blob: 2a7a9a508b1ffe197d4174729f3b714a46ef397d [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 Tarreau1db55792020-11-05 17:20:35 +01005 version 2.4
Willy Tarreau6a06a402007-07-15 20:15:28 +02006 willy tarreau
Willy Tarreau1c0a7222020-11-05 17:04:53 +01007 2020/11/05
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
William Lallemandc9515522019-06-12 16:32:11 +0200543.7. Programs
Christopher Faulet76edc0f2020-01-13 15:52:01 +0100553.8. HTTP-errors
Emeric Brun99c453d2020-05-25 15:01:04 +0200563.9. Rings
Willy Tarreauc57f0e22009-05-10 13:12:33 +020057
584. Proxies
594.1. Proxy keywords matrix
604.2. Alphabetically sorted keywords reference
61
Davor Ocelice9ed2812017-12-25 17:49:28 +0100625. Bind and server options
Willy Tarreau086fbf52012-09-24 20:34:51 +0200635.1. Bind options
645.2. Server and default-server options
Baptiste Assmann1fa66662015-04-14 00:28:47 +0200655.3. Server DNS resolution
665.3.1. Global overview
675.3.2. The resolvers section
Willy Tarreauc57f0e22009-05-10 13:12:33 +020068
Julien Pivotto6ccee412019-11-27 15:49:54 +0100696. Cache
706.1. Limitation
716.2. Setup
726.2.1. Cache section
736.2.2. Proxy section
74
Willy Tarreau74ca5042013-06-11 23:12:07 +0200757. Using ACLs and fetching samples
767.1. ACL basics
777.1.1. Matching booleans
787.1.2. Matching integers
797.1.3. Matching strings
807.1.4. Matching regular expressions (regexes)
817.1.5. Matching arbitrary data blocks
827.1.6. Matching IPv4 and IPv6 addresses
837.2. Using ACLs to form conditions
847.3. Fetching samples
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200857.3.1. Converters
867.3.2. Fetching samples from internal states
877.3.3. Fetching samples at Layer 4
887.3.4. Fetching samples at Layer 5
897.3.5. Fetching samples from buffer contents (Layer 6)
907.3.6. Fetching HTTP samples (Layer 7)
Christopher Faulete596d182020-05-05 17:46:34 +0200917.3.7. Fetching samples for developers
Willy Tarreau74ca5042013-06-11 23:12:07 +0200927.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020093
948. Logging
958.1. Log levels
968.2. Log formats
978.2.1. Default log format
988.2.2. TCP log format
998.2.3. HTTP log format
William Lallemand48940402012-01-30 16:47:22 +01001008.2.4. Custom log format
Willy Tarreau5f51e1a2012-12-03 18:40:10 +01001018.2.5. Error log format
Willy Tarreauc57f0e22009-05-10 13:12:33 +02001028.3. Advanced logging options
1038.3.1. Disabling logging of external tests
1048.3.2. Logging before waiting for the session to terminate
1058.3.3. Raising log level upon errors
1068.3.4. Disabling logging of successful connections
1078.4. Timing events
1088.5. Session state at disconnection
1098.6. Non-printable characters
1108.7. Capturing HTTP cookies
1118.8. Capturing HTTP headers
1128.9. Examples of logs
113
Christopher Fauletc3fe5332016-04-07 15:30:10 +02001149. Supported filters
1159.1. Trace
1169.2. HTTP compression
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +02001179.3. Stream Processing Offload Engine (SPOE)
Christopher Faulet99a17a22018-12-11 09:18:27 +01001189.4. Cache
Christopher Fauletb30b3102019-09-12 23:03:09 +02001199.5. fcgi-app
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200120
Christopher Fauletb30b3102019-09-12 23:03:09 +020012110. FastCGI applications
12210.1. Setup
12310.1.1. Fcgi-app section
12410.1.2. Proxy section
12510.1.3. Example
12610.2. Default parameters
12710.3. Limitations
128
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200129
1301. Quick reminder about HTTP
131----------------------------
132
Davor Ocelice9ed2812017-12-25 17:49:28 +0100133When HAProxy is running in HTTP mode, both the request and the response are
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200134fully analyzed and indexed, thus it becomes possible to build matching criteria
135on almost anything found in the contents.
136
137However, it is important to understand how HTTP requests and responses are
138formed, and how HAProxy decomposes them. It will then become easier to write
139correct rules and to debug existing configurations.
140
141
1421.1. The HTTP transaction model
143-------------------------------
144
145The HTTP protocol is transaction-driven. This means that each request will lead
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100146to one and only one response. Traditionally, a TCP connection is established
Davor Ocelice9ed2812017-12-25 17:49:28 +0100147from the client to the server, a request is sent by the client through the
148connection, the server responds, and the connection is closed. A new request
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200149will involve a new connection :
150
151 [CON1] [REQ1] ... [RESP1] [CLO1] [CON2] [REQ2] ... [RESP2] [CLO2] ...
152
153In this mode, called the "HTTP close" mode, there are as many connection
154establishments as there are HTTP transactions. Since the connection is closed
155by the server after the response, the client does not need to know the content
156length.
157
158Due to the transactional nature of the protocol, it was possible to improve it
159to avoid closing a connection between two subsequent transactions. In this mode
160however, it is mandatory that the server indicates the content length for each
161response so that the client does not wait indefinitely. For this, a special
162header is used: "Content-length". This mode is called the "keep-alive" mode :
163
164 [CON] [REQ1] ... [RESP1] [REQ2] ... [RESP2] [CLO] ...
165
166Its advantages are a reduced latency between transactions, and less processing
167power required on the server side. It is generally better than the close mode,
168but not always because the clients often limit their concurrent connections to
Patrick Mezard9ec2ec42010-06-12 17:02:45 +0200169a smaller value.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200170
Willy Tarreau95c4e142017-11-26 12:18:55 +0100171Another improvement in the communications is the pipelining mode. It still uses
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200172keep-alive, but the client does not wait for the first response to send the
173second request. This is useful for fetching large number of images composing a
174page :
175
176 [CON] [REQ1] [REQ2] ... [RESP1] [RESP2] [CLO] ...
177
178This can obviously have a tremendous benefit on performance because the network
179latency is eliminated between subsequent requests. Many HTTP agents do not
180correctly support pipelining since there is no way to associate a response with
181the corresponding request in HTTP. For this reason, it is mandatory for the
Cyril Bonté78caf842010-03-10 22:41:43 +0100182server to reply in the exact same order as the requests were received.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200183
Willy Tarreau95c4e142017-11-26 12:18:55 +0100184The next improvement is the multiplexed mode, as implemented in HTTP/2. This
185time, each transaction is assigned a single stream identifier, and all streams
186are multiplexed over an existing connection. Many requests can be sent in
187parallel by the client, and responses can arrive in any order since they also
188carry the stream identifier.
189
Willy Tarreau70dffda2014-01-30 03:07:23 +0100190By default HAProxy operates in keep-alive mode with regards to persistent
191connections: for each connection it processes each request and response, and
192leaves the connection idle on both sides between the end of a response and the
Willy Tarreau95c4e142017-11-26 12:18:55 +0100193start of a new request. When it receives HTTP/2 connections from a client, it
194processes all the requests in parallel and leaves the connection idling,
195waiting for new requests, just as if it was a keep-alive HTTP connection.
Patrick Mezard9ec2ec42010-06-12 17:02:45 +0200196
Christopher Faulet315b39c2018-09-21 16:26:19 +0200197HAProxy supports 4 connection modes :
Willy Tarreau70dffda2014-01-30 03:07:23 +0100198 - keep alive : all requests and responses are processed (default)
199 - tunnel : only the first request and response are processed,
Christopher Faulet6c9bbb22019-03-26 21:37:23 +0100200 everything else is forwarded with no analysis (deprecated).
Willy Tarreau70dffda2014-01-30 03:07:23 +0100201 - server close : the server-facing connection is closed after the response.
Christopher Faulet315b39c2018-09-21 16:26:19 +0200202 - close : the connection is actively closed after end of response.
Willy Tarreau70dffda2014-01-30 03:07:23 +0100203
Willy Tarreau95c4e142017-11-26 12:18:55 +0100204
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200205
2061.2. HTTP request
207-----------------
208
209First, let's consider this HTTP request :
210
211 Line Contents
Willy Tarreaud72758d2010-01-12 10:42:19 +0100212 number
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200213 1 GET /serv/login.php?lang=en&profile=2 HTTP/1.1
214 2 Host: www.mydomain.com
215 3 User-agent: my small browser
216 4 Accept: image/jpeg, image/gif
217 5 Accept: image/png
218
219
2201.2.1. The Request line
221-----------------------
222
223Line 1 is the "request line". It is always composed of 3 fields :
224
225 - a METHOD : GET
226 - a URI : /serv/login.php?lang=en&profile=2
227 - a version tag : HTTP/1.1
228
229All of them are delimited by what the standard calls LWS (linear white spaces),
230which are commonly spaces, but can also be tabs or line feeds/carriage returns
231followed by spaces/tabs. The method itself cannot contain any colon (':') and
232is limited to alphabetic letters. All those various combinations make it
233desirable that HAProxy performs the splitting itself rather than leaving it to
234the user to write a complex or inaccurate regular expression.
235
236The URI itself can have several forms :
237
238 - A "relative URI" :
239
240 /serv/login.php?lang=en&profile=2
241
242 It is a complete URL without the host part. This is generally what is
243 received by servers, reverse proxies and transparent proxies.
244
245 - An "absolute URI", also called a "URL" :
246
247 http://192.168.0.12:8080/serv/login.php?lang=en&profile=2
248
249 It is composed of a "scheme" (the protocol name followed by '://'), a host
250 name or address, optionally a colon (':') followed by a port number, then
251 a relative URI beginning at the first slash ('/') after the address part.
252 This is generally what proxies receive, but a server supporting HTTP/1.1
253 must accept this form too.
254
255 - a star ('*') : this form is only accepted in association with the OPTIONS
256 method and is not relayable. It is used to inquiry a next hop's
257 capabilities.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100258
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200259 - an address:port combination : 192.168.0.12:80
260 This is used with the CONNECT method, which is used to establish TCP
261 tunnels through HTTP proxies, generally for HTTPS, but sometimes for
262 other protocols too.
263
264In a relative URI, two sub-parts are identified. The part before the question
265mark is called the "path". It is typically the relative path to static objects
266on the server. The part after the question mark is called the "query string".
267It is mostly used with GET requests sent to dynamic scripts and is very
268specific to the language, framework or application in use.
269
Willy Tarreau95c4e142017-11-26 12:18:55 +0100270HTTP/2 doesn't convey a version information with the request, so the version is
Davor Ocelice9ed2812017-12-25 17:49:28 +0100271assumed to be the same as the one of the underlying protocol (i.e. "HTTP/2").
Willy Tarreau95c4e142017-11-26 12:18:55 +0100272
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200273
2741.2.2. The request headers
275--------------------------
276
277The headers start at the second line. They are composed of a name at the
278beginning of the line, immediately followed by a colon (':'). Traditionally,
279an LWS is added after the colon but that's not required. Then come the values.
280Multiple identical headers may be folded into one single line, delimiting the
281values with commas, provided that their order is respected. This is commonly
282encountered in the "Cookie:" field. A header may span over multiple lines if
283the subsequent lines begin with an LWS. In the example in 1.2, lines 4 and 5
284define a total of 3 values for the "Accept:" header.
285
Davor Ocelice9ed2812017-12-25 17:49:28 +0100286Contrary to a common misconception, header names are not case-sensitive, and
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200287their values are not either if they refer to other header names (such as the
Willy Tarreau95c4e142017-11-26 12:18:55 +0100288"Connection:" header). In HTTP/2, header names are always sent in lower case,
Willy Tarreau253c2512020-07-07 15:55:23 +0200289as can be seen when running in debug mode. Internally, all header names are
290normalized to lower case so that HTTP/1.x and HTTP/2 use the exact same
291representation, and they are sent as-is on the other side. This explains why an
292HTTP/1.x request typed with camel case is delivered in lower case.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200293
294The end of the headers is indicated by the first empty line. People often say
295that it's a double line feed, which is not exact, even if a double line feed
296is one valid form of empty line.
297
298Fortunately, HAProxy takes care of all these complex combinations when indexing
299headers, checking values and counting them, so there is no reason to worry
300about the way they could be written, but it is important not to accuse an
301application of being buggy if it does unusual, valid things.
302
303Important note:
Lukas Tribus23953682017-04-28 13:24:30 +0000304 As suggested by RFC7231, HAProxy normalizes headers by replacing line breaks
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200305 in the middle of headers by LWS in order to join multi-line headers. This
306 is necessary for proper analysis and helps less capable HTTP parsers to work
307 correctly and not to be fooled by such complex constructs.
