +020012915This one matches SSL versions between 3.0 and 3.1 (inclusive) :
12916
12917 acl sslv3 req_ssl_ver 3:3.1
12918
Willy Tarreau6a06a402007-07-15 20:15:28 +020012919
Willy Tarreau74ca5042013-06-11 23:12:07 +0200129207.1.3. Matching strings
12921-----------------------
12922
12923String matching applies to string or binary fetch methods, and exists in 6
12924different forms :
12925
12926 - exact match (-m str) : the extracted string must exactly match the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012927 patterns;
Willy Tarreau74ca5042013-06-11 23:12:07 +020012928
12929 - substring match (-m sub) : the patterns are looked up inside the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012930 extracted string, and the ACL matches if any of them is found inside;
Willy Tarreau74ca5042013-06-11 23:12:07 +020012931
12932 - prefix match (-m beg) : the patterns are compared with the beginning of
12933 the extracted string, and the ACL matches if any of them matches.
12934
12935 - suffix match (-m end) : the patterns are compared with the end of the
12936 extracted string, and the ACL matches if any of them matches.
12937
Baptiste Assmann33db6002016-03-06 23:32:10 +010012938 - subdir match (-m dir) : the patterns are looked up inside the extracted
Willy Tarreau74ca5042013-06-11 23:12:07 +020012939 string, delimited with slashes ("/"), and the ACL matches if any of them
12940 matches.
12941
12942 - domain match (-m dom) : the patterns are looked up inside the extracted
12943 string, delimited with dots ("."), and the ACL matches if any of them
12944 matches.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012945
12946String matching applies to verbatim strings as they are passed, with the
12947exception of the backslash ("\") which makes it possible to escape some
12948characters such as the space. If the "-i" flag is passed before the first
12949string, then the matching will be performed ignoring the case. In order
12950to match the string "-i", either set it second, or pass the "--" flag
Willy Tarreau0ba27502007-12-24 16:55:16 +010012951before the first string. Same applies of course to match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020012952
12953
Willy Tarreau74ca5042013-06-11 23:12:07 +0200129547.1.4. Matching regular expressions (regexes)
12955---------------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020012956
12957Just like with string matching, regex matching applies to verbatim strings as
12958they are passed, with the exception of the backslash ("\") which makes it
12959possible to escape some characters such as the space. If the "-i" flag is
12960passed before the first regex, then the matching will be performed ignoring
12961the case. In order to match the string "-i", either set it second, or pass
Willy Tarreau0ba27502007-12-24 16:55:16 +010012962the "--" flag before the first string. Same principle applies of course to
12963match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020012964
12965
Willy Tarreau74ca5042013-06-11 23:12:07 +0200129667.1.5. Matching arbitrary data blocks
12967-------------------------------------
12968
12969It is possible to match some extracted samples against a binary block which may
12970not safely be represented as a string. For this, the patterns must be passed as
12971a series of hexadecimal digits in an even number, when the match method is set
12972to binary. Each sequence of two digits will represent a byte. The hexadecimal
12973digits may be used upper or lower case.
12974
12975Example :
12976 # match "Hello\n" in the input stream (\x48 \x65 \x6c \x6c \x6f \x0a)
12977 acl hello payload(0,6) -m bin 48656c6c6f0a
12978
12979
129807.1.6. Matching IPv4 and IPv6 addresses
12981---------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020012982
12983IPv4 addresses values can be specified either as plain addresses or with a
12984netmask appended, in which case the IPv4 address matches whenever it is
12985within the network. Plain addresses may also be replaced with a resolvable
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010012986host name, but this practice is generally discouraged as it makes it more
Willy Tarreau0ba27502007-12-24 16:55:16 +010012987difficult to read and debug configurations. If hostnames are used, you should
12988at least ensure that they are present in /etc/hosts so that the configuration
12989does not depend on any random DNS match at the moment the configuration is
12990parsed.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012991
Daniel Schnellereba56342016-04-13 00:26:52 +020012992The dotted IPv4 address notation is supported in both regular as well as the
12993abbreviated form with all-0-octets omitted:
12994
12995 +------------------+------------------+------------------+
12996 | Example 1 | Example 2 | Example 3 |
12997 +------------------+------------------+------------------+
12998 | 192.168.0.1 | 10.0.0.12 | 127.0.0.1 |
12999 | 192.168.1 | 10.12 | 127.1 |
13000 | 192.168.0.1/22 | 10.0.0.12/8 | 127.0.0.1/8 |
13001 | 192.168.1/22 | 10.12/8 | 127.1/8 |
13002 +------------------+------------------+------------------+
13003
13004Notice that this is different from RFC 4632 CIDR address notation in which
13005192.168.42/24 would be equivalent to 192.168.42.0/24.
13006
Willy Tarreauceb4ac92012-04-28 00:41:46 +020013007IPv6 may be entered in their usual form, with or without a netmask appended.
13008Only bit counts are accepted for IPv6 netmasks. In order to avoid any risk of
13009trouble with randomly resolved IP addresses, host names are never allowed in
13010IPv6 patterns.
13011
13012HAProxy is also able to match IPv4 addresses with IPv6 addresses in the
13013following situations :
13014 - tested address is IPv4, pattern address is IPv4, the match applies
13015 in IPv4 using the supplied mask if any.
13016 - tested address is IPv6, pattern address is IPv6, the match applies
13017 in IPv6 using the supplied mask if any.
13018 - tested address is IPv6, pattern address is IPv4, the match applies in IPv4
13019 using the pattern's mask if the IPv6 address matches with 2002:IPV4::,
13020 ::IPV4 or ::ffff:IPV4, otherwise it fails.
13021 - tested address is IPv4, pattern address is IPv6, the IPv4 address is first
13022 converted to IPv6 by prefixing ::ffff: in front of it, then the match is
13023 applied in IPv6 using the supplied IPv6 mask.
13024
Willy Tarreau74ca5042013-06-11 23:12:07 +020013025
130267.2. Using ACLs to form conditions
13027----------------------------------
13028
13029Some actions are only performed upon a valid condition. A condition is a
13030combination of ACLs with operators. 3 operators are supported :
13031
13032 - AND (implicit)
13033 - OR (explicit with the "or" keyword or the "||" operator)
13034 - Negation with the exclamation mark ("!")
Willy Tarreau6a06a402007-07-15 20:15:28 +020013035
Willy Tarreau74ca5042013-06-11 23:12:07 +020013036A condition is formed as a disjunctive form:
Willy Tarreau6a06a402007-07-15 20:15:28 +020013037
Willy Tarreau74ca5042013-06-11 23:12:07 +020013038 [!]acl1 [!]acl2 ... [!]acln { or [!]acl1 [!]acl2 ... [!]acln } ...
Willy Tarreaubef91e72013-03-31 23:14:46 +020013039
Willy Tarreau74ca5042013-06-11 23:12:07 +020013040Such conditions are generally used after an "if" or "unless" statement,
13041indicating when the condition will trigger the action.
Willy Tarreaubef91e72013-03-31 23:14:46 +020013042
Willy Tarreau74ca5042013-06-11 23:12:07 +020013043For instance, to block HTTP requests to the "*" URL with methods other than
13044"OPTIONS", as well as POST requests without content-length, and GET or HEAD
13045requests with a content-length greater than 0, and finally every request which
13046is not either GET/HEAD/POST/OPTIONS !
13047
13048 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013049 http-request deny if HTTP_URL_STAR !METH_OPTIONS || METH_POST missing_cl
13050 http-request deny if METH_GET HTTP_CONTENT
13051 http-request deny unless METH_GET or METH_POST or METH_OPTIONS
Willy Tarreau74ca5042013-06-11 23:12:07 +020013052
13053To select a different backend for requests to static contents on the "www" site
13054and to every request on the "img", "video", "download" and "ftp" hosts :
13055
13056 acl url_static path_beg /static /images /img /css
13057 acl url_static path_end .gif .png .jpg .css .js
13058 acl host_www hdr_beg(host) -i www
13059 acl host_static hdr_beg(host) -i img. video. download. ftp.
13060
Davor Ocelice9ed2812017-12-25 17:49:28 +010013061 # now use backend "static" for all static-only hosts, and for static URLs
Willy Tarreau74ca5042013-06-11 23:12:07 +020013062 # of host "www". Use backend "www" for the rest.
13063 use_backend static if host_static or host_www url_static
13064 use_backend www if host_www
13065
13066It is also possible to form rules using "anonymous ACLs". Those are unnamed ACL
13067expressions that are built on the fly without needing to be declared. They must
13068be enclosed between braces, with a space before and after each brace (because
13069the braces must be seen as independent words). Example :
13070
13071 The following rule :
13072
13073 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013074 http-request deny if METH_POST missing_cl
Willy Tarreau74ca5042013-06-11 23:12:07 +020013075
13076 Can also be written that way :
13077
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013078 http-request deny if METH_POST { hdr_cnt(Content-length) eq 0 }
Willy Tarreau74ca5042013-06-11 23:12:07 +020013079
13080It is generally not recommended to use this construct because it's a lot easier
13081to leave errors in the configuration when written that way. However, for very
13082simple rules matching only one source IP address for instance, it can make more
13083sense to use them than to declare ACLs with random names. Another example of
13084good use is the following :
13085
13086 With named ACLs :
13087
13088 acl site_dead nbsrv(dynamic) lt 2
13089 acl site_dead nbsrv(static) lt 2
13090 monitor fail if site_dead
13091
13092 With anonymous ACLs :
13093
13094 monitor fail if { nbsrv(dynamic) lt 2 } || { nbsrv(static) lt 2 }
13095
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013096See section 4.2 for detailed help on the "http-request deny" and "use_backend"
13097keywords.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013098
13099
131007.3. Fetching samples
13101---------------------
13102
13103Historically, sample fetch methods were only used to retrieve data to match
13104against patterns using ACLs. With the arrival of stick-tables, a new class of
13105sample fetch methods was created, most often sharing the same syntax as their
13106ACL counterpart. These sample fetch methods are also known as "fetches". As
13107of now, ACLs and fetches have converged. All ACL fetch methods have been made
13108available as fetch methods, and ACLs may use any sample fetch method as well.
13109
13110This section details all available sample fetch methods and their output type.
13111Some sample fetch methods have deprecated aliases that are used to maintain
13112compatibility with existing configurations. They are then explicitly marked as
13113deprecated and should not be used in new setups.
13114
13115The ACL derivatives are also indicated when available, with their respective
13116matching methods. These ones all have a well defined default pattern matching
13117method, so it is never necessary (though allowed) to pass the "-m" option to
13118indicate how the sample will be matched using ACLs.
13119
13120As indicated in the sample type versus matching compatibility matrix above,
13121when using a generic sample fetch method in an ACL, the "-m" option is
13122mandatory unless the sample type is one of boolean, integer, IPv4 or IPv6. When
13123the same keyword exists as an ACL keyword and as a standard fetch method, the
13124ACL engine will automatically pick the ACL-only one by default.
13125
13126Some of these keywords support one or multiple mandatory arguments, and one or
13127multiple optional arguments. These arguments are strongly typed and are checked
13128when the configuration is parsed so that there is no risk of running with an
Davor Ocelice9ed2812017-12-25 17:49:28 +010013129incorrect argument (e.g. an unresolved backend name). Fetch function arguments
13130are passed between parenthesis and are delimited by commas. When an argument
Willy Tarreau74ca5042013-06-11 23:12:07 +020013131is optional, it will be indicated below between square brackets ('[ ]'). When
13132all arguments are optional, the parenthesis may be omitted.
13133
13134Thus, the syntax of a standard sample fetch method is one of the following :
13135 - name
13136 - name(arg1)
13137 - name(arg1,arg2)
13138
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013139
131407.3.1. Converters
13141-----------------
13142
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013143Sample fetch methods may be combined with transformations to be applied on top
13144of the fetched sample (also called "converters"). These combinations form what
13145is called "sample expressions" and the result is a "sample". Initially this
13146was only supported by "stick on" and "stick store-request" directives but this
Davor Ocelice9ed2812017-12-25 17:49:28 +010013147has now be extended to all places where samples may be used (ACLs, log-format,
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013148unique-id-format, add-header, ...).
13149
13150These transformations are enumerated as a series of specific keywords after the
13151sample fetch method. These keywords may equally be appended immediately after
13152the fetch keyword's argument, delimited by a comma. These keywords can also
Davor Ocelice9ed2812017-12-25 17:49:28 +010013153support some arguments (e.g. a netmask) which must be passed in parenthesis.
Willy Tarreau0ba27502007-12-24 16:55:16 +010013154
Willy Tarreau97707872015-01-27 15:12:13 +010013155A certain category of converters are bitwise and arithmetic operators which
13156support performing basic operations on integers. Some bitwise operations are
13157supported (and, or, xor, cpl) and some arithmetic operations are supported
13158(add, sub, mul, div, mod, neg). Some comparators are provided (odd, even, not,
13159bool) which make it possible to report a match without having to write an ACL.
13160
Willy Tarreau74ca5042013-06-11 23:12:07 +020013161The currently available list of transformation keywords include :
Willy Tarreau0ba27502007-12-24 16:55:16 +010013162
Ben Shillitof25e8e52016-12-02 14:25:37 +00001316351d.single(<prop>[,<prop>*])
13164 Returns values for the properties requested as a string, where values are
13165 separated by the delimiter specified with "51degrees-property-separator".
13166 The device is identified using the User-Agent header passed to the
13167 converter. The function can be passed up to five property names, and if a
13168 property name can't be found, the value "NoData" is returned.
13169
13170 Example :
Davor Ocelice9ed2812017-12-25 17:49:28 +010013171 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request,
13172 # containing values for the three properties requested by using the
Ben Shillitof25e8e52016-12-02 14:25:37 +000013173 # User-Agent passed to the converter.
13174 frontend http-in
13175 bind *:8081
13176 default_backend servers
13177 http-request set-header X-51D-DeviceTypeMobileTablet \
13178 %[req.fhdr(User-Agent),51d.single(DeviceType,IsMobile,IsTablet)]
13179
Willy Tarreau97707872015-01-27 15:12:13 +010013180add(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013181 Adds <value> to the input value of type signed integer, and returns the
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013182 result as a signed integer. <value> can be a numeric value or a variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013183 name. The name of the variable starts with an indication about its scope. The
13184 scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013185 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013186 "sess" : the variable is shared with the whole session
13187 "txn" : the variable is shared with the transaction (request and response)
13188 "req" : the variable is shared only during request processing
13189 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013190 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013191 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013192
13193and(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013194 Performs a bitwise "AND" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013195 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010013196 numeric value or a variable name. The name of the variable starts with an
13197 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013198 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013199 "sess" : the variable is shared with the whole session
13200 "txn" : the variable is shared with the transaction (request and response)
13201 "req" : the variable is shared only during request processing
13202 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013203 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013204 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013205
Holger Just1bfc24b2017-05-06 00:56:53 +020013206b64dec
13207 Converts (decodes) a base64 encoded input string to its binary
13208 representation. It performs the inverse operation of base64().
13209
Emeric Brun53d1a982014-04-30 18:21:37 +020013210base64
13211 Converts a binary input sample to a base64 string. It is used to log or
Davor Ocelice9ed2812017-12-25 17:49:28 +010013212 transfer binary content in a way that can be reliably transferred (e.g.
Emeric Brun53d1a982014-04-30 18:21:37 +020013213 an SSL ID can be copied in a header).
13214
Willy Tarreau97707872015-01-27 15:12:13 +010013215bool
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013216 Returns a boolean TRUE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010013217 non-null, otherwise returns FALSE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010013218 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010013219 presence of a flag).
13220
Emeric Brun54c4ac82014-11-03 15:32:43 +010013221bytes(<offset>[,<length>])
13222 Extracts some bytes from an input binary sample. The result is a binary
13223 sample starting at an offset (in bytes) of the original sample and
Tim Düsterhus4896c442016-11-29 02:15:19 +010013224 optionally truncated at the given length.
Emeric Brun54c4ac82014-11-03 15:32:43 +010013225
Willy Tarreau280f42b2018-02-19 15:34:12 +010013226concat([<start>],[<var>],[<end>])
13227 Concatenates up to 3 fields after the current sample which is then turned to
13228 a string. The first one, <start>, is a constant string, that will be appended
13229 immediately after the existing sample. It may be omitted if not used. The
13230 second one, <var>, is a variable name. The variable will be looked up, its
13231 contents converted to a string, and it will be appended immediately after the
13232 <first> part. If the variable is not found, nothing is appended. It may be
13233 omitted as well. The third field, <end> is a constant string that will be
13234 appended after the variable. It may also be omitted. Together, these elements
13235 allow to concatenate variables with delimiters to an existing set of
13236 variables. This can be used to build new variables made of a succession of
13237 other variables, such as colon-delimited varlues. Note that due to the config
13238 parser, it is not possible to use a comma nor a closing parenthesis as
13239 delimitors.
13240
13241 Example:
13242 tcp-request session set-var(sess.src) src
13243 tcp-request session set-var(sess.dn) ssl_c_s_dn
13244 tcp-request session set-var(txn.sig) str(),concat(<ip=,sess.ip,>),concat(<dn=,sess.dn,>)
13245 http-request set-header x-hap-sig %[var(txn.sig)]
13246
Willy Tarreau97707872015-01-27 15:12:13 +010013247cpl
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013248 Takes the input value of type signed integer, applies a ones-complement
13249 (flips all bits) and returns the result as an signed integer.
Willy Tarreau97707872015-01-27 15:12:13 +010013250
Willy Tarreau80599772015-01-20 19:35:24 +010013251crc32([<avalanche>])
13252 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32
13253 hash function. Optionally, it is possible to apply a full avalanche hash
13254 function to the output if the optional <avalanche> argument equals 1. This
13255 converter uses the same functions as used by the various hash-based load
13256 balancing algorithms, so it will provide exactly the same results. It is
13257 provided for compatibility with other software which want a CRC32 to be
13258 computed on some input keys, so it follows the most common implementation as
13259 found in Ethernet, Gzip, PNG, etc... It is slower than the other algorithms
13260 but may provide a better or at least less predictable distribution. It must
13261 not be used for security purposes as a 32-bit hash is trivial to break. See
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013262 also "djb2", "sdbm", "wt6", "crc32c" and the "hash-type" directive.
13263
13264crc32c([<avalanche>])
13265 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32C
13266 hash function. Optionally, it is possible to apply a full avalanche hash
13267 function to the output if the optional <avalanche> argument equals 1. This
13268 converter uses the same functions as described in RFC4960, Appendix B [8].
13269 It is provided for compatibility with other software which want a CRC32C to be
13270 computed on some input keys. It is slower than the other algorithms and it must
13271 not be used for security purposes as a 32-bit hash is trivial to break. See
13272 also "djb2", "sdbm", "wt6", "crc32" and the "hash-type" directive.
Willy Tarreau80599772015-01-20 19:35:24 +010013273
David Carlier29b3ca32015-09-25 14:09:21 +010013274da-csv-conv(<prop>[,<prop>*])
David Carlier4542b102015-06-01 13:54:29 +020013275 Asks the DeviceAtlas converter to identify the User Agent string passed on
13276 input, and to emit a string made of the concatenation of the properties
13277 enumerated in argument, delimited by the separator defined by the global
13278 keyword "deviceatlas-property-separator", or by default the pipe character
David Carlier840b0242016-03-16 10:09:55 +000013279 ('|'). There's a limit of 12 different properties imposed by the haproxy
David Carlier4542b102015-06-01 13:54:29 +020013280 configuration language.
13281
13282 Example:
13283 frontend www
Cyril Bonté307ee1e2015-09-28 23:16:06 +020013284 bind *:8881
13285 default_backend servers
David Carlier840b0242016-03-16 10:09:55 +000013286 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 +020013287
Thierry FOURNIER9687c772015-05-07 15:46:29 +020013288debug
13289 This converter is used as debug tool. It dumps on screen the content and the
13290 type of the input sample. The sample is returned as is on its output. This
13291 converter only exists when haproxy was built with debugging enabled.
13292
Willy Tarreau97707872015-01-27 15:12:13 +010013293div(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013294 Divides the input value of type signed integer by <value>, and returns the
13295 result as an signed integer. If <value> is null, the largest unsigned
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013296 integer is returned (typically 2^63-1). <value> can be a numeric value or a
Daniel Schneller0b547052016-03-21 20:46:57 +010013297 variable name. The name of the variable starts with an indication about its
13298 scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013299 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013300 "sess" : the variable is shared with the whole session
13301 "txn" : the variable is shared with the transaction (request and response)
13302 "req" : the variable is shared only during request processing
13303 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013304 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013305 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013306
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013307djb2([<avalanche>])
13308 Hashes a binary input sample into an unsigned 32-bit quantity using the DJB2
13309 hash function. Optionally, it is possible to apply a full avalanche hash
13310 function to the output if the optional <avalanche> argument equals 1. This
13311 converter uses the same functions as used by the various hash-based load
13312 balancing algorithms, so it will provide exactly the same results. It is
13313 mostly intended for debugging, but can be used as a stick-table entry to
13314 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013315 32-bit hash is trivial to break. See also "crc32", "sdbm", "wt6", "crc32c",
13316 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013317
Willy Tarreau97707872015-01-27 15:12:13 +010013318even
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013319 Returns a boolean TRUE if the input value of type signed integer is even
Willy Tarreau97707872015-01-27 15:12:13 +010013320 otherwise returns FALSE. It is functionally equivalent to "not,and(1),bool".
13321
Marcin Deranek9631a282018-04-16 14:30:46 +020013322field(<index>,<delimiters>[,<count>])
13323 Extracts the substring at the given index counting from the beginning
13324 (positive index) or from the end (negative index) considering given delimiters
13325 from an input string. Indexes start at 1 or -1 and delimiters are a string
13326 formatted list of chars. Optionally you can specify <count> of fields to
13327 extract (default: 1). Value of 0 indicates extraction of all remaining
13328 fields.
13329
13330 Example :
13331 str(f1_f2_f3__f5),field(5,_) # f5
13332 str(f1_f2_f3__f5),field(2,_,0) # f2_f3__f5
13333 str(f1_f2_f3__f5),field(2,_,2) # f2_f3
13334 str(f1_f2_f3__f5),field(-2,_,3) # f2_f3_
13335 str(f1_f2_f3__f5),field(-3,_,0) # f1_f2_f3
Emeric Brunf399b0d2014-11-03 17:07:03 +010013336
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013337hex
Davor Ocelice9ed2812017-12-25 17:49:28 +010013338 Converts a binary input sample to a hex string containing two hex digits per
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013339 input byte. It is used to log or transfer hex dumps of some binary input data
Davor Ocelice9ed2812017-12-25 17:49:28 +010013340 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 +020013341 header).
Thierry FOURNIER2f49d6d2014-03-12 15:01:52 +010013342
Dragan Dosen3f957b22017-10-24 09:27:34 +020013343hex2i
13344 Converts a hex string containing two hex digits per input byte to an
13345 integer. If the input value can not be converted, then zero is returned.
13346
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013347http_date([<offset>])
13348 Converts an integer supposed to contain a date since epoch to a string
13349 representing this date in a format suitable for use in HTTP header fields. If
13350 an offset value is specified, then it is a number of seconds that is added to
13351 the date before the conversion is operated. This is particularly useful to
13352 emit Date header fields, Expires values in responses when combined with a
13353 positive offset, or Last-Modified values when the offset is negative.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013354
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013355in_table(<table>)
13356 Uses the string representation of the input sample to perform a look up in
13357 the specified table. If the key is not found in the table, a boolean false
13358 is returned. Otherwise a boolean true is returned. This can be used to verify
Davor Ocelice9ed2812017-12-25 17:49:28 +010013359 the presence of a certain key in a table tracking some elements (e.g. whether
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013360 or not a source IP address or an Authorization header was already seen).
13361
Tim Duesterhus1478aa72018-01-25 16:24:51 +010013362ipmask(<mask4>, [<mask6>])
13363 Apply a mask to an IP address, and use the result for lookups and storage.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013364 This can be used to make all hosts within a certain mask to share the same
Tim Duesterhus1478aa72018-01-25 16:24:51 +010013365 table entries and as such use the same server. The mask4 can be passed in
13366 dotted form (e.g. 255.255.255.0) or in CIDR form (e.g. 24). The mask6 can
13367 be passed in quadruplet form (e.g. ffff:ffff::) or in CIDR form (e.g. 64).
13368 If no mask6 is given IPv6 addresses will fail to convert for backwards
13369 compatibility reasons.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013370
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013371json([<input-code>])
Davor Ocelice9ed2812017-12-25 17:49:28 +010013372 Escapes the input string and produces an ASCII output string ready to use as a
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013373 JSON string. The converter tries to decode the input string according to the
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020013374 <input-code> parameter. It can be "ascii", "utf8", "utf8s", "utf8p" or
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013375 "utf8ps". The "ascii" decoder never fails. The "utf8" decoder detects 3 types
13376 of errors:
13377 - bad UTF-8 sequence (lone continuation byte, bad number of continuation
13378 bytes, ...)
13379 - invalid range (the decoded value is within a UTF-8 prohibited range),
13380 - code overlong (the value is encoded with more bytes than necessary).
13381
13382 The UTF-8 JSON encoding can produce a "too long value" error when the UTF-8
13383 character is greater than 0xffff because the JSON string escape specification
13384 only authorizes 4 hex digits for the value encoding. The UTF-8 decoder exists
13385 in 4 variants designated by a combination of two suffix letters : "p" for
13386 "permissive" and "s" for "silently ignore". The behaviors of the decoders
13387 are :
Davor Ocelice9ed2812017-12-25 17:49:28 +010013388 - "ascii" : never fails;
13389 - "utf8" : fails on any detected errors;
13390 - "utf8s" : never fails, but removes characters corresponding to errors;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013391 - "utf8p" : accepts and fixes the overlong errors, but fails on any other
Davor Ocelice9ed2812017-12-25 17:49:28 +010013392 error;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013393 - "utf8ps" : never fails, accepts and fixes the overlong errors, but removes
13394 characters corresponding to the other errors.
13395
13396 This converter is particularly useful for building properly escaped JSON for
Davor Ocelice9ed2812017-12-25 17:49:28 +010013397 logging to servers which consume JSON-formatted traffic logs.
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013398
13399 Example:
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013400 capture request header Host len 15
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020013401 capture request header user-agent len 150
13402 log-format '{"ip":"%[src]","user-agent":"%[capture.req.hdr(1),json(utf8s)]"}'
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013403
13404 Input request from client 127.0.0.1:
13405 GET / HTTP/1.0
13406 User-Agent: Very "Ugly" UA 1/2
13407
13408 Output log:
13409 {"ip":"127.0.0.1","user-agent":"Very \"Ugly\" UA 1\/2"}
13410
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013411language(<value>[,<default>])
13412 Returns the value with the highest q-factor from a list as extracted from the
13413 "accept-language" header using "req.fhdr". Values with no q-factor have a
13414 q-factor of 1. Values with a q-factor of 0 are dropped. Only values which
13415 belong to the list of semi-colon delimited <values> will be considered. The
13416 argument <value> syntax is "lang[;lang[;lang[;...]]]". If no value matches the
13417 given list and a default value is provided, it is returned. Note that language
13418 names may have a variant after a dash ('-'). If this variant is present in the
13419 list, it will be matched, but if it is not, only the base language is checked.
13420 The match is case-sensitive, and the output string is always one of those
Davor Ocelice9ed2812017-12-25 17:49:28 +010013421 provided in arguments. The ordering of arguments is meaningless, only the
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013422 ordering of the values in the request counts, as the first value among
13423 multiple sharing the same q-factor is used.
Thierry FOURNIERad903512014-04-11 17:51:01 +020013424
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013425 Example :
Thierry FOURNIERad903512014-04-11 17:51:01 +020013426
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013427 # this configuration switches to the backend matching a
13428 # given language based on the request :
Thierry FOURNIERad903512014-04-11 17:51:01 +020013429
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013430 acl es req.fhdr(accept-language),language(es;fr;en) -m str es
13431 acl fr req.fhdr(accept-language),language(es;fr;en) -m str fr
13432 acl en req.fhdr(accept-language),language(es;fr;en) -m str en
13433 use_backend spanish if es
13434 use_backend french if fr
13435 use_backend english if en
13436 default_backend choose_your_language
Thierry FOURNIERad903512014-04-11 17:51:01 +020013437
Willy Tarreau60a2ee72017-12-15 07:13:48 +010013438length
Etienne Carriereed0d24e2017-12-13 13:41:34 +010013439 Get the length of the string. This can only be placed after a string
13440 sample fetch function or after a transformation keyword returning a string
13441 type. The result is of type integer.
13442
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013443lower
13444 Convert a string sample to lower case. This can only be placed after a string
13445 sample fetch function or after a transformation keyword returning a string
13446 type. The result is of type string.
13447
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013448ltime(<format>[,<offset>])
13449 Converts an integer supposed to contain a date since epoch to a string
13450 representing this date in local time using a format defined by the <format>
13451 string using strftime(3). The purpose is to allow any date format to be used
13452 in logs. An optional <offset> in seconds may be applied to the input date
13453 (positive or negative). See the strftime() man page for the format supported
13454 by your operating system. See also the utime converter.
13455
13456 Example :
13457
13458 # Emit two colons, one with the local time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010013459 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013460 log-format %[date,ltime(%Y%m%d%H%M%S)]\ %ci:%cp
13461
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013462map(<map_file>[,<default_value>])
13463map_<match_type>(<map_file>[,<default_value>])
13464map_<match_type>_<output_type>(<map_file>[,<default_value>])
13465 Search the input value from <map_file> using the <match_type> matching method,
13466 and return the associated value converted to the type <output_type>. If the
13467 input value cannot be found in the <map_file>, the converter returns the
13468 <default_value>. If the <default_value> is not set, the converter fails and
13469 acts as if no input value could be fetched. If the <match_type> is not set, it
13470 defaults to "str". Likewise, if the <output_type> is not set, it defaults to
13471 "str". For convenience, the "map" keyword is an alias for "map_str" and maps a
13472 string to another string.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013473
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013474 It is important to avoid overlapping between the keys : IP addresses and
13475 strings are stored in trees, so the first of the finest match will be used.
13476 Other keys are stored in lists, so the first matching occurrence will be used.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013477
Tim Düsterhus4896c442016-11-29 02:15:19 +010013478 The following array contains the list of all map functions available sorted by
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013479 input type, match type and output type.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013480
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013481 input type | match method | output type str | output type int | output type ip
13482 -----------+--------------+-----------------+-----------------+---------------
13483 str | str | map_str | map_str_int | map_str_ip
13484 -----------+--------------+-----------------+-----------------+---------------
Willy Tarreau787a4c02014-05-10 07:55:30 +020013485 str | beg | map_beg | map_beg_int | map_end_ip
13486 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013487 str | sub | map_sub | map_sub_int | map_sub_ip
13488 -----------+--------------+-----------------+-----------------+---------------
13489 str | dir | map_dir | map_dir_int | map_dir_ip
13490 -----------+--------------+-----------------+-----------------+---------------
13491 str | dom | map_dom | map_dom_int | map_dom_ip
13492 -----------+--------------+-----------------+-----------------+---------------
13493 str | end | map_end | map_end_int | map_end_ip
13494 -----------+--------------+-----------------+-----------------+---------------
Ruoshan Huang3c5e3742016-12-02 16:25:31 +080013495 str | reg | map_reg | map_reg_int | map_reg_ip
13496 -----------+--------------+-----------------+-----------------+---------------
13497 str | reg | map_regm | map_reg_int | map_reg_ip
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013498 -----------+--------------+-----------------+-----------------+---------------
13499 int | int | map_int | map_int_int | map_int_ip
13500 -----------+--------------+-----------------+-----------------+---------------
13501 ip | ip | map_ip | map_ip_int | map_ip_ip
13502 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013503
Thierry Fournier8feaa662016-02-10 22:55:20 +010013504 The special map called "map_regm" expect matching zone in the regular
13505 expression and modify the output replacing back reference (like "\1") by
13506 the corresponding match text.
13507
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013508 The file contains one key + value per line. Lines which start with '#' are
13509 ignored, just like empty lines. Leading tabs and spaces are stripped. The key
13510 is then the first "word" (series of non-space/tabs characters), and the value
13511 is what follows this series of space/tab till the end of the line excluding
13512 trailing spaces/tabs.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013513
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013514 Example :
13515
13516 # this is a comment and is ignored
13517 2.22.246.0/23 United Kingdom \n
13518 <-><-----------><--><------------><---->
13519 | | | | `- trailing spaces ignored
13520 | | | `---------- value
13521 | | `-------------------- middle spaces ignored
13522 | `---------------------------- key
13523 `------------------------------------ leading spaces ignored
13524
Willy Tarreau97707872015-01-27 15:12:13 +010013525mod(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013526 Divides the input value of type signed integer by <value>, and returns the
13527 remainder as an signed integer. If <value> is null, then zero is returned.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013528 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013529 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013530 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013531 "sess" : the variable is shared with the whole session
13532 "txn" : the variable is shared with the transaction (request and response)
13533 "req" : the variable is shared only during request processing
13534 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013535 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013536 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013537
13538mul(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013539 Multiplies the input value of type signed integer by <value>, and returns
Thierry FOURNIER00c005c2015-07-08 01:10:21 +020013540 the product as an signed integer. In case of overflow, the largest possible
13541 value for the sign is returned so that the operation doesn't wrap around.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013542 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013543 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013544 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013545 "sess" : the variable is shared with the whole session
13546 "txn" : the variable is shared with the transaction (request and response)
13547 "req" : the variable is shared only during request processing
13548 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013549 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013550 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013551
Nenad Merdanovicb7e7c472017-03-12 21:56:55 +010013552nbsrv
13553 Takes an input value of type string, interprets it as a backend name and
13554 returns the number of usable servers in that backend. Can be used in places
13555 where we want to look up a backend from a dynamic name, like a result of a
13556 map lookup.