308
309
3101.3. HTTP response
311------------------
312
313An HTTP response looks very much like an HTTP request. Both are called HTTP
314messages. Let's consider this HTTP response :
315
316 Line Contents
Willy Tarreaud72758d2010-01-12 10:42:19 +0100317 number
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200318 1 HTTP/1.1 200 OK
319 2 Content-length: 350
320 3 Content-Type: text/html
321
Willy Tarreau816b9792009-09-15 21:25:21 +0200322As a special case, HTTP supports so called "Informational responses" as status
323codes 1xx. These messages are special in that they don't convey any part of the
324response, they're just used as sort of a signaling message to ask a client to
Willy Tarreau5843d1a2010-02-01 15:13:32 +0100325continue to post its request for instance. In the case of a status 100 response
326the requested information will be carried by the next non-100 response message
327following the informational one. This implies that multiple responses may be
328sent to a single request, and that this only works when keep-alive is enabled
329(1xx messages are HTTP/1.1 only). HAProxy handles these messages and is able to
330correctly forward and skip them, and only process the next non-100 response. As
331such, these messages are neither logged nor transformed, unless explicitly
332state otherwise. Status 101 messages indicate that the protocol is changing
333over the same connection and that haproxy must switch to tunnel mode, just as
334if a CONNECT had occurred. Then the Upgrade header would contain additional
335information about the type of protocol the connection is switching to.
Willy Tarreau816b9792009-09-15 21:25:21 +0200336
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200337
Davor Ocelice9ed2812017-12-25 17:49:28 +01003381.3.1. The response line
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200339------------------------
340
341Line 1 is the "response line". It is always composed of 3 fields :
342
343 - a version tag : HTTP/1.1
344 - a status code : 200
345 - a reason : OK
346
347The status code is always 3-digit. The first digit indicates a general status :
Davor Ocelice9ed2812017-12-25 17:49:28 +0100348 - 1xx = informational message to be skipped (e.g. 100, 101)
349 - 2xx = OK, content is following (e.g. 200, 206)
350 - 3xx = OK, no content following (e.g. 302, 304)
351 - 4xx = error caused by the client (e.g. 401, 403, 404)
352 - 5xx = error caused by the server (e.g. 500, 502, 503)
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200353
Lukas Tribus23953682017-04-28 13:24:30 +0000354Please refer to RFC7231 for the detailed meaning of all such codes. The
Willy Tarreaud72758d2010-01-12 10:42:19 +0100355"reason" field is just a hint, but is not parsed by clients. Anything can be
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200356found there, but it's a common practice to respect the well-established
357messages. It can be composed of one or multiple words, such as "OK", "Found",
358or "Authentication Required".
359
Davor Ocelice9ed2812017-12-25 17:49:28 +0100360HAProxy may emit the following status codes by itself :
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200361
362 Code When / reason
363 200 access to stats page, and when replying to monitoring requests
364 301 when performing a redirection, depending on the configured code
365 302 when performing a redirection, depending on the configured code
366 303 when performing a redirection, depending on the configured code
Willy Tarreaub67fdc42013-03-29 19:28:11 +0100367 307 when performing a redirection, depending on the configured code
368 308 when performing a redirection, depending on the configured code
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200369 400 for an invalid or too large request
370 401 when an authentication is required to perform the action (when
371 accessing the stats page)
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200372 403 when a request is forbidden by a "http-request deny" rule
Florian Tham9205fea2020-01-08 13:35:30 +0100373 404 when the requested resource could not be found
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200374 408 when the request timeout strikes before the request is complete
Florian Tham272e29b2020-01-08 10:19:05 +0100375 410 when the requested resource is no longer available and will not
376 be available again
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200377 500 when haproxy encounters an unrecoverable internal error, such as a
378 memory allocation failure, which should never happen
379 502 when the server returns an empty, invalid or incomplete response, or
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200380 when an "http-response deny" rule blocks the response.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200381 503 when no server was available to handle the request, or in response to
382 monitoring requests which match the "monitor fail" condition
383 504 when the response timeout strikes before the server responds
384
385The error 4xx and 5xx codes above may be customized (see "errorloc" in section
3864.2).
387
388
3891.3.2. The response headers
390---------------------------
391
392Response headers work exactly like request headers, and as such, HAProxy uses
393the same parsing function for both. Please refer to paragraph 1.2.2 for more
394details.
395
396
3972. Configuring HAProxy
398----------------------
399
4002.1. Configuration file format
401------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +0200402
403HAProxy's configuration process involves 3 major sources of parameters :
404
405 - the arguments from the command-line, which always take precedence
406 - the "global" section, which sets process-wide parameters
407 - the proxies sections which can take form of "defaults", "listen",
408 "frontend" and "backend".
409
Willy Tarreau0ba27502007-12-24 16:55:16 +0100410The configuration file syntax consists in lines beginning with a keyword
411referenced in this manual, optionally followed by one or several parameters
William Lallemandf9873ba2015-05-05 17:37:14 +0200412delimited by spaces.
Willy Tarreau0ba27502007-12-24 16:55:16 +0100413
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200414
William Lallemandf9873ba2015-05-05 17:37:14 +02004152.2. Quoting and escaping
416-------------------------
417
418HAProxy's configuration introduces a quoting and escaping system similar to
419many programming languages. The configuration file supports 3 types: escaping
420with a backslash, weak quoting with double quotes, and strong quoting with
421single quotes.
422
423If spaces have to be entered in strings, then they must be escaped by preceding
424them by a backslash ('\') or by quoting them. Backslashes also have to be
425escaped by doubling or strong quoting them.
426
427Escaping is achieved by preceding a special character by a backslash ('\'):
428
429 \ to mark a space and differentiate it from a delimiter
430 \# to mark a hash and differentiate it from a comment
431 \\ to use a backslash
432 \' to use a single quote and differentiate it from strong quoting
433 \" to use a double quote and differentiate it from weak quoting
434
435Weak quoting is achieved by using double quotes (""). Weak quoting prevents
436the interpretation of:
437
438 space as a parameter separator
439 ' single quote as a strong quoting delimiter
440 # hash as a comment start
441
William Lallemandb2f07452015-05-12 14:27:13 +0200442Weak quoting permits the interpretation of variables, if you want to use a non
443-interpreted dollar within a double quoted string, you should escape it with a
444backslash ("\$"), it does not work outside weak quoting.
445
446Interpretation of escaping and special characters are not prevented by weak
William Lallemandf9873ba2015-05-05 17:37:14 +0200447quoting.
448
449Strong quoting is achieved by using single quotes (''). Inside single quotes,
450nothing is interpreted, it's the efficient way to quote regexes.
451
452Quoted and escaped strings are replaced in memory by their interpreted
453equivalent, it allows you to perform concatenation.
454
455 Example:
456 # those are equivalents:
457 log-format %{+Q}o\ %t\ %s\ %{-Q}r
458 log-format "%{+Q}o %t %s %{-Q}r"
459 log-format '%{+Q}o %t %s %{-Q}r'
460 log-format "%{+Q}o %t"' %s %{-Q}r'
461 log-format "%{+Q}o %t"' %s'\ %{-Q}r
462
463 # those are equivalents:
464 reqrep "^([^\ :]*)\ /static/(.*)" \1\ /\2
465 reqrep "^([^ :]*)\ /static/(.*)" '\1 /\2'
466 reqrep "^([^ :]*)\ /static/(.*)" "\1 /\2"
467 reqrep "^([^ :]*)\ /static/(.*)" "\1\ /\2"
468
469
William Lallemandb2f07452015-05-12 14:27:13 +02004702.3. Environment variables
471--------------------------
472
473HAProxy's configuration supports environment variables. Those variables are
474interpreted only within double quotes. Variables are expanded during the
475configuration parsing. Variable names must be preceded by a dollar ("$") and
476optionally enclosed with braces ("{}") similarly to what is done in Bourne
477shell. Variable names can contain alphanumerical characters or the character
Amaury Denoyellefa41cb62020-10-01 14:32:35 +0200478underscore ("_") but should not start with a digit. If the variable contains a
479list of several values separated by spaces, it can be expanded as individual
480arguments by enclosing the variable with braces and appending the suffix '[*]'
481before the closing brace.
William Lallemandb2f07452015-05-12 14:27:13 +0200482
483 Example:
484
485 bind "fd@${FD_APP1}"
486
487 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
488
489 user "$HAPROXY_USER"
490
William Lallemand4d03e432019-06-14 15:35:37 +0200491Some variables are defined by HAProxy, they can be used in the configuration
492file, or could be inherited by a program (See 3.7. Programs):
William Lallemanddaf4cd22018-04-17 16:46:13 +0200493
William Lallemand4d03e432019-06-14 15:35:37 +0200494* HAPROXY_LOCALPEER: defined at the startup of the process which contains the
495 name of the local peer. (See "-L" in the management guide.)
496
497* HAPROXY_CFGFILES: list of the configuration files loaded by HAProxy,
498 separated by semicolons. Can be useful in the case you specified a
499 directory.
500
501* HAPROXY_MWORKER: In master-worker mode, this variable is set to 1.
502
John Roeslerfb2fce12019-07-10 15:45:51 -0500503* HAPROXY_CLI: configured listeners addresses of the stats socket for every
William Lallemand4d03e432019-06-14 15:35:37 +0200504 processes, separated by semicolons.
505
John Roeslerfb2fce12019-07-10 15:45:51 -0500506* HAPROXY_MASTER_CLI: In master-worker mode, listeners addresses of the master
William Lallemand4d03e432019-06-14 15:35:37 +0200507 CLI, separated by semicolons.
508
509See also "external-check command" for other variables.
William Lallemandb2f07452015-05-12 14:27:13 +0200510
5112.4. Time format
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200512----------------
513
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100514Some parameters involve values representing time, such as timeouts. These
Willy Tarreau0ba27502007-12-24 16:55:16 +0100515values are generally expressed in milliseconds (unless explicitly stated
516otherwise) but may be expressed in any other unit by suffixing the unit to the
517numeric value. It is important to consider this because it will not be repeated
518for every keyword. Supported units are :
519
520 - us : microseconds. 1 microsecond = 1/1000000 second
521 - ms : milliseconds. 1 millisecond = 1/1000 second. This is the default.
522 - s : seconds. 1s = 1000ms
523 - m : minutes. 1m = 60s = 60000ms
524 - h : hours. 1h = 60m = 3600s = 3600000ms
525 - d : days. 1d = 24h = 1440m = 86400s = 86400000ms
526
527
Lukas Tribusaa83a312017-03-21 09:25:09 +00005282.5. Examples
Patrick Mezard35da19c2010-06-12 17:02:47 +0200529-------------
530
531 # Simple configuration for an HTTP proxy listening on port 80 on all
532 # interfaces and forwarding requests to a single backend "servers" with a
533 # single server "server1" listening on 127.0.0.1:8000
534 global
535 daemon
536 maxconn 256
537
538 defaults
539 mode http
540 timeout connect 5000ms
541 timeout client 50000ms
542 timeout server 50000ms
543
544 frontend http-in
545 bind *:80
546 default_backend servers
547
548 backend servers
549 server server1 127.0.0.1:8000 maxconn 32
550
551
552 # The same configuration defined with a single listen block. Shorter but
553 # less expressive, especially in HTTP mode.
554 global
555 daemon
556 maxconn 256
557
558 defaults
559 mode http
560 timeout connect 5000ms
561 timeout client 50000ms
562 timeout server 50000ms
563
564 listen http-in
565 bind *:80
566 server server1 127.0.0.1:8000 maxconn 32
567
568
569Assuming haproxy is in $PATH, test these configurations in a shell with:
570
Willy Tarreauccb289d2010-12-11 20:19:38 +0100571 $ sudo haproxy -f configuration.conf -c
Patrick Mezard35da19c2010-06-12 17:02:47 +0200572
573
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005743. Global parameters
Willy Tarreau6a06a402007-07-15 20:15:28 +0200575--------------------
576
577Parameters in the "global" section are process-wide and often OS-specific. They
578are generally set once for all and do not need being changed once correct. Some
579of them have command-line equivalents.