13557
Willy Tarreau97707872015-01-27 15:12:13 +010013558neg
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013559 Takes the input value of type signed integer, computes the opposite value,
13560 and returns the remainder as an signed integer. 0 is identity. This operator
13561 is provided for reversed subtracts : in order to subtract the input from a
13562 constant, simply perform a "neg,add(value)".
Willy Tarreau97707872015-01-27 15:12:13 +010013563
13564not
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013565 Returns a boolean FALSE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010013566 non-null, otherwise returns TRUE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010013567 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010013568 absence of a flag).
13569
13570odd
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013571 Returns a boolean TRUE if the input value of type signed integer is odd
Willy Tarreau97707872015-01-27 15:12:13 +010013572 otherwise returns FALSE. It is functionally equivalent to "and(1),bool".
13573
13574or(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013575 Performs a bitwise "OR" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013576 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010013577 numeric value or a variable name. The name of the variable starts with an
13578 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013579 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013580 "sess" : the variable is shared with the whole session
13581 "txn" : the variable is shared with the transaction (request and response)
13582 "req" : the variable is shared only during request processing
13583 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013584 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013585 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013586
Willy Tarreauc4dc3502015-01-23 20:39:28 +010013587regsub(<regex>,<subst>[,<flags>])
Willy Tarreau7eda8492015-01-20 19:47:06 +010013588 Applies a regex-based substitution to the input string. It does the same
13589 operation as the well-known "sed" utility with "s/<regex>/<subst>/". By
13590 default it will replace in the input string the first occurrence of the
13591 largest part matching the regular expression <regex> with the substitution
13592 string <subst>. It is possible to replace all occurrences instead by adding
13593 the flag "g" in the third argument <flags>. It is also possible to make the
13594 regex case insensitive by adding the flag "i" in <flags>. Since <flags> is a
13595 string, it is made up from the concatenation of all desired flags. Thus if
13596 both "i" and "g" are desired, using "gi" or "ig" will have the same effect.
13597 It is important to note that due to the current limitations of the
Baptiste Assmann66025d82016-03-06 23:36:48 +010013598 configuration parser, some characters such as closing parenthesis, closing
13599 square brackets or comma are not possible to use in the arguments. The first
13600 use of this converter is to replace certain characters or sequence of
13601 characters with other ones.
Willy Tarreau7eda8492015-01-20 19:47:06 +010013602
13603 Example :
13604
13605 # de-duplicate "/" in header "x-path".
13606 # input: x-path: /////a///b/c/xzxyz/
13607 # output: x-path: /a/b/c/xzxyz/
13608 http-request set-header x-path %[hdr(x-path),regsub(/+,/,g)]
13609
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020013610capture-req(<id>)
13611 Capture the string entry in the request slot <id> and returns the entry as
13612 is. If the slot doesn't exist, the capture fails silently.
13613
13614 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020013615 "http-response capture", "capture.req.hdr" and
13616 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020013617
13618capture-res(<id>)
13619 Capture the string entry in the response slot <id> and returns the entry as
13620 is. If the slot doesn't exist, the capture fails silently.
13621
13622 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020013623 "http-response capture", "capture.req.hdr" and
13624 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020013625
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013626sdbm([<avalanche>])
13627 Hashes a binary input sample into an unsigned 32-bit quantity using the SDBM
13628 hash function. Optionally, it is possible to apply a full avalanche hash
13629 function to the output if the optional <avalanche> argument equals 1. This
13630 converter uses the same functions as used by the various hash-based load
13631 balancing algorithms, so it will provide exactly the same results. It is
13632 mostly intended for debugging, but can be used as a stick-table entry to
13633 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013634 32-bit hash is trivial to break. See also "crc32", "djb2", "wt6", "crc32c",
13635 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013636
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013637set-var(<var name>)
Davor Ocelice9ed2812017-12-25 17:49:28 +010013638 Sets a variable with the input content and returns the content on the output
13639 as-is. The variable keeps the value and the associated input type. The name of
13640 the variable starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013641 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013642 "sess" : the variable is shared with the whole session
13643 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013644 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010013645 "req" : the variable is shared only during request processing,
13646 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013647 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013648 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013649
Dragan Dosen6e5a9ca2017-10-24 09:18:23 +020013650sha1
13651 Converts a binary input sample to a SHA1 digest. The result is a binary
13652 sample with length of 20 bytes.
13653
Tim Duesterhusca097c12018-04-27 21:18:45 +020013654strcmp(<var>)
13655 Compares the contents of <var> with the input value of type string. Returns
13656 the result as a signed integer compatible with strcmp(3): 0 if both strings
13657 are identical. A value less than 0 if the left string is lexicographically
13658 smaller than the right string or if the left string is shorter. A value greater
13659 than 0 otherwise (right string greater than left string or the right string is
13660 shorter).
13661
13662 Example :
13663
13664 http-request set-var(txn.host) hdr(host)
13665 # Check whether the client is attempting domain fronting.
13666 acl ssl_sni_http_host_match ssl_fc_sni,strcmp(txn.host) eq 0
13667
13668
Willy Tarreau97707872015-01-27 15:12:13 +010013669sub(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013670 Subtracts <value> from the input value of type signed integer, and returns
13671 the result as an signed integer. Note: in order to subtract the input from
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013672 a constant, simply perform a "neg,add(value)". <value> can be a numeric value
Daniel Schneller0b547052016-03-21 20:46:57 +010013673 or a variable name. The name of the variable starts with an indication about
13674 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013675 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013676 "sess" : the variable is shared with the whole session
13677 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013678 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010013679 "req" : the variable is shared only during request processing,
13680 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013681 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013682 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013683
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013684table_bytes_in_rate(<table>)
13685 Uses the string representation of the input sample to perform a look up in
13686 the specified table. If the key is not found in the table, integer value zero
13687 is returned. Otherwise the converter returns the average client-to-server
13688 bytes rate associated with the input sample in the designated table, measured
13689 in amount of bytes over the period configured in the table. See also the
13690 sc_bytes_in_rate sample fetch keyword.
13691
13692
13693table_bytes_out_rate(<table>)
13694 Uses the string representation of the input sample to perform a look up in
13695 the specified table. If the key is not found in the table, integer value zero
13696 is returned. Otherwise the converter returns the average server-to-client
13697 bytes rate associated with the input sample in the designated table, measured
13698 in amount of bytes over the period configured in the table. See also the
13699 sc_bytes_out_rate sample fetch keyword.
13700
13701table_conn_cnt(<table>)
13702 Uses the string representation of the input sample to perform a look up in
13703 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013704 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013705 connections associated with the input sample in the designated table. See
13706 also the sc_conn_cnt sample fetch keyword.
13707
13708table_conn_cur(<table>)
13709 Uses the string representation of the input sample to perform a look up in
13710 the specified table. If the key is not found in the table, integer value zero
13711 is returned. Otherwise the converter returns the current amount of concurrent
13712 tracked connections associated with the input sample in the designated table.
13713 See also the sc_conn_cur sample fetch keyword.
13714
13715table_conn_rate(<table>)
13716 Uses the string representation of the input sample to perform a look up in
13717 the specified table. If the key is not found in the table, integer value zero
13718 is returned. Otherwise the converter returns the average incoming connection
13719 rate associated with the input sample in the designated table. See also the
13720 sc_conn_rate sample fetch keyword.
13721
Thierry FOURNIER236657b2015-08-19 08:25:14 +020013722table_gpt0(<table>)
13723 Uses the string representation of the input sample to perform a look up in
13724 the specified table. If the key is not found in the table, boolean value zero
13725 is returned. Otherwise the converter returns the current value of the first
13726 general purpose tag associated with the input sample in the designated table.
13727 See also the sc_get_gpt0 sample fetch keyword.
13728
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013729table_gpc0(<table>)
13730 Uses the string representation of the input sample to perform a look up in
13731 the specified table. If the key is not found in the table, integer value zero
13732 is returned. Otherwise the converter returns the current value of the first
13733 general purpose counter associated with the input sample in the designated
13734 table. See also the sc_get_gpc0 sample fetch keyword.
13735
13736table_gpc0_rate(<table>)
13737 Uses the string representation of the input sample to perform a look up in
13738 the specified table. If the key is not found in the table, integer value zero
13739 is returned. Otherwise the converter returns the frequency which the gpc0
13740 counter was incremented over the configured period in the table, associated
13741 with the input sample in the designated table. See also the sc_get_gpc0_rate
13742 sample fetch keyword.
13743
Frédéric Lécaille6778b272018-01-29 15:22:53 +010013744table_gpc1(<table>)
13745 Uses the string representation of the input sample to perform a look up in
13746 the specified table. If the key is not found in the table, integer value zero
13747 is returned. Otherwise the converter returns the current value of the second
13748 general purpose counter associated with the input sample in the designated
13749 table. See also the sc_get_gpc1 sample fetch keyword.
13750
13751table_gpc1_rate(<table>)
13752 Uses the string representation of the input sample to perform a look up in
13753 the specified table. If the key is not found in the table, integer value zero
13754 is returned. Otherwise the converter returns the frequency which the gpc1
13755 counter was incremented over the configured period in the table, associated
13756 with the input sample in the designated table. See also the sc_get_gpc1_rate
13757 sample fetch keyword.
13758
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013759table_http_err_cnt(<table>)
13760 Uses the string representation of the input sample to perform a look up in
13761 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013762 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013763 errors associated with the input sample in the designated table. See also the
13764 sc_http_err_cnt sample fetch keyword.
13765
13766table_http_err_rate(<table>)
13767 Uses the string representation of the input sample to perform a look up in
13768 the specified table. If the key is not found in the table, integer value zero
13769 is returned. Otherwise the average rate of HTTP errors associated with the
13770 input sample in the designated table, measured in amount of errors over the
13771 period configured in the table. See also the sc_http_err_rate sample fetch
13772 keyword.
13773
13774table_http_req_cnt(<table>)
13775 Uses the string representation of the input sample to perform a look up in
13776 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013777 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013778 requests associated with the input sample in the designated table. See also
13779 the sc_http_req_cnt sample fetch keyword.
13780
13781table_http_req_rate(<table>)
13782 Uses the string representation of the input sample to perform a look up in
13783 the specified table. If the key is not found in the table, integer value zero
13784 is returned. Otherwise the average rate of HTTP requests associated with the
13785 input sample in the designated table, measured in amount of requests over the
13786 period configured in the table. See also the sc_http_req_rate sample fetch
13787 keyword.
13788
13789table_kbytes_in(<table>)
13790 Uses the string representation of the input sample to perform a look up in
13791 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013792 is returned. Otherwise the converter returns the cumulative number of client-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013793 to-server data associated with the input sample in the designated table,
13794 measured in kilobytes. The test is currently performed on 32-bit integers,
13795 which limits values to 4 terabytes. See also the sc_kbytes_in sample fetch
13796 keyword.
13797
13798table_kbytes_out(<table>)
13799 Uses the string representation of the input sample to perform a look up in
13800 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013801 is returned. Otherwise the converter returns the cumulative number of server-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013802 to-client data associated with the input sample in the designated table,
13803 measured in kilobytes. The test is currently performed on 32-bit integers,
13804 which limits values to 4 terabytes. See also the sc_kbytes_out sample fetch
13805 keyword.
13806
13807table_server_id(<table>)
13808 Uses the string representation of the input sample to perform a look up in
13809 the specified table. If the key is not found in the table, integer value zero
13810 is returned. Otherwise the converter returns the server ID associated with
13811 the input sample in the designated table. A server ID is associated to a
13812 sample by a "stick" rule when a connection to a server succeeds. A server ID
13813 zero means that no server is associated with this key.
13814
13815table_sess_cnt(<table>)
13816 Uses the string representation of the input sample to perform a look up in
13817 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013818 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013819 sessions associated with the input sample in the designated table. Note that
13820 a session here refers to an incoming connection being accepted by the
13821 "tcp-request connection" rulesets. See also the sc_sess_cnt sample fetch
13822 keyword.
13823
13824table_sess_rate(<table>)
13825 Uses the string representation of the input sample to perform a look up in
13826 the specified table. If the key is not found in the table, integer value zero
13827 is returned. Otherwise the converter returns the average incoming session
13828 rate associated with the input sample in the designated table. Note that a
13829 session here refers to an incoming connection being accepted by the
13830 "tcp-request connection" rulesets. See also the sc_sess_rate sample fetch
13831 keyword.
13832
13833table_trackers(<table>)
13834 Uses the string representation of the input sample to perform a look up in
13835 the specified table. If the key is not found in the table, integer value zero
13836 is returned. Otherwise the converter returns the current amount of concurrent
13837 connections tracking the same key as the input sample in the designated
13838 table. It differs from table_conn_cur in that it does not rely on any stored
13839 information but on the table's reference count (the "use" value which is
13840 returned by "show table" on the CLI). This may sometimes be more suited for
13841 layer7 tracking. It can be used to tell a server how many concurrent
13842 connections there are from a given address for example. See also the
13843 sc_trackers sample fetch keyword.
13844
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013845upper
13846 Convert a string sample to upper case. This can only be placed after a string
13847 sample fetch function or after a transformation keyword returning a string
13848 type. The result is of type string.
13849
Thierry FOURNIER82ff3c92015-05-07 15:46:20 +020013850url_dec
13851 Takes an url-encoded string provided as input and returns the decoded
13852 version as output. The input and the output are of type string.
13853
Christopher Faulet85d79c92016-11-09 16:54:56 +010013854unset-var(<var name>)
13855 Unsets a variable if the input content is defined. The name of the variable
13856 starts with an indication about its scope. The scopes allowed are:
13857 "proc" : the variable is shared with the whole process
13858 "sess" : the variable is shared with the whole session
13859 "txn" : the variable is shared with the transaction (request and
13860 response),
13861 "req" : the variable is shared only during request processing,
13862 "res" : the variable is shared only during response processing.
13863 This prefix is followed by a name. The separator is a '.'. The name may only
13864 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
13865
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013866utime(<format>[,<offset>])
13867 Converts an integer supposed to contain a date since epoch to a string
13868 representing this date in UTC time using a format defined by the <format>
13869 string using strftime(3). The purpose is to allow any date format to be used
13870 in logs. An optional <offset> in seconds may be applied to the input date
13871 (positive or negative). See the strftime() man page for the format supported
13872 by your operating system. See also the ltime converter.
13873
13874 Example :
13875
13876 # Emit two colons, one with the UTC time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010013877 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013878 log-format %[date,utime(%Y%m%d%H%M%S)]\ %ci:%cp
13879
Marcin Deranek9631a282018-04-16 14:30:46 +020013880word(<index>,<delimiters>[,<count>])
13881 Extracts the nth word counting from the beginning (positive index) or from
13882 the end (negative index) considering given delimiters from an input string.
13883 Indexes start at 1 or -1 and delimiters are a string formatted list of chars.
13884 Optionally you can specify <count> of words to extract (default: 1).
13885 Value of 0 indicates extraction of all remaining words.
13886
13887 Example :
13888 str(f1_f2_f3__f5),word(4,_) # f5
13889 str(f1_f2_f3__f5),word(2,_,0) # f2_f3__f5
13890 str(f1_f2_f3__f5),word(3,_,2) # f3__f5
13891 str(f1_f2_f3__f5),word(-2,_,3) # f1_f2_f3
13892 str(f1_f2_f3__f5),word(-3,_,0) # f1_f2
Emeric Brunc9a0f6d2014-11-25 14:09:01 +010013893
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013894wt6([<avalanche>])
13895 Hashes a binary input sample into an unsigned 32-bit quantity using the WT6
13896 hash function. Optionally, it is possible to apply a full avalanche hash
13897 function to the output if the optional <avalanche> argument equals 1. This
13898 converter uses the same functions as used by the various hash-based load
13899 balancing algorithms, so it will provide exactly the same results. It is
13900 mostly intended for debugging, but can be used as a stick-table entry to
13901 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013902 32-bit hash is trivial to break. See also "crc32", "djb2", "sdbm", "crc32c",
13903 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013904
Willy Tarreau97707872015-01-27 15:12:13 +010013905xor(<value>)
13906 Performs a bitwise "XOR" (exclusive OR) between <value> and the input value
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013907 of type signed integer, and returns the result as an signed integer.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013908 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013909 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013910 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013911 "sess" : the variable is shared with the whole session
13912 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013913 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010013914 "req" : the variable is shared only during request processing,
13915 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013916 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013917 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013918
Thierry FOURNIER01e09742016-12-26 11:46:11 +010013919xxh32([<seed>])
13920 Hashes a binary input sample into an unsigned 32-bit quantity using the 32-bit
13921 variant of the XXHash hash function. This hash supports a seed which defaults
13922 to zero but a different value maybe passed as the <seed> argument. This hash
13923 is known to be very good and very fast so it can be used to hash URLs and/or
13924 URL parameters for use as stick-table keys to collect statistics with a low
13925 collision rate, though care must be taken as the algorithm is not considered
13926 as cryptographically secure.
13927
13928xxh64([<seed>])
13929 Hashes a binary input sample into a signed 64-bit quantity using the 64-bit
13930 variant of the XXHash hash function. This hash supports a seed which defaults
13931 to zero but a different value maybe passed as the <seed> argument. This hash
13932 is known to be very good and very fast so it can be used to hash URLs and/or
13933 URL parameters for use as stick-table keys to collect statistics with a low
13934 collision rate, though care must be taken as the algorithm is not considered
13935 as cryptographically secure.
13936
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013937
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200139387.3.2. Fetching samples from internal states
Willy Tarreau74ca5042013-06-11 23:12:07 +020013939--------------------------------------------
13940
13941A first set of sample fetch methods applies to internal information which does
13942not even relate to any client information. These ones are sometimes used with
13943"monitor-fail" directives to report an internal status to external watchers.
13944The sample fetch methods described in this section are usable anywhere.
13945
13946always_false : boolean
13947 Always returns the boolean "false" value. It may be used with ACLs as a
13948 temporary replacement for another one when adjusting configurations.
13949
13950always_true : boolean
13951 Always returns the boolean "true" value. It may be used with ACLs as a
13952 temporary replacement for another one when adjusting configurations.
13953
13954avg_queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010013955 Returns the total number of queued connections of the designated backend
Willy Tarreau74ca5042013-06-11 23:12:07 +020013956 divided by the number of active servers. The current backend is used if no
13957 backend is specified. This is very similar to "queue" except that the size of
13958 the farm is considered, in order to give a more accurate measurement of the
13959 time it may take for a new connection to be processed. The main usage is with
13960 ACL to return a sorry page to new users when it becomes certain they will get
13961 a degraded service, or to pass to the backend servers in a header so that
13962 they decide to work in degraded mode or to disable some functions to speed up
13963 the processing a bit. Note that in the event there would not be any active
13964 server anymore, twice the number of queued connections would be considered as
13965 the measured value. This is a fair estimate, as we expect one server to get
13966 back soon anyway, but we still prefer to send new traffic to another backend
13967 if in better shape. See also the "queue", "be_conn", and "be_sess_rate"
13968 sample fetches.
Krzysztof Piotr Oledzki346f76d2010-01-12 21:59:30 +010013969
Willy Tarreau74ca5042013-06-11 23:12:07 +020013970be_conn([<backend>]) : integer
Willy Tarreaua36af912009-10-10 12:02:45 +020013971 Applies to the number of currently established connections on the backend,
13972 possibly including the connection being evaluated. If no backend name is
13973 specified, the current one is used. But it is also possible to check another
13974 backend. It can be used to use a specific farm when the nominal one is full.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040013975 See also the "fe_conn", "queue", "be_conn_free", and "be_sess_rate" criteria.
13976
13977be_conn_free([<backend>]) : integer
13978 Returns an integer value corresponding to the number of available connections
13979 across available servers in the backend. Queue slots are not included. Backup
13980 servers are also not included, unless all other servers are down. If no
13981 backend name is specified, the current one is used. But it is also possible
13982 to check another backend. It can be used to use a specific farm when the
Patrick Hemmer155e93e2018-06-14 18:01:35 -040013983 nominal one is full. See also the "be_conn", "connslots", and "srv_conn_free"
13984 criteria.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040013985
13986 OTHER CAVEATS AND NOTES: if any of the server maxconn, or maxqueue is 0
13987 (meaning unlimited), then this fetch clearly does not make sense, in which
13988 case the value returned will be -1.
Willy Tarreau6a06a402007-07-15 20:15:28 +020013989
Willy Tarreau74ca5042013-06-11 23:12:07 +020013990be_sess_rate([<backend>]) : integer
13991 Returns an integer value corresponding to the sessions creation rate on the
13992 backend, in number of new sessions per second. This is used with ACLs to
13993 switch to an alternate backend when an expensive or fragile one reaches too
Davor Ocelice9ed2812017-12-25 17:49:28 +010013994 high a session rate, or to limit abuse of service (e.g. prevent sucking of an
Willy Tarreau74ca5042013-06-11 23:12:07 +020013995 online dictionary). It can also be useful to add this element to logs using a
13996 log-format directive.
Willy Tarreaud63335a2010-02-26 12:56:52 +010013997
13998 Example :
13999 # Redirect to an error page if the dictionary is requested too often
14000 backend dynamic
14001 mode http
14002 acl being_scanned be_sess_rate gt 100
14003 redirect location /denied.html if being_scanned
Willy Tarreau0ba27502007-12-24 16:55:16 +010014004
Davor Ocelice9ed2812017-12-25 17:49:28 +010014005bin(<hex>) : bin
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014006 Returns a binary chain. The input is the hexadecimal representation
14007 of the string.
14008
14009bool(<bool>) : bool
14010 Returns a boolean value. <bool> can be 'true', 'false', '1' or '0'.
14011 'false' and '0' are the same. 'true' and '1' are the same.
14012
Willy Tarreau74ca5042013-06-11 23:12:07 +020014013connslots([<backend>]) : integer
14014 Returns an integer value corresponding to the number of connection slots
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014015 still available in the backend, by totaling the maximum amount of
Willy Tarreau74ca5042013-06-11 23:12:07 +020014016 connections on all servers and the maximum queue size. This is probably only
14017 used with ACLs.
Tait Clarridge7896d522012-12-05 21:39:31 -050014018
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014019 The basic idea here is to be able to measure the number of connection "slots"
Willy Tarreau55165fe2009-05-10 12:02:55 +020014020 still available (connection + queue), so that anything beyond that (intended
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014021 usage; see "use_backend" keyword) can be redirected to a different backend.
14022
Willy Tarreau55165fe2009-05-10 12:02:55 +020014023 'connslots' = number of available server connection slots, + number of
14024 available server queue slots.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014025
Willy Tarreaua36af912009-10-10 12:02:45 +020014026 Note that while "fe_conn" may be used, "connslots" comes in especially
Willy Tarreau55165fe2009-05-10 12:02:55 +020014027 useful when you have a case of traffic going to one single ip, splitting into
Willy Tarreau74ca5042013-06-11 23:12:07 +020014028 multiple backends (perhaps using ACLs to do name-based load balancing) and
Willy Tarreau55165fe2009-05-10 12:02:55 +020014029 you want to be able to differentiate between different backends, and their
Davor Ocelice9ed2812017-12-25 17:49:28 +010014030 available "connslots". Also, whereas "nbsrv" only measures servers that are
Willy Tarreau74ca5042013-06-11 23:12:07 +020014031 actually *down*, this fetch is more fine-grained and looks into the number of
Willy Tarreaua36af912009-10-10 12:02:45 +020014032 available connection slots as well. See also "queue" and "avg_queue".
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014033
Willy Tarreau55165fe2009-05-10 12:02:55 +020014034 OTHER CAVEATS AND NOTES: at this point in time, the code does not take care
14035 of dynamic connections. Also, if any of the server maxconn, or maxqueue is 0,
Willy Tarreau74ca5042013-06-11 23:12:07 +020014036 then this fetch clearly does not make sense, in which case the value returned
Willy Tarreau55165fe2009-05-10 12:02:55 +020014037 will be -1.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014038
Willy Tarreau70fe9442018-11-22 16:07:39 +010014039cpu_calls : integer
14040 Returns the number of calls to the task processing the stream or current
14041 request since it was allocated. This number is reset for each new request on
14042 the same connections in case of HTTP keep-alive. This value should usually be
14043 low and stable (around 2 calls for a typically simple request) but may become
14044 high if some processing (compression, caching or analysis) is performed. This
14045 is purely for performance monitoring purposes.
14046
14047cpu_ns_avg : integer
14048 Returns the average number of nanoseconds spent in each call to the task
14049 processing the stream or current request. This number is reset for each new
14050 request on the same connections in case of HTTP keep-alive. This value
14051 indicates the overall cost of processing the request or the connection for
14052 each call. There is no good nor bad value but the time spent in a call
14053 automatically causes latency for other processing (see lat_ns_avg below),
14054 and may affect other connection's apparent response time. Certain operations
14055 like compression, complex regex matching or heavy Lua operations may directly
14056 affect this value, and having it in the logs will make it easier to spot the
14057 faulty processing that needs to be fixed to recover decent performance.
14058 Note: this value is exactly cpu_ns_tot divided by cpu_calls.
14059
14060cpu_ns_tot : integer
14061 Returns the total number of nanoseconds spent in each call to the task
14062 processing the stream or current request. This number is reset for each new
14063 request on the same connections in case of HTTP keep-alive. This value
14064 indicates the overall cost of processing the request or the connection for
14065 each call. There is no good nor bad value but the time spent in a call
14066 automatically causes latency for other processing (see lat_ns_avg below),
14067 induces CPU costs on the machine, and may affect other connection's apparent
14068 response time. Certain operations like compression, complex regex matching or
14069 heavy Lua operations may directly affect this value, and having it in the
14070 logs will make it easier to spot the faulty processing that needs to be fixed
14071 to recover decent performance. The value may be artificially high due to a
14072 high cpu_calls count, for example when processing many HTTP chunks, and for
14073 this reason it is often preferred to log cpu_ns_avg instead.
14074
Willy Tarreau6236d3a2013-07-25 14:28:25 +020014075date([<offset>]) : integer
14076 Returns the current date as the epoch (number of seconds since 01/01/1970).
14077 If an offset value is specified, then it is a number of seconds that is added
14078 to the current date before returning the value. This is particularly useful
14079 to compute relative dates, as both positive and negative offsets are allowed.
Willy Tarreau276fae92013-07-25 14:36:01 +020014080 It is useful combined with the http_date converter.
14081
14082 Example :
14083
14084 # set an expires header to now+1 hour in every response
14085 http-response set-header Expires %[date(3600),http_date]
Willy Tarreau6236d3a2013-07-25 14:28:25 +020014086
Etienne Carrierea792a0a2018-01-17 13:43:24 +010014087date_us : integer
14088 Return the microseconds part of the date (the "second" part is returned by
14089 date sample). This sample is coherent with the date sample as it is comes
14090 from the same timeval structure.
14091
Willy Tarreaud716f9b2017-10-13 11:03:15 +020014092distcc_body(<token>[,<occ>]) : binary
14093 Parses a distcc message and returns the body associated to occurrence #<occ>
14094 of the token <token>. Occurrences start at 1, and when unspecified, any may
14095 match though in practice only the first one is checked for now. This can be
14096 used to extract file names or arguments in files built using distcc through
14097 haproxy. Please refer to distcc's protocol documentation for the complete
14098 list of supported tokens.
14099
14100distcc_param(<token>[,<occ>]) : integer
14101 Parses a distcc message and returns the parameter associated to occurrence
14102 #<occ> of the token <token>. Occurrences start at 1, and when unspecified,
14103 any may match though in practice only the first one is checked for now. This
14104 can be used to extract certain information such as the protocol version, the
14105 file size or the argument in files built using distcc through haproxy.
14106 Another use case consists in waiting for the start of the preprocessed file
14107 contents before connecting to the server to avoid keeping idle connections.
14108 Please refer to distcc's protocol documentation for the complete list of
14109 supported tokens.
14110
14111 Example :
14112 # wait up to 20s for the pre-processed file to be uploaded
14113 tcp-request inspect-delay 20s
14114 tcp-request content accept if { distcc_param(DOTI) -m found }
14115 # send large files to the big farm
14116 use_backend big_farm if { distcc_param(DOTI) gt 1000000 }
14117
Willy Tarreau595ec542013-06-12 21:34:28 +020014118env(<name>) : string
14119 Returns a string containing the value of environment variable <name>. As a
14120 reminder, environment variables are per-process and are sampled when the
14121 process starts. This can be useful to pass some information to a next hop
14122 server, or with ACLs to take specific action when the process is started a
14123 certain way.
14124
14125 Examples :
14126 # Pass the Via header to next hop with the local hostname in it
14127 http-request add-header Via 1.1\ %[env(HOSTNAME)]
14128
14129 # reject cookie-less requests when the STOP environment variable is set
14130 http-request deny if !{ cook(SESSIONID) -m found } { env(STOP) -m found }
14131
Willy Tarreau74ca5042013-06-11 23:12:07 +020014132fe_conn([<frontend>]) : integer
14133 Returns the number of currently established connections on the frontend,
Willy Tarreaud63335a2010-02-26 12:56:52 +010014134 possibly including the connection being evaluated. If no frontend name is
14135 specified, the current one is used. But it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020014136 frontend. It can be used to return a sorry page before hard-blocking, or to
14137 use a specific backend to drain new requests when the farm is considered
Davor Ocelice9ed2812017-12-25 17:49:28 +010014138 full. This is mostly used with ACLs but can also be used to pass some
Willy Tarreau74ca5042013-06-11 23:12:07 +020014139 statistics to servers in HTTP headers. See also the "dst_conn", "be_conn",
14140 "fe_sess_rate" fetches.
Willy Tarreaua36af912009-10-10 12:02:45 +020014141
Nenad Merdanovicad9a7e92016-10-03 04:57:37 +020014142fe_req_rate([<frontend>]) : integer
14143 Returns an integer value corresponding to the number of HTTP requests per
14144 second sent to a frontend. This number can differ from "fe_sess_rate" in
14145 situations where client-side keep-alive is enabled.
14146
Willy Tarreau74ca5042013-06-11 23:12:07 +020014147fe_sess_rate([<frontend>]) : integer
14148 Returns an integer value corresponding to the sessions creation rate on the
14149 frontend, in number of new sessions per second. This is used with ACLs to
14150 limit the incoming session rate to an acceptable range in order to prevent
14151 abuse of service at the earliest moment, for example when combined with other
14152 layer 4 ACLs in order to force the clients to wait a bit for the rate to go
14153 down below the limit. It can also be useful to add this element to logs using
14154 a log-format directive. See also the "rate-limit sessions" directive for use
14155 in frontends.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010014156
14157 Example :
14158 # This frontend limits incoming mails to 10/s with a max of 100
14159 # concurrent connections. We accept any connection below 10/s, and
14160 # force excess clients to wait for 100 ms. Since clients are limited to
14161 # 100 max, there cannot be more than 10 incoming mails per second.
14162 frontend mail
14163 bind :25
14164 mode tcp
14165 maxconn 100
14166 acl too_fast fe_sess_rate ge 10
14167 tcp-request inspect-delay 100ms
14168 tcp-request content accept if ! too_fast
14169 tcp-request content accept if WAIT_END
Willy Tarreaud72758d2010-01-12 10:42:19 +010014170
Nenad Merdanovic807a6e72017-03-12 22:00:00 +010014171hostname : string
14172 Returns the system hostname.
14173
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014174int(<integer>) : signed integer
14175 Returns a signed integer.
14176
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014177ipv4(<ipv4>) : ipv4
14178 Returns an ipv4.
14179
14180ipv6(<ipv6>) : ipv6
14181 Returns an ipv6.
14182
Willy Tarreau70fe9442018-11-22 16:07:39 +010014183lat_ns_avg : integer
14184 Returns the average number of nanoseconds spent between the moment the task
14185 handling the stream is woken up and the moment it is effectively called. This
14186 number is reset for each new request on the same connections in case of HTTP
14187 keep-alive. This value indicates the overall latency inflicted to the current
14188 request by all other requests being processed in parallel, and is a direct
14189 indicator of perceived performance due to noisy neighbours. In order to keep
14190 the value low, it is possible to reduce the scheduler's run queue depth using
14191 "tune.runqueue-depth", to reduce the number of concurrent events processed at
14192 once using "tune.maxpollevents", to decrease the stream's nice value using
14193 the "nice" option on the "bind" lines or in the frontend, or to look for
14194 other heavy requests in logs (those exhibiting large values of "cpu_ns_avg"),
14195 whose processing needs to be adjusted or fixed. Compression of large buffers
14196 could be a culprit, like heavy regex or long lists of regex.