580
581The following keywords are supported in the "global" section :
582
583 * Process management and security
Emeric Brunc8e8d122012-10-02 18:42:10 +0200584 - ca-base
Willy Tarreau6a06a402007-07-15 20:15:28 +0200585 - chroot
Emeric Brunc8e8d122012-10-02 18:42:10 +0200586 - crt-base
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200587 - cpu-map
Willy Tarreau6a06a402007-07-15 20:15:28 +0200588 - daemon
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200589 - description
590 - deviceatlas-json-file
591 - deviceatlas-log-level
592 - deviceatlas-separator
593 - deviceatlas-properties-cookie
Simon Horman98637e52014-06-20 12:30:16 +0900594 - external-check
Willy Tarreau6a06a402007-07-15 20:15:28 +0200595 - gid
596 - group
Cyril Bonté203ec5a2017-03-23 22:44:13 +0100597 - hard-stop-after
Christopher Faulet98fbe952019-07-22 16:18:24 +0200598 - h1-case-adjust
599 - h1-case-adjust-file
Willy Tarreaud96f1122019-12-03 07:07:36 +0100600 - insecure-fork-wanted
Willy Tarreaua45a8b52019-12-06 16:31:45 +0100601 - insecure-setuid-wanted
Emmanuel Hocdet70df7bf2019-01-04 11:08:20 +0100602 - issuers-chain-path
Dragan Dosen13cd54c2020-06-18 18:24:05 +0200603 - localpeer
Willy Tarreau6a06a402007-07-15 20:15:28 +0200604 - log
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200605 - log-tag
Joe Williamsdf5b38f2010-12-29 17:05:48 +0100606 - log-send-hostname
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200607 - lua-load
Tim Duesterhusdd74b5f2020-01-12 13:55:40 +0100608 - lua-prepend-path
William Lallemand27edc4b2019-05-07 17:49:33 +0200609 - mworker-max-reloads
Willy Tarreau6a06a402007-07-15 20:15:28 +0200610 - nbproc
Christopher Fauletbe0faa22017-08-29 15:37:10 +0200611 - nbthread
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200612 - node
Willy Tarreau6a06a402007-07-15 20:15:28 +0200613 - pidfile
Willy Tarreau119e50e2020-05-22 13:53:29 +0200614 - pp2-never-send-local
Willy Tarreau1d549722016-02-16 12:41:57 +0100615 - presetenv
616 - resetenv
Willy Tarreau6a06a402007-07-15 20:15:28 +0200617 - uid
618 - ulimit-n
619 - user
Willy Tarreau636848a2019-04-15 19:38:50 +0200620 - set-dumpable
Willy Tarreau1d549722016-02-16 12:41:57 +0100621 - setenv
Willy Tarreaufbee7132007-10-18 13:53:22 +0200622 - stats
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200623 - ssl-default-bind-ciphers
Dirkjan Bussink415150f2018-09-14 11:14:21 +0200624 - ssl-default-bind-ciphersuites
Jerome Magninb203ff62020-04-03 15:28:22 +0200625 - ssl-default-bind-curves
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200626 - ssl-default-bind-options
627 - ssl-default-server-ciphers
Dirkjan Bussink415150f2018-09-14 11:14:21 +0200628 - ssl-default-server-ciphersuites
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200629 - ssl-default-server-options
630 - ssl-dh-param-file
Emeric Brun850efd52014-01-29 12:24:34 +0100631 - ssl-server-verify
Emmanuel Hocdetc3b7e742020-04-22 11:06:19 +0200632 - ssl-skip-self-issued-ca
Willy Tarreauceb24bc2010-11-09 12:46:41 +0100633 - unix-bind
Willy Tarreau1d549722016-02-16 12:41:57 +0100634 - unsetenv
Thomas Holmesdb04f192015-05-18 13:21:39 +0100635 - 51degrees-data-file
636 - 51degrees-property-name-list
Dragan Dosen93b38d92015-06-29 16:43:25 +0200637 - 51degrees-property-separator
Dragan Dosenae6d39a2015-06-29 16:43:27 +0200638 - 51degrees-cache-size
Willy Tarreaub3cc9f22019-04-19 16:03:32 +0200639 - wurfl-data-file
640 - wurfl-information-list
641 - wurfl-information-list-separator
Willy Tarreaub3cc9f22019-04-19 16:03:32 +0200642 - wurfl-cache-size
William Dauchy0fec3ab2019-10-27 20:08:11 +0100643 - strict-limits
Willy Tarreaud72758d2010-01-12 10:42:19 +0100644
Willy Tarreau6a06a402007-07-15 20:15:28 +0200645 * Performance tuning
William Dauchy0a8824f2019-10-27 20:08:09 +0100646 - busy-polling
Willy Tarreau1746eec2014-04-25 10:46:47 +0200647 - max-spread-checks
Willy Tarreau6a06a402007-07-15 20:15:28 +0200648 - maxconn
Willy Tarreau81c25d02011-09-07 15:17:21 +0200649 - maxconnrate
William Lallemandd85f9172012-11-09 17:05:39 +0100650 - maxcomprate
William Lallemand072a2bf2012-11-20 17:01:01 +0100651 - maxcompcpuusage
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100652 - maxpipes
Willy Tarreau93e7c002013-10-07 18:51:07 +0200653 - maxsessrate
Willy Tarreau403edff2012-09-06 11:58:37 +0200654 - maxsslconn
Willy Tarreaue43d5322013-10-07 20:01:52 +0200655 - maxsslrate
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200656 - maxzlibmem
Willy Tarreau6a06a402007-07-15 20:15:28 +0200657 - noepoll
658 - nokqueue
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +0000659 - noevports
Willy Tarreau6a06a402007-07-15 20:15:28 +0200660 - nopoll
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100661 - nosplice
Jarno Huuskonen0e82b922014-04-12 18:22:19 +0300662 - nogetaddrinfo
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +0000663 - noreuseport
Willy Tarreau75c62c22018-11-22 11:02:09 +0100664 - profiling.tasks
Willy Tarreaufe255b72007-10-14 23:09:26 +0200665 - spread-checks
Baptiste Assmann5626f482015-08-23 10:00:10 +0200666 - server-state-base
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +0200667 - server-state-file
Grant Zhang872f9c22017-01-21 01:10:18 +0000668 - ssl-engine
Grant Zhangfa6c7ee2017-01-14 01:42:15 +0000669 - ssl-mode-async
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200670 - tune.buffers.limit
671 - tune.buffers.reserve
Willy Tarreau27a674e2009-08-17 07:23:33 +0200672 - tune.bufsize
Willy Tarreau43961d52010-10-04 20:39:20 +0200673 - tune.chksize
William Lallemandf3747832012-11-09 12:33:10 +0100674 - tune.comp.maxlevel
Willy Tarreaubc52bec2020-06-18 08:58:47 +0200675 - tune.fd.edge-triggered
Willy Tarreaufe20e5b2017-07-27 11:42:14 +0200676 - tune.h2.header-table-size
Willy Tarreaue6baec02017-07-27 11:45:11 +0200677 - tune.h2.initial-window-size
Willy Tarreau5242ef82017-07-27 11:47:28 +0200678 - tune.h2.max-concurrent-streams
Willy Tarreau193b8c62012-11-22 00:17:38 +0100679 - tune.http.cookielen
Stéphane Cottin23e9e932017-05-18 08:58:41 +0200680 - tune.http.logurilen
Willy Tarreauac1932d2011-10-24 19:14:41 +0200681 - tune.http.maxhdr
Willy Tarreau76cc6992020-07-01 18:49:24 +0200682 - tune.idle-pool.shared
Willy Tarreau7e312732014-02-12 16:35:14 +0100683 - tune.idletimer
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100684 - tune.lua.forced-yield
Willy Tarreau32f61e22015-03-18 17:54:59 +0100685 - tune.lua.maxmem
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100686 - tune.lua.session-timeout
687 - tune.lua.task-timeout
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +0200688 - tune.lua.service-timeout
Willy Tarreaua0250ba2008-01-06 11:22:57 +0100689 - tune.maxaccept
690 - tune.maxpollevents
Willy Tarreau27a674e2009-08-17 07:23:33 +0200691 - tune.maxrewrite
Willy Tarreauf3045d22015-04-29 16:24:50 +0200692 - tune.pattern.cache-size
Willy Tarreaubd9a0a72011-10-23 21:14:29 +0200693 - tune.pipesize
Willy Tarreaua8e2d972020-07-01 18:27:16 +0200694 - tune.pool-high-fd-ratio
695 - tune.pool-low-fd-ratio
Willy Tarreaue803de22010-01-21 17:43:04 +0100696 - tune.rcvbuf.client
697 - tune.rcvbuf.server
Willy Tarreaub22fc302015-12-14 12:04:35 +0100698 - tune.recv_enough
Olivier Houchard1599b802018-05-24 18:59:04 +0200699 - tune.runqueue-depth
Willy Tarreaue7723bd2020-06-24 11:11:02 +0200700 - tune.sched.low-latency
Willy Tarreaue803de22010-01-21 17:43:04 +0100701 - tune.sndbuf.client
702 - tune.sndbuf.server
Willy Tarreau6ec58db2012-11-16 16:32:15 +0100703 - tune.ssl.cachesize
William Lallemand7d42ef52020-07-06 11:41:30 +0200704 - tune.ssl.keylog
Willy Tarreaubfd59462013-02-21 07:46:09 +0100705 - tune.ssl.lifetime
Emeric Brun8dc60392014-05-09 13:52:00 +0200706 - tune.ssl.force-private-cache
Willy Tarreaubfd59462013-02-21 07:46:09 +0100707 - tune.ssl.maxrecord
Remi Gacognef46cd6e2014-06-12 14:58:40 +0200708 - tune.ssl.default-dh-param
Christopher Faulet31af49d2015-06-09 17:29:50 +0200709 - tune.ssl.ssl-ctx-cache-size
Thierry FOURNIER5bf77322017-02-25 12:45:22 +0100710 - tune.ssl.capture-cipherlist-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200711 - tune.vars.global-max-size
Christopher Fauletff2613e2016-11-09 11:36:17 +0100712 - tune.vars.proc-max-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200713 - tune.vars.reqres-max-size
714 - tune.vars.sess-max-size
715 - tune.vars.txn-max-size
William Lallemanda509e4c2012-11-07 16:54:34 +0100716 - tune.zlib.memlevel
717 - tune.zlib.windowsize
Willy Tarreaud72758d2010-01-12 10:42:19 +0100718
Willy Tarreau6a06a402007-07-15 20:15:28 +0200719 * Debugging
Willy Tarreau6a06a402007-07-15 20:15:28 +0200720 - quiet
Willy Tarreau3eb10b82020-04-15 16:42:39 +0200721 - zero-warning
Willy Tarreau6a06a402007-07-15 20:15:28 +0200722
723
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007243.1. Process management and security
Willy Tarreau6a06a402007-07-15 20:15:28 +0200725------------------------------------
726
Emeric Brunc8e8d122012-10-02 18:42:10 +0200727ca-base <dir>
728 Assigns a default directory to fetch SSL CA certificates and CRLs from when a
Emmanuel Hocdet842e94e2019-12-16 16:39:17 +0100729 relative path is used with "ca-file", "ca-verify-file" or "crl-file"
730 directives. Absolute locations specified in "ca-file", "ca-verify-file" and
731 "crl-file" prevail and ignore "ca-base".
Emeric Brunc8e8d122012-10-02 18:42:10 +0200732
Willy Tarreau6a06a402007-07-15 20:15:28 +0200733chroot <jail dir>
734 Changes current directory to <jail dir> and performs a chroot() there before
735 dropping privileges. This increases the security level in case an unknown
736 vulnerability would be exploited, since it would make it very hard for the
737 attacker to exploit the system. This only works when the process is started
738 with superuser privileges. It is important to ensure that <jail_dir> is both
Davor Ocelice9ed2812017-12-25 17:49:28 +0100739 empty and non-writable to anyone.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100740
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100741cpu-map [auto:]<process-set>[/<thread-set>] <cpu-set>...
742 On Linux 2.6 and above, it is possible to bind a process or a thread to a
743 specific CPU set. This means that the process or the thread will never run on
744 other CPUs. The "cpu-map" directive specifies CPU sets for process or thread
745 sets. The first argument is a process set, eventually followed by a thread
746 set. These sets have the format
747
748 all | odd | even | number[-[number]]
749
750 <number>> must be a number between 1 and 32 or 64, depending on the machine's
Davor Ocelice9ed2812017-12-25 17:49:28 +0100751 word size. Any process IDs above nbproc and any thread IDs above nbthread are
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100752 ignored. It is possible to specify a range with two such number delimited by
753 a dash ('-'). It also is possible to specify all processes at once using
Christopher Faulet1dcb9cb2017-11-22 10:24:40 +0100754 "all", only odd numbers using "odd" or even numbers using "even", just like
755 with the "bind-process" directive. The second and forthcoming arguments are
Davor Ocelice9ed2812017-12-25 17:49:28 +0100756 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 +0100757 range with two such numbers delimited by a dash ('-'). Multiple CPU numbers
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100758 or ranges may be specified, and the processes or threads will be allowed to
Davor Ocelice9ed2812017-12-25 17:49:28 +0100759 bind to all of them. Obviously, multiple "cpu-map" directives may be
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100760 specified. Each "cpu-map" directive will replace the previous ones when they
761 overlap. A thread will be bound on the intersection of its mapping and the
762 one of the process on which it is attached. If the intersection is null, no
763 specific binding will be set for the thread.
Willy Tarreaufc6c0322012-11-16 16:12:27 +0100764
Christopher Fauletff4121f2017-11-22 16:38:49 +0100765 Ranges can be partially defined. The higher bound can be omitted. In such
766 case, it is replaced by the corresponding maximum value, 32 or 64 depending
767 on the machine's word size.
768
Christopher Faulet26028f62017-11-22 15:01:51 +0100769 The prefix "auto:" can be added before the process set to let HAProxy
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100770 automatically bind a process or a thread to a CPU by incrementing
771 process/thread and CPU sets. To be valid, both sets must have the same
772 size. No matter the declaration order of the CPU sets, it will be bound from
773 the lowest to the highest bound. Having a process and a thread range with the
774 "auto:" prefix is not supported. Only one range is supported, the other one
775 must be a fixed number.
Christopher Faulet26028f62017-11-22 15:01:51 +0100776
777 Examples:
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100778 cpu-map 1-4 0-3 # bind processes 1 to 4 on the first 4 CPUs
779
780 cpu-map 1/all 0-3 # bind all threads of the first process on the
781 # first 4 CPUs
782
783 cpu-map 1- 0- # will be replaced by "cpu-map 1-64 0-63"
784 # or "cpu-map 1-32 0-31" depending on the machine's
785 # word size.
786
Christopher Faulet26028f62017-11-22 15:01:51 +0100787 # all these lines bind the process 1 to the cpu 0, the process 2 to cpu 1
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100788 # and so on.
Christopher Faulet26028f62017-11-22 15:01:51 +0100789 cpu-map auto:1-4 0-3
790 cpu-map auto:1-4 0-1 2-3
791 cpu-map auto:1-4 3 2 1 0
792
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100793 # all these lines bind the thread 1 to the cpu 0, the thread 2 to cpu 1
794 # and so on.
795 cpu-map auto:1/1-4 0-3
796 cpu-map auto:1/1-4 0-1 2-3
797 cpu-map auto:1/1-4 3 2 1 0
798
Davor Ocelice9ed2812017-12-25 17:49:28 +0100799 # bind each process to exactly one CPU using all/odd/even keyword
Christopher Faulet26028f62017-11-22 15:01:51 +0100800 cpu-map auto:all 0-63
801 cpu-map auto:even 0-31
802 cpu-map auto:odd 32-63
803
804 # invalid cpu-map because process and CPU sets have different sizes.
805 cpu-map auto:1-4 0 # invalid
806 cpu-map auto:1 0-3 # invalid
807
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100808 # invalid cpu-map because automatic binding is used with a process range
809 # and a thread range.
810 cpu-map auto:all/all 0 # invalid
811 cpu-map auto:all/1-4 0 # invalid
812 cpu-map auto:1-4/all 0 # invalid
813
Emeric Brunc8e8d122012-10-02 18:42:10 +0200814crt-base <dir>
815 Assigns a default directory to fetch SSL certificates from when a relative
William Dauchy238ea3b2020-01-11 13:09:12 +0100816 path is used with "crtfile" or "crt" directives. Absolute locations specified
817 prevail and ignore "crt-base".
Emeric Brunc8e8d122012-10-02 18:42:10 +0200818
Willy Tarreau6a06a402007-07-15 20:15:28 +0200819daemon
820 Makes the process fork into background. This is the recommended mode of
821 operation. It is equivalent to the command line "-D" argument. It can be
Lukas Tribusf46bf952017-11-21 12:39:34 +0100822 disabled by the command line "-db" argument. This option is ignored in
823 systemd mode.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200824
David Carlier8167f302015-06-01 13:50:06 +0200825deviceatlas-json-file <path>
826 Sets the path of the DeviceAtlas JSON data file to be loaded by the API.
Davor Ocelice9ed2812017-12-25 17:49:28 +0100827 The path must be a valid JSON data file and accessible by HAProxy process.
David Carlier8167f302015-06-01 13:50:06 +0200828
829deviceatlas-log-level <value>
Davor Ocelice9ed2812017-12-25 17:49:28 +0100830 Sets the level of information returned by the API. This directive is
David Carlier8167f302015-06-01 13:50:06 +0200831 optional and set to 0 by default if not set.
832
833deviceatlas-separator <char>
834 Sets the character separator for the API properties results. This directive
835 is optional and set to | by default if not set.
836
Cyril Bonté0306c4a2015-10-26 22:37:38 +0100837deviceatlas-properties-cookie <name>
Cyril Bonté307ee1e2015-09-28 23:16:06 +0200838 Sets the client cookie's name used for the detection if the DeviceAtlas
839 Client-side component was used during the request. This directive is optional
840 and set to DAPROPS by default if not set.
David Carlier29b3ca32015-09-25 14:09:21 +0100841
Simon Horman98637e52014-06-20 12:30:16 +0900842external-check
Willy Tarreaud96f1122019-12-03 07:07:36 +0100843 Allows the use of an external agent to perform health checks. This is
844 disabled by default as a security precaution, and even when enabled, checks
Willy Tarreaua45a8b52019-12-06 16:31:45 +0100845 may still fail unless "insecure-fork-wanted" is enabled as well. If the
846 program launched makes use of a setuid executable (it should really not),
847 you may also need to set "insecure-setuid-wanted" in the global section.
848 See "option external-check", and "insecure-fork-wanted", and
849 "insecure-setuid-wanted".
Simon Horman98637e52014-06-20 12:30:16 +0900850
Willy Tarreau6a06a402007-07-15 20:15:28 +0200851gid <number>
852 Changes the process' group ID to <number>. It is recommended that the group
853 ID is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
854 be started with a user belonging to this group, or with superuser privileges.
Michael Schererab012dd2013-01-12 18:35:19 +0100855 Note that if haproxy is started from a user having supplementary groups, it
856 will only be able to drop these groups if started with superuser privileges.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200857 See also "group" and "uid".
Willy Tarreaud72758d2010-01-12 10:42:19 +0100858
Willy Tarreau11770ce2019-12-03 08:29:22 +0100859group <group name>
860 Similar to "gid" but uses the GID of group name <group name> from /etc/group.
861 See also "gid" and "user".
862
Cyril Bonté203ec5a2017-03-23 22:44:13 +0100863hard-stop-after <time>
864 Defines the maximum time allowed to perform a clean soft-stop.
865
866 Arguments :
867 <time> is the maximum time (by default in milliseconds) for which the
868 instance will remain alive when a soft-stop is received via the
869 SIGUSR1 signal.
870
871 This may be used to ensure that the instance will quit even if connections
872 remain opened during a soft-stop (for example with long timeouts for a proxy
873 in tcp mode). It applies both in TCP and HTTP mode.
874
875 Example:
876 global
877 hard-stop-after 30s
878
Christopher Faulet98fbe952019-07-22 16:18:24 +0200879h1-case-adjust <from> <to>
880 Defines the case adjustment to apply, when enabled, to the header name
881 <from>, to change it to <to> before sending it to HTTP/1 clients or
882 servers. <from> must be in lower case, and <from> and <to> must not differ
883 except for their case. It may be repeated if several header names need to be
Ilya Shipitsin8525fd92020-02-29 12:34:59 +0500884 adjusted. Duplicate entries are not allowed. If a lot of header names have to
Christopher Faulet98fbe952019-07-22 16:18:24 +0200885 be adjusted, it might be more convenient to use "h1-case-adjust-file".
886 Please note that no transformation will be applied unless "option
887 h1-case-adjust-bogus-client" or "option h1-case-adjust-bogus-server" is
888 specified in a proxy.
889
890 There is no standard case for header names because, as stated in RFC7230,
891 they are case-insensitive. So applications must handle them in a case-
892 insensitive manner. But some bogus applications violate the standards and
893 erroneously rely on the cases most commonly used by browsers. This problem
894 becomes critical with HTTP/2 because all header names must be exchanged in
895 lower case, and HAProxy follows the same convention. All header names are
896 sent in lower case to clients and servers, regardless of the HTTP version.
897
898 Applications which fail to properly process requests or responses may require
899 to temporarily use such workarounds to adjust header names sent to them for
900 the time it takes the application to be fixed. Please note that an
901 application which requires such workarounds might be vulnerable to content
902 smuggling attacks and must absolutely be fixed.
903
904 Example:
905 global
906 h1-case-adjust content-length Content-Length
907
908 See "h1-case-adjust-file", "option h1-case-adjust-bogus-client" and
909 "option h1-case-adjust-bogus-server".
910
911h1-case-adjust-file <hdrs-file>
912 Defines a file containing a list of key/value pairs used to adjust the case
913 of some header names before sending them to HTTP/1 clients or servers. The
914 file <hdrs-file> must contain 2 header names per line. The first one must be
915 in lower case and both must not differ except for their case. Lines which
916 start with '#' are ignored, just like empty lines. Leading and trailing tabs
917 and spaces are stripped. Duplicate entries are not allowed. Please note that
918 no transformation will be applied unless "option h1-case-adjust-bogus-client"
919 or "option h1-case-adjust-bogus-server" is specified in a proxy.
920
921 If this directive is repeated, only the last one will be processed. It is an
922 alternative to the directive "h1-case-adjust" if a lot of header names need
923 to be adjusted. Please read the risks associated with using this.
924
925 See "h1-case-adjust", "option h1-case-adjust-bogus-client" and
926 "option h1-case-adjust-bogus-server".
927
Willy Tarreaud96f1122019-12-03 07:07:36 +0100928insecure-fork-wanted
929 By default haproxy tries hard to prevent any thread and process creation
930 after it starts. Doing so is particularly important when using Lua files of
931 uncertain origin, and when experimenting with development versions which may
932 still contain bugs whose exploitability is uncertain. And generally speaking
933 it's good hygiene to make sure that no unexpected background activity can be
934 triggered by traffic. But this prevents external checks from working, and may
935 break some very specific Lua scripts which actively rely on the ability to
936 fork. This option is there to disable this protection. Note that it is a bad
937 idea to disable it, as a vulnerability in a library or within haproxy itself
938 will be easier to exploit once disabled. In addition, forking from Lua or
939 anywhere else is not reliable as the forked process may randomly embed a lock
940 set by another thread and never manage to finish an operation. As such it is
941 highly recommended that this option is never used and that any workload
942 requiring such a fork be reconsidered and moved to a safer solution (such as
943 agents instead of external checks). This option supports the "no" prefix to
944 disable it.
945
Willy Tarreaua45a8b52019-12-06 16:31:45 +0100946insecure-setuid-wanted
947 HAProxy doesn't need to call executables at run time (except when using
948 external checks which are strongly recommended against), and is even expected
949 to isolate itself into an empty chroot. As such, there basically is no valid
950 reason to allow a setuid executable to be called without the user being fully
951 aware of the risks. In a situation where haproxy would need to call external
952 checks and/or disable chroot, exploiting a vulnerability in a library or in
953 haproxy itself could lead to the execution of an external program. On Linux
954 it is possible to lock the process so that any setuid bit present on such an
955 executable is ignored. This significantly reduces the risk of privilege
956 escalation in such a situation. This is what haproxy does by default. In case
957 this causes a problem to an external check (for example one which would need
958 the "ping" command), then it is possible to disable this protection by
959 explicitly adding this directive in the global section. If enabled, it is
960 possible to turn it back off by prefixing it with the "no" keyword.
961
Emmanuel Hocdet70df7bf2019-01-04 11:08:20 +0100962issuers-chain-path <dir>
963 Assigns a directory to load certificate chain for issuer completion. All
964 files must be in PEM format. For certificates loaded with "crt" or "crt-list",
965 if certificate chain is not included in PEM (also commonly known as
966 intermediate certificate), haproxy will complete chain if the issuer of the
967 certificate corresponds to the first certificate of the chain loaded with
968 "issuers-chain-path".
969 A "crt" file with PrivateKey+Certificate+IntermediateCA2+IntermediateCA1
970 could be replaced with PrivateKey+Certificate. HAProxy will complete the
971 chain if a file with IntermediateCA2+IntermediateCA1 is present in
972 "issuers-chain-path" directory. All other certificates with the same issuer
973 will share the chain in memory.
974
Dragan Dosen13cd54c2020-06-18 18:24:05 +0200975localpeer <name>
976 Sets the local instance's peer name. It will be ignored if the "-L"
977 command line argument is specified or if used after "peers" section
978 definitions. In such cases, a warning message will be emitted during
979 the configuration parsing.
980
981 This option will also set the HAPROXY_LOCALPEER environment variable.
982 See also "-L" in the management guide and "peers" section below.
983
Frédéric Lécailled690dfa2019-04-25 10:52:17 +0200984log <address> [len <length>] [format <format>] [sample <ranges>:<smp_size>]
985 <facility> [max level [min level]]
Cyril Bonté3e954872018-03-20 23:30:27 +0100986 Adds a global syslog server. Several global servers can be defined. They
Davor Ocelice9ed2812017-12-25 17:49:28 +0100987 will receive logs for starts and exits, as well as all logs from proxies
Robert Tsai81ae1952007-12-05 10:47:29 +0100988 configured with "log global".