14197 Note: this value is exactly lat_ns_tot divided by cpu_calls.
14198
14199lat_ns_tot : integer
14200 Returns the total number of nanoseconds spent between the moment the task
14201 handling the stream is woken up and the moment it is effectively called. This
14202 number is reset for each new request on the same connections in case of HTTP
14203 keep-alive. This value indicates the overall latency inflicted to the current
14204 request by all other requests being processed in parallel, and is a direct
14205 indicator of perceived performance due to noisy neighbours. In order to keep
14206 the value low, it is possible to reduce the scheduler's run queue depth using
14207 "tune.runqueue-depth", to reduce the number of concurrent events processed at
14208 once using "tune.maxpollevents", to decrease the stream's nice value using
14209 the "nice" option on the "bind" lines or in the frontend, or to look for
14210 other heavy requests in logs (those exhibiting large values of "cpu_ns_avg"),
14211 whose processing needs to be adjusted or fixed. Compression of large buffers
14212 could be a culprit, like heavy regex or long lists of regex. Note: while it
14213 may intuitively seem that the total latency adds to a transfer time, it is
14214 almost never true because while a task waits for the CPU, network buffers
14215 continue to fill up and the next call will process more at once. The value
14216 may be artificially high due to a high cpu_calls count, for example when
14217 processing many HTTP chunks, and for this reason it is often preferred to log
14218 lat_ns_avg instead, which is a more relevant performance indicator.
14219
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014220meth(<method>) : method
14221 Returns a method.
14222
Willy Tarreau0f30d262014-11-24 16:02:05 +010014223nbproc : integer
14224 Returns an integer value corresponding to the number of processes that were
14225 started (it equals the global "nbproc" setting). This is useful for logging
14226 and debugging purposes.
14227
Willy Tarreau74ca5042013-06-11 23:12:07 +020014228nbsrv([<backend>]) : integer
14229 Returns an integer value corresponding to the number of usable servers of
14230 either the current backend or the named backend. This is mostly used with
14231 ACLs but can also be useful when added to logs. This is normally used to
Willy Tarreaud63335a2010-02-26 12:56:52 +010014232 switch to an alternate backend when the number of servers is too low to
14233 to handle some load. It is useful to report a failure when combined with
14234 "monitor fail".
Willy Tarreau079ff0a2009-03-05 21:34:28 +010014235
Patrick Hemmerfabb24f2018-08-13 14:07:57 -040014236prio_class : integer
14237 Returns the priority class of the current session for http mode or connection
14238 for tcp mode. The value will be that set by the last call to "http-request
14239 set-priority-class" or "tcp-request content set-priority-class".
14240
14241prio_offset : integer
14242 Returns the priority offset of the current session for http mode or
14243 connection for tcp mode. The value will be that set by the last call to
14244 "http-request set-priority-offset" or "tcp-request content
14245 set-priority-offset".
14246
Willy Tarreau0f30d262014-11-24 16:02:05 +010014247proc : integer
14248 Returns an integer value corresponding to the position of the process calling
14249 the function, between 1 and global.nbproc. This is useful for logging and
14250 debugging purposes.
14251
Willy Tarreau74ca5042013-06-11 23:12:07 +020014252queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010014253 Returns the total number of queued connections of the designated backend,
14254 including all the connections in server queues. If no backend name is
14255 specified, the current one is used, but it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020014256 one. This is useful with ACLs or to pass statistics to backend servers. This
14257 can be used to take actions when queuing goes above a known level, generally
14258 indicating a surge of traffic or a massive slowdown on the servers. One
14259 possible action could be to reject new users but still accept old ones. See
14260 also the "avg_queue", "be_conn", and "be_sess_rate" fetches.
14261
Willy Tarreau84310e22014-02-14 11:59:04 +010014262rand([<range>]) : integer
14263 Returns a random integer value within a range of <range> possible values,
14264 starting at zero. If the range is not specified, it defaults to 2^32, which
14265 gives numbers between 0 and 4294967295. It can be useful to pass some values
14266 needed to take some routing decisions for example, or just for debugging
14267 purposes. This random must not be used for security purposes.
14268
Willy Tarreau74ca5042013-06-11 23:12:07 +020014269srv_conn([<backend>/]<server>) : integer
14270 Returns an integer value corresponding to the number of currently established
14271 connections on the designated server, possibly including the connection being
14272 evaluated. If <backend> is omitted, then the server is looked up in the
14273 current backend. It can be used to use a specific farm when one server is
14274 full, or to inform the server about our view of the number of active
Patrick Hemmer155e93e2018-06-14 18:01:35 -040014275 connections with it. See also the "fe_conn", "be_conn", "queue", and
14276 "srv_conn_free" fetch methods.
14277
14278srv_conn_free([<backend>/]<server>) : integer
14279 Returns an integer value corresponding to the number of available connections
14280 on the designated server, possibly including the connection being evaluated.
14281 The value does not include queue slots. If <backend> is omitted, then the
14282 server is looked up in the current backend. It can be used to use a specific
14283 farm when one server is full, or to inform the server about our view of the
14284 number of active connections with it. See also the "be_conn_free" and
14285 "srv_conn" fetch methods.
14286
14287 OTHER CAVEATS AND NOTES: If the server maxconn is 0, then this fetch clearly
14288 does not make sense, in which case the value returned will be -1.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014289
14290srv_is_up([<backend>/]<server>) : boolean
14291 Returns true when the designated server is UP, and false when it is either
14292 DOWN or in maintenance mode. If <backend> is omitted, then the server is
14293 looked up in the current backend. It is mainly used to take action based on
Davor Ocelice9ed2812017-12-25 17:49:28 +010014294 an external status reported via a health check (e.g. a geographical site's
Willy Tarreau74ca5042013-06-11 23:12:07 +020014295 availability). Another possible use which is more of a hack consists in
14296 using dummy servers as boolean variables that can be enabled or disabled from
14297 the CLI, so that rules depending on those ACLs can be tweaked in realtime.
14298
Willy Tarreauff2b7af2017-10-13 11:46:26 +020014299srv_queue([<backend>/]<server>) : integer
14300 Returns an integer value corresponding to the number of connections currently
14301 pending in the designated server's queue. If <backend> is omitted, then the
14302 server is looked up in the current backend. It can sometimes be used together
14303 with the "use-server" directive to force to use a known faster server when it
14304 is not much loaded. See also the "srv_conn", "avg_queue" and "queue" sample
14305 fetch methods.
14306
Willy Tarreau74ca5042013-06-11 23:12:07 +020014307srv_sess_rate([<backend>/]<server>) : integer
14308 Returns an integer corresponding to the sessions creation rate on the
14309 designated server, in number of new sessions per second. If <backend> is
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014310 omitted, then the server is looked up in the current backend. This is mostly
Willy Tarreau74ca5042013-06-11 23:12:07 +020014311 used with ACLs but can make sense with logs too. This is used to switch to an
14312 alternate backend when an expensive or fragile one reaches too high a session
Davor Ocelice9ed2812017-12-25 17:49:28 +010014313 rate, or to limit abuse of service (e.g. prevent latent requests from
Willy Tarreau74ca5042013-06-11 23:12:07 +020014314 overloading servers).
14315
14316 Example :
14317 # Redirect to a separate back
14318 acl srv1_full srv_sess_rate(be1/srv1) gt 50
14319 acl srv2_full srv_sess_rate(be1/srv2) gt 50
14320 use_backend be2 if srv1_full or srv2_full
14321
Willy Tarreau0f30d262014-11-24 16:02:05 +010014322stopping : boolean
14323 Returns TRUE if the process calling the function is currently stopping. This
14324 can be useful for logging, or for relaxing certain checks or helping close
14325 certain connections upon graceful shutdown.
14326
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014327str(<string>) : string
14328 Returns a string.
14329
Willy Tarreau74ca5042013-06-11 23:12:07 +020014330table_avl([<table>]) : integer
14331 Returns the total number of available entries in the current proxy's
14332 stick-table or in the designated stick-table. See also table_cnt.
14333
14334table_cnt([<table>]) : integer
14335 Returns the total number of entries currently in use in the current proxy's
14336 stick-table or in the designated stick-table. See also src_conn_cnt and
14337 table_avl for other entry counting methods.
14338
Christopher Faulet34adb2a2017-11-21 21:45:38 +010014339thread : integer
14340 Returns an integer value corresponding to the position of the thread calling
14341 the function, between 0 and (global.nbthread-1). This is useful for logging
14342 and debugging purposes.
14343
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014344var(<var-name>) : undefined
14345 Returns a variable with the stored type. If the variable is not set, the
Daniel Schneller0b547052016-03-21 20:46:57 +010014346 sample fetch fails. The name of the variable starts with an indication
14347 about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010014348 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010014349 "sess" : the variable is shared with the whole session
14350 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014351 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010014352 "req" : the variable is shared only during request processing,
14353 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014354 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014355 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014356
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200143577.3.3. Fetching samples at Layer 4
Willy Tarreau74ca5042013-06-11 23:12:07 +020014358----------------------------------
14359
14360The layer 4 usually describes just the transport layer which in haproxy is
14361closest to the connection, where no content is yet made available. The fetch
14362methods described here are usable as low as the "tcp-request connection" rule
14363sets unless they require some future information. Those generally include
14364TCP/IP addresses and ports, as well as elements from stick-tables related to
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014365the incoming connection. For retrieving a value from a sticky counters, the
14366counter number can be explicitly set as 0, 1, or 2 using the pre-defined
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020014367"sc0_", "sc1_", or "sc2_" prefix. These three pre-defined prefixes can only be
14368used if MAX_SESS_STKCTR value does not exceed 3, otherwise the counter number
14369can be specified as the first integer argument when using the "sc_" prefix.
14370Starting from "sc_0" to "sc_N" where N is (MAX_SESS_STKCTR-1). An optional
14371table may be specified with the "sc*" form, in which case the currently
14372tracked key will be looked up into this alternate table instead of the table
14373currently being tracked.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014374
Jérôme Magnin86577422018-12-07 09:03:11 +010014375bc_http_major: integer
14376 Returns the backend connection's HTTP major version encoding, which may be 1
14377 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
14378 encoding and not the version present in the request header.
14379
Willy Tarreau74ca5042013-06-11 23:12:07 +020014380be_id : integer
14381 Returns an integer containing the current backend's id. It can be used in
14382 frontends with responses to check which backend processed the request.
14383
Marcin Deranekd2471c22016-12-12 14:08:05 +010014384be_name : string
14385 Returns a string containing the current backend's name. It can be used in
14386 frontends with responses to check which backend processed the request.
14387
Willy Tarreau74ca5042013-06-11 23:12:07 +020014388dst : ip
14389 This is the destination IPv4 address of the connection on the client side,
14390 which is the address the client connected to. It can be useful when running
14391 in transparent mode. It is of type IP and works on both IPv4 and IPv6 tables.
14392 On IPv6 tables, IPv4 address is mapped to its IPv6 equivalent, according to
14393 RFC 4291.
14394
14395dst_conn : integer
14396 Returns an integer value corresponding to the number of currently established
14397 connections on the same socket including the one being evaluated. It is
14398 normally used with ACLs but can as well be used to pass the information to
14399 servers in an HTTP header or in logs. It can be used to either return a sorry
14400 page before hard-blocking, or to use a specific backend to drain new requests
14401 when the socket is considered saturated. This offers the ability to assign
14402 different limits to different listening ports or addresses. See also the
14403 "fe_conn" and "be_conn" fetches.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014404
Willy Tarreau16e01562016-08-09 16:46:18 +020014405dst_is_local : boolean
14406 Returns true if the destination address of the incoming connection is local
14407 to the system, or false if the address doesn't exist on the system, meaning
14408 that it was intercepted in transparent mode. It can be useful to apply
14409 certain rules by default to forwarded traffic and other rules to the traffic
Davor Ocelice9ed2812017-12-25 17:49:28 +010014410 targeting the real address of the machine. For example the stats page could
Willy Tarreau16e01562016-08-09 16:46:18 +020014411 be delivered only on this address, or SSH access could be locally redirected.
14412 Please note that the check involves a few system calls, so it's better to do
14413 it only once per connection.
14414
Willy Tarreau74ca5042013-06-11 23:12:07 +020014415dst_port : integer
14416 Returns an integer value corresponding to the destination TCP port of the
14417 connection on the client side, which is the port the client connected to.
14418 This might be used when running in transparent mode, when assigning dynamic
14419 ports to some clients for a whole application session, to stick all users to
14420 a same server, or to pass the destination port information to a server using
14421 an HTTP header.
14422
Willy Tarreau60ca10a2017-08-18 15:26:54 +020014423fc_http_major : integer
14424 Reports the front connection's HTTP major version encoding, which may be 1
14425 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
14426 encoding and not on the version present in the request header.
14427
Emeric Brun4f603012017-01-05 15:11:44 +010014428fc_rcvd_proxy : boolean
14429 Returns true if the client initiated the connection with a PROXY protocol
14430 header.
14431
Thierry Fournier / OZON.IO6310bef2016-07-24 20:16:50 +020014432fc_rtt(<unit>) : integer
14433 Returns the Round Trip Time (RTT) measured by the kernel for the client
14434 connection. <unit> is facultative, by default the unit is milliseconds. <unit>
14435 can be set to "ms" for milliseconds or "us" for microseconds. If the server
14436 connection is not established, if the connection is not TCP or if the
14437 operating system does not support TCP_INFO, for example Linux kernels before
14438 2.4, the sample fetch fails.
14439
14440fc_rttvar(<unit>) : integer
14441 Returns the Round Trip Time (RTT) variance measured by the kernel for the
14442 client connection. <unit> is facultative, by default the unit is milliseconds.
14443 <unit> can be set to "ms" for milliseconds or "us" for microseconds. If the
14444 server connection is not established, if the connection is not TCP or if the
14445 operating system does not support TCP_INFO, for example Linux kernels before
14446 2.4, the sample fetch fails.
14447
Joe Williams30fcd392016-08-10 07:06:44 -070014448fc_unacked(<unit>) : integer
14449 Returns the unacked counter measured by the kernel for the client connection.
14450 If the server connection is not established, if the connection is not TCP or
14451 if the operating system does not support TCP_INFO, for example Linux kernels
14452 before 2.4, the sample fetch fails.
14453
14454fc_sacked(<unit>) : integer
14455 Returns the sacked counter measured by the kernel for the client connection.
14456 If the server connection is not established, if the connection is not TCP or
14457 if the operating system does not support TCP_INFO, for example Linux kernels
14458 before 2.4, the sample fetch fails.
14459
14460fc_retrans(<unit>) : integer
14461 Returns the retransmits counter measured by the kernel for the client
14462 connection. If the server connection is not established, if the connection is
14463 not TCP or if the operating system does not support TCP_INFO, for example
14464 Linux kernels before 2.4, the sample fetch fails.
14465
14466fc_fackets(<unit>) : integer
14467 Returns the fack counter measured by the kernel for the client
14468 connection. If the server connection is not established, if the connection is
14469 not TCP or if the operating system does not support TCP_INFO, for example
14470 Linux kernels before 2.4, the sample fetch fails.
14471
14472fc_lost(<unit>) : integer
14473 Returns the lost counter measured by the kernel for the client
14474 connection. If the server connection is not established, if the connection is
14475 not TCP or if the operating system does not support TCP_INFO, for example
14476 Linux kernels before 2.4, the sample fetch fails.
14477
14478fc_reordering(<unit>) : integer
14479 Returns the reordering counter measured by the kernel for the client
14480 connection. If the server connection is not established, if the connection is
14481 not TCP or if the operating system does not support TCP_INFO, for example
14482 Linux kernels before 2.4, the sample fetch fails.
14483
Marcin Deranek9a66dfb2018-04-13 14:37:50 +020014484fe_defbe : string
14485 Returns a string containing the frontend's default backend name. It can be
14486 used in frontends to check which backend will handle requests by default.
14487
Willy Tarreau74ca5042013-06-11 23:12:07 +020014488fe_id : integer
14489 Returns an integer containing the current frontend's id. It can be used in
Marcin Deranek6e413ed2016-12-13 12:40:01 +010014490 backends to check from which frontend it was called, or to stick all users
Willy Tarreau74ca5042013-06-11 23:12:07 +020014491 coming via a same frontend to the same server.
14492
Marcin Deranekd2471c22016-12-12 14:08:05 +010014493fe_name : string
14494 Returns a string containing the current frontend's name. It can be used in
14495 backends to check from which frontend it was called, or to stick all users
14496 coming via a same frontend to the same server.
14497
Cyril Bonté62ba8702014-04-22 23:52:25 +020014498sc_bytes_in_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014499sc0_bytes_in_rate([<table>]) : integer
14500sc1_bytes_in_rate([<table>]) : integer
14501sc2_bytes_in_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014502 Returns the average client-to-server bytes rate from the currently tracked
14503 counters, measured in amount of bytes over the period configured in the
14504 table. See also src_bytes_in_rate.
14505
Cyril Bonté62ba8702014-04-22 23:52:25 +020014506sc_bytes_out_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014507sc0_bytes_out_rate([<table>]) : integer
14508sc1_bytes_out_rate([<table>]) : integer
14509sc2_bytes_out_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014510 Returns the average server-to-client bytes rate from the currently tracked
14511 counters, measured in amount of bytes over the period configured in the
14512 table. See also src_bytes_out_rate.
14513
Cyril Bonté62ba8702014-04-22 23:52:25 +020014514sc_clr_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014515sc0_clr_gpc0([<table>]) : integer
14516sc1_clr_gpc0([<table>]) : integer
14517sc2_clr_gpc0([<table>]) : integer
Willy Tarreauf73cd112011-08-13 01:45:16 +020014518 Clears the first General Purpose Counter associated to the currently tracked
14519 counters, and returns its previous value. Before the first invocation, the
Willy Tarreau869948b2013-01-04 14:14:57 +010014520 stored value is zero, so first invocation will always return zero. This is
14521 typically used as a second ACL in an expression in order to mark a connection
14522 when a first ACL was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020014523
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014524 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020014525 # block if 5 consecutive requests continue to come faster than 10 sess
14526 # per second, and reset the counter as soon as the traffic slows down.
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014527 acl abuse sc0_http_req_rate gt 10
14528 acl kill sc0_inc_gpc0 gt 5
14529 acl save sc0_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020014530 tcp-request connection accept if !abuse save
14531 tcp-request connection reject if abuse kill
14532
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014533sc_clr_gpc1(<ctr>[,<table>]) : integer
14534sc0_clr_gpc1([<table>]) : integer
14535sc1_clr_gpc1([<table>]) : integer
14536sc2_clr_gpc1([<table>]) : integer
14537 Clears the second General Purpose Counter associated to the currently tracked
14538 counters, and returns its previous value. Before the first invocation, the
14539 stored value is zero, so first invocation will always return zero. This is
14540 typically used as a second ACL in an expression in order to mark a connection
14541 when a first ACL was verified.
14542
Cyril Bonté62ba8702014-04-22 23:52:25 +020014543sc_conn_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014544sc0_conn_cnt([<table>]) : integer
14545sc1_conn_cnt([<table>]) : integer
14546sc2_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014547 Returns the cumulative number of incoming connections from currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020014548 counters. See also src_conn_cnt.
14549
Cyril Bonté62ba8702014-04-22 23:52:25 +020014550sc_conn_cur(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014551sc0_conn_cur([<table>]) : integer
14552sc1_conn_cur([<table>]) : integer
14553sc2_conn_cur([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014554 Returns the current amount of concurrent connections tracking the same
14555 tracked counters. This number is automatically incremented when tracking
14556 begins and decremented when tracking stops. See also src_conn_cur.
14557
Cyril Bonté62ba8702014-04-22 23:52:25 +020014558sc_conn_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014559sc0_conn_rate([<table>]) : integer
14560sc1_conn_rate([<table>]) : integer
14561sc2_conn_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014562 Returns the average connection rate from the currently tracked counters,
14563 measured in amount of connections over the period configured in the table.
14564 See also src_conn_rate.
14565
Cyril Bonté62ba8702014-04-22 23:52:25 +020014566sc_get_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014567sc0_get_gpc0([<table>]) : integer
14568sc1_get_gpc0([<table>]) : integer
14569sc2_get_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014570 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014571 currently tracked counters. See also src_get_gpc0 and sc/sc0/sc1/sc2_inc_gpc0.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014572
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014573sc_get_gpc1(<ctr>[,<table>]) : integer
14574sc0_get_gpc1([<table>]) : integer
14575sc1_get_gpc1([<table>]) : integer
14576sc2_get_gpc1([<table>]) : integer
14577 Returns the value of the second General Purpose Counter associated to the
14578 currently tracked counters. See also src_get_gpc1 and sc/sc0/sc1/sc2_inc_gpc1.
14579
Thierry FOURNIER236657b2015-08-19 08:25:14 +020014580sc_get_gpt0(<ctr>[,<table>]) : integer
14581sc0_get_gpt0([<table>]) : integer
14582sc1_get_gpt0([<table>]) : integer
14583sc2_get_gpt0([<table>]) : integer
14584 Returns the value of the first General Purpose Tag associated to the
14585 currently tracked counters. See also src_get_gpt0.
14586
Cyril Bonté62ba8702014-04-22 23:52:25 +020014587sc_gpc0_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014588sc0_gpc0_rate([<table>]) : integer
14589sc1_gpc0_rate([<table>]) : integer
14590sc2_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014591 Returns the average increment rate of the first General Purpose Counter
14592 associated to the currently tracked counters. It reports the frequency
14593 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014594 src_gpc0_rate, sc/sc0/sc1/sc2_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
14595 that the "gpc0_rate" counter must be stored in the stick-table for a value to
14596 be returned, as "gpc0" only holds the event count.
Willy Tarreaue9656522010-08-17 15:40:09 +020014597
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014598sc_gpc1_rate(<ctr>[,<table>]) : integer
14599sc0_gpc1_rate([<table>]) : integer
14600sc1_gpc1_rate([<table>]) : integer
14601sc2_gpc1_rate([<table>]) : integer
14602 Returns the average increment rate of the second General Purpose Counter
14603 associated to the currently tracked counters. It reports the frequency
14604 which the gpc1 counter was incremented over the configured period. See also
14605 src_gpcA_rate, sc/sc0/sc1/sc2_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
14606 that the "gpc1_rate" counter must be stored in the stick-table for a value to
14607 be returned, as "gpc1" only holds the event count.
14608
Cyril Bonté62ba8702014-04-22 23:52:25 +020014609sc_http_err_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014610sc0_http_err_cnt([<table>]) : integer
14611sc1_http_err_cnt([<table>]) : integer
14612sc2_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014613 Returns the cumulative number of HTTP errors from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020014614 counters. This includes the both request errors and 4xx error responses.
14615 See also src_http_err_cnt.
14616
Cyril Bonté62ba8702014-04-22 23:52:25 +020014617sc_http_err_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014618sc0_http_err_rate([<table>]) : integer
14619sc1_http_err_rate([<table>]) : integer
14620sc2_http_err_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014621 Returns the average rate of HTTP errors from the currently tracked counters,
14622 measured in amount of errors over the period configured in the table. This
14623 includes the both request errors and 4xx error responses. See also
14624 src_http_err_rate.
14625
Cyril Bonté62ba8702014-04-22 23:52:25 +020014626sc_http_req_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014627sc0_http_req_cnt([<table>]) : integer
14628sc1_http_req_cnt([<table>]) : integer
14629sc2_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014630 Returns the cumulative number of HTTP requests from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020014631 counters. This includes every started request, valid or not. See also
14632 src_http_req_cnt.
14633
Cyril Bonté62ba8702014-04-22 23:52:25 +020014634sc_http_req_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014635sc0_http_req_rate([<table>]) : integer
14636sc1_http_req_rate([<table>]) : integer
14637sc2_http_req_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014638 Returns the average rate of HTTP requests from the currently tracked
14639 counters, measured in amount of requests over the period configured in
14640 the table. This includes every started request, valid or not. See also
14641 src_http_req_rate.
14642
Cyril Bonté62ba8702014-04-22 23:52:25 +020014643sc_inc_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014644sc0_inc_gpc0([<table>]) : integer
14645sc1_inc_gpc0([<table>]) : integer
14646sc2_inc_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014647 Increments the first General Purpose Counter associated to the currently
Willy Tarreau869948b2013-01-04 14:14:57 +010014648 tracked counters, and returns its new value. Before the first invocation,
14649 the stored value is zero, so first invocation will increase it to 1 and will
14650 return 1. This is typically used as a second ACL in an expression in order
14651 to mark a connection when a first ACL was verified :
Willy Tarreaue9656522010-08-17 15:40:09 +020014652
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014653 Example:
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014654 acl abuse sc0_http_req_rate gt 10
14655 acl kill sc0_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020014656 tcp-request connection reject if abuse kill
14657
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014658sc_inc_gpc1(<ctr>[,<table>]) : integer
14659sc0_inc_gpc1([<table>]) : integer
14660sc1_inc_gpc1([<table>]) : integer
14661sc2_inc_gpc1([<table>]) : integer
14662 Increments the second General Purpose Counter associated to the currently
14663 tracked counters, and returns its new value. Before the first invocation,
14664 the stored value is zero, so first invocation will increase it to 1 and will
14665 return 1. This is typically used as a second ACL in an expression in order
14666 to mark a connection when a first ACL was verified.
14667
Cyril Bonté62ba8702014-04-22 23:52:25 +020014668sc_kbytes_in(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014669sc0_kbytes_in([<table>]) : integer
14670sc1_kbytes_in([<table>]) : integer
14671sc2_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014672 Returns the total amount of client-to-server data from the currently tracked
14673 counters, measured in kilobytes. The test is currently performed on 32-bit
14674 integers, which limits values to 4 terabytes. See also src_kbytes_in.
Willy Tarreaue9656522010-08-17 15:40:09 +020014675
Cyril Bonté62ba8702014-04-22 23:52:25 +020014676sc_kbytes_out(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014677sc0_kbytes_out([<table>]) : integer
14678sc1_kbytes_out([<table>]) : integer
14679sc2_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014680 Returns the total amount of server-to-client data from the currently tracked
14681 counters, measured in kilobytes. The test is currently performed on 32-bit
14682 integers, which limits values to 4 terabytes. See also src_kbytes_out.
Willy Tarreaue9656522010-08-17 15:40:09 +020014683
Cyril Bonté62ba8702014-04-22 23:52:25 +020014684sc_sess_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014685sc0_sess_cnt([<table>]) : integer
14686sc1_sess_cnt([<table>]) : integer
14687sc2_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014688 Returns the cumulative number of incoming connections that were transformed
Willy Tarreaue9656522010-08-17 15:40:09 +020014689 into sessions, which means that they were accepted by a "tcp-request
14690 connection" rule, from the currently tracked counters. A backend may count
14691 more sessions than connections because each connection could result in many
Jamie Gloudonaaa21002012-08-25 00:18:33 -040014692 backend sessions if some HTTP keep-alive is performed over the connection
Willy Tarreaue9656522010-08-17 15:40:09 +020014693 with the client. See also src_sess_cnt.
14694
Cyril Bonté62ba8702014-04-22 23:52:25 +020014695sc_sess_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014696sc0_sess_rate([<table>]) : integer
14697sc1_sess_rate([<table>]) : integer
14698sc2_sess_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014699 Returns the average session rate from the currently tracked counters,
14700 measured in amount of sessions over the period configured in the table. A
14701 session is a connection that got past the early "tcp-request connection"
14702 rules. A backend may count more sessions than connections because each
14703 connection could result in many backend sessions if some HTTP keep-alive is
Jamie Gloudonaaa21002012-08-25 00:18:33 -040014704 performed over the connection with the client. See also src_sess_rate.
Willy Tarreaue9656522010-08-17 15:40:09 +020014705
Cyril Bonté62ba8702014-04-22 23:52:25 +020014706sc_tracked(<ctr>[,<table>]) : boolean
Willy Tarreau0f791d42013-07-23 19:56:43 +020014707sc0_tracked([<table>]) : boolean
14708sc1_tracked([<table>]) : boolean
14709sc2_tracked([<table>]) : boolean
Willy Tarreau6f1615f2013-06-03 15:15:22 +020014710 Returns true if the designated session counter is currently being tracked by
14711 the current session. This can be useful when deciding whether or not we want
14712 to set some values in a header passed to the server.
14713
Cyril Bonté62ba8702014-04-22 23:52:25 +020014714sc_trackers(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014715sc0_trackers([<table>]) : integer
14716sc1_trackers([<table>]) : integer
14717sc2_trackers([<table>]) : integer
Willy Tarreau2406db42012-12-09 12:16:43 +010014718 Returns the current amount of concurrent connections tracking the same
14719 tracked counters. This number is automatically incremented when tracking
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014720 begins and decremented when tracking stops. It differs from sc0_conn_cur in
Willy Tarreau2406db42012-12-09 12:16:43 +010014721 that it does not rely on any stored information but on the table's reference
14722 count (the "use" value which is returned by "show table" on the CLI). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020014723 may sometimes be more suited for layer7 tracking. It can be used to tell a
14724 server how many concurrent connections there are from a given address for
14725 example.
Willy Tarreau2406db42012-12-09 12:16:43 +010014726
Willy Tarreau74ca5042013-06-11 23:12:07 +020014727so_id : integer
14728 Returns an integer containing the current listening socket's id. It is useful
14729 in frontends involving many "bind" lines, or to stick all users coming via a
14730 same socket to the same server.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014731
Willy Tarreau74ca5042013-06-11 23:12:07 +020014732src : ip
Davor Ocelice9ed2812017-12-25 17:49:28 +010014733 This is the source IPv4 address of the client of the session. It is of type
Willy Tarreau74ca5042013-06-11 23:12:07 +020014734 IP and works on both IPv4 and IPv6 tables. On IPv6 tables, IPv4 addresses are
14735 mapped to their IPv6 equivalent, according to RFC 4291. Note that it is the
14736 TCP-level source address which is used, and not the address of a client
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010014737 behind a proxy. However if the "accept-proxy" or "accept-netscaler-cip" bind
14738 directive is used, it can be the address of a client behind another
14739 PROXY-protocol compatible component for all rule sets except
14740 "tcp-request connection" which sees the real address.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014741
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010014742 Example:
14743 # add an HTTP header in requests with the originating address' country
14744 http-request set-header X-Country %[src,map_ip(geoip.lst)]
14745
Willy Tarreau74ca5042013-06-11 23:12:07 +020014746src_bytes_in_rate([<table>]) : integer
14747 Returns the average bytes rate from the incoming connection's source address
14748 in the current proxy's stick-table or in the designated stick-table, measured
14749 in amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014750 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_in_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014751
Willy Tarreau74ca5042013-06-11 23:12:07 +020014752src_bytes_out_rate([<table>]) : integer
14753 Returns the average bytes rate to the incoming connection's source address in
14754 the current proxy's stick-table or in the designated stick-table, measured in
Willy Tarreauc9705a12010-07-27 20:05:50 +020014755 amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014756 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_out_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014757
Willy Tarreau74ca5042013-06-11 23:12:07 +020014758src_clr_gpc0([<table>]) : integer
14759 Clears the first General Purpose Counter associated to the incoming
14760 connection's source address in the current proxy's stick-table or in the
14761 designated stick-table, and returns its previous value. If the address is not
14762 found, an entry is created and 0 is returned. This is typically used as a
14763 second ACL in an expression in order to mark a connection when a first ACL
14764 was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020014765
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014766 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020014767 # block if 5 consecutive requests continue to come faster than 10 sess
14768 # per second, and reset the counter as soon as the traffic slows down.
14769 acl abuse src_http_req_rate gt 10
14770 acl kill src_inc_gpc0 gt 5
Willy Tarreau869948b2013-01-04 14:14:57 +010014771 acl save src_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020014772 tcp-request connection accept if !abuse save
14773 tcp-request connection reject if abuse kill
14774
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014775src_clr_gpc1([<table>]) : integer
14776 Clears the second General Purpose Counter associated to the incoming
14777 connection's source address in the current proxy's stick-table or in the
14778 designated stick-table, and returns its previous value. If the address is not
14779 found, an entry is created and 0 is returned. This is typically used as a
14780 second ACL in an expression in order to mark a connection when a first ACL
14781 was verified.
14782
Willy Tarreau74ca5042013-06-11 23:12:07 +020014783src_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014784 Returns the cumulative number of connections initiated from the current
Willy Tarreau74ca5042013-06-11 23:12:07 +020014785 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020014786 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014787 See also sc/sc0/sc1/sc2_conn_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014788
Willy Tarreau74ca5042013-06-11 23:12:07 +020014789src_conn_cur([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020014790 Returns the current amount of concurrent connections initiated from the
Willy Tarreau74ca5042013-06-11 23:12:07 +020014791 current incoming connection's source address in the current proxy's
14792 stick-table or in the designated stick-table. If the address is not found,
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014793 zero is returned. See also sc/sc0/sc1/sc2_conn_cur.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014794
Willy Tarreau74ca5042013-06-11 23:12:07 +020014795src_conn_rate([<table>]) : integer
14796 Returns the average connection rate from the incoming connection's source
14797 address in the current proxy's stick-table or in the designated stick-table,
14798 measured in amount of connections over the period configured in the table. If
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014799 the address is not found, zero is returned. See also sc/sc0/sc1/sc2_conn_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014800
Willy Tarreau74ca5042013-06-11 23:12:07 +020014801src_get_gpc0([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020014802 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau74ca5042013-06-11 23:12:07 +020014803 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020014804 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014805 See also sc/sc0/sc1/sc2_get_gpc0 and src_inc_gpc0.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014806
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014807src_get_gpc1([<table>]) : integer
14808 Returns the value of the second General Purpose Counter associated to the
14809 incoming connection's source address in the current proxy's stick-table or in
14810 the designated stick-table. If the address is not found, zero is returned.