989
990 <address> can be one of:
991
Willy Tarreau2769aa02007-12-27 18:26:09 +0100992 - An IPv4 address optionally followed by a colon and a UDP port. If
Robert Tsai81ae1952007-12-05 10:47:29 +0100993 no port is specified, 514 is used by default (the standard syslog
994 port).
995
David du Colombier24bb5f52011-03-17 10:40:23 +0100996 - An IPv6 address followed by a colon and optionally a UDP port. If
997 no port is specified, 514 is used by default (the standard syslog
998 port).
999
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01001000 - A filesystem path to a datagram UNIX domain socket, keeping in mind
Robert Tsai81ae1952007-12-05 10:47:29 +01001001 considerations for chroot (be sure the path is accessible inside
1002 the chroot) and uid/gid (be sure the path is appropriately
Davor Ocelice9ed2812017-12-25 17:49:28 +01001003 writable).
Robert Tsai81ae1952007-12-05 10:47:29 +01001004
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01001005 - A file descriptor number in the form "fd@<number>", which may point
1006 to a pipe, terminal, or socket. In this case unbuffered logs are used
1007 and one writev() call per log is performed. This is a bit expensive
1008 but acceptable for most workloads. Messages sent this way will not be
1009 truncated but may be dropped, in which case the DroppedLogs counter
1010 will be incremented. The writev() call is atomic even on pipes for
1011 messages up to PIPE_BUF size, which POSIX recommends to be at least
1012 512 and which is 4096 bytes on most modern operating systems. Any
1013 larger message may be interleaved with messages from other processes.
1014 Exceptionally for debugging purposes the file descriptor may also be
1015 directed to a file, but doing so will significantly slow haproxy down
1016 as non-blocking calls will be ignored. Also there will be no way to
1017 purge nor rotate this file without restarting the process. Note that
1018 the configured syslog format is preserved, so the output is suitable
Willy Tarreauc1b06452018-11-12 11:57:56 +01001019 for use with a TCP syslog server. See also the "short" and "raw"
1020 format below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01001021
1022 - "stdout" / "stderr", which are respectively aliases for "fd@1" and
1023 "fd@2", see above.
1024
Willy Tarreauc046d162019-08-30 15:24:59 +02001025 - A ring buffer in the form "ring@<name>", which will correspond to an
1026 in-memory ring buffer accessible over the CLI using the "show events"
1027 command, which will also list existing rings and their sizes. Such
1028 buffers are lost on reload or restart but when used as a complement
1029 this can help troubleshooting by having the logs instantly available.
1030
William Lallemandb2f07452015-05-12 14:27:13 +02001031 You may want to reference some environment variables in the address
1032 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01001033
Willy Tarreau18324f52014-06-27 18:10:07 +02001034 <length> is an optional maximum line length. Log lines larger than this value
1035 will be truncated before being sent. The reason is that syslog
1036 servers act differently on log line length. All servers support the
1037 default value of 1024, but some servers simply drop larger lines
1038 while others do log them. If a server supports long lines, it may
1039 make sense to set this value here in order to avoid truncating long
1040 lines. Similarly, if a server drops long lines, it is preferable to
1041 truncate them before sending them. Accepted values are 80 to 65535
1042 inclusive. The default value of 1024 is generally fine for all
1043 standard usages. Some specific cases of long captures or
Davor Ocelice9ed2812017-12-25 17:49:28 +01001044 JSON-formatted logs may require larger values. You may also need to
1045 increase "tune.http.logurilen" if your request URIs are truncated.
Willy Tarreau18324f52014-06-27 18:10:07 +02001046
Dragan Dosen7ad31542015-09-28 17:16:47 +02001047 <format> is the log format used when generating syslog messages. It may be
1048 one of the following :
1049
1050 rfc3164 The RFC3164 syslog message format. This is the default.
1051 (https://tools.ietf.org/html/rfc3164)
1052
1053 rfc5424 The RFC5424 syslog message format.
1054 (https://tools.ietf.org/html/rfc5424)
1055
Emeric Brun54648852020-07-06 15:54:06 +02001056 priority A message containing only a level plus syslog facility between
1057 angle brackets such as '<63>', followed by the text. The PID,
1058 date, time, process name and system name are omitted. This is
1059 designed to be used with a local log server.
1060
Willy Tarreaue8746a02018-11-12 08:45:00 +01001061 short A message containing only a level between angle brackets such as
1062 '<3>', followed by the text. The PID, date, time, process name
1063 and system name are omitted. This is designed to be used with a
1064 local log server. This format is compatible with what the systemd
1065 logger consumes.
1066
Emeric Brun54648852020-07-06 15:54:06 +02001067 timed A message containing only a level between angle brackets such as
1068 '<3>', followed by ISO date and by the text. The PID, process
1069 name and system name are omitted. This is designed to be
1070 used with a local log server.
1071
1072 iso A message containing only the ISO date, followed by the text.
1073 The PID, process name and system name are omitted. This is
1074 designed to be used with a local log server.
1075
Willy Tarreauc1b06452018-11-12 11:57:56 +01001076 raw A message containing only the text. The level, PID, date, time,
1077 process name and system name are omitted. This is designed to be
1078 used in containers or during development, where the severity only
1079 depends on the file descriptor used (stdout/stderr).
1080
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02001081 <ranges> A list of comma-separated ranges to identify the logs to sample.
1082 This is used to balance the load of the logs to send to the log
1083 server. The limits of the ranges cannot be null. They are numbered
1084 from 1. The size or period (in number of logs) of the sample must be
1085 set with <sample_size> parameter.
1086
1087 <sample_size>
1088 The size of the sample in number of logs to consider when balancing
1089 their logging loads. It is used to balance the load of the logs to
1090 send to the syslog server. This size must be greater or equal to the
1091 maximum of the high limits of the ranges.
1092 (see also <ranges> parameter).
1093
Robert Tsai81ae1952007-12-05 10:47:29 +01001094 <facility> must be one of the 24 standard syslog facilities :
Willy Tarreau6a06a402007-07-15 20:15:28 +02001095
Willy Tarreaue8746a02018-11-12 08:45:00 +01001096 kern user mail daemon auth syslog lpr news
1097 uucp cron auth2 ftp ntp audit alert cron2
1098 local0 local1 local2 local3 local4 local5 local6 local7
1099
Willy Tarreauc1b06452018-11-12 11:57:56 +01001100 Note that the facility is ignored for the "short" and "raw"
1101 formats, but still required as a positional field. It is
1102 recommended to use "daemon" in this case to make it clear that
1103 it's only supposed to be used locally.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001104
1105 An optional level can be specified to filter outgoing messages. By default,
Willy Tarreauf7edefa2009-05-10 17:20:05 +02001106 all messages are sent. If a maximum level is specified, only messages with a
1107 severity at least as important as this level will be sent. An optional minimum
1108 level can be specified. If it is set, logs emitted with a more severe level
1109 than this one will be capped to this level. This is used to avoid sending
1110 "emerg" messages on all terminals on some default syslog configurations.
1111 Eight levels are known :
Willy Tarreau6a06a402007-07-15 20:15:28 +02001112
Cyril Bontédc4d9032012-04-08 21:57:39 +02001113 emerg alert crit err warning notice info debug
Willy Tarreau6a06a402007-07-15 20:15:28 +02001114
Joe Williamsdf5b38f2010-12-29 17:05:48 +01001115log-send-hostname [<string>]
1116 Sets the hostname field in the syslog header. If optional "string" parameter
1117 is set the header is set to the string contents, otherwise uses the hostname
1118 of the system. Generally used if one is not relaying logs through an
1119 intermediate syslog server or for simply customizing the hostname printed in
1120 the logs.
1121
Kevinm48936af2010-12-22 16:08:21 +00001122log-tag <string>
1123 Sets the tag field in the syslog header to this string. It defaults to the
1124 program name as launched from the command line, which usually is "haproxy".
1125 Sometimes it can be useful to differentiate between multiple processes
Willy Tarreau094af4e2015-01-07 15:03:42 +01001126 running on the same host. See also the per-proxy "log-tag" directive.
Kevinm48936af2010-12-22 16:08:21 +00001127
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001128lua-load <file>
1129 This global directive loads and executes a Lua file. This directive can be
1130 used multiple times.
1131
Tim Duesterhusdd74b5f2020-01-12 13:55:40 +01001132lua-prepend-path <string> [<type>]
1133 Prepends the given string followed by a semicolon to Lua's package.<type>
1134 variable.
1135 <type> must either be "path" or "cpath". If <type> is not given it defaults
1136 to "path".
1137
1138 Lua's paths are semicolon delimited lists of patterns that specify how the
1139 `require` function attempts to find the source file of a library. Question
1140 marks (?) within a pattern will be replaced by module name. The path is
1141 evaluated left to right. This implies that paths that are prepended later
1142 will be checked earlier.
1143
1144 As an example by specifying the following path:
1145
1146 lua-prepend-path /usr/share/haproxy-lua/?/init.lua
1147 lua-prepend-path /usr/share/haproxy-lua/?.lua
1148
1149 When `require "example"` is being called Lua will first attempt to load the
1150 /usr/share/haproxy-lua/example.lua script, if that does not exist the
1151 /usr/share/haproxy-lua/example/init.lua will be attempted and the default
1152 paths if that does not exist either.
1153
1154 See https://www.lua.org/pil/8.1.html for the details within the Lua
1155 documentation.
1156
William Lallemand4cfede82017-11-24 22:02:34 +01001157master-worker [no-exit-on-failure]
William Lallemande202b1e2017-06-01 17:38:56 +02001158 Master-worker mode. It is equivalent to the command line "-W" argument.
1159 This mode will launch a "master" which will monitor the "workers". Using
1160 this mode, you can reload HAProxy directly by sending a SIGUSR2 signal to
Davor Ocelice9ed2812017-12-25 17:49:28 +01001161 the master. The master-worker mode is compatible either with the foreground
William Lallemande202b1e2017-06-01 17:38:56 +02001162 or daemon mode. It is recommended to use this mode with multiprocess and
1163 systemd.
William Lallemand4cfede82017-11-24 22:02:34 +01001164 By default, if a worker exits with a bad return code, in the case of a
1165 segfault for example, all workers will be killed, and the master will leave.
1166 It is convenient to combine this behavior with Restart=on-failure in a
1167 systemd unit file in order to relaunch the whole process. If you don't want
1168 this behavior, you must use the keyword "no-exit-on-failure".
William Lallemande202b1e2017-06-01 17:38:56 +02001169
William Lallemand4cfede82017-11-24 22:02:34 +01001170 See also "-W" in the management guide.
William Lallemande202b1e2017-06-01 17:38:56 +02001171
William Lallemand27edc4b2019-05-07 17:49:33 +02001172mworker-max-reloads <number>
1173 In master-worker mode, this option limits the number of time a worker can
John Roeslerfb2fce12019-07-10 15:45:51 -05001174 survive to a reload. If the worker did not leave after a reload, once its
William Lallemand27edc4b2019-05-07 17:49:33 +02001175 number of reloads is greater than this number, the worker will receive a
1176 SIGTERM. This option helps to keep under control the number of workers.
1177 See also "show proc" in the Management Guide.
1178
Willy Tarreauf42d7942020-10-20 11:54:49 +02001179nbproc <number> (deprecated)
Willy Tarreau6a06a402007-07-15 20:15:28 +02001180 Creates <number> processes when going daemon. This requires the "daemon"
1181 mode. By default, only one process is created, which is the recommended mode
1182 of operation. For systems limited to small sets of file descriptors per
Willy Tarreau149ab772019-01-26 14:27:06 +01001183 process, it may be needed to fork multiple daemons. When set to a value
1184 larger than 1, threads are automatically disabled. USING MULTIPLE PROCESSES
Willy Tarreauf42d7942020-10-20 11:54:49 +02001185 IS HARDER TO DEBUG AND IS REALLY DISCOURAGED. This directive is deprecated
1186 and scheduled for removal in 2.5. Please use "nbthread" instead. See also
1187 "daemon" and "nbthread".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001188
Christopher Fauletbe0faa22017-08-29 15:37:10 +02001189nbthread <number>
1190 This setting is only available when support for threads was built in. It
Willy Tarreau26f6ae12019-02-02 12:56:15 +01001191 makes haproxy run on <number> threads. This is exclusive with "nbproc". While
1192 "nbproc" historically used to be the only way to use multiple processors, it
1193 also involved a number of shortcomings related to the lack of synchronization
1194 between processes (health-checks, peers, stick-tables, stats, ...) which do
1195 not affect threads. As such, any modern configuration is strongly encouraged
Willy Tarreau149ab772019-01-26 14:27:06 +01001196 to migrate away from "nbproc" to "nbthread". "nbthread" also works when
1197 HAProxy is started in foreground. On some platforms supporting CPU affinity,
1198 when nbproc is not used, the default "nbthread" value is automatically set to
1199 the number of CPUs the process is bound to upon startup. This means that the
1200 thread count can easily be adjusted from the calling process using commands
1201 like "taskset" or "cpuset". Otherwise, this value defaults to 1. The default
1202 value is reported in the output of "haproxy -vv". See also "nbproc".