14811 See also sc/sc0/sc1/sc2_get_gpc1 and src_inc_gpc1.
14812
Thierry FOURNIER236657b2015-08-19 08:25:14 +020014813src_get_gpt0([<table>]) : integer
14814 Returns the value of the first General Purpose Tag associated to the
14815 incoming connection's source address in the current proxy's stick-table or in
14816 the designated stick-table. If the address is not found, zero is returned.
14817 See also sc/sc0/sc1/sc2_get_gpt0.
14818
Willy Tarreau74ca5042013-06-11 23:12:07 +020014819src_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014820 Returns the average increment rate of the first General Purpose Counter
Willy Tarreau74ca5042013-06-11 23:12:07 +020014821 associated to the incoming connection's source address in the current proxy's
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014822 stick-table or in the designated stick-table. It reports the frequency
14823 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014824 sc/sc0/sc1/sc2_gpc0_rate, src_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
14825 that the "gpc0_rate" counter must be stored in the stick-table for a value to
14826 be returned, as "gpc0" only holds the event count.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014827
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014828src_gpc1_rate([<table>]) : integer
14829 Returns the average increment rate of the second General Purpose Counter
14830 associated to the incoming connection's source address in the current proxy's
14831 stick-table or in the designated stick-table. It reports the frequency
14832 which the gpc1 counter was incremented over the configured period. See also
14833 sc/sc0/sc1/sc2_gpc1_rate, src_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
14834 that the "gpc1_rate" counter must be stored in the stick-table for a value to
14835 be returned, as "gpc1" only holds the event count.
14836
Willy Tarreau74ca5042013-06-11 23:12:07 +020014837src_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014838 Returns the cumulative number of HTTP errors from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020014839 source address in the current proxy's stick-table or in the designated
Willy Tarreauc9705a12010-07-27 20:05:50 +020014840 stick-table. This includes the both request errors and 4xx error responses.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014841 See also sc/sc0/sc1/sc2_http_err_cnt. If the address is not found, zero is
Willy Tarreau74ca5042013-06-11 23:12:07 +020014842 returned.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014843
Willy Tarreau74ca5042013-06-11 23:12:07 +020014844src_http_err_rate([<table>]) : integer
14845 Returns the average rate of HTTP errors from the incoming connection's source
14846 address in the current proxy's stick-table or in the designated stick-table,
14847 measured in amount of errors over the period configured in the table. This
14848 includes the both request errors and 4xx error responses. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014849 not found, zero is returned. See also sc/sc0/sc1/sc2_http_err_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014850
Willy Tarreau74ca5042013-06-11 23:12:07 +020014851src_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014852 Returns the cumulative number of HTTP requests from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020014853 source address in the current proxy's stick-table or in the designated stick-
14854 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014855 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014856
Willy Tarreau74ca5042013-06-11 23:12:07 +020014857src_http_req_rate([<table>]) : integer
14858 Returns the average rate of HTTP requests from the incoming connection's
14859 source address in the current proxy's stick-table or in the designated stick-
14860 table, measured in amount of requests over the period configured in the
Willy Tarreauc9705a12010-07-27 20:05:50 +020014861 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014862 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014863
Willy Tarreau74ca5042013-06-11 23:12:07 +020014864src_inc_gpc0([<table>]) : integer
14865 Increments the first General Purpose Counter associated to the incoming
14866 connection's source address in the current proxy's stick-table or in the
14867 designated stick-table, and returns its new value. If the address is not
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014868 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc0.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014869 This is typically used as a second ACL in an expression in order to mark a
14870 connection when a first ACL was verified :
Willy Tarreauc9705a12010-07-27 20:05:50 +020014871
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014872 Example:
Willy Tarreauc9705a12010-07-27 20:05:50 +020014873 acl abuse src_http_req_rate gt 10
Willy Tarreau869948b2013-01-04 14:14:57 +010014874 acl kill src_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020014875 tcp-request connection reject if abuse kill
Willy Tarreauc9705a12010-07-27 20:05:50 +020014876
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014877src_inc_gpc1([<table>]) : integer
14878 Increments the second General Purpose Counter associated to the incoming
14879 connection's source address in the current proxy's stick-table or in the
14880 designated stick-table, and returns its new value. If the address is not
14881 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc1.
14882 This is typically used as a second ACL in an expression in order to mark a
14883 connection when a first ACL was verified.
14884
Willy Tarreau16e01562016-08-09 16:46:18 +020014885src_is_local : boolean
14886 Returns true if the source address of the incoming connection is local to the
14887 system, or false if the address doesn't exist on the system, meaning that it
14888 comes from a remote machine. Note that UNIX addresses are considered local.
14889 It can be useful to apply certain access restrictions based on where the
Davor Ocelice9ed2812017-12-25 17:49:28 +010014890 client comes from (e.g. require auth or https for remote machines). Please
Willy Tarreau16e01562016-08-09 16:46:18 +020014891 note that the check involves a few system calls, so it's better to do it only
14892 once per connection.
14893
Willy Tarreau74ca5042013-06-11 23:12:07 +020014894src_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014895 Returns the total amount of data received from the incoming connection's
14896 source address in the current proxy's stick-table or in the designated
14897 stick-table, measured in kilobytes. If the address is not found, zero is
14898 returned. The test is currently performed on 32-bit integers, which limits
14899 values to 4 terabytes. See also sc/sc0/sc1/sc2_kbytes_in.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014900
Willy Tarreau74ca5042013-06-11 23:12:07 +020014901src_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014902 Returns the total amount of data sent to the incoming connection's source
14903 address in the current proxy's stick-table or in the designated stick-table,
14904 measured in kilobytes. If the address is not found, zero is returned. The
14905 test is currently performed on 32-bit integers, which limits values to 4
14906 terabytes. See also sc/sc0/sc1/sc2_kbytes_out.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020014907
Willy Tarreau74ca5042013-06-11 23:12:07 +020014908src_port : integer
14909 Returns an integer value corresponding to the TCP source port of the
14910 connection on the client side, which is the port the client connected from.
14911 Usage of this function is very limited as modern protocols do not care much
14912 about source ports nowadays.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010014913
Willy Tarreau74ca5042013-06-11 23:12:07 +020014914src_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014915 Returns the cumulative number of connections initiated from the incoming
Willy Tarreauc9705a12010-07-27 20:05:50 +020014916 connection's source IPv4 address in the current proxy's stick-table or in the
14917 designated stick-table, that were transformed into sessions, which means that
14918 they were accepted by "tcp-request" rules. If the address is not found, zero
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014919 is returned. See also sc/sc0/sc1/sc2_sess_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014920
Willy Tarreau74ca5042013-06-11 23:12:07 +020014921src_sess_rate([<table>]) : integer
14922 Returns the average session rate from the incoming connection's source
14923 address in the current proxy's stick-table or in the designated stick-table,
14924 measured in amount of sessions over the period configured in the table. A
14925 session is a connection that went past the early "tcp-request" rules. If the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014926 address is not found, zero is returned. See also sc/sc0/sc1/sc2_sess_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014927
Willy Tarreau74ca5042013-06-11 23:12:07 +020014928src_updt_conn_cnt([<table>]) : integer
14929 Creates or updates the entry associated to the incoming connection's source
14930 address in the current proxy's stick-table or in the designated stick-table.
14931 This table must be configured to store the "conn_cnt" data type, otherwise
14932 the match will be ignored. The current count is incremented by one, and the
14933 expiration timer refreshed. The updated count is returned, so this match
14934 can't return zero. This was used to reject service abusers based on their
14935 source address. Note: it is recommended to use the more complete "track-sc*"
14936 actions in "tcp-request" rules instead.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020014937
14938 Example :
14939 # This frontend limits incoming SSH connections to 3 per 10 second for
14940 # each source address, and rejects excess connections until a 10 second
14941 # silence is observed. At most 20 addresses are tracked.
14942 listen ssh
14943 bind :22
14944 mode tcp
14945 maxconn 100
Willy Tarreauc9705a12010-07-27 20:05:50 +020014946 stick-table type ip size 20 expire 10s store conn_cnt
Willy Tarreau74ca5042013-06-11 23:12:07 +020014947 tcp-request content reject if { src_updt_conn_cnt gt 3 }
Willy Tarreaua975b8f2010-06-05 19:13:27 +020014948 server local 127.0.0.1:22
14949
Willy Tarreau74ca5042013-06-11 23:12:07 +020014950srv_id : integer
14951 Returns an integer containing the server's id when processing the response.
14952 While it's almost only used with ACLs, it may be used for logging or
14953 debugging.
Hervé COMMOWICKdaa824e2011-08-05 12:09:44 +020014954
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200149557.3.4. Fetching samples at Layer 5
Willy Tarreau74ca5042013-06-11 23:12:07 +020014956----------------------------------
Willy Tarreau0b1cd942010-05-16 22:18:27 +020014957
Willy Tarreau74ca5042013-06-11 23:12:07 +020014958The layer 5 usually describes just the session layer which in haproxy is
14959closest to the session once all the connection handshakes are finished, but
14960when no content is yet made available. The fetch methods described here are
14961usable as low as the "tcp-request content" rule sets unless they require some
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014962future information. Those generally include the results of SSL negotiations.
Willy Tarreauc735a072011-03-29 00:57:02 +020014963
Ben Shillitof25e8e52016-12-02 14:25:37 +00001496451d.all(<prop>[,<prop>*]) : string
14965 Returns values for the properties requested as a string, where values are
14966 separated by the delimiter specified with "51degrees-property-separator".
14967 The device is identified using all the important HTTP headers from the
14968 request. The function can be passed up to five property names, and if a
14969 property name can't be found, the value "NoData" is returned.
14970
14971 Example :
14972 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request
14973 # containing the three properties requested using all relevant headers from
14974 # the request.
14975 frontend http-in
14976 bind *:8081
14977 default_backend servers
14978 http-request set-header X-51D-DeviceTypeMobileTablet \
14979 %[51d.all(DeviceType,IsMobile,IsTablet)]
14980
Emeric Brun645ae792014-04-30 14:21:06 +020014981ssl_bc : boolean
14982 Returns true when the back connection was made via an SSL/TLS transport
14983 layer and is locally deciphered. This means the outgoing connection was made
14984 other a server with the "ssl" option.
14985
14986ssl_bc_alg_keysize : integer
14987 Returns the symmetric cipher key size supported in bits when the outgoing
14988 connection was made over an SSL/TLS transport layer.
14989
Olivier Houchard6b77f492018-11-22 18:18:29 +010014990ssl_bc_alpn : string
14991 This extracts the Application Layer Protocol Negotiation field from an
14992 outgoing connection made via a TLS transport layer.
14993 The result is a string containing the protocol name negociated with the
14994 server. The SSL library must have been built with support for TLS
14995 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
14996 not advertised unless the "alpn" keyword on the "server" line specifies a
14997 protocol list. Also, nothing forces the server to pick a protocol from this
14998 list, any other one may be requested. The TLS ALPN extension is meant to
14999 replace the TLS NPN extension. See also "ssl_bc_npn".
15000
Emeric Brun645ae792014-04-30 14:21:06 +020015001ssl_bc_cipher : string
15002 Returns the name of the used cipher when the outgoing connection was made
15003 over an SSL/TLS transport layer.
15004
Emeric Brun74f7ffa2018-02-19 16:14:12 +010015005ssl_bc_is_resumed : boolean
15006 Returns true when the back connection was made over an SSL/TLS transport
15007 layer and the newly created SSL session was resumed using a cached
15008 session or a TLS ticket.
15009
Olivier Houchard6b77f492018-11-22 18:18:29 +010015010ssl_bc_npn : string
15011 This extracts the Next Protocol Negotiation field from an outgoing connection
15012 made via a TLS transport layer. The result is a string containing the
15013 protocol name negociated with the server . The SSL library must have been
15014 built with support for TLS extensions enabled (check haproxy -vv). Note that
15015 the TLS NPN extension is not advertised unless the "npn" keyword on the
15016 "server" line specifies a protocol list. Also, nothing forces the server to
15017 pick a protocol from this list, any other one may be used. Please note that
15018 the TLS NPN extension was replaced with ALPN.
15019
Emeric Brun645ae792014-04-30 14:21:06 +020015020ssl_bc_protocol : string
15021 Returns the name of the used protocol when the outgoing connection was made
15022 over an SSL/TLS transport layer.
15023
Emeric Brunb73a9b02014-04-30 18:49:19 +020015024ssl_bc_unique_id : binary
Emeric Brun645ae792014-04-30 14:21:06 +020015025 When the outgoing connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020015026 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
15027 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
Emeric Brun645ae792014-04-30 14:21:06 +020015028
15029ssl_bc_session_id : binary
15030 Returns the SSL ID of the back connection when the outgoing connection was
15031 made over an SSL/TLS transport layer. It is useful to log if we want to know
15032 if session was reused or not.
15033
Patrick Hemmere0275472018-04-28 19:15:51 -040015034ssl_bc_session_key : binary
15035 Returns the SSL session master key of the back connection when the outgoing
15036 connection was made over an SSL/TLS transport layer. It is useful to decrypt
15037 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
15038 BoringSSL.
15039
Emeric Brun645ae792014-04-30 14:21:06 +020015040ssl_bc_use_keysize : integer
15041 Returns the symmetric cipher key size used in bits when the outgoing
15042 connection was made over an SSL/TLS transport layer.
15043
Willy Tarreau74ca5042013-06-11 23:12:07 +020015044ssl_c_ca_err : integer
15045 When the incoming connection was made over an SSL/TLS transport layer,
15046 returns the ID of the first error detected during verification of the client
15047 certificate at depth > 0, or 0 if no error was encountered during this
15048 verification process. Please refer to your SSL library's documentation to
15049 find the exhaustive list of error codes.
Willy Tarreauc735a072011-03-29 00:57:02 +020015050
Willy Tarreau74ca5042013-06-11 23:12:07 +020015051ssl_c_ca_err_depth : integer
15052 When the incoming connection was made over an SSL/TLS transport layer,
15053 returns the depth in the CA chain of the first error detected during the
15054 verification of the client certificate. If no error is encountered, 0 is
15055 returned.
Willy Tarreau0ba27502007-12-24 16:55:16 +010015056
Emeric Brun43e79582014-10-29 19:03:26 +010015057ssl_c_der : binary
15058 Returns the DER formatted certificate presented by the client when the
15059 incoming connection was made over an SSL/TLS transport layer. When used for
15060 an ACL, the value(s) to match against can be passed in hexadecimal form.
15061
Willy Tarreau74ca5042013-06-11 23:12:07 +020015062ssl_c_err : integer
15063 When the incoming connection was made over an SSL/TLS transport layer,
15064 returns the ID of the first error detected during verification at depth 0, or
15065 0 if no error was encountered during this verification process. Please refer
15066 to your SSL library's documentation to find the exhaustive list of error
15067 codes.
Willy Tarreau62644772008-07-16 18:36:06 +020015068
Willy Tarreau74ca5042013-06-11 23:12:07 +020015069ssl_c_i_dn([<entry>[,<occ>]]) : string
15070 When the incoming connection was made over an SSL/TLS transport layer,
15071 returns the full distinguished name of the issuer of the certificate
15072 presented by the client when no <entry> is specified, or the value of the
15073 first given entry found from the beginning of the DN. If a positive/negative
15074 occurrence number is specified as the optional second argument, it returns
15075 the value of the nth given entry value from the beginning/end of the DN.
15076 For instance, "ssl_c_i_dn(OU,2)" the second organization unit, and
15077 "ssl_c_i_dn(CN)" retrieves the common name.
Willy Tarreau62644772008-07-16 18:36:06 +020015078
Willy Tarreau74ca5042013-06-11 23:12:07 +020015079ssl_c_key_alg : string
15080 Returns the name of the algorithm used to generate the key of the certificate
15081 presented by the client when the incoming connection was made over an SSL/TLS
15082 transport layer.
Willy Tarreau62644772008-07-16 18:36:06 +020015083
Willy Tarreau74ca5042013-06-11 23:12:07 +020015084ssl_c_notafter : string
15085 Returns the end date presented by the client as a formatted string
15086 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15087 transport layer.
Emeric Brunbede3d02009-06-30 17:54:00 +020015088
Willy Tarreau74ca5042013-06-11 23:12:07 +020015089ssl_c_notbefore : string
15090 Returns the start date presented by the client as a formatted string
15091 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15092 transport layer.
Willy Tarreaub6672b52011-12-12 17:23:41 +010015093
Willy Tarreau74ca5042013-06-11 23:12:07 +020015094ssl_c_s_dn([<entry>[,<occ>]]) : string
15095 When the incoming connection was made over an SSL/TLS transport layer,
15096 returns the full distinguished name of the subject of the certificate
15097 presented by the client when no <entry> is specified, or the value of the
15098 first given entry found from the beginning of the DN. If a positive/negative
15099 occurrence number is specified as the optional second argument, it returns
15100 the value of the nth given entry value from the beginning/end of the DN.
15101 For instance, "ssl_c_s_dn(OU,2)" the second organization unit, and
15102 "ssl_c_s_dn(CN)" retrieves the common name.
Willy Tarreaub6672b52011-12-12 17:23:41 +010015103
Willy Tarreau74ca5042013-06-11 23:12:07 +020015104ssl_c_serial : binary
15105 Returns the serial of the certificate presented by the client when the
15106 incoming connection was made over an SSL/TLS transport layer. When used for
15107 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun2525b6b2012-10-18 15:59:43 +020015108
Willy Tarreau74ca5042013-06-11 23:12:07 +020015109ssl_c_sha1 : binary
15110 Returns the SHA-1 fingerprint of the certificate presented by the client when
15111 the incoming connection was made over an SSL/TLS transport layer. This can be
15112 used to stick a client to a server, or to pass this information to a server.
Willy Tarreau2d0caa32014-07-02 19:01:22 +020015113 Note that the output is binary, so if you want to pass that signature to the
15114 server, you need to encode it in hex or base64, such as in the example below:
15115
Jarno Huuskonen676f6222017-03-30 09:19:45 +030015116 Example:
Willy Tarreau2d0caa32014-07-02 19:01:22 +020015117 http-request set-header X-SSL-Client-SHA1 %[ssl_c_sha1,hex]
Emeric Brun2525b6b2012-10-18 15:59:43 +020015118
Willy Tarreau74ca5042013-06-11 23:12:07 +020015119ssl_c_sig_alg : string
15120 Returns the name of the algorithm used to sign the certificate presented by
15121 the client when the incoming connection was made over an SSL/TLS transport
15122 layer.
Emeric Brun87855892012-10-17 17:39:35 +020015123
Willy Tarreau74ca5042013-06-11 23:12:07 +020015124ssl_c_used : boolean
15125 Returns true if current SSL session uses a client certificate even if current
15126 connection uses SSL session resumption. See also "ssl_fc_has_crt".
Emeric Brun7f56e742012-10-19 18:15:40 +020015127
Willy Tarreau74ca5042013-06-11 23:12:07 +020015128ssl_c_verify : integer
15129 Returns the verify result error ID when the incoming connection was made over
15130 an SSL/TLS transport layer, otherwise zero if no error is encountered. Please
15131 refer to your SSL library's documentation for an exhaustive list of error
15132 codes.
Emeric Brunce5ad802012-10-22 14:11:22 +020015133
Willy Tarreau74ca5042013-06-11 23:12:07 +020015134ssl_c_version : integer
15135 Returns the version of the certificate presented by the client when the
15136 incoming connection was made over an SSL/TLS transport layer.
Emeric Brunce5ad802012-10-22 14:11:22 +020015137
Emeric Brun43e79582014-10-29 19:03:26 +010015138ssl_f_der : binary
15139 Returns the DER formatted certificate presented by the frontend when the
15140 incoming connection was made over an SSL/TLS transport layer. When used for
15141 an ACL, the value(s) to match against can be passed in hexadecimal form.
15142
Willy Tarreau74ca5042013-06-11 23:12:07 +020015143ssl_f_i_dn([<entry>[,<occ>]]) : string
15144 When the incoming connection was made over an SSL/TLS transport layer,
15145 returns the full distinguished name of the issuer of the certificate
15146 presented by the frontend when no <entry> is specified, or the value of the
15147 first given entry found from the beginning of the DN. If a positive/negative
Emeric Brun87855892012-10-17 17:39:35 +020015148 occurrence number is specified as the optional second argument, it returns
Willy Tarreau74ca5042013-06-11 23:12:07 +020015149 the value of the nth given entry value from the beginning/end of the DN.
15150 For instance, "ssl_f_i_dn(OU,2)" the second organization unit, and
15151 "ssl_f_i_dn(CN)" retrieves the common name.
Emeric Brun87855892012-10-17 17:39:35 +020015152
Willy Tarreau74ca5042013-06-11 23:12:07 +020015153ssl_f_key_alg : string
15154 Returns the name of the algorithm used to generate the key of the certificate
15155 presented by the frontend when the incoming connection was made over an
15156 SSL/TLS transport layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020015157
Willy Tarreau74ca5042013-06-11 23:12:07 +020015158ssl_f_notafter : string
15159 Returns the end date presented by the frontend as a formatted string
15160 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15161 transport layer.
Emeric Brun2525b6b2012-10-18 15:59:43 +020015162
Willy Tarreau74ca5042013-06-11 23:12:07 +020015163ssl_f_notbefore : string
15164 Returns the start date presented by the frontend as a formatted string
15165 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15166 transport layer.
Emeric Brun87855892012-10-17 17:39:35 +020015167
Willy Tarreau74ca5042013-06-11 23:12:07 +020015168ssl_f_s_dn([<entry>[,<occ>]]) : string
15169 When the incoming connection was made over an SSL/TLS transport layer,
15170 returns the full distinguished name of the subject of the certificate
15171 presented by the frontend when no <entry> is specified, or the value of the
15172 first given entry found from the beginning of the DN. If a positive/negative
15173 occurrence number is specified as the optional second argument, it returns
15174 the value of the nth given entry value from the beginning/end of the DN.
15175 For instance, "ssl_f_s_dn(OU,2)" the second organization unit, and
15176 "ssl_f_s_dn(CN)" retrieves the common name.
Emeric Brunce5ad802012-10-22 14:11:22 +020015177
Willy Tarreau74ca5042013-06-11 23:12:07 +020015178ssl_f_serial : binary
15179 Returns the serial of the certificate presented by the frontend when the
15180 incoming connection was made over an SSL/TLS transport layer. When used for
15181 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun87855892012-10-17 17:39:35 +020015182
Emeric Brun55f4fa82014-04-30 17:11:25 +020015183ssl_f_sha1 : binary
15184 Returns the SHA-1 fingerprint of the certificate presented by the frontend
15185 when the incoming connection was made over an SSL/TLS transport layer. This
15186 can be used to know which certificate was chosen using SNI.
15187
Willy Tarreau74ca5042013-06-11 23:12:07 +020015188ssl_f_sig_alg : string
15189 Returns the name of the algorithm used to sign the certificate presented by
15190 the frontend when the incoming connection was made over an SSL/TLS transport
15191 layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020015192
Willy Tarreau74ca5042013-06-11 23:12:07 +020015193ssl_f_version : integer
15194 Returns the version of the certificate presented by the frontend when the
15195 incoming connection was made over an SSL/TLS transport layer.
15196
15197ssl_fc : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020015198 Returns true when the front connection was made via an SSL/TLS transport
15199 layer and is locally deciphered. This means it has matched a socket declared
15200 with a "bind" line having the "ssl" option.
15201
Willy Tarreau74ca5042013-06-11 23:12:07 +020015202 Example :
15203 # This passes "X-Proto: https" to servers when client connects over SSL
15204 listen http-https
15205 bind :80
15206 bind :443 ssl crt /etc/haproxy.pem
15207 http-request add-header X-Proto https if { ssl_fc }
15208
15209ssl_fc_alg_keysize : integer
15210 Returns the symmetric cipher key size supported in bits when the incoming
15211 connection was made over an SSL/TLS transport layer.
15212
15213ssl_fc_alpn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015214 This extracts the Application Layer Protocol Negotiation field from an
Willy Tarreau74ca5042013-06-11 23:12:07 +020015215 incoming connection made via a TLS transport layer and locally deciphered by
15216 haproxy. The result is a string containing the protocol name advertised by
15217 the client. The SSL library must have been built with support for TLS
15218 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
15219 not advertised unless the "alpn" keyword on the "bind" line specifies a
15220 protocol list. Also, nothing forces the client to pick a protocol from this
15221 list, any other one may be requested. The TLS ALPN extension is meant to
15222 replace the TLS NPN extension. See also "ssl_fc_npn".
15223
Willy Tarreau74ca5042013-06-11 23:12:07 +020015224ssl_fc_cipher : string
15225 Returns the name of the used cipher when the incoming connection was made
15226 over an SSL/TLS transport layer.
Willy Tarreauab861d32013-04-02 02:30:41 +020015227
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015228ssl_fc_cipherlist_bin : binary
15229 Returns the binary form of the client hello cipher list. The maximum returned
15230 value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010015231 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015232
15233ssl_fc_cipherlist_hex : string
15234 Returns the binary form of the client hello cipher list encoded as
15235 hexadecimal. The maximum returned value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010015236 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015237
15238ssl_fc_cipherlist_str : string
15239 Returns the decoded text form of the client hello cipher list. The maximum
15240 number of ciphers returned is according with the value of
15241 "tune.ssl.capture-cipherlist-size". Note that this sample-fetch is only
Davor Ocelice9ed2812017-12-25 17:49:28 +010015242 available with OpenSSL >= 1.0.2. If the function is not enabled, this
Emmanuel Hocdetddcde192017-09-01 17:32:08 +020015243 sample-fetch returns the hash like "ssl_fc_cipherlist_xxh".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015244
15245ssl_fc_cipherlist_xxh : integer
15246 Returns a xxh64 of the cipher list. This hash can be return only is the value
15247 "tune.ssl.capture-cipherlist-size" is set greater than 0, however the hash
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010015248 take in account all the data of the cipher list.
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015249
Willy Tarreau74ca5042013-06-11 23:12:07 +020015250ssl_fc_has_crt : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020015251 Returns true if a client certificate is present in an incoming connection over
15252 SSL/TLS transport layer. Useful if 'verify' statement is set to 'optional'.
Emeric Brun9143d372012-12-20 15:44:16 +010015253 Note: on SSL session resumption with Session ID or TLS ticket, client
15254 certificate is not present in the current connection but may be retrieved
15255 from the cache or the ticket. So prefer "ssl_c_used" if you want to check if
15256 current SSL session uses a client certificate.
Emeric Brun2525b6b2012-10-18 15:59:43 +020015257
Olivier Houchardccaa7de2017-10-02 11:51:03 +020015258ssl_fc_has_early : boolean
15259 Returns true if early data were sent, and the handshake didn't happen yet. As
15260 it has security implications, it is useful to be able to refuse those, or
15261 wait until the handshake happened.
15262
Willy Tarreau74ca5042013-06-11 23:12:07 +020015263ssl_fc_has_sni : boolean
15264 This checks for the presence of a Server Name Indication TLS extension (SNI)
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020015265 in an incoming connection was made over an SSL/TLS transport layer. Returns
15266 true when the incoming connection presents a TLS SNI field. This requires
15267 that the SSL library is build with support for TLS extensions enabled (check
15268 haproxy -vv).
Willy Tarreau7875d092012-09-10 08:20:03 +020015269
Nenad Merdanovic1516fe32016-05-17 03:31:21 +020015270ssl_fc_is_resumed : boolean
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020015271 Returns true if the SSL/TLS session has been resumed through the use of
Jérôme Magnin4a326cb2018-01-15 14:01:17 +010015272 SSL session cache or TLS tickets on an incoming connection over an SSL/TLS
15273 transport layer.
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020015274
Willy Tarreau74ca5042013-06-11 23:12:07 +020015275ssl_fc_npn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015276 This extracts the Next Protocol Negotiation field from an incoming connection
Willy Tarreau74ca5042013-06-11 23:12:07 +020015277 made via a TLS transport layer and locally deciphered by haproxy. The result
15278 is a string containing the protocol name advertised by the client. The SSL
15279 library must have been built with support for TLS extensions enabled (check
15280 haproxy -vv). Note that the TLS NPN extension is not advertised unless the
15281 "npn" keyword on the "bind" line specifies a protocol list. Also, nothing
15282 forces the client to pick a protocol from this list, any other one may be
15283 requested. Please note that the TLS NPN extension was replaced with ALPN.
Willy Tarreaua33c6542012-10-15 13:19:06 +020015284
Willy Tarreau74ca5042013-06-11 23:12:07 +020015285ssl_fc_protocol : string
15286 Returns the name of the used protocol when the incoming connection was made
15287 over an SSL/TLS transport layer.
Willy Tarreau7875d092012-09-10 08:20:03 +020015288
Emeric Brunb73a9b02014-04-30 18:49:19 +020015289ssl_fc_unique_id : binary
David Sc1ad52e2014-04-08 18:48:47 -040015290 When the incoming connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020015291 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
15292 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
David Sc1ad52e2014-04-08 18:48:47 -040015293
Willy Tarreau74ca5042013-06-11 23:12:07 +020015294ssl_fc_session_id : binary
15295 Returns the SSL ID of the front connection when the incoming connection was
15296 made over an SSL/TLS transport layer. It is useful to stick a given client to
15297 a server. It is important to note that some browsers refresh their session ID
15298 every few minutes.
Willy Tarreau7875d092012-09-10 08:20:03 +020015299
Patrick Hemmere0275472018-04-28 19:15:51 -040015300ssl_fc_session_key : binary
15301 Returns the SSL session master key of the front connection when the incoming
15302 connection was made over an SSL/TLS transport layer. It is useful to decrypt
15303 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
15304 BoringSSL.
15305
15306
Willy Tarreau74ca5042013-06-11 23:12:07 +020015307ssl_fc_sni : string
15308 This extracts the Server Name Indication TLS extension (SNI) field from an
15309 incoming connection made via an SSL/TLS transport layer and locally
15310 deciphered by haproxy. The result (when present) typically is a string
15311 matching the HTTPS host name (253 chars or less). The SSL library must have
15312 been built with support for TLS extensions enabled (check haproxy -vv).
15313
15314 This fetch is different from "req_ssl_sni" above in that it applies to the
15315 connection being deciphered by haproxy and not to SSL contents being blindly
15316 forwarded. See also "ssl_fc_sni_end" and "ssl_fc_sni_reg" below. This
Cyril Bonté9c1eb1e2012-10-09 22:45:34 +020015317 requires that the SSL library is build with support for TLS extensions
15318 enabled (check haproxy -vv).
Willy Tarreau62644772008-07-16 18:36:06 +020015319
Willy Tarreau74ca5042013-06-11 23:12:07 +020015320 ACL derivatives :
Willy Tarreau74ca5042013-06-11 23:12:07 +020015321 ssl_fc_sni_end : suffix match
15322 ssl_fc_sni_reg : regex match
Emeric Brun589fcad2012-10-16 14:13:26 +020015323
Willy Tarreau74ca5042013-06-11 23:12:07 +020015324ssl_fc_use_keysize : integer
15325 Returns the symmetric cipher key size used in bits when the incoming
15326 connection was made over an SSL/TLS transport layer.
Willy Tarreaub6fb4202008-07-20 11:18:28 +020015327
Willy Tarreaub6fb4202008-07-20 11:18:28 +020015328
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200153297.3.5. Fetching samples from buffer contents (Layer 6)
Willy Tarreau74ca5042013-06-11 23:12:07 +020015330------------------------------------------------------
Willy Tarreaub6fb4202008-07-20 11:18:28 +020015331
Willy Tarreau74ca5042013-06-11 23:12:07 +020015332Fetching samples from buffer contents is a bit different from the previous
15333sample fetches above because the sampled data are ephemeral. These data can
15334only be used when they're available and will be lost when they're forwarded.
15335For this reason, samples fetched from buffer contents during a request cannot
15336be used in a response for example. Even while the data are being fetched, they
15337can change. Sometimes it is necessary to set some delays or combine multiple
15338sample fetch methods to ensure that the expected data are complete and usable,
15339for example through TCP request content inspection. Please see the "tcp-request
15340content" keyword for more detailed information on the subject.
Willy Tarreau62644772008-07-16 18:36:06 +020015341
Willy Tarreau74ca5042013-06-11 23:12:07 +020015342payload(<offset>,<length>) : binary (deprecated)
Davor Ocelice9ed2812017-12-25 17:49:28 +010015343 This is an alias for "req.payload" when used in the context of a request (e.g.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015344 "stick on", "stick match"), and for "res.payload" when used in the context of
15345 a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010015346
Willy Tarreau74ca5042013-06-11 23:12:07 +020015347payload_lv(<offset1>,<length>[,<offset2>]) : binary (deprecated)
15348 This is an alias for "req.payload_lv" when used in the context of a request
Davor Ocelice9ed2812017-12-25 17:49:28 +010015349 (e.g. "stick on", "stick match"), and for "res.payload_lv" when used in the
Willy Tarreau74ca5042013-06-11 23:12:07 +020015350 context of a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010015351
Thierry FOURNIERd7d88812017-04-19 15:15:14 +020015352req.hdrs : string
15353 Returns the current request headers as string including the last empty line
15354 separating headers from the request body. The last empty line can be used to
15355 detect a truncated header block. This sample fetch is useful for some SPOE
15356 headers analyzers and for advanced logging.
15357
Thierry FOURNIER5617dce2017-04-09 05:38:19 +020015358req.hdrs_bin : binary
15359 Returns the current request headers contained in preparsed binary form. This
15360 is useful for offloading some processing with SPOE. Each string is described
15361 by a length followed by the number of bytes indicated in the length. The
15362 length is represented using the variable integer encoding detailed in the
15363 SPOE documentation. The end of the list is marked by a couple of empty header
15364 names and values (length of 0 for both).
15365
15366 *(<str:header-name><str:header-value>)<empty string><empty string>
15367
15368 int: refer to the SPOE documentation for the encoding
15369 str: <int:length><bytes>
15370
Willy Tarreau74ca5042013-06-11 23:12:07 +020015371req.len : integer
15372req_len : integer (deprecated)
15373 Returns an integer value corresponding to the number of bytes present in the
15374 request buffer. This is mostly used in ACL. It is important to understand
15375 that this test does not return false as long as the buffer is changing. This
15376 means that a check with equality to zero will almost always immediately match
15377 at the beginning of the session, while a test for more data will wait for
15378 that data to come in and return false only when haproxy is certain that no
15379 more data will come in. This test was designed to be used with TCP request
15380 content inspection.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015381
Willy Tarreau74ca5042013-06-11 23:12:07 +020015382req.payload(<offset>,<length>) : binary
15383 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020015384 in the request buffer. As a special case, if the <length> argument is zero,
15385 the the whole buffer from <offset> to the end is extracted. This can be used
15386 with ACLs in order to check for the presence of some content in a buffer at
15387 any location.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015388
Willy Tarreau74ca5042013-06-11 23:12:07 +020015389 ACL alternatives :
15390 payload(<offset>,<length>) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015391
Willy Tarreau74ca5042013-06-11 23:12:07 +020015392req.payload_lv(<offset1>,<length>[,<offset2>]) : binary
15393 This extracts a binary block whose size is specified at <offset1> for <length>
15394 bytes, and which starts at <offset2> if specified or just after the length in
15395 the request buffer. The <offset2> parameter also supports relative offsets if
15396 prepended with a '+' or '-' sign.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015397
Willy Tarreau74ca5042013-06-11 23:12:07 +020015398 ACL alternatives :
15399 payload_lv(<offset1>,<length>[,<offset2>]) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015400
Willy Tarreau74ca5042013-06-11 23:12:07 +020015401 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015402
Willy Tarreau74ca5042013-06-11 23:12:07 +020015403req.proto_http : boolean
15404req_proto_http : boolean (deprecated)
15405 Returns true when data in the request buffer look like HTTP and correctly
15406 parses as such. It is the same parser as the common HTTP request parser which
15407 is used so there should be no surprises. The test does not match until the
15408 request is complete, failed or timed out. This test may be used to report the
15409 protocol in TCP logs, but the biggest use is to block TCP request analysis
15410 until a complete HTTP request is present in the buffer, for example to track
15411 a header.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015412
Willy Tarreau74ca5042013-06-11 23:12:07 +020015413 Example:
15414 # track request counts per "base" (concatenation of Host+URL)
15415 tcp-request inspect-delay 10s
15416 tcp-request content reject if !HTTP
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020015417 tcp-request content track-sc0 base table req-rate
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015418
Willy Tarreau74ca5042013-06-11 23:12:07 +020015419req.rdp_cookie([<name>]) : string
15420rdp_cookie([<name>]) : string (deprecated)
15421 When the request buffer looks like the RDP protocol, extracts the RDP cookie
15422 <name>, or any cookie if unspecified. The parser only checks for the first
15423 cookie, as illustrated in the RDP protocol specification. The cookie name is
15424 case insensitive. Generally the "MSTS" cookie name will be used, as it can
15425 contain the user name of the client connecting to the server if properly
15426 configured on the client. The "MSTSHASH" cookie is often used as well for
15427 session stickiness to servers.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015428
Willy Tarreau74ca5042013-06-11 23:12:07 +020015429 This differs from "balance rdp-cookie" in that any balancing algorithm may be
15430 used and thus the distribution of clients to backend servers is not linked to
15431 a hash of the RDP cookie. It is envisaged that using a balancing algorithm
15432 such as "balance roundrobin" or "balance leastconn" will lead to a more even
15433 distribution of clients to backend servers than the hash used by "balance
15434 rdp-cookie".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015435
Willy Tarreau74ca5042013-06-11 23:12:07 +020015436 ACL derivatives :
15437 req_rdp_cookie([<name>]) : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015438
Willy Tarreau74ca5042013-06-11 23:12:07 +020015439 Example :
15440 listen tse-farm
15441 bind 0.0.0.0:3389
15442 # wait up to 5s for an RDP cookie in the request
15443 tcp-request inspect-delay 5s
15444 tcp-request content accept if RDP_COOKIE
15445 # apply RDP cookie persistence
15446 persist rdp-cookie
15447 # Persist based on the mstshash cookie
15448 # This is only useful makes sense if
15449 # balance rdp-cookie is not used
15450 stick-table type string size 204800
15451 stick on req.rdp_cookie(mstshash)
15452 server srv1 1.1.1.1:3389
15453 server srv1 1.1.1.2:3389
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015454
Willy Tarreau74ca5042013-06-11 23:12:07 +020015455 See also : "balance rdp-cookie", "persist rdp-cookie", "tcp-request" and the
15456 "req_rdp_cookie" ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015457
Willy Tarreau74ca5042013-06-11 23:12:07 +020015458req.rdp_cookie_cnt([name]) : integer
15459rdp_cookie_cnt([name]) : integer (deprecated)
15460 Tries to parse the request buffer as RDP protocol, then returns an integer
15461 corresponding to the number of RDP cookies found. If an optional cookie name
15462 is passed, only cookies matching this name are considered. This is mostly
15463 used in ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015464
Willy Tarreau74ca5042013-06-11 23:12:07 +020015465 ACL derivatives :
15466 req_rdp_cookie_cnt([<name>]) : integer match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015467
Alex Zorin4afdd132018-12-30 13:56:28 +110015468req.ssl_alpn : string
15469 Returns a string containing the values of the Application-Layer Protocol
15470 Negotiation (ALPN) TLS extension (RFC7301), sent by the client within the SSL
15471 ClientHello message. Note that this only applies to raw contents found in the
15472 request buffer and not to the contents deciphered via an SSL data layer, so
15473 this will not work with "bind" lines having the "ssl" option. This is useful
15474 in ACL to make a routing decision based upon the ALPN preferences of a TLS
15475 client, like in the example below.
15476
15477 Examples :
15478 # Wait for a client hello for at most 5 seconds
15479 tcp-request inspect-delay 5s
15480 tcp-request content accept if { req_ssl_hello_type 1 }
15481 use_backend bk_acme if { req_ssl.alpn acme-tls/1 }
15482 default_backend bk_default
15483
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020015484req.ssl_ec_ext : boolean
15485 Returns a boolean identifying if client sent the Supported Elliptic Curves
15486 Extension as defined in RFC4492, section 5.1. within the SSL ClientHello
Cyril Bonté307ee1e2015-09-28 23:16:06 +020015487 message. This can be used to present ECC compatible clients with EC
15488 certificate and to use RSA for all others, on the same IP address. Note that
15489 this only applies to raw contents found in the request buffer and not to
15490 contents deciphered via an SSL data layer, so this will not work with "bind"
15491 lines having the "ssl" option.
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020015492
Willy Tarreau74ca5042013-06-11 23:12:07 +020015493req.ssl_hello_type : integer
15494req_ssl_hello_type : integer (deprecated)
15495 Returns an integer value containing the type of the SSL hello message found
15496 in the request buffer if the buffer contains data that parse as a complete
15497 SSL (v3 or superior) client hello message. Note that this only applies to raw
15498 contents found in the request buffer and not to contents deciphered via an
15499 SSL data layer, so this will not work with "bind" lines having the "ssl"
15500 option. This is mostly used in ACL to detect presence of an SSL hello message
15501 that is supposed to contain an SSL session ID usable for stickiness.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015502
Willy Tarreau74ca5042013-06-11 23:12:07 +020015503req.ssl_sni : string
15504req_ssl_sni : string (deprecated)
15505 Returns a string containing the value of the Server Name TLS extension sent
15506 by a client in a TLS stream passing through the request buffer if the buffer
15507 contains data that parse as a complete SSL (v3 or superior) client hello
15508 message. Note that this only applies to raw contents found in the request
15509 buffer and not to contents deciphered via an SSL data layer, so this will not
15510 work with "bind" lines having the "ssl" option. SNI normally contains the
15511 name of the host the client tries to connect to (for recent browsers). SNI is
15512 useful for allowing or denying access to certain hosts when SSL/TLS is used
15513 by the client. This test was designed to be used with TCP request content
15514 inspection. If content switching is needed, it is recommended to first wait
15515 for a complete client hello (type 1), like in the example below. See also
15516 "ssl_fc_sni".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015517
Willy Tarreau74ca5042013-06-11 23:12:07 +020015518 ACL derivatives :
15519 req_ssl_sni : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015520
Willy Tarreau74ca5042013-06-11 23:12:07 +020015521 Examples :
15522 # Wait for a client hello for at most 5 seconds
15523 tcp-request inspect-delay 5s
15524 tcp-request content accept if { req_ssl_hello_type 1 }
15525 use_backend bk_allow if { req_ssl_sni -f allowed_sites }
15526 default_backend bk_sorry_page
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015527
Pradeep Jindalbb2acf52015-09-29 10:12:57 +053015528req.ssl_st_ext : integer
15529 Returns 0 if the client didn't send a SessionTicket TLS Extension (RFC5077)
15530 Returns 1 if the client sent SessionTicket TLS Extension
15531 Returns 2 if the client also sent non-zero length TLS SessionTicket
15532 Note that this only applies to raw contents found in the request buffer and
15533 not to contents deciphered via an SSL data layer, so this will not work with
15534 "bind" lines having the "ssl" option. This can for example be used to detect
15535 whether the client sent a SessionTicket or not and stick it accordingly, if
15536 no SessionTicket then stick on SessionID or don't stick as there's no server
15537 side state is there when SessionTickets are in use.
15538
Willy Tarreau74ca5042013-06-11 23:12:07 +020015539req.ssl_ver : integer
15540req_ssl_ver : integer (deprecated)
15541 Returns an integer value containing the version of the SSL/TLS protocol of a
15542 stream present in the request buffer. Both SSLv2 hello messages and SSLv3
15543 messages are supported. TLSv1 is announced as SSL version 3.1. The value is
15544 composed of the major version multiplied by 65536, added to the minor
15545 version. Note that this only applies to raw contents found in the request
15546 buffer and not to contents deciphered via an SSL data layer, so this will not
15547 work with "bind" lines having the "ssl" option. The ACL version of the test
Davor Ocelice9ed2812017-12-25 17:49:28 +010015548 matches against a decimal notation in the form MAJOR.MINOR (e.g. 3.1). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020015549 fetch is mostly used in ACL.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015550
Willy Tarreau74ca5042013-06-11 23:12:07 +020015551 ACL derivatives :
15552 req_ssl_ver : decimal match
Willy Tarreaud63335a2010-02-26 12:56:52 +010015553
Willy Tarreau47e8eba2013-09-11 23:28:46 +020015554res.len : integer
15555 Returns an integer value corresponding to the number of bytes present in the
15556 response buffer. This is mostly used in ACL. It is important to understand
15557 that this test does not return false as long as the buffer is changing. This
15558 means that a check with equality to zero will almost always immediately match
15559 at the beginning of the session, while a test for more data will wait for
15560 that data to come in and return false only when haproxy is certain that no
15561 more data will come in. This test was designed to be used with TCP response
15562 content inspection.
15563
Willy Tarreau74ca5042013-06-11 23:12:07 +020015564res.payload(<offset>,<length>) : binary
15565 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020015566 in the response buffer. As a special case, if the <length> argument is zero,
15567 the the whole buffer from <offset> to the end is extracted. This can be used
15568 with ACLs in order to check for the presence of some content in a buffer at
15569 any location.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015570
Willy Tarreau74ca5042013-06-11 23:12:07 +020015571res.payload_lv(<offset1>,<length>[,<offset2>]) : binary
15572 This extracts a binary block whose size is specified at <offset1> for <length>
15573 bytes, and which starts at <offset2> if specified or just after the length in
15574 the response buffer. The <offset2> parameter also supports relative offsets
15575 if prepended with a '+' or '-' sign.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015576
Willy Tarreau74ca5042013-06-11 23:12:07 +020015577 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015578
Willy Tarreau971f7b62015-09-29 14:06:59 +020015579res.ssl_hello_type : integer
15580rep_ssl_hello_type : integer (deprecated)
15581 Returns an integer value containing the type of the SSL hello message found
15582 in the response buffer if the buffer contains data that parses as a complete
15583 SSL (v3 or superior) hello message. Note that this only applies to raw
15584 contents found in the response buffer and not to contents deciphered via an
15585 SSL data layer, so this will not work with "server" lines having the "ssl"
15586 option. This is mostly used in ACL to detect presence of an SSL hello message
15587 that is supposed to contain an SSL session ID usable for stickiness.
15588
Willy Tarreau74ca5042013-06-11 23:12:07 +020015589wait_end : boolean
15590 This fetch either returns true when the inspection period is over, or does
15591 not fetch. It is only used in ACLs, in conjunction with content analysis to
Davor Ocelice9ed2812017-12-25 17:49:28 +010015592 avoid returning a wrong verdict early. It may also be used to delay some
Willy Tarreau74ca5042013-06-11 23:12:07 +020015593 actions, such as a delayed reject for some special addresses. Since it either
15594 stops the rules evaluation or immediately returns true, it is recommended to
Davor Ocelice9ed2812017-12-25 17:49:28 +010015595 use this acl as the last one in a rule. Please note that the default ACL
Willy Tarreau74ca5042013-06-11 23:12:07 +020015596 "WAIT_END" is always usable without prior declaration. This test was designed
15597 to be used with TCP request content inspection.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015598
Willy Tarreau74ca5042013-06-11 23:12:07 +020015599 Examples :
15600 # delay every incoming request by 2 seconds
15601 tcp-request inspect-delay 2s
15602 tcp-request content accept if WAIT_END
Willy Tarreaud63335a2010-02-26 12:56:52 +010015603
Willy Tarreau74ca5042013-06-11 23:12:07 +020015604 # don't immediately tell bad guys they are rejected
15605 tcp-request inspect-delay 10s
15606 acl goodguys src 10.0.0.0/24
15607 acl badguys src 10.0.1.0/24
15608 tcp-request content accept if goodguys
15609 tcp-request content reject if badguys WAIT_END
15610 tcp-request content reject
15611
15612
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200156137.3.6. Fetching HTTP samples (Layer 7)
Willy Tarreau74ca5042013-06-11 23:12:07 +020015614--------------------------------------
15615
15616It is possible to fetch samples from HTTP contents, requests and responses.
15617This application layer is also called layer 7. It is only possible to fetch the
15618data in this section when a full HTTP request or response has been parsed from
15619its respective request or response buffer. This is always the case with all
15620HTTP specific rules and for sections running with "mode http". When using TCP
15621content inspection, it may be necessary to support an inspection delay in order
15622to let the request or response come in first. These fetches may require a bit
15623more CPU resources than the layer 4 ones, but not much since the request and
15624response are indexed.
15625
15626base : string
15627 This returns the concatenation of the first Host header and the path part of
15628 the request, which starts at the first slash and ends before the question
15629 mark. It can be useful in virtual hosted environments to detect URL abuses as
15630 well as to improve shared caches efficiency. Using this with a limited size
15631 stick table also allows one to collect statistics about most commonly
15632 requested objects by host/path. With ACLs it can allow simple content
15633 switching rules involving the host and the path at the same time, such as
15634 "www.example.com/favicon.ico". See also "path" and "uri".
15635
15636 ACL derivatives :
15637 base : exact string match
15638 base_beg : prefix match
15639 base_dir : subdir match
15640 base_dom : domain match
15641 base_end : suffix match
15642 base_len : length match
15643 base_reg : regex match
15644 base_sub : substring match
15645
15646base32 : integer
15647 This returns a 32-bit hash of the value returned by the "base" fetch method
15648 above. This is useful to track per-URL activity on high traffic sites without
15649 having to store all URLs. Instead a shorter hash is stored, saving a lot of
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015650 memory. The output type is an unsigned integer. The hash function used is
15651 SDBM with full avalanche on the output. Technically, base32 is exactly equal
15652 to "base,sdbm(1)".
Willy Tarreau74ca5042013-06-11 23:12:07 +020015653
15654base32+src : binary
15655 This returns the concatenation of the base32 fetch above and the src fetch
15656 below. The resulting type is of type binary, with a size of 8 or 20 bytes
15657 depending on the source address family. This can be used to track per-IP,
15658 per-URL counters.
15659
William Lallemand65ad6e12014-01-31 15:08:02 +010015660capture.req.hdr(<idx>) : string
15661 This extracts the content of the header captured by the "capture request
15662 header", idx is the position of the capture keyword in the configuration.
15663 The first entry is an index of 0. See also: "capture request header".
15664
15665capture.req.method : string
15666 This extracts the METHOD of an HTTP request. It can be used in both request
15667 and response. Unlike "method", it can be used in both request and response
15668 because it's allocated.
15669
15670capture.req.uri : string
15671 This extracts the request's URI, which starts at the first slash and ends
15672 before the first space in the request (without the host part). Unlike "path"
15673 and "url", it can be used in both request and response because it's
15674 allocated.
15675
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020015676capture.req.ver : string
15677 This extracts the request's HTTP version and returns either "HTTP/1.0" or
15678 "HTTP/1.1". Unlike "req.ver", it can be used in both request, response, and
15679 logs because it relies on a persistent flag.
15680
William Lallemand65ad6e12014-01-31 15:08:02 +010015681capture.res.hdr(<idx>) : string
15682 This extracts the content of the header captured by the "capture response
15683 header", idx is the position of the capture keyword in the configuration.
15684 The first entry is an index of 0.
15685 See also: "capture response header"
15686
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020015687capture.res.ver : string
15688 This extracts the response's HTTP version and returns either "HTTP/1.0" or
15689 "HTTP/1.1". Unlike "res.ver", it can be used in logs because it relies on a
15690 persistent flag.
15691
Willy Tarreaua5910cc2015-05-02 00:46:08 +020015692req.body : binary
15693 This returns the HTTP request's available body as a block of data. It
15694 requires that the request body has been buffered made available using
15695 "option http-buffer-request". In case of chunked-encoded body, currently only
15696 the first chunk is analyzed.
15697
Thierry FOURNIER9826c772015-05-20 15:50:54 +020015698req.body_param([<name>) : string
15699 This fetch assumes that the body of the POST request is url-encoded. The user
15700 can check if the "content-type" contains the value
15701 "application/x-www-form-urlencoded". This extracts the first occurrence of the
15702 parameter <name> in the body, which ends before '&'. The parameter name is
15703 case-sensitive. If no name is given, any parameter will match, and the first
15704 one will be returned. The result is a string corresponding to the value of the
15705 parameter <name> as presented in the request body (no URL decoding is
15706 performed). Note that the ACL version of this fetch iterates over multiple
15707 parameters and will iteratively report all parameters values if no name is
15708 given.
15709
Willy Tarreaua5910cc2015-05-02 00:46:08 +020015710req.body_len : integer
15711 This returns the length of the HTTP request's available body in bytes. It may
15712 be lower than the advertised length if the body is larger than the buffer. It
15713 requires that the request body has been buffered made available using
15714 "option http-buffer-request".
15715
15716req.body_size : integer
15717 This returns the advertised length of the HTTP request's body in bytes. It
15718 will represent the advertised Content-Length header, or the size of the first
15719 chunk in case of chunked encoding. In order to parse the chunks, it requires
15720 that the request body has been buffered made available using
15721 "option http-buffer-request".
15722
Willy Tarreau74ca5042013-06-11 23:12:07 +020015723req.cook([<name>]) : string
15724cook([<name>]) : string (deprecated)
15725 This extracts the last occurrence of the cookie name <name> on a "Cookie"
15726 header line from the request, and returns its value as string. If no name is
15727 specified, the first cookie value is returned. When used with ACLs, all
15728 matching cookies are evaluated. Spaces around the name and the value are
15729 ignored as requested by the Cookie header specification (RFC6265). The cookie
15730 name is case-sensitive. Empty cookies are valid, so an empty cookie may very
15731 well return an empty value if it is present. Use the "found" match to detect
15732 presence. Use the res.cook() variant for response cookies sent by the server.
15733
15734 ACL derivatives :
15735 cook([<name>]) : exact string match
15736 cook_beg([<name>]) : prefix match
15737 cook_dir([<name>]) : subdir match
15738 cook_dom([<name>]) : domain match
15739 cook_end([<name>]) : suffix match
15740 cook_len([<name>]) : length match
15741 cook_reg([<name>]) : regex match
15742 cook_sub([<name>]) : substring match
Willy Tarreaud63335a2010-02-26 12:56:52 +010015743
Willy Tarreau74ca5042013-06-11 23:12:07 +020015744req.cook_cnt([<name>]) : integer
15745cook_cnt([<name>]) : integer (deprecated)
15746 Returns an integer value representing the number of occurrences of the cookie
15747 <name> in the request, or all cookies if <name> is not specified.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015748
Willy Tarreau74ca5042013-06-11 23:12:07 +020015749req.cook_val([<name>]) : integer
15750cook_val([<name>]) : integer (deprecated)
15751 This extracts the last occurrence of the cookie name <name> on a "Cookie"
15752 header line from the request, and converts its value to an integer which is
15753 returned. If no name is specified, the first cookie value is returned. When
15754 used in ACLs, all matching names are iterated over until a value matches.
Willy Tarreau0e698542011-09-16 08:32:32 +020015755
Willy Tarreau74ca5042013-06-11 23:12:07 +020015756cookie([<name>]) : string (deprecated)
15757 This extracts the last occurrence of the cookie name <name> on a "Cookie"
15758 header line from the request, or a "Set-Cookie" header from the response, and
15759 returns its value as a string. A typical use is to get multiple clients
15760 sharing a same profile use the same server. This can be similar to what
Willy Tarreau294d0f02015-08-10 19:40:12 +020015761 "appsession" did with the "request-learn" statement, but with support for
Willy Tarreau74ca5042013-06-11 23:12:07 +020015762 multi-peer synchronization and state keeping across restarts. If no name is
15763 specified, the first cookie value is returned. This fetch should not be used
15764 anymore and should be replaced by req.cook() or res.cook() instead as it
15765 ambiguously uses the direction based on the context where it is used.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015766
Willy Tarreau74ca5042013-06-11 23:12:07 +020015767hdr([<name>[,<occ>]]) : string
15768 This is equivalent to req.hdr() when used on requests, and to res.hdr() when
15769 used on responses. Please refer to these respective fetches for more details.
15770 In case of doubt about the fetch direction, please use the explicit ones.
15771 Note that contrary to the hdr() sample fetch method, the hdr_* ACL keywords
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015772 unambiguously apply to the request headers.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015773
Willy Tarreau74ca5042013-06-11 23:12:07 +020015774req.fhdr(<name>[,<occ>]) : string
15775 This extracts the last occurrence of header <name> in an HTTP request. When
15776 used from an ACL, all occurrences are iterated over until a match is found.
15777 Optionally, a specific occurrence might be specified as a position number.
15778 Positive values indicate a position from the first occurrence, with 1 being
15779 the first one. Negative values indicate positions relative to the last one,
15780 with -1 being the last one. It differs from req.hdr() in that any commas
15781 present in the value are returned and are not used as delimiters. This is
15782 sometimes useful with headers such as User-Agent.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015783
Willy Tarreau74ca5042013-06-11 23:12:07 +020015784req.fhdr_cnt([<name>]) : integer
15785 Returns an integer value representing the number of occurrences of request
15786 header field name <name>, or the total number of header fields if <name> is
15787 not specified. Contrary to its req.hdr_cnt() cousin, this function returns
15788 the number of full line headers and does not stop on commas.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015789
Willy Tarreau74ca5042013-06-11 23:12:07 +020015790req.hdr([<name>[,<occ>]]) : string
15791 This extracts the last occurrence of header <name> in an HTTP request. When
15792 used from an ACL, all occurrences are iterated over until a match is found.
15793 Optionally, a specific occurrence might be specified as a position number.
15794 Positive values indicate a position from the first occurrence, with 1 being
15795 the first one. Negative values indicate positions relative to the last one,
15796 with -1 being the last one. A typical use is with the X-Forwarded-For header
15797 once converted to IP, associated with an IP stick-table. The function
15798 considers any comma as a delimiter for distinct values. If full-line headers
Lukas Tribus23953682017-04-28 13:24:30 +000015799 are desired instead, use req.fhdr(). Please carefully check RFC7231 to know
Willy Tarreau74ca5042013-06-11 23:12:07 +020015800 how certain headers are supposed to be parsed. Also, some of them are case
Davor Ocelice9ed2812017-12-25 17:49:28 +010015801 insensitive (e.g. Connection).
Willy Tarreaud63335a2010-02-26 12:56:52 +010015802
Willy Tarreau74ca5042013-06-11 23:12:07 +020015803 ACL derivatives :
15804 hdr([<name>[,<occ>]]) : exact string match
15805 hdr_beg([<name>[,<occ>]]) : prefix match
15806 hdr_dir([<name>[,<occ>]]) : subdir match
15807 hdr_dom([<name>[,<occ>]]) : domain match
15808 hdr_end([<name>[,<occ>]]) : suffix match
15809 hdr_len([<name>[,<occ>]]) : length match
15810 hdr_reg([<name>[,<occ>]]) : regex match
15811 hdr_sub([<name>[,<occ>]]) : substring match
15812
15813req.hdr_cnt([<name>]) : integer
15814hdr_cnt([<header>]) : integer (deprecated)
15815 Returns an integer value representing the number of occurrences of request
15816 header field name <name>, or the total number of header field values if
15817 <name> is not specified. It is important to remember that one header line may
15818 count as several headers if it has several values. The function considers any
15819 comma as a delimiter for distinct values. If full-line headers are desired
15820 instead, req.fhdr_cnt() should be used instead. With ACLs, it can be used to
15821 detect presence, absence or abuse of a specific header, as well as to block
15822 request smuggling attacks by rejecting requests which contain more than one
15823 of certain headers. See "req.hdr" for more information on header matching.
15824
15825req.hdr_ip([<name>[,<occ>]]) : ip
15826hdr_ip([<name>[,<occ>]]) : ip (deprecated)
15827 This extracts the last occurrence of header <name> in an HTTP request,
15828 converts it to an IPv4 or IPv6 address and returns this address. When used
15829 with ACLs, all occurrences are checked, and if <name> is omitted, every value
15830 of every header is checked. Optionally, a specific occurrence might be
15831 specified as a position number. Positive values indicate a position from the
Davor Ocelice9ed2812017-12-25 17:49:28 +010015832 first occurrence, with 1 being the first one. Negative values indicate
Willy Tarreau74ca5042013-06-11 23:12:07 +020015833 positions relative to the last one, with -1 being the last one. A typical use
15834 is with the X-Forwarded-For and X-Client-IP headers.
15835
15836req.hdr_val([<name>[,<occ>]]) : integer
15837hdr_val([<name>[,<occ>]]) : integer (deprecated)
15838 This extracts the last occurrence of header <name> in an HTTP request, and
15839 converts it to an integer value. When used with ACLs, all occurrences are
15840 checked, and if <name> is omitted, every value of every header is checked.
15841 Optionally, a specific occurrence might be specified as a position number.
15842 Positive values indicate a position from the first occurrence, with 1 being
15843 the first one. Negative values indicate positions relative to the last one,
15844 with -1 being the last one. A typical use is with the X-Forwarded-For header.
15845
15846http_auth(<userlist>) : boolean
15847 Returns a boolean indicating whether the authentication data received from
15848 the client match a username & password stored in the specified userlist. This
15849 fetch function is not really useful outside of ACLs. Currently only http
15850 basic auth is supported.
15851
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010015852http_auth_group(<userlist>) : string
15853 Returns a string corresponding to the user name found in the authentication
15854 data received from the client if both the user name and password are valid
15855 according to the specified userlist. The main purpose is to use it in ACLs
15856 where it is then checked whether the user belongs to any group within a list.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015857 This fetch function is not really useful outside of ACLs. Currently only http
15858 basic auth is supported.
15859
15860 ACL derivatives :
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010015861 http_auth_group(<userlist>) : group ...
15862 Returns true when the user extracted from the request and whose password is
15863 valid according to the specified userlist belongs to at least one of the
15864 groups.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015865
15866http_first_req : boolean
Willy Tarreau7f18e522010-10-22 20:04:13 +020015867 Returns true when the request being processed is the first one of the
15868 connection. This can be used to add or remove headers that may be missing
Willy Tarreau74ca5042013-06-11 23:12:07 +020015869 from some requests when a request is not the first one, or to help grouping
15870 requests in the logs.
Willy Tarreau7f18e522010-10-22 20:04:13 +020015871
Willy Tarreau74ca5042013-06-11 23:12:07 +020015872method : integer + string
15873 Returns an integer value corresponding to the method in the HTTP request. For
15874 example, "GET" equals 1 (check sources to establish the matching). Value 9
15875 means "other method" and may be converted to a string extracted from the
15876 stream. This should not be used directly as a sample, this is only meant to
15877 be used from ACLs, which transparently convert methods from patterns to these
15878 integer + string values. Some predefined ACL already check for most common
15879 methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015880
Willy Tarreau74ca5042013-06-11 23:12:07 +020015881 ACL derivatives :
15882 method : case insensitive method match
Willy Tarreau6a06a402007-07-15 20:15:28 +020015883
Willy Tarreau74ca5042013-06-11 23:12:07 +020015884 Example :
15885 # only accept GET and HEAD requests
15886 acl valid_method method GET HEAD
15887 http-request deny if ! valid_method
Willy Tarreau6a06a402007-07-15 20:15:28 +020015888
Willy Tarreau74ca5042013-06-11 23:12:07 +020015889path : string
15890 This extracts the request's URL path, which starts at the first slash and
15891 ends before the question mark (without the host part). A typical use is with
15892 prefetch-capable caches, and with portals which need to aggregate multiple
15893 information from databases and keep them in caches. Note that with outgoing
15894 caches, it would be wiser to use "url" instead. With ACLs, it's typically
Davor Ocelice9ed2812017-12-25 17:49:28 +010015895 used to match exact file names (e.g. "/login.php"), or directory parts using
Willy Tarreau74ca5042013-06-11 23:12:07 +020015896 the derivative forms. See also the "url" and "base" fetch methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015897
Willy Tarreau74ca5042013-06-11 23:12:07 +020015898 ACL derivatives :
15899 path : exact string match
15900 path_beg : prefix match
15901 path_dir : subdir match
15902 path_dom : domain match
15903 path_end : suffix match
15904 path_len : length match
15905 path_reg : regex match
15906 path_sub : substring match
Willy Tarreau6a06a402007-07-15 20:15:28 +020015907
Willy Tarreau49ad95c2015-01-19 15:06:26 +010015908query : string
15909 This extracts the request's query string, which starts after the first
15910 question mark. If no question mark is present, this fetch returns nothing. If
15911 a question mark is present but nothing follows, it returns an empty string.
15912 This means it's possible to easily know whether a query string is present
Tim Düsterhus4896c442016-11-29 02:15:19 +010015913 using the "found" matching method. This fetch is the complement of "path"
Willy Tarreau49ad95c2015-01-19 15:06:26 +010015914 which stops before the question mark.
15915
Willy Tarreaueb27ec72015-02-20 13:55:29 +010015916req.hdr_names([<delim>]) : string
15917 This builds a string made from the concatenation of all header names as they
15918 appear in the request when the rule is evaluated. The default delimiter is
15919 the comma (',') but it may be overridden as an optional argument <delim>. In
15920 this case, only the first character of <delim> is considered.
15921
Willy Tarreau74ca5042013-06-11 23:12:07 +020015922req.ver : string
15923req_ver : string (deprecated)
15924 Returns the version string from the HTTP request, for example "1.1". This can
15925 be useful for logs, but is mostly there for ACL. Some predefined ACL already
15926 check for versions 1.0 and 1.1.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015927
Willy Tarreau74ca5042013-06-11 23:12:07 +020015928 ACL derivatives :
15929 req_ver : exact string match
Willy Tarreau0e698542011-09-16 08:32:32 +020015930
Willy Tarreau74ca5042013-06-11 23:12:07 +020015931res.comp : boolean
15932 Returns the boolean "true" value if the response has been compressed by
15933 HAProxy, otherwise returns boolean "false". This may be used to add
15934 information in the logs.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015935
Willy Tarreau74ca5042013-06-11 23:12:07 +020015936res.comp_algo : string
15937 Returns a string containing the name of the algorithm used if the response
15938 was compressed by HAProxy, for example : "deflate". This may be used to add
15939 some information in the logs.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015940
Willy Tarreau74ca5042013-06-11 23:12:07 +020015941res.cook([<name>]) : string
15942scook([<name>]) : string (deprecated)
15943 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
15944 header line from the response, and returns its value as string. If no name is
15945 specified, the first cookie value is returned.