Christopher Fauletbe0faa22017-08-29 15:37:10 +02001203
Willy Tarreau6a06a402007-07-15 20:15:28 +02001204pidfile <pidfile>
MIZUTA Takeshic32f3942020-08-26 13:46:19 +09001205 Writes PIDs of all daemons into file <pidfile> when daemon mode or writes PID
1206 of master process into file <pidfile> when master-worker mode. This option is
1207 equivalent to the "-p" command line argument. The file must be accessible to
1208 the user starting the process. See also "daemon" and "master-worker".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001209
Willy Tarreau119e50e2020-05-22 13:53:29 +02001210pp2-never-send-local
1211 A bug in the PROXY protocol v2 implementation was present in HAProxy up to
1212 version 2.1, causing it to emit a PROXY command instead of a LOCAL command
1213 for health checks. This is particularly minor but confuses some servers'
1214 logs. Sadly, the bug was discovered very late and revealed that some servers
1215 which possibly only tested their PROXY protocol implementation against
1216 HAProxy fail to properly handle the LOCAL command, and permanently remain in
1217 the "down" state when HAProxy checks them. When this happens, it is possible
1218 to enable this global option to revert to the older (bogus) behavior for the
1219 time it takes to contact the affected components' vendors and get them fixed.
1220 This option is disabled by default and acts on all servers having the
1221 "send-proxy-v2" statement.
1222
Willy Tarreau1d549722016-02-16 12:41:57 +01001223presetenv <name> <value>
1224 Sets environment variable <name> to value <value>. If the variable exists, it
1225 is NOT overwritten. The changes immediately take effect so that the next line
1226 in the configuration file sees the new value. See also "setenv", "resetenv",
1227 and "unsetenv".
1228
1229resetenv [<name> ...]
1230 Removes all environment variables except the ones specified in argument. It
1231 allows to use a clean controlled environment before setting new values with
1232 setenv or unsetenv. Please note that some internal functions may make use of
1233 some environment variables, such as time manipulation functions, but also
1234 OpenSSL or even external checks. This must be used with extreme care and only
1235 after complete validation. The changes immediately take effect so that the
1236 next line in the configuration file sees the new environment. See also
1237 "setenv", "presetenv", and "unsetenv".
1238
Christopher Fauletff4121f2017-11-22 16:38:49 +01001239stats bind-process [ all | odd | even | <process_num>[-[process_num>]] ] ...
Willy Tarreau35b7b162012-10-22 23:17:18 +02001240 Limits the stats socket to a certain set of processes numbers. By default the
1241 stats socket is bound to all processes, causing a warning to be emitted when
1242 nbproc is greater than 1 because there is no way to select the target process
1243 when connecting. However, by using this setting, it becomes possible to pin
1244 the stats socket to a specific set of processes, typically the first one. The
1245 warning will automatically be disabled when this setting is used, whatever
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001246 the number of processes used. The maximum process ID depends on the machine's
Christopher Fauletff4121f2017-11-22 16:38:49 +01001247 word size (32 or 64). Ranges can be partially defined. The higher bound can
1248 be omitted. In such case, it is replaced by the corresponding maximum
1249 value. A better option consists in using the "process" setting of the "stats
1250 socket" line to force the process on each line.
Willy Tarreau35b7b162012-10-22 23:17:18 +02001251
Baptiste Assmann5626f482015-08-23 10:00:10 +02001252server-state-base <directory>
1253 Specifies the directory prefix to be prepended in front of all servers state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001254 file names which do not start with a '/'. See also "server-state-file",
1255 "load-server-state-from-file" and "server-state-file-name".
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +02001256
1257server-state-file <file>
1258 Specifies the path to the file containing state of servers. If the path starts
1259 with a slash ('/'), it is considered absolute, otherwise it is considered
1260 relative to the directory specified using "server-state-base" (if set) or to
1261 the current directory. Before reloading HAProxy, it is possible to save the
1262 servers' current state using the stats command "show servers state". The
1263 output of this command must be written in the file pointed by <file>. When
1264 starting up, before handling traffic, HAProxy will read, load and apply state
1265 for each server found in the file and available in its current running
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001266 configuration. See also "server-state-base" and "show servers state",
1267 "load-server-state-from-file" and "server-state-file-name"
Baptiste Assmann5626f482015-08-23 10:00:10 +02001268
Willy Tarreau1d549722016-02-16 12:41:57 +01001269setenv <name> <value>
1270 Sets environment variable <name> to value <value>. If the variable exists, it
1271 is overwritten. The changes immediately take effect so that the next line in
1272 the configuration file sees the new value. See also "presetenv", "resetenv",
1273 and "unsetenv".
1274
Willy Tarreau636848a2019-04-15 19:38:50 +02001275set-dumpable
1276 This option is better left disabled by default and enabled only upon a
William Dauchyec730982019-10-27 20:08:10 +01001277 developer's request. If it has been enabled, it may still be forcibly
1278 disabled by prefixing it with the "no" keyword. It has no impact on
1279 performance nor stability but will try hard to re-enable core dumps that were
1280 possibly disabled by file size limitations (ulimit -f), core size limitations
1281 (ulimit -c), or "dumpability" of a process after changing its UID/GID (such
1282 as /proc/sys/fs/suid_dumpable on Linux). Core dumps might still be limited by
1283 the current directory's permissions (check what directory the file is started
1284 from), the chroot directory's permission (it may be needed to temporarily
1285 disable the chroot directive or to move it to a dedicated writable location),
1286 or any other system-specific constraint. For example, some Linux flavours are
1287 notorious for replacing the default core file with a path to an executable
1288 not even installed on the system (check /proc/sys/kernel/core_pattern). Often,
1289 simply writing "core", "core.%p" or "/var/log/core/core.%p" addresses the
1290 issue. When trying to enable this option waiting for a rare issue to
1291 re-appear, it's often a good idea to first try to obtain such a dump by
1292 issuing, for example, "kill -11" to the haproxy process and verify that it
1293 leaves a core where expected when dying.
Willy Tarreau636848a2019-04-15 19:38:50 +02001294
Willy Tarreau610f04b2014-02-13 11:36:41 +01001295ssl-default-bind-ciphers <ciphers>
1296 This setting is only available when support for OpenSSL was built in. It sets
1297 the default string describing the list of cipher algorithms ("cipher suite")
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +00001298 that are negotiated during the SSL/TLS handshake up to TLSv1.2 for all
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001299 "bind" lines which do not explicitly define theirs. The format of the string
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001300 is defined in "man 1 ciphers" from OpenSSL man pages. For background
1301 information and recommendations see e.g.
1302 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
1303 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
1304 cipher configuration, please check the "ssl-default-bind-ciphersuites" keyword.
1305 Please check the "bind" keyword for more information.
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001306
1307ssl-default-bind-ciphersuites <ciphersuites>
1308 This setting is only available when support for OpenSSL was built in and
1309 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default string
1310 describing the list of cipher algorithms ("cipher suite") that are negotiated
1311 during the TLSv1.3 handshake for all "bind" lines which do not explicitly define
1312 theirs. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001313 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites". For
1314 cipher configuration for TLSv1.2 and earlier, please check the
1315 "ssl-default-bind-ciphers" keyword. Please check the "bind" keyword for more
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001316 information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001317
Jerome Magninb203ff62020-04-03 15:28:22 +02001318ssl-default-bind-curves <curves>
1319 This setting is only available when support for OpenSSL was built in. It sets
1320 the default string describing the list of elliptic curves algorithms ("curve
1321 suite") that are negotiated during the SSL/TLS handshake with ECDHE. The format
1322 of the string is a colon-delimited list of curve name.
1323 Please check the "bind" keyword for more information.
1324
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001325ssl-default-bind-options [<option>]...
1326 This setting is only available when support for OpenSSL was built in. It sets
1327 default ssl-options to force on all "bind" lines. Please check the "bind"
1328 keyword to see available options.
1329
1330 Example:
1331 global
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +02001332 ssl-default-bind-options ssl-min-ver TLSv1.0 no-tls-tickets
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001333
Willy Tarreau610f04b2014-02-13 11:36:41 +01001334ssl-default-server-ciphers <ciphers>
1335 This setting is only available when support for OpenSSL was built in. It
1336 sets the default string describing the list of cipher algorithms that are
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +00001337 negotiated during the SSL/TLS handshake up to TLSv1.2 with the server,
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001338 for all "server" lines which do not explicitly define theirs. The format of
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001339 the string is defined in "man 1 ciphers" from OpenSSL man pages. For background
1340 information and recommendations see e.g.
1341 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
1342 (https://mozilla.github.io/server-side-tls/ssl-config-generator/).
1343 For TLSv1.3 cipher configuration, please check the
1344 "ssl-default-server-ciphersuites" keyword. Please check the "server" keyword
1345 for more information.
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001346
1347ssl-default-server-ciphersuites <ciphersuites>
1348 This setting is only available when support for OpenSSL was built in and
1349 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default
1350 string describing the list of cipher algorithms that are negotiated during
1351 the TLSv1.3 handshake with the server, for all "server" lines which do not
1352 explicitly define theirs. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001353 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites". For
1354 cipher configuration for TLSv1.2 and earlier, please check the
1355 "ssl-default-server-ciphers" keyword. Please check the "server" keyword for
1356 more information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001357
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001358ssl-default-server-options [<option>]...
1359 This setting is only available when support for OpenSSL was built in. It sets
1360 default ssl-options to force on all "server" lines. Please check the "server"
1361 keyword to see available options.
1362
Remi Gacogne47783ef2015-05-29 15:53:22 +02001363ssl-dh-param-file <file>
1364 This setting is only available when support for OpenSSL was built in. It sets
1365 the default DH parameters that are used during the SSL/TLS handshake when
1366 ephemeral Diffie-Hellman (DHE) key exchange is used, for all "bind" lines
Davor Ocelice9ed2812017-12-25 17:49:28 +01001367 which do not explicitly define theirs. It will be overridden by custom DH
Remi Gacogne47783ef2015-05-29 15:53:22 +02001368 parameters found in a bind certificate file if any. If custom DH parameters
Cyril Bonté307ee1e2015-09-28 23:16:06 +02001369 are not specified either by using ssl-dh-param-file or by setting them
1370 directly in the certificate file, pre-generated DH parameters of the size
1371 specified by tune.ssl.default-dh-param will be used. Custom parameters are
1372 known to be more secure and therefore their use is recommended.
Remi Gacogne47783ef2015-05-29 15:53:22 +02001373 Custom DH parameters may be generated by using the OpenSSL command
1374 "openssl dhparam <size>", where size should be at least 2048, as 1024-bit DH
1375 parameters should not be considered secure anymore.
1376
William Lallemand8e8581e2020-10-20 17:36:46 +02001377ssl-load-extra-del-ext
1378 This setting allows to configure the way HAProxy does the lookup for the
1379 extra SSL files. By default HAProxy adds a new extension to the filename.
William Lallemand089c1382020-10-23 17:35:12 +02001380 (ex: with "foobar.crt" load "foobar.crt.key"). With this option enabled,
William Lallemand8e8581e2020-10-20 17:36:46 +02001381 HAProxy removes the extension before adding the new one (ex: with
William Lallemand089c1382020-10-23 17:35:12 +02001382 "foobar.crt" load "foobar.key").
1383
1384 Your crt file must have a ".crt" extension for this option to work.
William Lallemand8e8581e2020-10-20 17:36:46 +02001385
1386 This option is not compatible with bundle extensions (.ecdsa, .rsa. .dsa)
1387 and won't try to remove them.
1388
1389 This option is disabled by default. See also "ssl-load-extra-files".
1390
William Lallemand4c5adbf2020-02-24 14:23:22 +01001391ssl-load-extra-files <none|all|bundle|sctl|ocsp|issuer|key>*
William Lallemand3af48e72020-02-03 17:15:52 +01001392 This setting alters the way HAProxy will look for unspecified files during
Jerome Magnin587be9c2020-09-07 11:55:57 +02001393 the loading of the SSL certificates associated to "bind" lines. It does not
1394 apply to certificates used for client authentication on "server" lines.
William Lallemand3af48e72020-02-03 17:15:52 +01001395
1396 By default, HAProxy discovers automatically a lot of files not specified in
1397 the configuration, and you may want to disable this behavior if you want to
1398 optimize the startup time.
1399
1400 "none": Only load the files specified in the configuration. Don't try to load
1401 a certificate bundle if the file does not exist. In the case of a directory,
1402 it won't try to bundle the certificates if they have the same basename.
1403
1404 "all": This is the default behavior, it will try to load everything,
William Lallemand4c5adbf2020-02-24 14:23:22 +01001405 bundles, sctl, ocsp, issuer, key.