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020015946
Willy Tarreau74ca5042013-06-11 23:12:07 +020015947 ACL derivatives :
15948 scook([<name>] : exact string match
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020015949
Willy Tarreau74ca5042013-06-11 23:12:07 +020015950res.cook_cnt([<name>]) : integer
15951scook_cnt([<name>]) : integer (deprecated)
15952 Returns an integer value representing the number of occurrences of the cookie
15953 <name> in the response, or all cookies if <name> is not specified. This is
15954 mostly useful when combined with ACLs to detect suspicious responses.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015955
Willy Tarreau74ca5042013-06-11 23:12:07 +020015956res.cook_val([<name>]) : integer
15957scook_val([<name>]) : integer (deprecated)
15958 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
15959 header line from the response, and converts its value to an integer which is
15960 returned. If no name is specified, the first cookie value is returned.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015961
Willy Tarreau74ca5042013-06-11 23:12:07 +020015962res.fhdr([<name>[,<occ>]]) : string
15963 This extracts the last occurrence of header <name> in an HTTP response, or of
15964 the last header if no <name> is specified. Optionally, a specific occurrence
15965 might be specified as a position number. Positive values indicate a position
15966 from the first occurrence, with 1 being the first one. Negative values
15967 indicate positions relative to the last one, with -1 being the last one. It
15968 differs from res.hdr() in that any commas present in the value are returned
15969 and are not used as delimiters. If this is not desired, the res.hdr() fetch
15970 should be used instead. This is sometimes useful with headers such as Date or
15971 Expires.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015972
Willy Tarreau74ca5042013-06-11 23:12:07 +020015973res.fhdr_cnt([<name>]) : integer
15974 Returns an integer value representing the number of occurrences of response
15975 header field name <name>, or the total number of header fields if <name> is
15976 not specified. Contrary to its res.hdr_cnt() cousin, this function returns
15977 the number of full line headers and does not stop on commas. If this is not
15978 desired, the res.hdr_cnt() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015979
Willy Tarreau74ca5042013-06-11 23:12:07 +020015980res.hdr([<name>[,<occ>]]) : string
15981shdr([<name>[,<occ>]]) : string (deprecated)
15982 This extracts the last occurrence of header <name> in an HTTP response, or of
15983 the last header if no <name> is specified. Optionally, a specific occurrence
15984 might be specified as a position number. Positive values indicate a position
15985 from the first occurrence, with 1 being the first one. Negative values
15986 indicate positions relative to the last one, with -1 being the last one. This
15987 can be useful to learn some data into a stick-table. The function considers
15988 any comma as a delimiter for distinct values. If this is not desired, the
15989 res.fhdr() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015990
Willy Tarreau74ca5042013-06-11 23:12:07 +020015991 ACL derivatives :
15992 shdr([<name>[,<occ>]]) : exact string match
15993 shdr_beg([<name>[,<occ>]]) : prefix match
15994 shdr_dir([<name>[,<occ>]]) : subdir match
15995 shdr_dom([<name>[,<occ>]]) : domain match
15996 shdr_end([<name>[,<occ>]]) : suffix match
15997 shdr_len([<name>[,<occ>]]) : length match
15998 shdr_reg([<name>[,<occ>]]) : regex match
15999 shdr_sub([<name>[,<occ>]]) : substring match
16000
16001res.hdr_cnt([<name>]) : integer
16002shdr_cnt([<name>]) : integer (deprecated)
16003 Returns an integer value representing the number of occurrences of response
16004 header field name <name>, or the total number of header fields if <name> is
16005 not specified. The function considers any comma as a delimiter for distinct
16006 values. If this is not desired, the res.fhdr_cnt() fetch should be used
16007 instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016008
Willy Tarreau74ca5042013-06-11 23:12:07 +020016009res.hdr_ip([<name>[,<occ>]]) : ip
16010shdr_ip([<name>[,<occ>]]) : ip (deprecated)
16011 This extracts the last occurrence of header <name> in an HTTP response,
16012 convert it to an IPv4 or IPv6 address and returns this address. Optionally, a
16013 specific occurrence might be specified as a position number. Positive values
16014 indicate a position from the first occurrence, with 1 being the first one.
16015 Negative values indicate positions relative to the last one, with -1 being
16016 the last one. This can be useful to learn some data into a stick table.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016017
Willy Tarreaueb27ec72015-02-20 13:55:29 +010016018res.hdr_names([<delim>]) : string
16019 This builds a string made from the concatenation of all header names as they
16020 appear in the response when the rule is evaluated. The default delimiter is
16021 the comma (',') but it may be overridden as an optional argument <delim>. In
16022 this case, only the first character of <delim> is considered.
16023
Willy Tarreau74ca5042013-06-11 23:12:07 +020016024res.hdr_val([<name>[,<occ>]]) : integer
16025shdr_val([<name>[,<occ>]]) : integer (deprecated)
16026 This extracts the last occurrence of header <name> in an HTTP response, and
16027 converts it to an integer value. Optionally, a specific occurrence might be
16028 specified as a position number. Positive values indicate a position from the
16029 first occurrence, with 1 being the first one. Negative values indicate
16030 positions relative to the last one, with -1 being the last one. This can be
16031 useful to learn some data into a stick table.
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010016032
Willy Tarreau74ca5042013-06-11 23:12:07 +020016033res.ver : string
16034resp_ver : string (deprecated)
16035 Returns the version string from the HTTP response, for example "1.1". This
16036 can be useful for logs, but is mostly there for ACL.
Willy Tarreau0e698542011-09-16 08:32:32 +020016037
Willy Tarreau74ca5042013-06-11 23:12:07 +020016038 ACL derivatives :
16039 resp_ver : exact string match
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010016040
Willy Tarreau74ca5042013-06-11 23:12:07 +020016041set-cookie([<name>]) : string (deprecated)
16042 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
16043 header line from the response and uses the corresponding value to match. This
Willy Tarreau294d0f02015-08-10 19:40:12 +020016044 can be comparable to what "appsession" did with default options, but with
Willy Tarreau74ca5042013-06-11 23:12:07 +020016045 support for multi-peer synchronization and state keeping across restarts.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010016046
Willy Tarreau74ca5042013-06-11 23:12:07 +020016047 This fetch function is deprecated and has been superseded by the "res.cook"
16048 fetch. This keyword will disappear soon.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010016049
Willy Tarreau74ca5042013-06-11 23:12:07 +020016050status : integer
16051 Returns an integer containing the HTTP status code in the HTTP response, for
16052 example, 302. It is mostly used within ACLs and integer ranges, for example,
16053 to remove any Location header if the response is not a 3xx.
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016054
Thierry Fournier0e00dca2016-04-07 15:47:40 +020016055unique-id : string
16056 Returns the unique-id attached to the request. The directive
16057 "unique-id-format" must be set. If it is not set, the unique-id sample fetch
16058 fails. Note that the unique-id is usually used with HTTP requests, however this
16059 sample fetch can be used with other protocols. Obviously, if it is used with
16060 other protocols than HTTP, the unique-id-format directive must not contain
16061 HTTP parts. See: unique-id-format and unique-id-header
16062
Willy Tarreau74ca5042013-06-11 23:12:07 +020016063url : string
16064 This extracts the request's URL as presented in the request. A typical use is
16065 with prefetch-capable caches, and with portals which need to aggregate
16066 multiple information from databases and keep them in caches. With ACLs, using
16067 "path" is preferred over using "url", because clients may send a full URL as
16068 is normally done with proxies. The only real use is to match "*" which does
16069 not match in "path", and for which there is already a predefined ACL. See
16070 also "path" and "base".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016071
Willy Tarreau74ca5042013-06-11 23:12:07 +020016072 ACL derivatives :
16073 url : exact string match
16074 url_beg : prefix match
16075 url_dir : subdir match
16076 url_dom : domain match
16077 url_end : suffix match
16078 url_len : length match
16079 url_reg : regex match
16080 url_sub : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016081
Willy Tarreau74ca5042013-06-11 23:12:07 +020016082url_ip : ip
16083 This extracts the IP address from the request's URL when the host part is
16084 presented as an IP address. Its use is very limited. For instance, a
16085 monitoring system might use this field as an alternative for the source IP in
16086 order to test what path a given source address would follow, or to force an
16087 entry in a table for a given source address. With ACLs it can be used to
16088 restrict access to certain systems through a proxy, for example when combined
16089 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016090
Willy Tarreau74ca5042013-06-11 23:12:07 +020016091url_port : integer
16092 This extracts the port part from the request's URL. Note that if the port is
16093 not specified in the request, port 80 is assumed. With ACLs it can be used to
16094 restrict access to certain systems through a proxy, for example when combined
16095 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016096
Willy Tarreau1ede1da2015-05-07 16:06:18 +020016097urlp([<name>[,<delim>]]) : string
16098url_param([<name>[,<delim>]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020016099 This extracts the first occurrence of the parameter <name> in the query
16100 string, which begins after either '?' or <delim>, and which ends before '&',
Willy Tarreau1ede1da2015-05-07 16:06:18 +020016101 ';' or <delim>. The parameter name is case-sensitive. If no name is given,
16102 any parameter will match, and the first one will be returned. The result is
16103 a string corresponding to the value of the parameter <name> as presented in
16104 the request (no URL decoding is performed). This can be used for session
Willy Tarreau74ca5042013-06-11 23:12:07 +020016105 stickiness based on a client ID, to extract an application cookie passed as a
16106 URL parameter, or in ACLs to apply some checks. Note that the ACL version of
Willy Tarreau1ede1da2015-05-07 16:06:18 +020016107 this fetch iterates over multiple parameters and will iteratively report all
16108 parameters values if no name is given
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016109
Willy Tarreau74ca5042013-06-11 23:12:07 +020016110 ACL derivatives :
16111 urlp(<name>[,<delim>]) : exact string match
16112 urlp_beg(<name>[,<delim>]) : prefix match
16113 urlp_dir(<name>[,<delim>]) : subdir match
16114 urlp_dom(<name>[,<delim>]) : domain match
16115 urlp_end(<name>[,<delim>]) : suffix match
16116 urlp_len(<name>[,<delim>]) : length match
16117 urlp_reg(<name>[,<delim>]) : regex match
16118 urlp_sub(<name>[,<delim>]) : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016119
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016120
Willy Tarreau74ca5042013-06-11 23:12:07 +020016121 Example :
16122 # match http://example.com/foo?PHPSESSIONID=some_id
16123 stick on urlp(PHPSESSIONID)
16124 # match http://example.com/foo;JSESSIONID=some_id
16125 stick on urlp(JSESSIONID,;)
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016126
Jarno Huuskonen676f6222017-03-30 09:19:45 +030016127urlp_val([<name>[,<delim>]]) : integer
Willy Tarreau74ca5042013-06-11 23:12:07 +020016128 See "urlp" above. This one extracts the URL parameter <name> in the request
16129 and converts it to an integer value. This can be used for session stickiness
16130 based on a user ID for example, or with ACLs to match a page number or price.
Willy Tarreaua9fddca2012-07-31 07:51:48 +020016131
Dragan Dosen0070cd52016-06-16 12:19:49 +020016132url32 : integer
16133 This returns a 32-bit hash of the value obtained by concatenating the first
16134 Host header and the whole URL including parameters (not only the path part of
16135 the request, as in the "base32" fetch above). This is useful to track per-URL
16136 activity. A shorter hash is stored, saving a lot of memory. The output type
16137 is an unsigned integer.
16138
16139url32+src : binary
16140 This returns the concatenation of the "url32" fetch and the "src" fetch. The
16141 resulting type is of type binary, with a size of 8 or 20 bytes depending on
16142 the source address family. This can be used to track per-IP, per-URL counters.
16143
Willy Tarreau198a7442008-01-17 12:05:32 +010016144
Willy Tarreau74ca5042013-06-11 23:12:07 +0200161457.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016146---------------------
Willy Tarreauced27012008-01-17 20:35:34 +010016147
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016148Some predefined ACLs are hard-coded so that they do not have to be declared in
16149every frontend which needs them. They all have their names in upper case in
Patrick Mézard2382ad62010-05-09 10:43:32 +020016150order to avoid confusion. Their equivalence is provided below.
Willy Tarreauced27012008-01-17 20:35:34 +010016151
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016152ACL name Equivalent to Usage
16153---------------+-----------------------------+---------------------------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016154FALSE always_false never match
Willy Tarreau2492d5b2009-07-11 00:06:00 +020016155HTTP req_proto_http match if protocol is valid HTTP
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016156HTTP_1.0 req_ver 1.0 match HTTP version 1.0
16157HTTP_1.1 req_ver 1.1 match HTTP version 1.1
Willy Tarreaud63335a2010-02-26 12:56:52 +010016158HTTP_CONTENT hdr_val(content-length) gt 0 match an existing content-length
16159HTTP_URL_ABS url_reg ^[^/:]*:// match absolute URL with scheme
16160HTTP_URL_SLASH url_beg / match URL beginning with "/"
16161HTTP_URL_STAR url * match URL equal to "*"
16162LOCALHOST src 127.0.0.1/8 match connection from local host
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016163METH_CONNECT method CONNECT match HTTP CONNECT method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020016164METH_DELETE method DELETE match HTTP DELETE method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016165METH_GET method GET HEAD match HTTP GET or HEAD method
16166METH_HEAD method HEAD match HTTP HEAD method
16167METH_OPTIONS method OPTIONS match HTTP OPTIONS method
16168METH_POST method POST match HTTP POST method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020016169METH_PUT method PUT match HTTP PUT method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016170METH_TRACE method TRACE match HTTP TRACE method
Emeric Brunbede3d02009-06-30 17:54:00 +020016171RDP_COOKIE req_rdp_cookie_cnt gt 0 match presence of an RDP cookie
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016172REQ_CONTENT req_len gt 0 match data in the request buffer
Willy Tarreaud63335a2010-02-26 12:56:52 +010016173TRUE always_true always match
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016174WAIT_END wait_end wait for end of content analysis
16175---------------+-----------------------------+---------------------------------
Willy Tarreauced27012008-01-17 20:35:34 +010016176
Willy Tarreaub937b7e2010-01-12 15:27:54 +010016177
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200161788. Logging
16179----------
Willy Tarreau844e3c52008-01-11 16:28:18 +010016180
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016181One of HAProxy's strong points certainly lies is its precise logs. It probably
16182provides the finest level of information available for such a product, which is
16183very important for troubleshooting complex environments. Standard information
16184provided in logs include client ports, TCP/HTTP state timers, precise session
16185state at termination and precise termination cause, information about decisions
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010016186to direct traffic to a server, and of course the ability to capture arbitrary
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016187headers.
16188
16189In order to improve administrators reactivity, it offers a great transparency
16190about encountered problems, both internal and external, and it is possible to
16191send logs to different sources at the same time with different level filters :
16192
16193 - global process-level logs (system errors, start/stop, etc..)
16194 - per-instance system and internal errors (lack of resource, bugs, ...)
16195 - per-instance external troubles (servers up/down, max connections)
16196 - per-instance activity (client connections), either at the establishment or
16197 at the termination.
Davor Ocelice9ed2812017-12-25 17:49:28 +010016198 - per-request control of log-level, e.g.
Jim Freeman9e8714b2015-05-26 09:16:34 -060016199 http-request set-log-level silent if sensitive_request
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016200
16201The ability to distribute different levels of logs to different log servers
16202allow several production teams to interact and to fix their problems as soon
16203as possible. For example, the system team might monitor system-wide errors,
16204while the application team might be monitoring the up/down for their servers in
16205real time, and the security team might analyze the activity logs with one hour
16206delay.
16207
16208
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200162098.1. Log levels
16210---------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016211
Simon Hormandf791f52011-05-29 15:01:10 +090016212TCP and HTTP connections can be logged with information such as the date, time,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016213source IP address, destination address, connection duration, response times,
Simon Hormandf791f52011-05-29 15:01:10 +090016214HTTP request, HTTP return code, number of bytes transmitted, conditions
16215in which the session ended, and even exchanged cookies values. For example
16216track a particular user's problems. All messages may be sent to up to two
16217syslog servers. Check the "log" keyword in section 4.2 for more information
16218about log facilities.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016219
16220
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200162218.2. Log formats
16222----------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016223
William Lallemand48940402012-01-30 16:47:22 +010016224HAProxy supports 5 log formats. Several fields are common between these formats
Simon Hormandf791f52011-05-29 15:01:10 +090016225and will be detailed in the following sections. A few of them may vary
16226slightly with the configuration, due to indicators specific to certain
16227options. The supported formats are as follows :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016228
16229 - the default format, which is very basic and very rarely used. It only
16230 provides very basic information about the incoming connection at the moment
16231 it is accepted : source IP:port, destination IP:port, and frontend-name.
16232 This mode will eventually disappear so it will not be described to great
16233 extents.
16234
16235 - the TCP format, which is more advanced. This format is enabled when "option
16236 tcplog" is set on the frontend. HAProxy will then usually wait for the
16237 connection to terminate before logging. This format provides much richer
16238 information, such as timers, connection counts, queue size, etc... This
16239 format is recommended for pure TCP proxies.
16240
16241 - the HTTP format, which is the most advanced for HTTP proxying. This format
16242 is enabled when "option httplog" is set on the frontend. It provides the
16243 same information as the TCP format with some HTTP-specific fields such as
16244 the request, the status code, and captures of headers and cookies. This
16245 format is recommended for HTTP proxies.
16246
Emeric Brun3a058f32009-06-30 18:26:00 +020016247 - the CLF HTTP format, which is equivalent to the HTTP format, but with the
16248 fields arranged in the same order as the CLF format. In this mode, all
16249 timers, captures, flags, etc... appear one per field after the end of the
16250 common fields, in the same order they appear in the standard HTTP format.
16251
William Lallemand48940402012-01-30 16:47:22 +010016252 - the custom log format, allows you to make your own log line.
16253
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016254Next sections will go deeper into details for each of these formats. Format
16255specification will be performed on a "field" basis. Unless stated otherwise, a
16256field is a portion of text delimited by any number of spaces. Since syslog
16257servers are susceptible of inserting fields at the beginning of a line, it is
16258always assumed that the first field is the one containing the process name and
16259identifier.
16260
16261Note : Since log lines may be quite long, the log examples in sections below
16262 might be broken into multiple lines. The example log lines will be
16263 prefixed with 3 closing angle brackets ('>>>') and each time a log is
16264 broken into multiple lines, each non-final line will end with a
16265 backslash ('\') and the next line will start indented by two characters.
16266
16267
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200162688.2.1. Default log format
16269-------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016270
16271This format is used when no specific option is set. The log is emitted as soon
16272as the connection is accepted. One should note that this currently is the only
16273format which logs the request's destination IP and ports.
16274
16275 Example :
16276 listen www
16277 mode http
16278 log global
16279 server srv1 127.0.0.1:8000
16280
16281 >>> Feb 6 12:12:09 localhost \
16282 haproxy[14385]: Connect from 10.0.1.2:33312 to 10.0.3.31:8012 \
16283 (www/HTTP)
16284
16285 Field Format Extract from the example above
16286 1 process_name '[' pid ']:' haproxy[14385]:
16287 2 'Connect from' Connect from
16288 3 source_ip ':' source_port 10.0.1.2:33312
16289 4 'to' to
16290 5 destination_ip ':' destination_port 10.0.3.31:8012
16291 6 '(' frontend_name '/' mode ')' (www/HTTP)
16292
16293Detailed fields description :
16294 - "source_ip" is the IP address of the client which initiated the connection.
16295 - "source_port" is the TCP port of the client which initiated the connection.
16296 - "destination_ip" is the IP address the client connected to.
16297 - "destination_port" is the TCP port the client connected to.
16298 - "frontend_name" is the name of the frontend (or listener) which received
16299 and processed the connection.
16300 - "mode is the mode the frontend is operating (TCP or HTTP).
16301
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016302In case of a UNIX socket, the source and destination addresses are marked as
16303"unix:" and the ports reflect the internal ID of the socket which accepted the
16304connection (the same ID as reported in the stats).
16305
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016306It is advised not to use this deprecated format for newer installations as it
16307will eventually disappear.
16308
16309
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200163108.2.2. TCP log format
16311---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016312
16313The TCP format is used when "option tcplog" is specified in the frontend, and
16314is the recommended format for pure TCP proxies. It provides a lot of precious
16315information for troubleshooting. Since this format includes timers and byte
16316counts, the log is normally emitted at the end of the session. It can be
16317emitted earlier if "option logasap" is specified, which makes sense in most
16318environments with long sessions such as remote terminals. Sessions which match
16319the "monitor" rules are never logged. It is also possible not to emit logs for
16320sessions for which no data were exchanged between the client and the server, by
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020016321specifying "option dontlognull" in the frontend. Successful connections will
16322not be logged if "option dontlog-normal" is specified in the frontend. A few
16323fields may slightly vary depending on some configuration options, those are
16324marked with a star ('*') after the field name below.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016325
16326 Example :
16327 frontend fnt
16328 mode tcp
16329 option tcplog
16330 log global
16331 default_backend bck
16332
16333 backend bck
16334 server srv1 127.0.0.1:8000
16335
16336 >>> Feb 6 12:12:56 localhost \
16337 haproxy[14387]: 10.0.1.2:33313 [06/Feb/2009:12:12:51.443] fnt \
16338 bck/srv1 0/0/5007 212 -- 0/0/0/0/3 0/0
16339
16340 Field Format Extract from the example above
16341 1 process_name '[' pid ']:' haproxy[14387]:
16342 2 client_ip ':' client_port 10.0.1.2:33313
16343 3 '[' accept_date ']' [06/Feb/2009:12:12:51.443]
16344 4 frontend_name fnt
16345 5 backend_name '/' server_name bck/srv1
16346 6 Tw '/' Tc '/' Tt* 0/0/5007
16347 7 bytes_read* 212
16348 8 termination_state --
16349 9 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 0/0/0/0/3
16350 10 srv_queue '/' backend_queue 0/0
16351
16352Detailed fields description :
16353 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016354 connection to haproxy. If the connection was accepted on a UNIX socket
16355 instead, the IP address would be replaced with the word "unix". Note that
16356 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016357 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010016358 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016359 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016360
16361 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016362 If the connection was accepted on a UNIX socket instead, the port would be
16363 replaced with the ID of the accepting socket, which is also reported in the
16364 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016365
16366 - "accept_date" is the exact date when the connection was received by haproxy
16367 (which might be very slightly different from the date observed on the
16368 network if there was some queuing in the system's backlog). This is usually
Willy Tarreau590a0512018-09-05 11:56:48 +020016369 the same date which may appear in any upstream firewall's log. When used in
16370 HTTP mode, the accept_date field will be reset to the first moment the
16371 connection is ready to receive a new request (end of previous response for
16372 HTTP/1, immediately after previous request for HTTP/2).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016373
16374 - "frontend_name" is the name of the frontend (or listener) which received
16375 and processed the connection.
16376
16377 - "backend_name" is the name of the backend (or listener) which was selected
16378 to manage the connection to the server. This will be the same as the
16379 frontend if no switching rule has been applied, which is common for TCP
16380 applications.
16381
16382 - "server_name" is the name of the last server to which the connection was
16383 sent, which might differ from the first one if there were connection errors
16384 and a redispatch occurred. Note that this server belongs to the backend
16385 which processed the request. If the connection was aborted before reaching
16386 a server, "<NOSRV>" is indicated instead of a server name.
16387
16388 - "Tw" is the total time in milliseconds spent waiting in the various queues.
16389 It can be "-1" if the connection was aborted before reaching the queue.
16390 See "Timers" below for more details.
16391
16392 - "Tc" is the total time in milliseconds spent waiting for the connection to
16393 establish to the final server, including retries. It can be "-1" if the
16394 connection was aborted before a connection could be established. See
16395 "Timers" below for more details.
16396
16397 - "Tt" is the total time in milliseconds elapsed between the accept and the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016398 last close. It covers all possible processing. There is one exception, if
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016399 "option logasap" was specified, then the time counting stops at the moment
16400 the log is emitted. In this case, a '+' sign is prepended before the value,
16401 indicating that the final one will be larger. See "Timers" below for more
16402 details.
16403
16404 - "bytes_read" is the total number of bytes transmitted from the server to
16405 the client when the log is emitted. If "option logasap" is specified, the
16406 this value will be prefixed with a '+' sign indicating that the final one
16407 may be larger. Please note that this value is a 64-bit counter, so log
16408 analysis tools must be able to handle it without overflowing.
16409
16410 - "termination_state" is the condition the session was in when the session
16411 ended. This indicates the session state, which side caused the end of
16412 session to happen, and for what reason (timeout, error, ...). The normal
16413 flags should be "--", indicating the session was closed by either end with
16414 no data remaining in buffers. See below "Session state at disconnection"
16415 for more details.
16416
16417 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016418 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016419 limits have been reached. For instance, if actconn is close to 512 when
16420 multiple connection errors occur, chances are high that the system limits
16421 the process to use a maximum of 1024 file descriptors and that all of them
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016422 are used. See section 3 "Global parameters" to find how to tune the system.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016423
16424 - "feconn" is the total number of concurrent connections on the frontend when
16425 the session was logged. It is useful to estimate the amount of resource
16426 required to sustain high loads, and to detect when the frontend's "maxconn"
16427 has been reached. Most often when this value increases by huge jumps, it is
16428 because there is congestion on the backend servers, but sometimes it can be
16429 caused by a denial of service attack.
16430
16431 - "beconn" is the total number of concurrent connections handled by the
16432 backend when the session was logged. It includes the total number of
16433 concurrent connections active on servers as well as the number of
16434 connections pending in queues. It is useful to estimate the amount of
16435 additional servers needed to support high loads for a given application.
16436 Most often when this value increases by huge jumps, it is because there is
16437 congestion on the backend servers, but sometimes it can be caused by a
16438 denial of service attack.
16439
16440 - "srv_conn" is the total number of concurrent connections still active on
16441 the server when the session was logged. It can never exceed the server's
16442 configured "maxconn" parameter. If this value is very often close or equal
16443 to the server's "maxconn", it means that traffic regulation is involved a
16444 lot, meaning that either the server's maxconn value is too low, or that
16445 there aren't enough servers to process the load with an optimal response
16446 time. When only one of the server's "srv_conn" is high, it usually means
16447 that this server has some trouble causing the connections to take longer to
16448 be processed than on other servers.
16449
16450 - "retries" is the number of connection retries experienced by this session
16451 when trying to connect to the server. It must normally be zero, unless a
16452 server is being stopped at the same moment the connection was attempted.
16453 Frequent retries generally indicate either a network problem between
16454 haproxy and the server, or a misconfigured system backlog on the server
16455 preventing new connections from being queued. This field may optionally be
16456 prefixed with a '+' sign, indicating that the session has experienced a
16457 redispatch after the maximal retry count has been reached on the initial
16458 server. In this case, the server name appearing in the log is the one the
16459 connection was redispatched to, and not the first one, though both may
16460 sometimes be the same in case of hashing for instance. So as a general rule
16461 of thumb, when a '+' is present in front of the retry count, this count
16462 should not be attributed to the logged server.
16463
16464 - "srv_queue" is the total number of requests which were processed before
16465 this one in the server queue. It is zero when the request has not gone
16466 through the server queue. It makes it possible to estimate the approximate
16467 server's response time by dividing the time spent in queue by the number of
16468 requests in the queue. It is worth noting that if a session experiences a
16469 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010016470 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016471 backend queue unless a redispatch occurs.
16472
16473 - "backend_queue" is the total number of requests which were processed before
16474 this one in the backend's global queue. It is zero when the request has not
16475 gone through the global queue. It makes it possible to estimate the average
16476 queue length, which easily translates into a number of missing servers when
16477 divided by a server's "maxconn" parameter. It is worth noting that if a
16478 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010016479 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016480 through both the server queue and the backend queue unless a redispatch
16481 occurs.
16482
16483
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200164848.2.3. HTTP log format
16485----------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016486
16487The HTTP format is the most complete and the best suited for HTTP proxies. It
16488is enabled by when "option httplog" is specified in the frontend. It provides
16489the same level of information as the TCP format with additional features which
16490are specific to the HTTP protocol. Just like the TCP format, the log is usually
16491emitted at the end of the session, unless "option logasap" is specified, which
16492generally only makes sense for download sites. A session which matches the
16493"monitor" rules will never logged. It is also possible not to log sessions for
16494which no data were sent by the client by specifying "option dontlognull" in the
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020016495frontend. Successful connections will not be logged if "option dontlog-normal"
16496is specified in the frontend.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016497
16498Most fields are shared with the TCP log, some being different. A few fields may
16499slightly vary depending on some configuration options. Those ones are marked
16500with a star ('*') after the field name below.
16501
16502 Example :
16503 frontend http-in
16504 mode http
16505 option httplog
16506 log global
16507 default_backend bck
16508
16509 backend static
16510 server srv1 127.0.0.1:8000
16511
16512 >>> Feb 6 12:14:14 localhost \
16513 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
16514 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 +010016515 {} "GET /index.html HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016516
16517 Field Format Extract from the example above
16518 1 process_name '[' pid ']:' haproxy[14389]:
16519 2 client_ip ':' client_port 10.0.1.2:33317
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016520 3 '[' request_date ']' [06/Feb/2009:12:14:14.655]
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016521 4 frontend_name http-in
16522 5 backend_name '/' server_name static/srv1
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016523 6 TR '/' Tw '/' Tc '/' Tr '/' Ta* 10/0/30/69/109
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016524 7 status_code 200
16525 8 bytes_read* 2750
16526 9 captured_request_cookie -
16527 10 captured_response_cookie -
16528 11 termination_state ----
16529 12 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 1/1/1/1/0
16530 13 srv_queue '/' backend_queue 0/0
16531 14 '{' captured_request_headers* '}' {haproxy.1wt.eu}
16532 15 '{' captured_response_headers* '}' {}
16533 16 '"' http_request '"' "GET /index.html HTTP/1.1"
Willy Tarreaud72758d2010-01-12 10:42:19 +010016534
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016535Detailed fields description :
16536 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016537 connection to haproxy. If the connection was accepted on a UNIX socket
16538 instead, the IP address would be replaced with the word "unix". Note that
16539 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016540 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010016541 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016542 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016543
16544 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016545 If the connection was accepted on a UNIX socket instead, the port would be
16546 replaced with the ID of the accepting socket, which is also reported in the
16547 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016548
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016549 - "request_date" is the exact date when the first byte of the HTTP request
16550 was received by haproxy (log field %tr).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016551
16552 - "frontend_name" is the name of the frontend (or listener) which received
16553 and processed the connection.
16554
16555 - "backend_name" is the name of the backend (or listener) which was selected
16556 to manage the connection to the server. This will be the same as the
16557 frontend if no switching rule has been applied.
16558
16559 - "server_name" is the name of the last server to which the connection was
16560 sent, which might differ from the first one if there were connection errors
16561 and a redispatch occurred. Note that this server belongs to the backend
16562 which processed the request. If the request was aborted before reaching a
16563 server, "<NOSRV>" is indicated instead of a server name. If the request was
16564 intercepted by the stats subsystem, "<STATS>" is indicated instead.
16565
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016566 - "TR" is the total time in milliseconds spent waiting for a full HTTP
16567 request from the client (not counting body) after the first byte was
16568 received. It can be "-1" if the connection was aborted before a complete
16569 request could be received or the a bad request was received. It should
16570 always be very small because a request generally fits in one single packet.
16571 Large times here generally indicate network issues between the client and
Willy Tarreau590a0512018-09-05 11:56:48 +020016572 haproxy or requests being typed by hand. See section 8.4 "Timing Events"
16573 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016574
16575 - "Tw" is the total time in milliseconds spent waiting in the various queues.
16576 It can be "-1" if the connection was aborted before reaching the queue.
Willy Tarreau590a0512018-09-05 11:56:48 +020016577 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016578
16579 - "Tc" is the total time in milliseconds spent waiting for the connection to
16580 establish to the final server, including retries. It can be "-1" if the
Willy Tarreau590a0512018-09-05 11:56:48 +020016581 request was aborted before a connection could be established. See section
16582 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016583
16584 - "Tr" is the total time in milliseconds spent waiting for the server to send
16585 a full HTTP response, not counting data. It can be "-1" if the request was
16586 aborted before a complete response could be received. It generally matches
16587 the server's processing time for the request, though it may be altered by
16588 the amount of data sent by the client to the server. Large times here on
Willy Tarreau590a0512018-09-05 11:56:48 +020016589 "GET" requests generally indicate an overloaded server. See section 8.4
16590 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016591
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016592 - "Ta" is the time the request remained active in haproxy, which is the total
16593 time in milliseconds elapsed between the first byte of the request was
16594 received and the last byte of response was sent. It covers all possible
16595 processing except the handshake (see Th) and idle time (see Ti). There is
16596 one exception, if "option logasap" was specified, then the time counting
16597 stops at the moment the log is emitted. In this case, a '+' sign is
16598 prepended before the value, indicating that the final one will be larger.
Willy Tarreau590a0512018-09-05 11:56:48 +020016599 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016600
16601 - "status_code" is the HTTP status code returned to the client. This status
16602 is generally set by the server, but it might also be set by haproxy when
16603 the server cannot be reached or when its response is blocked by haproxy.
16604
16605 - "bytes_read" is the total number of bytes transmitted to the client when
16606 the log is emitted. This does include HTTP headers. If "option logasap" is
16607 specified, the this value will be prefixed with a '+' sign indicating that
16608 the final one may be larger. Please note that this value is a 64-bit
16609 counter, so log analysis tools must be able to handle it without
16610 overflowing.
16611
16612 - "captured_request_cookie" is an optional "name=value" entry indicating that
16613 the client had this cookie in the request. The cookie name and its maximum
16614 length are defined by the "capture cookie" statement in the frontend
16615 configuration. The field is a single dash ('-') when the option is not
16616 set. Only one cookie may be captured, it is generally used to track session
16617 ID exchanges between a client and a server to detect session crossing
16618 between clients due to application bugs. For more details, please consult
16619 the section "Capturing HTTP headers and cookies" below.
16620
16621 - "captured_response_cookie" is an optional "name=value" entry indicating
16622 that the server has returned a cookie with its response. The cookie name
16623 and its maximum length are defined by the "capture cookie" statement in the
16624 frontend configuration. The field is a single dash ('-') when the option is
16625 not set. Only one cookie may be captured, it is generally used to track
16626 session ID exchanges between a client and a server to detect session
16627 crossing between clients due to application bugs. For more details, please
16628 consult the section "Capturing HTTP headers and cookies" below.
16629
16630 - "termination_state" is the condition the session was in when the session
16631 ended. This indicates the session state, which side caused the end of
16632 session to happen, for what reason (timeout, error, ...), just like in TCP
16633 logs, and information about persistence operations on cookies in the last
16634 two characters. The normal flags should begin with "--", indicating the
16635 session was closed by either end with no data remaining in buffers. See
16636 below "Session state at disconnection" for more details.
16637
16638 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016639 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016640 limits have been reached. For instance, if actconn is close to 512 or 1024
16641 when multiple connection errors occur, chances are high that the system
16642 limits the process to use a maximum of 1024 file descriptors and that all
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016643 of them are used. See section 3 "Global parameters" to find how to tune the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016644 system.
16645
16646 - "feconn" is the total number of concurrent connections on the frontend when
16647 the session was logged. It is useful to estimate the amount of resource
16648 required to sustain high loads, and to detect when the frontend's "maxconn"
16649 has been reached. Most often when this value increases by huge jumps, it is
16650 because there is congestion on the backend servers, but sometimes it can be
16651 caused by a denial of service attack.
16652
16653 - "beconn" is the total number of concurrent connections handled by the
16654 backend when the session was logged. It includes the total number of
16655 concurrent connections active on servers as well as the number of
16656 connections pending in queues. It is useful to estimate the amount of
16657 additional servers needed to support high loads for a given application.
16658 Most often when this value increases by huge jumps, it is because there is
16659 congestion on the backend servers, but sometimes it can be caused by a
16660 denial of service attack.
16661
16662 - "srv_conn" is the total number of concurrent connections still active on
16663 the server when the session was logged. It can never exceed the server's
16664 configured "maxconn" parameter. If this value is very often close or equal
16665 to the server's "maxconn", it means that traffic regulation is involved a
16666 lot, meaning that either the server's maxconn value is too low, or that
16667 there aren't enough servers to process the load with an optimal response
16668 time. When only one of the server's "srv_conn" is high, it usually means
16669 that this server has some trouble causing the requests to take longer to be
16670 processed than on other servers.
16671
16672 - "retries" is the number of connection retries experienced by this session
16673 when trying to connect to the server. It must normally be zero, unless a
16674 server is being stopped at the same moment the connection was attempted.
16675 Frequent retries generally indicate either a network problem between
16676 haproxy and the server, or a misconfigured system backlog on the server
16677 preventing new connections from being queued. This field may optionally be
16678 prefixed with a '+' sign, indicating that the session has experienced a
16679 redispatch after the maximal retry count has been reached on the initial
16680 server. In this case, the server name appearing in the log is the one the
16681 connection was redispatched to, and not the first one, though both may
16682 sometimes be the same in case of hashing for instance. So as a general rule
16683 of thumb, when a '+' is present in front of the retry count, this count
16684 should not be attributed to the logged server.
16685
16686 - "srv_queue" is the total number of requests which were processed before
16687 this one in the server queue. It is zero when the request has not gone
16688 through the server queue. It makes it possible to estimate the approximate
16689 server's response time by dividing the time spent in queue by the number of
16690 requests in the queue. It is worth noting that if a session experiences a
16691 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010016692 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016693 backend queue unless a redispatch occurs.
16694
16695 - "backend_queue" is the total number of requests which were processed before
16696 this one in the backend's global queue. It is zero when the request has not
16697 gone through the global queue. It makes it possible to estimate the average
16698 queue length, which easily translates into a number of missing servers when
16699 divided by a server's "maxconn" parameter. It is worth noting that if a
16700 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010016701 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016702 through both the server queue and the backend queue unless a redispatch
16703 occurs.
16704
16705 - "captured_request_headers" is a list of headers captured in the request due
16706 to the presence of the "capture request header" statement in the frontend.
16707 Multiple headers can be captured, they will be delimited by a vertical bar
16708 ('|'). When no capture is enabled, the braces do not appear, causing a
16709 shift of remaining fields. It is important to note that this field may
16710 contain spaces, and that using it requires a smarter log parser than when
16711 it's not used. Please consult the section "Capturing HTTP headers and
16712 cookies" below for more details.
16713
16714 - "captured_response_headers" is a list of headers captured in the response
16715 due to the presence of the "capture response header" statement in the
16716 frontend. Multiple headers can be captured, they will be delimited by a
16717 vertical bar ('|'). When no capture is enabled, the braces do not appear,
16718 causing a shift of remaining fields. It is important to note that this
16719 field may contain spaces, and that using it requires a smarter log parser
16720 than when it's not used. Please consult the section "Capturing HTTP headers
16721 and cookies" below for more details.
16722
16723 - "http_request" is the complete HTTP request line, including the method,
16724 request and HTTP version string. Non-printable characters are encoded (see
16725 below the section "Non-printable characters"). This is always the last
16726 field, and it is always delimited by quotes and is the only one which can
16727 contain quotes. If new fields are added to the log format, they will be
16728 added before this field. This field might be truncated if the request is
16729 huge and does not fit in the standard syslog buffer (1024 characters). This
16730 is the reason why this field must always remain the last one.
16731
16732
Cyril Bontédc4d9032012-04-08 21:57:39 +0200167338.2.4. Custom log format
16734------------------------
William Lallemand48940402012-01-30 16:47:22 +010016735
Willy Tarreau2beef582012-12-20 17:22:52 +010016736The directive log-format allows you to customize the logs in http mode and tcp
William Lallemandbddd4fd2012-02-27 11:23:10 +010016737mode. It takes a string as argument.
William Lallemand48940402012-01-30 16:47:22 +010016738
Davor Ocelice9ed2812017-12-25 17:49:28 +010016739HAProxy understands some log format variables. % precedes log format variables.
William Lallemand48940402012-01-30 16:47:22 +010016740Variables can take arguments using braces ('{}'), and multiple arguments are
16741separated by commas within the braces. Flags may be added or removed by
16742prefixing them with a '+' or '-' sign.
16743
16744Special variable "%o" may be used to propagate its flags to all other
16745variables on the same format string. This is particularly handy with quoted
Dragan Dosen835b9212016-02-12 13:23:03 +010016746("Q") and escaped ("E") string formats.
William Lallemand48940402012-01-30 16:47:22 +010016747
Willy Tarreauc8368452012-12-21 00:09:23 +010016748If a variable is named between square brackets ('[' .. ']') then it is used
Willy Tarreaube722a22014-06-13 16:31:59 +020016749as a sample expression rule (see section 7.3). This it useful to add some
Willy Tarreauc8368452012-12-21 00:09:23 +010016750less common information such as the client's SSL certificate's DN, or to log
16751the key that would be used to store an entry into a stick table.
16752
William Lallemand48940402012-01-30 16:47:22 +010016753Note: spaces must be escaped. A space character is considered as a separator.
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016754In order to emit a verbatim '%', it must be preceded by another '%' resulting
Willy Tarreau06d97f92013-12-02 17:45:48 +010016755in '%%'. HAProxy will automatically merge consecutive separators.
William Lallemand48940402012-01-30 16:47:22 +010016756
Dragan Dosen835b9212016-02-12 13:23:03 +010016757Note: when using the RFC5424 syslog message format, the characters '"',
16758'\' and ']' inside PARAM-VALUE should be escaped with '\' as prefix (see
16759https://tools.ietf.org/html/rfc5424#section-6.3.3 for more details). In
16760such cases, the use of the flag "E" should be considered.
16761
William Lallemand48940402012-01-30 16:47:22 +010016762Flags are :
16763 * Q: quote a string
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016764 * X: hexadecimal representation (IPs, Ports, %Ts, %rt, %pid)
Dragan Dosen835b9212016-02-12 13:23:03 +010016765 * E: escape characters '"', '\' and ']' in a string with '\' as prefix
16766 (intended purpose is for the RFC5424 structured-data log formats)
William Lallemand48940402012-01-30 16:47:22 +010016767
16768 Example:
16769
16770 log-format %T\ %t\ Some\ Text
16771 log-format %{+Q}o\ %t\ %s\ %{-Q}r
16772
Dragan Dosen835b9212016-02-12 13:23:03 +010016773 log-format-sd %{+Q,+E}o\ [exampleSDID@1234\ header=%[capture.req.hdr(0)]]
16774
William Lallemand48940402012-01-30 16:47:22 +010016775At the moment, the default HTTP format is defined this way :
16776
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016777 log-format "%ci:%cp [%tr] %ft %b/%s %TR/%Tw/%Tc/%Tr/%Ta %ST %B %CC \
16778 %CS %tsc %ac/%fc/%bc/%sc/%rc %sq/%bq %hr %hs %{+Q}r"
William Lallemand48940402012-01-30 16:47:22 +010016779
William Lallemandbddd4fd2012-02-27 11:23:10 +010016780the default CLF format is defined this way :
William Lallemand48940402012-01-30 16:47:22 +010016781
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016782 log-format "%{+Q}o %{-Q}ci - - [%trg] %r %ST %B \"\" \"\" %cp \
16783 %ms %ft %b %s %TR %Tw %Tc %Tr %Ta %tsc %ac %fc \
16784 %bc %sc %rc %sq %bq %CC %CS %hrl %hsl"
William Lallemand48940402012-01-30 16:47:22 +010016785
William Lallemandbddd4fd2012-02-27 11:23:10 +010016786and the default TCP format is defined this way :
16787
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016788 log-format "%ci:%cp [%t] %ft %b/%s %Tw/%Tc/%Tt %B %ts \
16789 %ac/%fc/%bc/%sc/%rc %sq/%bq"
William Lallemandbddd4fd2012-02-27 11:23:10 +010016790
William Lallemand48940402012-01-30 16:47:22 +010016791Please refer to the table below for currently defined variables :
16792
William Lallemandbddd4fd2012-02-27 11:23:10 +010016793 +---+------+-----------------------------------------------+-------------+
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016794 | R | var | field name (8.2.2 and 8.2.3 for description) | type |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016795 +---+------+-----------------------------------------------+-------------+
16796 | | %o | special variable, apply flags on all next var | |
16797 +---+------+-----------------------------------------------+-------------+
Willy Tarreau2beef582012-12-20 17:22:52 +010016798 | | %B | bytes_read (from server to client) | numeric |
16799 | H | %CC | captured_request_cookie | string |
16800 | H | %CS | captured_response_cookie | string |
William Lallemand5f232402012-04-05 18:02:55 +020016801 | | %H | hostname | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000016802 | H | %HM | HTTP method (ex: POST) | string |
16803 | H | %HP | HTTP request URI without query string (path) | string |
Andrew Hayworthe63ac872015-07-31 16:14:16 +000016804 | H | %HQ | HTTP request URI query string (ex: ?bar=baz) | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000016805 | H | %HU | HTTP request URI (ex: /foo?bar=baz) | string |
16806 | H | %HV | HTTP version (ex: HTTP/1.0) | string |
William Lallemanda73203e2012-03-12 12:48:57 +010016807 | | %ID | unique-id | string |
Willy Tarreau4bf99632014-06-13 12:21:40 +020016808 | | %ST | status_code | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020016809 | | %T | gmt_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016810 | | %Ta | Active time of the request (from TR to end) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016811 | | %Tc | Tc | numeric |
Willy Tarreau27b639d2016-05-17 17:55:27 +020016812 | | %Td | Td = Tt - (Tq + Tw + Tc + Tr) | numeric |
Yuxans Yao4e25b012012-10-19 10:36:09 +080016813 | | %Tl | local_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016814 | | %Th | connection handshake time (SSL, PROXY proto) | numeric |
16815 | H | %Ti | idle time before the HTTP request | numeric |
16816 | H | %Tq | Th + Ti + TR | numeric |
16817 | H | %TR | time to receive the full request from 1st byte| numeric |
16818 | H | %Tr | Tr (response time) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020016819 | | %Ts | timestamp | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016820 | | %Tt | Tt | numeric |
16821 | | %Tw | Tw | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010016822 | | %U | bytes_uploaded (from client to server) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016823 | | %ac | actconn | numeric |
16824 | | %b | backend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016825 | | %bc | beconn (backend concurrent connections) | numeric |
16826 | | %bi | backend_source_ip (connecting address) | IP |
16827 | | %bp | backend_source_port (connecting address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016828 | | %bq | backend_queue | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010016829 | | %ci | client_ip (accepted address) | IP |
16830 | | %cp | client_port (accepted address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016831 | | %f | frontend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016832 | | %fc | feconn (frontend concurrent connections) | numeric |
16833 | | %fi | frontend_ip (accepting address) | IP |
16834 | | %fp | frontend_port (accepting address) | numeric |
Willy Tarreau773d65f2012-10-12 14:56:11 +020016835 | | %ft | frontend_name_transport ('~' suffix for SSL) | string |
Willy Tarreau7346acb2014-08-28 15:03:15 +020016836 | | %lc | frontend_log_counter | numeric |
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020016837 | | %hr | captured_request_headers default style | string |
16838 | | %hrl | captured_request_headers CLF style | string list |
16839 | | %hs | captured_response_headers default style | string |
16840 | | %hsl | captured_response_headers CLF style | string list |
Willy Tarreau812c88e2015-08-09 10:56:35 +020016841 | | %ms | accept date milliseconds (left-padded with 0) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020016842 | | %pid | PID | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016843 | H | %r | http_request | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016844 | | %rc | retries | numeric |
Willy Tarreau1f0da242014-01-25 11:01:50 +010016845 | | %rt | request_counter (HTTP req or TCP session) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016846 | | %s | server_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016847 | | %sc | srv_conn (server concurrent connections) | numeric |
16848 | | %si | server_IP (target address) | IP |
16849 | | %sp | server_port (target address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016850 | | %sq | srv_queue | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016851 | S | %sslc| ssl_ciphers (ex: AES-SHA) | string |
16852 | S | %sslv| ssl_version (ex: TLSv1) | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016853 | | %t | date_time (with millisecond resolution) | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016854 | H | %tr | date_time of HTTP request | date |
16855 | H | %trg | gmt_date_time of start of HTTP request | date |
Jens Bissinger15c64ff2018-08-23 14:11:27 +020016856 | H | %trl | local_date_time of start of HTTP request | date |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016857 | | %ts | termination_state | string |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016858 | H | %tsc | termination_state with cookie status | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016859 +---+------+-----------------------------------------------+-------------+
William Lallemand48940402012-01-30 16:47:22 +010016860
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016861 R = Restrictions : H = mode http only ; S = SSL only
William Lallemand48940402012-01-30 16:47:22 +010016862
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010016863
168648.2.5. Error log format
16865-----------------------
16866
16867When an incoming connection fails due to an SSL handshake or an invalid PROXY
16868protocol header, haproxy will log the event using a shorter, fixed line format.
16869By default, logs are emitted at the LOG_INFO level, unless the option
16870"log-separate-errors" is set in the backend, in which case the LOG_ERR level
Davor Ocelice9ed2812017-12-25 17:49:28 +010016871will be used. Connections on which no data are exchanged (e.g. probes) are not
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010016872logged if the "dontlognull" option is set.
16873
16874The format looks like this :
16875
16876 >>> Dec 3 18:27:14 localhost \
16877 haproxy[6103]: 127.0.0.1:56059 [03/Dec/2012:17:35:10.380] frt/f1: \
16878 Connection error during SSL handshake
16879
16880 Field Format Extract from the example above
16881 1 process_name '[' pid ']:' haproxy[6103]:
16882 2 client_ip ':' client_port 127.0.0.1:56059
16883 3 '[' accept_date ']' [03/Dec/2012:17:35:10.380]
16884 4 frontend_name "/" bind_name ":" frt/f1:
16885 5 message Connection error during SSL handshake
16886
16887These fields just provide minimal information to help debugging connection
16888failures.
16889
16890
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200168918.3. Advanced logging options
16892-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016893
16894Some advanced logging options are often looked for but are not easy to find out
16895just by looking at the various options. Here is an entry point for the few
16896options which can enable better logging. Please refer to the keywords reference
16897for more information about their usage.
16898
16899
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200169008.3.1. Disabling logging of external tests
16901------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016902
16903It is quite common to have some monitoring tools perform health checks on
16904haproxy. Sometimes it will be a layer 3 load-balancer such as LVS or any
16905commercial load-balancer, and sometimes it will simply be a more complete
16906monitoring system such as Nagios. When the tests are very frequent, users often
16907ask how to disable logging for those checks. There are three possibilities :
16908
16909 - if connections come from everywhere and are just TCP probes, it is often
16910 desired to simply disable logging of connections without data exchange, by
16911 setting "option dontlognull" in the frontend. It also disables logging of
16912 port scans, which may or may not be desired.
16913
16914 - if the connection come from a known source network, use "monitor-net" to
16915 declare this network as monitoring only. Any host in this network will then
16916 only be able to perform health checks, and their requests will not be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016917 logged. This is generally appropriate to designate a list of equipment
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016918 such as other load-balancers.
16919
16920 - if the tests are performed on a known URI, use "monitor-uri" to declare
16921 this URI as dedicated to monitoring. Any host sending this request will
16922 only get the result of a health-check, and the request will not be logged.
16923
16924
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200169258.3.2. Logging before waiting for the session to terminate
16926----------------------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016927
16928The problem with logging at end of connection is that you have no clue about
16929what is happening during very long sessions, such as remote terminal sessions
16930or large file downloads. This problem can be worked around by specifying
Davor Ocelice9ed2812017-12-25 17:49:28 +010016931"option logasap" in the frontend. HAProxy will then log as soon as possible,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016932just before data transfer begins. This means that in case of TCP, it will still
16933log the connection status to the server, and in case of HTTP, it will log just
16934after processing the server headers. In this case, the number of bytes reported
16935is the number of header bytes sent to the client. In order to avoid confusion
16936with normal logs, the total time field and the number of bytes are prefixed
16937with a '+' sign which means that real numbers are certainly larger.
16938
16939
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200169408.3.3. Raising log level upon errors
16941------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020016942
16943Sometimes it is more convenient to separate normal traffic from errors logs,
16944for instance in order to ease error monitoring from log files. When the option
16945"log-separate-errors" is used, connections which experience errors, timeouts,
16946retries, redispatches or HTTP status codes 5xx will see their syslog level
16947raised from "info" to "err". This will help a syslog daemon store the log in
16948a separate file. It is very important to keep the errors in the normal traffic
16949file too, so that log ordering is not altered. You should also be careful if
16950you already have configured your syslog daemon to store all logs higher than
16951"notice" in an "admin" file, because the "err" level is higher than "notice".
16952
16953
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200169548.3.4. Disabling logging of successful connections
16955--------------------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020016956
16957Although this may sound strange at first, some large sites have to deal with
16958multiple thousands of logs per second and are experiencing difficulties keeping
16959them intact for a long time or detecting errors within them. If the option
16960"dontlog-normal" is set on the frontend, all normal connections will not be
16961logged. In this regard, a normal connection is defined as one without any
16962error, timeout, retry nor redispatch. In HTTP, the status code is checked too,
16963and a response with a status 5xx is not considered normal and will be logged
16964too. Of course, doing is is really discouraged as it will remove most of the
16965useful information from the logs. Do this only if you have no other
16966alternative.
16967
16968
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200169698.4. Timing events
16970------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016971
16972Timers provide a great help in troubleshooting network problems. All values are
16973reported in milliseconds (ms). These timers should be used in conjunction with
16974the session termination flags. In TCP mode with "option tcplog" set on the
16975frontend, 3 control points are reported under the form "Tw/Tc/Tt", and in HTTP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016976mode, 5 control points are reported under the form "TR/Tw/Tc/Tr/Ta". In
16977addition, three other measures are provided, "Th", "Ti", and "Tq".
16978
Guillaume de Lafondf27cddc2016-12-23 17:32:43 +010016979Timings events in HTTP mode:
16980
16981 first request 2nd request
16982 |<-------------------------------->|<-------------- ...
16983 t tr t tr ...
16984 ---|----|----|----|----|----|----|----|----|--
16985 : Th Ti TR Tw Tc Tr Td : Ti ...
16986 :<---- Tq ---->: :
16987 :<-------------- Tt -------------->:
16988 :<--------- Ta --------->:
16989
16990Timings events in TCP mode:
16991
16992 TCP session
16993 |<----------------->|
16994 t t
16995 ---|----|----|----|----|---
16996 | Th Tw Tc Td |
16997 |<------ Tt ------->|
16998
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016999 - Th: total time to accept tcp connection and execute handshakes for low level
Davor Ocelice9ed2812017-12-25 17:49:28 +010017000 protocols. Currently, these protocols are proxy-protocol and SSL. This may
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017001 only happen once during the whole connection's lifetime. A large time here
17002 may indicate that the client only pre-established the connection without
17003 speaking, that it is experiencing network issues preventing it from
Davor Ocelice9ed2812017-12-25 17:49:28 +010017004 completing a handshake in a reasonable time (e.g. MTU issues), or that an
Willy Tarreau590a0512018-09-05 11:56:48 +020017005 SSL handshake was very expensive to compute. Please note that this time is
17006 reported only before the first request, so it is safe to average it over
17007 all request to calculate the amortized value. The second and subsequent
17008 request will always report zero here.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017009
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017010 - Ti: is the idle time before the HTTP request (HTTP mode only). This timer
17011 counts between the end of the handshakes and the first byte of the HTTP
17012 request. When dealing with a second request in keep-alive mode, it starts
Willy Tarreau590a0512018-09-05 11:56:48 +020017013 to count after the end of the transmission the previous response. When a
17014 multiplexed protocol such as HTTP/2 is used, it starts to count immediately
17015 after the previous request. Some browsers pre-establish connections to a
17016 server in order to reduce the latency of a future request, and keep them
17017 pending until they need it. This delay will be reported as the idle time. A
17018 value of -1 indicates that nothing was received on the connection.
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017019
17020 - TR: total time to get the client request (HTTP mode only). It's the time
17021 elapsed between the first bytes received and the moment the proxy received
17022 the empty line marking the end of the HTTP headers. The value "-1"
17023 indicates that the end of headers has never been seen. This happens when
17024 the client closes prematurely or times out. This time is usually very short
17025 since most requests fit in a single packet. A large time may indicate a
17026 request typed by hand during a test.
17027
17028 - Tq: total time to get the client request from the accept date or since the
17029 emission of the last byte of the previous response (HTTP mode only). It's
Davor Ocelice9ed2812017-12-25 17:49:28 +010017030 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 +020017031 returns -1 as well. This timer used to be very useful before the arrival of
17032 HTTP keep-alive and browsers' pre-connect feature. It's recommended to drop
17033 it in favor of TR nowadays, as the idle time adds a lot of noise to the
17034 reports.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017035
17036 - Tw: total time spent in the queues waiting for a connection slot. It
17037 accounts for backend queue as well as the server queues, and depends on the
17038 queue size, and the time needed for the server to complete previous
17039 requests. The value "-1" means that the request was killed before reaching
17040 the queue, which is generally what happens with invalid or denied requests.
17041
17042 - Tc: total time to establish the TCP connection to the server. It's the time
17043 elapsed between the moment the proxy sent the connection request, and the
17044 moment it was acknowledged by the server, or between the TCP SYN packet and
17045 the matching SYN/ACK packet in return. The value "-1" means that the
17046 connection never established.
17047
17048 - Tr: server response time (HTTP mode only). It's the time elapsed between
17049 the moment the TCP connection was established to the server and the moment
17050 the server sent its complete response headers. It purely shows its request
17051 processing time, without the network overhead due to the data transmission.
17052 It is worth noting that when the client has data to send to the server, for
17053 instance during a POST request, the time already runs, and this can distort
17054 apparent response time. For this reason, it's generally wise not to trust
17055 too much this field for POST requests initiated from clients behind an
17056 untrusted network. A value of "-1" here means that the last the response
17057 header (empty line) was never seen, most likely because the server timeout
17058 stroke before the server managed to process the request.
17059
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017060 - Ta: total active time for the HTTP request, between the moment the proxy
17061 received the first byte of the request header and the emission of the last
17062 byte of the response body. The exception is when the "logasap" option is
17063 specified. In this case, it only equals (TR+Tw+Tc+Tr), and is prefixed with
17064 a '+' sign. From this field, we can deduce "Td", the data transmission time,
17065 by subtracting other timers when valid :
17066
17067 Td = Ta - (TR + Tw + Tc + Tr)
17068
17069 Timers with "-1" values have to be excluded from this equation. Note that
17070 "Ta" can never be negative.
17071
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017072 - Tt: total session duration time, between the moment the proxy accepted it
17073 and the moment both ends were closed. The exception is when the "logasap"
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017074 option is specified. In this case, it only equals (Th+Ti+TR+Tw+Tc+Tr), and
17075 is prefixed with a '+' sign. From this field, we can deduce "Td", the data
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017076 transmission time, by subtracting other timers when valid :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017077
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017078 Td = Tt - (Th + Ti + TR + Tw + Tc + Tr)
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017079
17080 Timers with "-1" values have to be excluded from this equation. In TCP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017081 mode, "Ti", "Tq" and "Tr" have to be excluded too. Note that "Tt" can never
17082 be negative and that for HTTP, Tt is simply equal to (Th+Ti+Ta).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017083
17084These timers provide precious indications on trouble causes. Since the TCP
17085protocol defines retransmit delays of 3, 6, 12... seconds, we know for sure
17086that timers close to multiples of 3s are nearly always related to lost packets
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017087due to network problems (wires, negotiation, congestion). Moreover, if "Ta" or
17088"Tt" is close to a timeout value specified in the configuration, it often means
17089that a session has been aborted on timeout.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017090
17091Most common cases :
17092
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017093 - If "Th" or "Ti" are close to 3000, a packet has probably been lost between
17094 the client and the proxy. This is very rare on local networks but might
17095 happen when clients are on far remote networks and send large requests. It
17096 may happen that values larger than usual appear here without any network
17097 cause. Sometimes, during an attack or just after a resource starvation has
17098 ended, haproxy may accept thousands of connections in a few milliseconds.
17099 The time spent accepting these connections will inevitably slightly delay
17100 processing of other connections, and it can happen that request times in the
17101 order of a few tens of milliseconds are measured after a few thousands of
17102 new connections have been accepted at once. Using one of the keep-alive
17103 modes may display larger idle times since "Ti" measures the time spent
Patrick Mezard105faca2010-06-12 17:02:46 +020017104 waiting for additional requests.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017105
17106 - If "Tc" is close to 3000, a packet has probably been lost between the
17107 server and the proxy during the server connection phase. This value should
17108 always be very low, such as 1 ms on local networks and less than a few tens
17109 of ms on remote networks.
17110
Willy Tarreau55165fe2009-05-10 12:02:55 +020017111 - If "Tr" is nearly always lower than 3000 except some rare values which seem
17112 to be the average majored by 3000, there are probably some packets lost
17113 between the proxy and the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017114
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017115 - If "Ta" is large even for small byte counts, it generally is because
17116 neither the client nor the server decides to close the connection while
17117 haproxy is running in tunnel mode and both have agreed on a keep-alive
17118 connection mode. In order to solve this issue, it will be needed to specify
17119 one of the HTTP options to manipulate keep-alive or close options on either
17120 the frontend or the backend. Having the smallest possible 'Ta' or 'Tt' is
17121 important when connection regulation is used with the "maxconn" option on
17122 the servers, since no new connection will be sent to the server until
17123 another one is released.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017124
17125Other noticeable HTTP log cases ('xx' means any value to be ignored) :
17126
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017127 TR/Tw/Tc/Tr/+Ta The "option logasap" is present on the frontend and the log
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017128 was emitted before the data phase. All the timers are valid
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017129 except "Ta" which is shorter than reality.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017130
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017131 -1/xx/xx/xx/Ta The client was not able to send a complete request in time
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017132 or it aborted too early. Check the session termination flags
17133 then "timeout http-request" and "timeout client" settings.
17134
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017135 TR/-1/xx/xx/Ta It was not possible to process the request, maybe because
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017136 servers were out of order, because the request was invalid
17137 or forbidden by ACL rules. Check the session termination
17138 flags.
17139
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017140 TR/Tw/-1/xx/Ta The connection could not establish on the server. Either it
17141 actively refused it or it timed out after Ta-(TR+Tw) ms.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017142 Check the session termination flags, then check the
17143 "timeout connect" setting. Note that the tarpit action might
17144 return similar-looking patterns, with "Tw" equal to the time
17145 the client connection was maintained open.
17146
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017147 TR/Tw/Tc/-1/Ta The server has accepted the connection but did not return
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017148 a complete response in time, or it closed its connection
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017149 unexpectedly after Ta-(TR+Tw+Tc) ms. Check the session
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017150 termination flags, then check the "timeout server" setting.
17151
17152
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200171538.5. Session state at disconnection
17154-----------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017155
17156TCP and HTTP logs provide a session termination indicator in the
17157"termination_state" field, just before the number of active connections. It is
171582-characters long in TCP mode, and is extended to 4 characters in HTTP mode,
17159each of which has a special meaning :
17160
17161 - On the first character, a code reporting the first event which caused the
17162 session to terminate :
17163
17164 C : the TCP session was unexpectedly aborted by the client.
17165
17166 S : the TCP session was unexpectedly aborted by the server, or the
17167 server explicitly refused it.
17168
17169 P : the session was prematurely aborted by the proxy, because of a
17170 connection limit enforcement, because a DENY filter was matched,
17171 because of a security check which detected and blocked a dangerous
17172 error in server response which might have caused information leak
Davor Ocelice9ed2812017-12-25 17:49:28 +010017173 (e.g. cacheable cookie).
Willy Tarreau570f2212013-06-10 16:42:09 +020017174
17175 L : the session was locally processed by haproxy and was not passed to
17176 a server. This is what happens for stats and redirects.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017177
17178 R : a resource on the proxy has been exhausted (memory, sockets, source
17179 ports, ...). Usually, this appears during the connection phase, and
17180 system logs should contain a copy of the precise error. If this
17181 happens, it must be considered as a very serious anomaly which
17182 should be fixed as soon as possible by any means.
17183
17184 I : an internal error was identified by the proxy during a self-check.
17185 This should NEVER happen, and you are encouraged to report any log
17186 containing this, because this would almost certainly be a bug. It
17187 would be wise to preventively restart the process after such an
17188 event too, in case it would be caused by memory corruption.
17189
Simon Horman752dc4a2011-06-21 14:34:59 +090017190 D : the session was killed by haproxy because the server was detected
17191 as down and was configured to kill all connections when going down.
17192
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070017193 U : the session was killed by haproxy on this backup server because an
17194 active server was detected as up and was configured to kill all
17195 backup connections when going up.
17196
Willy Tarreaua2a64e92011-09-07 23:01:56 +020017197 K : the session was actively killed by an admin operating on haproxy.
17198
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017199 c : the client-side timeout expired while waiting for the client to
17200 send or receive data.
17201
17202 s : the server-side timeout expired while waiting for the server to
17203 send or receive data.
17204
17205 - : normal session completion, both the client and the server closed
17206 with nothing left in the buffers.
17207
17208 - on the second character, the TCP or HTTP session state when it was closed :
17209
Willy Tarreauf7b30a92010-12-06 22:59:17 +010017210 R : the proxy was waiting for a complete, valid REQUEST from the client
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017211 (HTTP mode only). Nothing was sent to any server.
17212
17213 Q : the proxy was waiting in the QUEUE for a connection slot. This can
17214 only happen when servers have a 'maxconn' parameter set. It can
17215 also happen in the global queue after a redispatch consecutive to
17216 a failed attempt to connect to a dying server. If no redispatch is
17217 reported, then no connection attempt was made to any server.
17218
17219 C : the proxy was waiting for the CONNECTION to establish on the
17220 server. The server might at most have noticed a connection attempt.
17221
17222 H : the proxy was waiting for complete, valid response HEADERS from the
17223 server (HTTP only).
17224
17225 D : the session was in the DATA phase.
17226
17227 L : the proxy was still transmitting LAST data to the client while the
17228 server had already finished. This one is very rare as it can only
17229 happen when the client dies while receiving the last packets.