William Lallemand3af48e72020-02-03 17:15:52 +01001406
1407 "bundle": When a file specified in the configuration does not exist, HAProxy
William Lallemandf9ff3ec2020-10-02 17:57:44 +02001408 will try to load a "cert bundle".
1409
1410 Starting from HAProxy 2.3, the bundles are not loaded in the same OpenSSL
1411 certificate store, instead it will loads each certificate in a separate
1412 store which is equivalent to declaring multiple "crt". OpenSSL 1.1.1 is
1413 required to achieve this. Which means that bundles are now used only for
1414 backward compatibility and are not mandatory anymore to do an hybrid RSA/ECC
1415 bind configuration..
1416
1417 To associate these PEM files into a "cert bundle" that is recognized by
1418 haproxy, they must be named in the following way: All PEM files that are to
1419 be bundled must have the same base name, with a suffix indicating the key
1420 type. Currently, three suffixes are supported: rsa, dsa and ecdsa. For
1421 example, if www.example.com has two PEM files, an RSA file and an ECDSA
1422 file, they must be named: "example.pem.rsa" and "example.pem.ecdsa". The
1423 first part of the filename is arbitrary; only the suffix matters. To load
1424 this bundle into haproxy, specify the base name only:
1425
1426 Example : bind :8443 ssl crt example.pem
1427
1428 Note that the suffix is not given to haproxy; this tells haproxy to look for
1429 a cert bundle.
1430
1431 HAProxy will load all PEM files in the bundle as if they were configured
1432 separately in several "crt".
1433
1434 The bundle loading does not have an impact anymore on the directory loading
1435 since files are loading separately.
1436
1437 On the CLI, bundles are seen as separate files, and the bundle extension is
1438 required to commit them.
1439
William Dauchy57dd6f12020-10-06 15:22:37 +02001440 OCSP files (.ocsp), issuer files (.issuer), Certificate Transparency (.sctl)
William Lallemandf9ff3ec2020-10-02 17:57:44 +02001441 as well as private keys (.key) are supported with multi-cert bundling.
William Lallemand3af48e72020-02-03 17:15:52 +01001442
1443 "sctl": Try to load "<basename>.sctl" for each crt keyword.
1444
1445 "ocsp": Try to load "<basename>.ocsp" for each crt keyword.
1446
1447 "issuer": Try to load "<basename>.issuer" if the issuer of the OCSP file is
1448 not provided in the PEM file.
1449
William Lallemand4c5adbf2020-02-24 14:23:22 +01001450 "key": If the private key was not provided by the PEM file, try to load a
1451 file "<basename>.key" containing a private key.
1452
William Lallemand3af48e72020-02-03 17:15:52 +01001453 The default behavior is "all".
1454
1455 Example:
1456 ssl-load-extra-files bundle sctl
1457 ssl-load-extra-files sctl ocsp issuer
1458 ssl-load-extra-files none
1459
1460 See also: "crt", section 5.1 about bind options.
1461
Emeric Brun850efd52014-01-29 12:24:34 +01001462ssl-server-verify [none|required]
1463 The default behavior for SSL verify on servers side. If specified to 'none',
1464 servers certificates are not verified. The default is 'required' except if
1465 forced using cmdline option '-dV'.
1466
Emmanuel Hocdetc3b7e742020-04-22 11:06:19 +02001467ssl-skip-self-issued-ca
Daniel Corbett67a82712020-07-06 23:01:19 -04001468 Self issued CA, aka x509 root CA, is the anchor for chain validation: as a
Emmanuel Hocdetc3b7e742020-04-22 11:06:19 +02001469 server is useless to send it, client must have it. Standard configuration
1470 need to not include such CA in PEM file. This option allows you to keep such
1471 CA in PEM file without sending it to the client. Use case is to provide
1472 issuer for ocsp without the need for '.issuer' file and be able to share it
1473 with 'issuers-chain-path'. This concerns all certificates without intermediate
1474 certificates. It's useless for BoringSSL, .issuer is ignored because ocsp
William Lallemand9a1d8392020-08-10 17:28:23 +02001475 bits does not need it. Requires at least OpenSSL 1.0.2.
Emmanuel Hocdetc3b7e742020-04-22 11:06:19 +02001476
Willy Tarreauabb175f2012-09-24 12:43:26 +02001477stats socket [<address:port>|<path>] [param*]
1478 Binds a UNIX socket to <path> or a TCPv4/v6 address to <address:port>.
1479 Connections to this socket will return various statistics outputs and even
1480 allow some commands to be issued to change some runtime settings. Please
Willy Tarreau1af20c72017-06-23 16:01:14 +02001481 consult section 9.3 "Unix Socket commands" of Management Guide for more
Kevin Decherf949c7202015-10-13 23:26:44 +02001482 details.
Willy Tarreau6162db22009-10-10 17:13:00 +02001483
Willy Tarreauabb175f2012-09-24 12:43:26 +02001484 All parameters supported by "bind" lines are supported, for instance to
1485 restrict access to some users or their access rights. Please consult
1486 section 5.1 for more information.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001487
1488stats timeout <timeout, in milliseconds>
1489 The default timeout on the stats socket is set to 10 seconds. It is possible
1490 to change this value with "stats timeout". The value must be passed in
Willy Tarreaubefdff12007-12-02 22:27:38 +01001491 milliseconds, or be suffixed by a time unit among { us, ms, s, m, h, d }.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001492
1493stats maxconn <connections>
1494 By default, the stats socket is limited to 10 concurrent connections. It is
1495 possible to change this value with "stats maxconn".
1496
Willy Tarreau6a06a402007-07-15 20:15:28 +02001497uid <number>
1498 Changes the process' user ID to <number>. It is recommended that the user ID
1499 is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
1500 be started with superuser privileges in order to be able to switch to another
1501 one. See also "gid" and "user".
1502
1503ulimit-n <number>
1504 Sets the maximum number of per-process file-descriptors to <number>. By
1505 default, it is automatically computed, so it is recommended not to use this
1506 option.
1507
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001508unix-bind [ prefix <prefix> ] [ mode <mode> ] [ user <user> ] [ uid <uid> ]
1509 [ group <group> ] [ gid <gid> ]
1510
1511 Fixes common settings to UNIX listening sockets declared in "bind" statements.
1512 This is mainly used to simplify declaration of those UNIX sockets and reduce
1513 the risk of errors, since those settings are most commonly required but are
1514 also process-specific. The <prefix> setting can be used to force all socket
1515 path to be relative to that directory. This might be needed to access another
1516 component's chroot. Note that those paths are resolved before haproxy chroots
1517 itself, so they are absolute. The <mode>, <user>, <uid>, <group> and <gid>
1518 all have the same meaning as their homonyms used by the "bind" statement. If
1519 both are specified, the "bind" statement has priority, meaning that the
1520 "unix-bind" settings may be seen as process-wide default settings.
1521
Willy Tarreau1d549722016-02-16 12:41:57 +01001522unsetenv [<name> ...]
1523 Removes environment variables specified in arguments. This can be useful to
1524 hide some sensitive information that are occasionally inherited from the
1525 user's environment during some operations. Variables which did not exist are
1526 silently ignored so that after the operation, it is certain that none of
1527 these variables remain. The changes immediately take effect so that the next
1528 line in the configuration file will not see these variables. See also
1529 "setenv", "presetenv", and "resetenv".
1530
Willy Tarreau6a06a402007-07-15 20:15:28 +02001531user <user name>
1532 Similar to "uid" but uses the UID of user name <user name> from /etc/passwd.
1533 See also "uid" and "group".
1534
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02001535node <name>
1536 Only letters, digits, hyphen and underscore are allowed, like in DNS names.
1537
1538 This statement is useful in HA configurations where two or more processes or
1539 servers share the same IP address. By setting a different node-name on all
1540 nodes, it becomes easy to immediately spot what server is handling the
1541 traffic.
1542
1543description <text>
1544 Add a text that describes the instance.
1545
1546 Please note that it is required to escape certain characters (# for example)
1547 and this text is inserted into a html page so you should avoid using
1548 "<" and ">" characters.
1549
Thomas Holmesdb04f192015-05-18 13:21:39 +0100155051degrees-data-file <file path>
1551 The path of the 51Degrees data file to provide device detection services. The
Davor Ocelice9ed2812017-12-25 17:49:28 +01001552 file should be unzipped and accessible by HAProxy with relevant permissions.
Thomas Holmesdb04f192015-05-18 13:21:39 +01001553
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001554 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001555 compiled with USE_51DEGREES.
1556
Ben Shillitof25e8e52016-12-02 14:25:37 +0000155751degrees-property-name-list [<string> ...]
Thomas Holmesdb04f192015-05-18 13:21:39 +01001558 A list of 51Degrees property names to be load from the dataset. A full list
1559 of names is available on the 51Degrees website:
1560 https://51degrees.com/resources/property-dictionary
1561
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001562 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001563 compiled with USE_51DEGREES.
1564
Dragan Dosen93b38d92015-06-29 16:43:25 +0200156551degrees-property-separator <char>
Thomas Holmesdb04f192015-05-18 13:21:39 +01001566 A char that will be appended to every property value in a response header
1567 containing 51Degrees results. If not set that will be set as ','.
1568
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001569 Please note that this option is only available when haproxy has been
1570 compiled with USE_51DEGREES.
1571
157251degrees-cache-size <number>
1573 Sets the size of the 51Degrees converter cache to <number> entries. This
1574 is an LRU cache which reminds previous device detections and their results.
1575 By default, this cache is disabled.
1576
1577 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001578 compiled with USE_51DEGREES.
1579
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001580wurfl-data-file <file path>
1581 The path of the WURFL data file to provide device detection services. The
1582 file should be accessible by HAProxy with relevant permissions.
1583
1584 Please note that this option is only available when haproxy has been compiled
1585 with USE_WURFL=1.
1586
1587wurfl-information-list [<capability>]*
1588 A space-delimited list of WURFL capabilities, virtual capabilities, property
1589 names we plan to use in injected headers. A full list of capability and
1590 virtual capability names is available on the Scientiamobile website :
1591
1592 https://www.scientiamobile.com/wurflCapability
1593
1594 Valid WURFL properties are:
1595 - wurfl_id Contains the device ID of the matched device.
1596
1597 - wurfl_root_id Contains the device root ID of the matched
1598 device.
1599
1600 - wurfl_isdevroot Tells if the matched device is a root device.
1601 Possible values are "TRUE" or "FALSE".
1602
1603 - wurfl_useragent The original useragent coming with this
1604 particular web request.
1605
1606 - wurfl_api_version Contains a string representing the currently
1607 used Libwurfl API version.
1608
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001609 - wurfl_info A string containing information on the parsed
1610 wurfl.xml and its full path.
1611
1612 - wurfl_last_load_time Contains the UNIX timestamp of the last time
1613 WURFL has been loaded successfully.
1614
1615 - wurfl_normalized_useragent The normalized useragent.
1616
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001617 Please note that this option is only available when haproxy has been compiled
1618 with USE_WURFL=1.
1619
1620wurfl-information-list-separator <char>
1621 A char that will be used to separate values in a response header containing
1622 WURFL results. If not set that a comma (',') will be used by default.
1623
1624 Please note that this option is only available when haproxy has been compiled
1625 with USE_WURFL=1.
1626
1627wurfl-patch-file [<file path>]
1628 A list of WURFL patch file paths. Note that patches are loaded during startup
1629 thus before the chroot.
1630
1631 Please note that this option is only available when haproxy has been compiled
1632 with USE_WURFL=1.
1633
paulborilebad132c2019-04-18 11:57:04 +02001634wurfl-cache-size <size>
1635 Sets the WURFL Useragent cache size. For faster lookups, already processed user
1636 agents are kept in a LRU cache :
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001637 - "0" : no cache is used.
paulborilebad132c2019-04-18 11:57:04 +02001638 - <size> : size of lru cache in elements.
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001639
1640 Please note that this option is only available when haproxy has been compiled
1641 with USE_WURFL=1.
1642
William Dauchy0fec3ab2019-10-27 20:08:11 +01001643strict-limits
William Dauchya5194602020-03-28 19:29:58 +01001644 Makes process fail at startup when a setrlimit fails. Haproxy tries to set the
1645 best setrlimit according to what has been calculated. If it fails, it will
1646 emit a warning. This option is here to guarantee an explicit failure of
1647 haproxy when those limits fail. It is enabled by default. It may still be
1648 forcibly disabled by prefixing it with the "no" keyword.