17230
17231 T : the request was tarpitted. It has been held open with the client
17232 during the whole "timeout tarpit" duration or until the client
17233 closed, both of which will be reported in the "Tw" timer.
17234
17235 - : normal session completion after end of data transfer.
17236
17237 - the third character tells whether the persistence cookie was provided by
17238 the client (only in HTTP mode) :
17239
17240 N : the client provided NO cookie. This is usually the case for new
17241 visitors, so counting the number of occurrences of this flag in the
17242 logs generally indicate a valid trend for the site frequentation.
17243
17244 I : the client provided an INVALID cookie matching no known server.
17245 This might be caused by a recent configuration change, mixed
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +020017246 cookies between HTTP/HTTPS sites, persistence conditionally
17247 ignored, or an attack.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017248
17249 D : the client provided a cookie designating a server which was DOWN,
17250 so either "option persist" was used and the client was sent to
17251 this server, or it was not set and the client was redispatched to
17252 another server.
17253
Willy Tarreau996a92c2010-10-13 19:30:47 +020017254 V : the client provided a VALID cookie, and was sent to the associated
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017255 server.
17256
Willy Tarreau996a92c2010-10-13 19:30:47 +020017257 E : the client provided a valid cookie, but with a last date which was
17258 older than what is allowed by the "maxidle" cookie parameter, so
17259 the cookie is consider EXPIRED and is ignored. The request will be
17260 redispatched just as if there was no cookie.
17261
17262 O : the client provided a valid cookie, but with a first date which was
17263 older than what is allowed by the "maxlife" cookie parameter, so
17264 the cookie is consider too OLD and is ignored. The request will be
17265 redispatched just as if there was no cookie.
17266
Willy Tarreauc89ccb62012-04-05 21:18:22 +020017267 U : a cookie was present but was not used to select the server because
17268 some other server selection mechanism was used instead (typically a
17269 "use-server" rule).
17270
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017271 - : does not apply (no cookie set in configuration).
17272
17273 - the last character reports what operations were performed on the persistence
17274 cookie returned by the server (only in HTTP mode) :
17275
17276 N : NO cookie was provided by the server, and none was inserted either.
17277
17278 I : no cookie was provided by the server, and the proxy INSERTED one.
17279 Note that in "cookie insert" mode, if the server provides a cookie,
17280 it will still be overwritten and reported as "I" here.
17281
Willy Tarreau996a92c2010-10-13 19:30:47 +020017282 U : the proxy UPDATED the last date in the cookie that was presented by
17283 the client. This can only happen in insert mode with "maxidle". It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017284 happens every time there is activity at a different date than the
Willy Tarreau996a92c2010-10-13 19:30:47 +020017285 date indicated in the cookie. If any other change happens, such as
17286 a redispatch, then the cookie will be marked as inserted instead.
17287
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017288 P : a cookie was PROVIDED by the server and transmitted as-is.
17289
17290 R : the cookie provided by the server was REWRITTEN by the proxy, which
17291 happens in "cookie rewrite" or "cookie prefix" modes.
17292
17293 D : the cookie provided by the server was DELETED by the proxy.
17294
17295 - : does not apply (no cookie set in configuration).
17296
Willy Tarreau996a92c2010-10-13 19:30:47 +020017297The combination of the two first flags gives a lot of information about what
17298was happening when the session terminated, and why it did terminate. It can be
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017299helpful to detect server saturation, network troubles, local system resource
17300starvation, attacks, etc...
17301
17302The most common termination flags combinations are indicated below. They are
17303alphabetically sorted, with the lowercase set just after the upper case for
17304easier finding and understanding.
17305
17306 Flags Reason
17307
17308 -- Normal termination.
17309
17310 CC The client aborted before the connection could be established to the
17311 server. This can happen when haproxy tries to connect to a recently
17312 dead (or unchecked) server, and the client aborts while haproxy is
17313 waiting for the server to respond or for "timeout connect" to expire.
17314
17315 CD The client unexpectedly aborted during data transfer. This can be
17316 caused by a browser crash, by an intermediate equipment between the
17317 client and haproxy which decided to actively break the connection,
17318 by network routing issues between the client and haproxy, or by a
17319 keep-alive session between the server and the client terminated first
17320 by the client.
Willy Tarreaud72758d2010-01-12 10:42:19 +010017321
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017322 cD The client did not send nor acknowledge any data for as long as the
17323 "timeout client" delay. This is often caused by network failures on
Cyril Bontédc4d9032012-04-08 21:57:39 +020017324 the client side, or the client simply leaving the net uncleanly.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017325
17326 CH The client aborted while waiting for the server to start responding.
17327 It might be the server taking too long to respond or the client
17328 clicking the 'Stop' button too fast.
17329
17330 cH The "timeout client" stroke while waiting for client data during a
17331 POST request. This is sometimes caused by too large TCP MSS values
17332 for PPPoE networks which cannot transport full-sized packets. It can
17333 also happen when client timeout is smaller than server timeout and
17334 the server takes too long to respond.
17335
17336 CQ The client aborted while its session was queued, waiting for a server
17337 with enough empty slots to accept it. It might be that either all the
17338 servers were saturated or that the assigned server was taking too
17339 long a time to respond.
17340
17341 CR The client aborted before sending a full HTTP request. Most likely
17342 the request was typed by hand using a telnet client, and aborted
17343 too early. The HTTP status code is likely a 400 here. Sometimes this
17344 might also be caused by an IDS killing the connection between haproxy
Willy Tarreau0f228a02015-05-01 15:37:53 +020017345 and the client. "option http-ignore-probes" can be used to ignore
17346 connections without any data transfer.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017347
17348 cR The "timeout http-request" stroke before the client sent a full HTTP
17349 request. This is sometimes caused by too large TCP MSS values on the
17350 client side for PPPoE networks which cannot transport full-sized
17351 packets, or by clients sending requests by hand and not typing fast
17352 enough, or forgetting to enter the empty line at the end of the
Willy Tarreau2705a612014-05-23 17:38:34 +020017353 request. The HTTP status code is likely a 408 here. Note: recently,
Willy Tarreau0f228a02015-05-01 15:37:53 +020017354 some browsers started to implement a "pre-connect" feature consisting
17355 in speculatively connecting to some recently visited web sites just
17356 in case the user would like to visit them. This results in many
17357 connections being established to web sites, which end up in 408
17358 Request Timeout if the timeout strikes first, or 400 Bad Request when
17359 the browser decides to close them first. These ones pollute the log
17360 and feed the error counters. Some versions of some browsers have even
17361 been reported to display the error code. It is possible to work
Davor Ocelice9ed2812017-12-25 17:49:28 +010017362 around the undesirable effects of this behavior by adding "option
Willy Tarreau0f228a02015-05-01 15:37:53 +020017363 http-ignore-probes" in the frontend, resulting in connections with
17364 zero data transfer to be totally ignored. This will definitely hide
17365 the errors of people experiencing connectivity issues though.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017366
17367 CT The client aborted while its session was tarpitted. It is important to
17368 check if this happens on valid requests, in order to be sure that no
Willy Tarreau55165fe2009-05-10 12:02:55 +020017369 wrong tarpit rules have been written. If a lot of them happen, it
17370 might make sense to lower the "timeout tarpit" value to something
17371 closer to the average reported "Tw" timer, in order not to consume
17372 resources for just a few attackers.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017373
Willy Tarreau570f2212013-06-10 16:42:09 +020017374 LR The request was intercepted and locally handled by haproxy. Generally
17375 it means that this was a redirect or a stats request.
17376
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010017377 SC The server or an equipment between it and haproxy explicitly refused
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017378 the TCP connection (the proxy received a TCP RST or an ICMP message
17379 in return). Under some circumstances, it can also be the network
Davor Ocelice9ed2812017-12-25 17:49:28 +010017380 stack telling the proxy that the server is unreachable (e.g. no route,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017381 or no ARP response on local network). When this happens in HTTP mode,
17382 the status code is likely a 502 or 503 here.
17383
17384 sC The "timeout connect" stroke before a connection to the server could
17385 complete. When this happens in HTTP mode, the status code is likely a
17386 503 or 504 here.
17387
17388 SD The connection to the server died with an error during the data
17389 transfer. This usually means that haproxy has received an RST from
17390 the server or an ICMP message from an intermediate equipment while
17391 exchanging data with the server. This can be caused by a server crash
17392 or by a network issue on an intermediate equipment.
17393
17394 sD The server did not send nor acknowledge any data for as long as the
17395 "timeout server" setting during the data phase. This is often caused
Davor Ocelice9ed2812017-12-25 17:49:28 +010017396 by too short timeouts on L4 equipment before the server (firewalls,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017397 load-balancers, ...), as well as keep-alive sessions maintained
17398 between the client and the server expiring first on haproxy.
17399
17400 SH The server aborted before sending its full HTTP response headers, or
17401 it crashed while processing the request. Since a server aborting at
17402 this moment is very rare, it would be wise to inspect its logs to
17403 control whether it crashed and why. The logged request may indicate a
17404 small set of faulty requests, demonstrating bugs in the application.
17405 Sometimes this might also be caused by an IDS killing the connection
17406 between haproxy and the server.
17407
17408 sH The "timeout server" stroke before the server could return its
17409 response headers. This is the most common anomaly, indicating too
17410 long transactions, probably caused by server or database saturation.
17411 The immediate workaround consists in increasing the "timeout server"
17412 setting, but it is important to keep in mind that the user experience
17413 will suffer from these long response times. The only long term
17414 solution is to fix the application.
17415
17416 sQ The session spent too much time in queue and has been expired. See
17417 the "timeout queue" and "timeout connect" settings to find out how to
17418 fix this if it happens too often. If it often happens massively in
17419 short periods, it may indicate general problems on the affected
17420 servers due to I/O or database congestion, or saturation caused by
17421 external attacks.
17422
17423 PC The proxy refused to establish a connection to the server because the
17424 process' socket limit has been reached while attempting to connect.
Cyril Bontédc4d9032012-04-08 21:57:39 +020017425 The global "maxconn" parameter may be increased in the configuration
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017426 so that it does not happen anymore. This status is very rare and
17427 might happen when the global "ulimit-n" parameter is forced by hand.
17428
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010017429 PD The proxy blocked an incorrectly formatted chunked encoded message in
17430 a request or a response, after the server has emitted its headers. In
17431 most cases, this will indicate an invalid message from the server to
Davor Ocelice9ed2812017-12-25 17:49:28 +010017432 the client. HAProxy supports chunk sizes of up to 2GB - 1 (2147483647
Willy Tarreauf3a3e132013-08-31 08:16:26 +020017433 bytes). Any larger size will be considered as an error.
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010017434
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017435 PH The proxy blocked the server's response, because it was invalid,
17436 incomplete, dangerous (cache control), or matched a security filter.
17437 In any case, an HTTP 502 error is sent to the client. One possible
17438 cause for this error is an invalid syntax in an HTTP header name
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010017439 containing unauthorized characters. It is also possible but quite
17440 rare, that the proxy blocked a chunked-encoding request from the
17441 client due to an invalid syntax, before the server responded. In this
17442 case, an HTTP 400 error is sent to the client and reported in the
17443 logs.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017444
17445 PR The proxy blocked the client's HTTP request, either because of an
17446 invalid HTTP syntax, in which case it returned an HTTP 400 error to
17447 the client, or because a deny filter matched, in which case it
17448 returned an HTTP 403 error.
17449
17450 PT The proxy blocked the client's request and has tarpitted its
17451 connection before returning it a 500 server error. Nothing was sent
17452 to the server. The connection was maintained open for as long as
17453 reported by the "Tw" timer field.
17454
17455 RC A local resource has been exhausted (memory, sockets, source ports)
17456 preventing the connection to the server from establishing. The error
17457 logs will tell precisely what was missing. This is very rare and can
17458 only be solved by proper system tuning.
17459
Willy Tarreau996a92c2010-10-13 19:30:47 +020017460The combination of the two last flags gives a lot of information about how
17461persistence was handled by the client, the server and by haproxy. This is very
17462important to troubleshoot disconnections, when users complain they have to
17463re-authenticate. The commonly encountered flags are :
17464
17465 -- Persistence cookie is not enabled.
17466
17467 NN No cookie was provided by the client, none was inserted in the
17468 response. For instance, this can be in insert mode with "postonly"
17469 set on a GET request.
17470
17471 II A cookie designating an invalid server was provided by the client,
17472 a valid one was inserted in the response. This typically happens when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017473 a "server" entry is removed from the configuration, since its cookie
Willy Tarreau996a92c2010-10-13 19:30:47 +020017474 value can be presented by a client when no other server knows it.
17475
17476 NI No cookie was provided by the client, one was inserted in the
17477 response. This typically happens for first requests from every user
17478 in "insert" mode, which makes it an easy way to count real users.
17479
17480 VN A cookie was provided by the client, none was inserted in the
17481 response. This happens for most responses for which the client has
17482 already got a cookie.
17483
17484 VU A cookie was provided by the client, with a last visit date which is
17485 not completely up-to-date, so an updated cookie was provided in
17486 response. This can also happen if there was no date at all, or if
17487 there was a date but the "maxidle" parameter was not set, so that the
17488 cookie can be switched to unlimited time.
17489
17490 EI A cookie was provided by the client, with a last visit date which is
17491 too old for the "maxidle" parameter, so the cookie was ignored and a
17492 new cookie was inserted in the response.
17493
17494 OI A cookie was provided by the client, with a first visit date which is
17495 too old for the "maxlife" parameter, so the cookie was ignored and a
17496 new cookie was inserted in the response.
17497
17498 DI The server designated by the cookie was down, a new server was
17499 selected and a new cookie was emitted in the response.
17500
17501 VI The server designated by the cookie was not marked dead but could not
17502 be reached. A redispatch happened and selected another one, which was
17503 then advertised in the response.
17504
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017505
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200175068.6. Non-printable characters
17507-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017508
17509In order not to cause trouble to log analysis tools or terminals during log
17510consulting, non-printable characters are not sent as-is into log files, but are
17511converted to the two-digits hexadecimal representation of their ASCII code,
17512prefixed by the character '#'. The only characters that can be logged without
17513being escaped are comprised between 32 and 126 (inclusive). Obviously, the
17514escape character '#' itself is also encoded to avoid any ambiguity ("#23"). It
17515is the same for the character '"' which becomes "#22", as well as '{', '|' and
17516'}' when logging headers.
17517
17518Note that the space character (' ') is not encoded in headers, which can cause
17519issues for tools relying on space count to locate fields. A typical header
17520containing spaces is "User-Agent".
17521
17522Last, it has been observed that some syslog daemons such as syslog-ng escape
17523the quote ('"') with a backslash ('\'). The reverse operation can safely be
17524performed since no quote may appear anywhere else in the logs.
17525
17526
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200175278.7. Capturing HTTP cookies
17528---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017529
17530Cookie capture simplifies the tracking a complete user session. This can be
17531achieved using the "capture cookie" statement in the frontend. Please refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017532section 4.2 for more details. Only one cookie can be captured, and the same
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017533cookie will simultaneously be checked in the request ("Cookie:" header) and in
17534the response ("Set-Cookie:" header). The respective values will be reported in
17535the HTTP logs at the "captured_request_cookie" and "captured_response_cookie"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017536locations (see section 8.2.3 about HTTP log format). When either cookie is
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017537not seen, a dash ('-') replaces the value. This way, it's easy to detect when a
17538user switches to a new session for example, because the server will reassign it
17539a new cookie. It is also possible to detect if a server unexpectedly sets a
17540wrong cookie to a client, leading to session crossing.
17541
17542 Examples :
17543 # capture the first cookie whose name starts with "ASPSESSION"
17544 capture cookie ASPSESSION len 32
17545
17546 # capture the first cookie whose name is exactly "vgnvisitor"
17547 capture cookie vgnvisitor= len 32
17548
17549
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200175508.8. Capturing HTTP headers
17551---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017552
17553Header captures are useful to track unique request identifiers set by an upper
17554proxy, virtual host names, user-agents, POST content-length, referrers, etc. In
17555the response, one can search for information about the response length, how the
17556server asked the cache to behave, or an object location during a redirection.
17557
17558Header captures are performed using the "capture request header" and "capture
17559response header" statements in the frontend. Please consult their definition in
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017560section 4.2 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017561
17562It is possible to include both request headers and response headers at the same
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010017563time. Non-existent headers are logged as empty strings, and if one header
17564appears more than once, only its last occurrence will be logged. Request headers
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017565are grouped within braces '{' and '}' in the same order as they were declared,
17566and delimited with a vertical bar '|' without any space. Response headers
17567follow the same representation, but are displayed after a space following the
17568request headers block. These blocks are displayed just before the HTTP request
17569in the logs.
17570
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020017571As a special case, it is possible to specify an HTTP header capture in a TCP
17572frontend. The purpose is to enable logging of headers which will be parsed in
17573an HTTP backend if the request is then switched to this HTTP backend.
17574
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017575 Example :
17576 # This instance chains to the outgoing proxy
17577 listen proxy-out
17578 mode http
17579 option httplog
17580 option logasap
17581 log global
17582 server cache1 192.168.1.1:3128
17583
17584 # log the name of the virtual server
17585 capture request header Host len 20
17586
17587 # log the amount of data uploaded during a POST
17588 capture request header Content-Length len 10
17589
17590 # log the beginning of the referrer
17591 capture request header Referer len 20
17592
17593 # server name (useful for outgoing proxies only)
17594 capture response header Server len 20
17595
17596 # logging the content-length is useful with "option logasap"
17597 capture response header Content-Length len 10
17598
Davor Ocelice9ed2812017-12-25 17:49:28 +010017599 # log the expected cache behavior on the response
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017600 capture response header Cache-Control len 8
17601
17602 # the Via header will report the next proxy's name
17603 capture response header Via len 20
17604
17605 # log the URL location during a redirection
17606 capture response header Location len 20
17607
17608 >>> Aug 9 20:26:09 localhost \
17609 haproxy[2022]: 127.0.0.1:34014 [09/Aug/2004:20:26:09] proxy-out \
17610 proxy-out/cache1 0/0/0/162/+162 200 +350 - - ---- 0/0/0/0/0 0/0 \
17611 {fr.adserver.yahoo.co||http://fr.f416.mail.} {|864|private||} \
17612 "GET http://fr.adserver.yahoo.com/"
17613
17614 >>> Aug 9 20:30:46 localhost \
17615 haproxy[2022]: 127.0.0.1:34020 [09/Aug/2004:20:30:46] proxy-out \
17616 proxy-out/cache1 0/0/0/182/+182 200 +279 - - ---- 0/0/0/0/0 0/0 \
17617 {w.ods.org||} {Formilux/0.1.8|3495|||} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017618 "GET http://trafic.1wt.eu/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017619
17620 >>> Aug 9 20:30:46 localhost \
17621 haproxy[2022]: 127.0.0.1:34028 [09/Aug/2004:20:30:46] proxy-out \
17622 proxy-out/cache1 0/0/2/126/+128 301 +223 - - ---- 0/0/0/0/0 0/0 \
17623 {www.sytadin.equipement.gouv.fr||http://trafic.1wt.eu/} \
17624 {Apache|230|||http://www.sytadin.} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017625 "GET http://www.sytadin.equipement.gouv.fr/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017626
17627
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200176288.9. Examples of logs
17629---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017630
17631These are real-world examples of logs accompanied with an explanation. Some of
17632them have been made up by hand. The syslog part has been removed for better
17633reading. Their sole purpose is to explain how to decipher them.
17634
17635 >>> haproxy[674]: 127.0.0.1:33318 [15/Oct/2003:08:31:57.130] px-http \
17636 px-http/srv1 6559/0/7/147/6723 200 243 - - ---- 5/3/3/1/0 0/0 \
17637 "HEAD / HTTP/1.0"
17638
17639 => long request (6.5s) entered by hand through 'telnet'. The server replied
17640 in 147 ms, and the session ended normally ('----')
17641
17642 >>> haproxy[674]: 127.0.0.1:33319 [15/Oct/2003:08:31:57.149] px-http \
17643 px-http/srv1 6559/1230/7/147/6870 200 243 - - ---- 324/239/239/99/0 \
17644 0/9 "HEAD / HTTP/1.0"
17645
17646 => Idem, but the request was queued in the global queue behind 9 other
17647 requests, and waited there for 1230 ms.
17648
17649 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.654] px-http \
17650 px-http/srv1 9/0/7/14/+30 200 +243 - - ---- 3/3/3/1/0 0/0 \
17651 "GET /image.iso HTTP/1.0"
17652
17653 => request for a long data transfer. The "logasap" option was specified, so
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010017654 the log was produced just before transferring data. The server replied in
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017655 14 ms, 243 bytes of headers were sent to the client, and total time from
17656 accept to first data byte is 30 ms.
17657
17658 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.925] px-http \
17659 px-http/srv1 9/0/7/14/30 502 243 - - PH-- 3/2/2/0/0 0/0 \
17660 "GET /cgi-bin/bug.cgi? HTTP/1.0"
17661
17662 => the proxy blocked a server response either because of an "rspdeny" or
17663 "rspideny" filter, or because the response was improperly formatted and
Willy Tarreau3c92c5f2011-08-28 09:45:47 +020017664 not HTTP-compliant, or because it blocked sensitive information which
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017665 risked being cached. In this case, the response is replaced with a "502
17666 bad gateway". The flags ("PH--") tell us that it was haproxy who decided
17667 to return the 502 and not the server.
17668
17669 >>> haproxy[18113]: 127.0.0.1:34548 [15/Oct/2003:15:18:55.798] px-http \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017670 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 +010017671
17672 => the client never completed its request and aborted itself ("C---") after
17673 8.5s, while the proxy was waiting for the request headers ("-R--").
17674 Nothing was sent to any server.
17675
17676 >>> haproxy[18113]: 127.0.0.1:34549 [15/Oct/2003:15:19:06.103] px-http \
17677 px-http/<NOSRV> -1/-1/-1/-1/50001 408 0 - - cR-- 2/2/2/0/0 0/0 ""
17678
17679 => The client never completed its request, which was aborted by the
17680 time-out ("c---") after 50s, while the proxy was waiting for the request
Davor Ocelice9ed2812017-12-25 17:49:28 +010017681 headers ("-R--"). Nothing was sent to any server, but the proxy could
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017682 send a 408 return code to the client.
17683
17684 >>> haproxy[18989]: 127.0.0.1:34550 [15/Oct/2003:15:24:28.312] px-tcp \
17685 px-tcp/srv1 0/0/5007 0 cD 0/0/0/0/0 0/0
17686
17687 => This log was produced with "option tcplog". The client timed out after
17688 5 seconds ("c----").
17689
17690 >>> haproxy[18989]: 10.0.0.1:34552 [15/Oct/2003:15:26:31.462] px-http \
17691 px-http/srv1 3183/-1/-1/-1/11215 503 0 - - SC-- 205/202/202/115/3 \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017692 0/0 "HEAD / HTTP/1.0"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017693
17694 => The request took 3s to complete (probably a network problem), and the
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017695 connection to the server failed ('SC--') after 4 attempts of 2 seconds
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017696 (config says 'retries 3'), and no redispatch (otherwise we would have
17697 seen "/+3"). Status code 503 was returned to the client. There were 115
17698 connections on this server, 202 connections on this proxy, and 205 on
17699 the global process. It is possible that the server refused the
17700 connection because of too many already established.
Willy Tarreau844e3c52008-01-11 16:28:18 +010017701
Willy Tarreau52b2d222011-09-07 23:48:48 +020017702
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200177039. Supported filters
17704--------------------
17705
17706Here are listed officially supported filters with the list of parameters they
17707accept. Depending on compile options, some of these filters might be
17708unavailable. The list of available filters is reported in haproxy -vv.
17709
17710See also : "filter"
17711
177129.1. Trace
17713----------
17714
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010017715filter trace [name <name>] [random-parsing] [random-forwarding] [hexdump]
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017716
17717 Arguments:
17718 <name> is an arbitrary name that will be reported in
17719 messages. If no name is provided, "TRACE" is used.
17720
17721 <random-parsing> enables the random parsing of data exchanged between
17722 the client and the server. By default, this filter
17723 parses all available data. With this parameter, it
17724 only parses a random amount of the available data.
17725
Davor Ocelice9ed2812017-12-25 17:49:28 +010017726 <random-forwarding> enables the random forwarding of parsed data. By
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017727 default, this filter forwards all previously parsed
17728 data. With this parameter, it only forwards a random
17729 amount of the parsed data.
17730
Davor Ocelice9ed2812017-12-25 17:49:28 +010017731 <hexdump> dumps all forwarded data to the server and the client.
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010017732
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017733This filter can be used as a base to develop new filters. It defines all
17734callbacks and print a message on the standard error stream (stderr) with useful
17735information for all of them. It may be useful to debug the activity of other
17736filters or, quite simply, HAProxy's activity.
17737
17738Using <random-parsing> and/or <random-forwarding> parameters is a good way to
17739tests the behavior of a filter that parses data exchanged between a client and
17740a server by adding some latencies in the processing.
17741
17742
177439.2. HTTP compression
17744---------------------
17745
17746filter compression
17747
17748The HTTP compression has been moved in a filter in HAProxy 1.7. "compression"
17749keyword must still be used to enable and configure the HTTP compression. And
Christopher Faulet27d93c32018-12-15 22:32:02 +010017750when no other filter is used, it is enough. When used with the cache enabled,
17751it is also enough. In this case, the compression is always done after the
17752response is stored in the cache. But it is mandatory to explicitly use a filter
17753line to enable the HTTP compression when at least one filter other than the
17754cache is used for the same listener/frontend/backend. This is important to know
17755the filters evaluation order.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017756
Christopher Faulet27d93c32018-12-15 22:32:02 +010017757See also : "compression" and section 9.4 about the cache filter.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017758
17759
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +0200177609.3. Stream Processing Offload Engine (SPOE)
17761--------------------------------------------
17762
17763filter spoe [engine <name>] config <file>
17764
17765 Arguments :
17766
17767 <name> is the engine name that will be used to find the right scope in
17768 the configuration file. If not provided, all the file will be
17769 parsed.
17770
17771 <file> is the path of the engine configuration file. This file can
17772 contain configuration of several engines. In this case, each
17773 part must be placed in its own scope.
17774
17775The Stream Processing Offload Engine (SPOE) is a filter communicating with
17776external components. It allows the offload of some specifics processing on the
Davor Ocelice9ed2812017-12-25 17:49:28 +010017777streams in tiered applications. These external components and information
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020017778exchanged with them are configured in dedicated files, for the main part. It
17779also requires dedicated backends, defined in HAProxy configuration.
17780
17781SPOE communicates with external components using an in-house binary protocol,
17782the Stream Processing Offload Protocol (SPOP).
17783
Tim Düsterhus4896c442016-11-29 02:15:19 +010017784For all information about the SPOE configuration and the SPOP specification, see
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020017785"doc/SPOE.txt".
17786
17787Important note:
17788 The SPOE filter is highly experimental for now and was not heavily
17789 tested. It is really not production ready. So use it carefully.
17790
Christopher Faulet99a17a22018-12-11 09:18:27 +0100177919.4. Cache
17792----------
17793
17794filter cache <name>
17795
17796 Arguments :
17797
17798 <name> is name of the cache section this filter will use.
17799
17800The cache uses a filter to store cacheable responses. The HTTP rules
17801"cache-store" and "cache-use" must be used to define how and when to use a
17802cache. By default the correpsonding filter is implicitly defined. And when no
Christopher Faulet27d93c32018-12-15 22:32:02 +010017803other filters than cache or compression are used, it is enough. In such case,
17804the compression filter is always evaluated after the cache filter. But it is
17805mandatory to explicitly use a filter line to use a cache when at least one
17806filter other than the compression is used for the same
17807listener/frontend/backend. This is important to know the filters evaluation
17808order.
Christopher Faulet99a17a22018-12-11 09:18:27 +010017809
Christopher Faulet27d93c32018-12-15 22:32:02 +010017810See also : section 9.2 about the compression filter and section 10 about cache.
Christopher Faulet99a17a22018-12-11 09:18:27 +010017811
William Lallemand86d0df02017-11-24 21:36:45 +01001781210. Cache
17813---------
17814
17815HAProxy provides a cache, which was designed to perform cache on small objects
17816(favicon, css...). This is a minimalist low-maintenance cache which runs in
17817RAM.
17818
17819The cache is based on a memory which is shared between processes and threads,
Cyril Bonté7b888f12017-11-26 22:24:31 +010017820this memory is split in blocks of 1k.
William Lallemand86d0df02017-11-24 21:36:45 +010017821
17822If an object is not used anymore, it can be deleted to store a new object
17823independently of its expiration date. The oldest objects are deleted first
17824when we try to allocate a new one.
17825
Cyril Bonté7b888f12017-11-26 22:24:31 +010017826The cache uses a hash of the host header and the URI as the key.
William Lallemand86d0df02017-11-24 21:36:45 +010017827
17828It's possible to view the status of a cache using the Unix socket command
17829"show cache" consult section 9.3 "Unix Socket commands" of Management Guide
17830for more details.
17831
17832When an object is delivered from the cache, the server name in the log is
17833replaced by "<CACHE>".
17834
Cyril Bonté7b888f12017-11-26 22:24:31 +01001783510.1. Limitation
17836----------------
William Lallemand86d0df02017-11-24 21:36:45 +010017837
17838The cache won't store and won't deliver objects in these cases:
17839
17840- If the response is not a 200
17841- If the response contains a Vary header
Frédéric Lécaille5f8bea62018-10-23 10:09:19 +020017842- If the Content-Length + the headers size is greater than "max-object-size"
William Lallemand86d0df02017-11-24 21:36:45 +010017843- If the response is not cacheable
17844
17845- If the request is not a GET
17846- If the HTTP version of the request is smaller than 1.1
William Lallemand8a16fe02018-05-22 11:04:33 +020017847- If the request contains an Authorization header
William Lallemand86d0df02017-11-24 21:36:45 +010017848
Christopher Faulet99a17a22018-12-11 09:18:27 +010017849Caution!: For HAProxy version prior to 1.9, due to the limitation of the
17850filters, it is not recommended to use the cache with other filters. Using them
17851can cause undefined behavior if they modify the response (compression for
17852example). For HAProxy 1.9 and greater, it is safe, for HTX proxies only (see
17853"option http-use-htx" for details).
William Lallemand86d0df02017-11-24 21:36:45 +010017854
Cyril Bonté7b888f12017-11-26 22:24:31 +01001785510.2. Setup
17856-----------
William Lallemand86d0df02017-11-24 21:36:45 +010017857
17858To setup a cache, you must define a cache section and use it in a proxy with
17859the corresponding http-request and response actions.
17860
Cyril Bonté7b888f12017-11-26 22:24:31 +01001786110.2.1. Cache section
17862---------------------
William Lallemand86d0df02017-11-24 21:36:45 +010017863
17864cache <name>
17865 Declare a cache section, allocate a shared cache memory named <name>, the
17866 size of cache is mandatory.
17867
17868total-max-size <megabytes>
Davor Ocelice9ed2812017-12-25 17:49:28 +010017869 Define the size in RAM of the cache in megabytes. This size is split in
Frédéric Lécaillee3c83d82018-10-25 10:46:40 +020017870 blocks of 1kB which are used by the cache entries. Its maximum value is 4095.
William Lallemand86d0df02017-11-24 21:36:45 +010017871
Frédéric Lécaille5f8bea62018-10-23 10:09:19 +020017872max-object-size <bytes>
Frédéric Lécaillee3c83d82018-10-25 10:46:40 +020017873 Define the maximum size of the objects to be cached. Must not be greater than
17874 an half of "total-max-size". If not set, it equals to a 256th of the cache size.
17875 All objects with sizes larger than "max-object-size" will not be cached.
Frédéric Lécaille5f8bea62018-10-23 10:09:19 +020017876
William Lallemand86d0df02017-11-24 21:36:45 +010017877max-age <seconds>
17878 Define the maximum expiration duration. The expiration is set has the lowest
17879 value between the s-maxage or max-age (in this order) directive in the
17880 Cache-Control response header and this value. The default value is 60
17881 seconds, which means that you can't cache an object more than 60 seconds by
17882 default.
17883
Cyril Bonté7b888f12017-11-26 22:24:31 +01001788410.2.2. Proxy section
17885---------------------
William Lallemand86d0df02017-11-24 21:36:45 +010017886
17887http-request cache-use <name>
17888 Try to deliver a cached object from the cache <name>. This directive is also
17889 mandatory to store the cache as it calculates the cache hash. If you want to
17890 use a condition for both storage and delivering that's a good idea to put it
17891 after this one.
17892
17893http-response cache-store <name>
17894 Store an http-response within the cache. The storage of the response headers
17895 is done at this step, which means you can use others http-response actions
17896 to modify headers before or after the storage of the response. This action
17897 is responsible for the setup of the cache storage filter.
17898
17899
17900Example:
17901
17902 backend bck1
17903 mode http
17904
17905 http-request cache-use foobar
17906 http-response cache-store foobar
17907 server srv1 127.0.0.1:80
17908
17909 cache foobar
17910 total-max-size 4
17911 max-age 240
17912
Willy Tarreau0ba27502007-12-24 16:55:16 +010017913/*
17914 * Local variables:
17915 * fill-column: 79
17916 * End:
17917 */