William Dauchy0fec3ab2019-10-27 20:08:11 +01001649
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016503.2. Performance tuning
Willy Tarreau6a06a402007-07-15 20:15:28 +02001651-----------------------
1652
Willy Tarreaubeb859a2018-11-22 18:07:59 +01001653busy-polling
1654 In some situations, especially when dealing with low latency on processors
1655 supporting a variable frequency or when running inside virtual machines, each
1656 time the process waits for an I/O using the poller, the processor goes back
1657 to sleep or is offered to another VM for a long time, and it causes
1658 excessively high latencies. This option provides a solution preventing the
1659 processor from sleeping by always using a null timeout on the pollers. This
1660 results in a significant latency reduction (30 to 100 microseconds observed)
1661 at the expense of a risk to overheat the processor. It may even be used with
1662 threads, in which case improperly bound threads may heavily conflict,
1663 resulting in a worse performance and high values for the CPU stolen fields
1664 in "show info" output, indicating which threads are misconfigured. It is
1665 important not to let the process run on the same processor as the network
1666 interrupts when this option is used. It is also better to avoid using it on
1667 multiple CPU threads sharing the same core. This option is disabled by
1668 default. If it has been enabled, it may still be forcibly disabled by
1669 prefixing it with the "no" keyword. It is ignored by the "select" and
1670 "poll" pollers.
1671
William Dauchy3894d972019-12-28 15:36:02 +01001672 This option is automatically disabled on old processes in the context of
1673 seamless reload; it avoids too much cpu conflicts when multiple processes
1674 stay around for some time waiting for the end of their current connections.
1675
Willy Tarreau1746eec2014-04-25 10:46:47 +02001676max-spread-checks <delay in milliseconds>
1677 By default, haproxy tries to spread the start of health checks across the
1678 smallest health check interval of all the servers in a farm. The principle is
1679 to avoid hammering services running on the same server. But when using large
1680 check intervals (10 seconds or more), the last servers in the farm take some
1681 time before starting to be tested, which can be a problem. This parameter is
1682 used to enforce an upper bound on delay between the first and the last check,
1683 even if the servers' check intervals are larger. When servers run with
1684 shorter intervals, their intervals will be respected though.
1685
Willy Tarreau6a06a402007-07-15 20:15:28 +02001686maxconn <number>
1687 Sets the maximum per-process number of concurrent connections to <number>. It
1688 is equivalent to the command-line argument "-n". Proxies will stop accepting
1689 connections when this limit is reached. The "ulimit-n" parameter is
Willy Tarreau8274e102014-06-19 15:31:25 +02001690 automatically adjusted according to this value. See also "ulimit-n". Note:
1691 the "select" poller cannot reliably use more than 1024 file descriptors on
1692 some platforms. If your platform only supports select and reports "select
1693 FAILED" on startup, you need to reduce maxconn until it works (slightly
Willy Tarreaub28f3442019-03-04 08:13:43 +01001694 below 500 in general). If this value is not set, it will automatically be
1695 calculated based on the current file descriptors limit reported by the
1696 "ulimit -n" command, possibly reduced to a lower value if a memory limit
1697 is enforced, based on the buffer size, memory allocated to compression, SSL
1698 cache size, and use or not of SSL and the associated maxsslconn (which can
1699 also be automatic).
Willy Tarreau6a06a402007-07-15 20:15:28 +02001700
Willy Tarreau81c25d02011-09-07 15:17:21 +02001701maxconnrate <number>
1702 Sets the maximum per-process number of connections per second to <number>.
1703 Proxies will stop accepting connections when this limit is reached. It can be
1704 used to limit the global capacity regardless of each frontend capacity. It is
1705 important to note that this can only be used as a service protection measure,
1706 as there will not necessarily be a fair share between frontends when the
1707 limit is reached, so it's a good idea to also limit each frontend to some
1708 value close to its expected share. Also, lowering tune.maxaccept can improve
1709 fairness.
1710
William Lallemandd85f9172012-11-09 17:05:39 +01001711maxcomprate <number>
1712 Sets the maximum per-process input compression rate to <number> kilobytes
Davor Ocelice9ed2812017-12-25 17:49:28 +01001713 per second. For each session, if the maximum is reached, the compression
William Lallemandd85f9172012-11-09 17:05:39 +01001714 level will be decreased during the session. If the maximum is reached at the
1715 beginning of a session, the session will not compress at all. If the maximum
1716 is not reached, the compression level will be increased up to
Davor Ocelice9ed2812017-12-25 17:49:28 +01001717 tune.comp.maxlevel. A value of zero means there is no limit, this is the
William Lallemandd85f9172012-11-09 17:05:39 +01001718 default value.
1719
William Lallemand072a2bf2012-11-20 17:01:01 +01001720maxcompcpuusage <number>
1721 Sets the maximum CPU usage HAProxy can reach before stopping the compression
1722 for new requests or decreasing the compression level of current requests.
1723 It works like 'maxcomprate' but measures CPU usage instead of incoming data
1724 bandwidth. The value is expressed in percent of the CPU used by haproxy. In
1725 case of multiple processes (nbproc > 1), each process manages its individual
1726 usage. A value of 100 disable the limit. The default value is 100. Setting
1727 a lower value will prevent the compression work from slowing the whole
1728 process down and from introducing high latencies.
1729
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001730maxpipes <number>
1731 Sets the maximum per-process number of pipes to <number>. Currently, pipes
1732 are only used by kernel-based tcp splicing. Since a pipe contains two file
1733 descriptors, the "ulimit-n" value will be increased accordingly. The default
1734 value is maxconn/4, which seems to be more than enough for most heavy usages.
1735 The splice code dynamically allocates and releases pipes, and can fall back
1736 to standard copy, so setting this value too low may only impact performance.
1737
Willy Tarreau93e7c002013-10-07 18:51:07 +02001738maxsessrate <number>
1739 Sets the maximum per-process number of sessions per second to <number>.
1740 Proxies will stop accepting connections when this limit is reached. It can be
1741 used to limit the global capacity regardless of each frontend capacity. It is
1742 important to note that this can only be used as a service protection measure,
1743 as there will not necessarily be a fair share between frontends when the
1744 limit is reached, so it's a good idea to also limit each frontend to some
1745 value close to its expected share. Also, lowering tune.maxaccept can improve
1746 fairness.
1747
Willy Tarreau403edff2012-09-06 11:58:37 +02001748maxsslconn <number>
1749 Sets the maximum per-process number of concurrent SSL connections to
1750 <number>. By default there is no SSL-specific limit, which means that the
1751 global maxconn setting will apply to all connections. Setting this limit
1752 avoids having openssl use too much memory and crash when malloc returns NULL
1753 (since it unfortunately does not reliably check for such conditions). Note
1754 that the limit applies both to incoming and outgoing connections, so one
1755 connection which is deciphered then ciphered accounts for 2 SSL connections.
Willy Tarreaud0256482015-01-15 21:45:22 +01001756 If this value is not set, but a memory limit is enforced, this value will be
1757 automatically computed based on the memory limit, maxconn, the buffer size,
1758 memory allocated to compression, SSL cache size, and use of SSL in either
1759 frontends, backends or both. If neither maxconn nor maxsslconn are specified
1760 when there is a memory limit, haproxy will automatically adjust these values
1761 so that 100% of the connections can be made over SSL with no risk, and will
1762 consider the sides where it is enabled (frontend, backend, both).
Willy Tarreau403edff2012-09-06 11:58:37 +02001763
Willy Tarreaue43d5322013-10-07 20:01:52 +02001764maxsslrate <number>
1765 Sets the maximum per-process number of SSL sessions per second to <number>.
1766 SSL listeners will stop accepting connections when this limit is reached. It
1767 can be used to limit the global SSL CPU usage regardless of each frontend
1768 capacity. It is important to note that this can only be used as a service
1769 protection measure, as there will not necessarily be a fair share between
1770 frontends when the limit is reached, so it's a good idea to also limit each
1771 frontend to some value close to its expected share. It is also important to
1772 note that the sessions are accounted before they enter the SSL stack and not
1773 after, which also protects the stack against bad handshakes. Also, lowering
1774 tune.maxaccept can improve fairness.
1775
William Lallemand9d5f5482012-11-07 16:12:57 +01001776maxzlibmem <number>
1777 Sets the maximum amount of RAM in megabytes per process usable by the zlib.
1778 When the maximum amount is reached, future sessions will not compress as long
1779 as RAM is unavailable. When sets to 0, there is no limit.
William Lallemande3a7d992012-11-20 11:25:20 +01001780 The default value is 0. The value is available in bytes on the UNIX socket
1781 with "show info" on the line "MaxZlibMemUsage", the memory used by zlib is
1782 "ZlibMemUsage" in bytes.
1783
Willy Tarreau6a06a402007-07-15 20:15:28 +02001784noepoll
1785 Disables the use of the "epoll" event polling system on Linux. It is
1786 equivalent to the command-line argument "-de". The next polling system
Willy Tarreaue9f49e72012-11-11 17:42:00 +01001787 used will generally be "poll". See also "nopoll".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001788
1789nokqueue
1790 Disables the use of the "kqueue" event polling system on BSD. It is
1791 equivalent to the command-line argument "-dk". The next polling system
1792 used will generally be "poll". See also "nopoll".
1793
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +00001794noevports
1795 Disables the use of the event ports event polling system on SunOS systems
1796 derived from Solaris 10 and later. It is equivalent to the command-line
1797 argument "-dv". The next polling system used will generally be "poll". See
1798 also "nopoll".
1799
Willy Tarreau6a06a402007-07-15 20:15:28 +02001800nopoll
1801 Disables the use of the "poll" event polling system. It is equivalent to the
1802 command-line argument "-dp". The next polling system used will be "select".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001803 It should never be needed to disable "poll" since it's available on all
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +00001804 platforms supported by HAProxy. See also "nokqueue", "noepoll" and
1805 "noevports".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001806
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001807nosplice
1808 Disables the use of kernel tcp splicing between sockets on Linux. It is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001809 equivalent to the command line argument "-dS". Data will then be copied
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001810 using conventional and more portable recv/send calls. Kernel tcp splicing is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001811 limited to some very recent instances of kernel 2.6. Most versions between
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001812 2.6.25 and 2.6.28 are buggy and will forward corrupted data, so they must not
1813 be used. This option makes it easier to globally disable kernel splicing in
1814 case of doubt. See also "option splice-auto", "option splice-request" and
1815 "option splice-response".
1816
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001817nogetaddrinfo
1818 Disables the use of getaddrinfo(3) for name resolving. It is equivalent to
1819 the command line argument "-dG". Deprecated gethostbyname(3) will be used.
1820
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +00001821noreuseport
1822 Disables the use of SO_REUSEPORT - see socket(7). It is equivalent to the
1823 command line argument "-dR".
1824
Willy Tarreaud2d33482019-04-25 17:09:07 +02001825profiling.tasks { auto | on | off }
1826 Enables ('on') or disables ('off') per-task CPU profiling. When set to 'auto'
1827 the profiling automatically turns on a thread when it starts to suffer from
1828 an average latency of 1000 microseconds or higher as reported in the
1829 "avg_loop_us" activity field, and automatically turns off when the latency
John Roeslerfb2fce12019-07-10 15:45:51 -05001830 returns below 990 microseconds (this value is an average over the last 1024
Willy Tarreaud2d33482019-04-25 17:09:07 +02001831 loops so it does not vary quickly and tends to significantly smooth short
1832 spikes). It may also spontaneously trigger from time to time on overloaded
1833 systems, containers, or virtual machines, or when the system swaps (which
1834 must absolutely never happen on a load balancer).
1835
1836 CPU profiling per task can be very convenient to report where the time is
1837 spent and which requests have what effect on which other request. Enabling
1838 it will typically affect the overall's performance by less than 1%, thus it
1839 is recommended to leave it to the default 'auto' value so that it only
1840 operates when a problem is identified. This feature requires a system
Willy Tarreau75c62c22018-11-22 11:02:09 +01001841 supporting the clock_gettime(2) syscall with clock identifiers
1842 CLOCK_MONOTONIC and CLOCK_THREAD_CPUTIME_ID, otherwise the reported time will
1843 be zero. This option may be changed at run time using "set profiling" on the
1844 CLI.
1845
Willy Tarreaufe255b72007-10-14 23:09:26 +02001846spread-checks <0..50, in percent>
Simon Hormand60d6912013-11-25 10:46:36 +09001847 Sometimes it is desirable to avoid sending agent and health checks to
1848 servers at exact intervals, for instance when many logical servers are
1849 located on the same physical server. With the help of this parameter, it
1850 becomes possible to add some randomness in the check interval between 0
1851 and +/- 50%. A value between 2 and 5 seems to show good results. The
1852 default value remains at 0.
Willy Tarreaufe255b72007-10-14 23:09:26 +02001853
Davor Ocelice9ed2812017-12-25 17:49:28 +01001854ssl-engine <name> [algo <comma-separated list of algorithms>]
Grant Zhang872f9c22017-01-21 01:10:18 +00001855 Sets the OpenSSL engine to <name>. List of valid values for <name> may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01001856 obtained using the command "openssl engine". This statement may be used
Grant Zhang872f9c22017-01-21 01:10:18 +00001857 multiple times, it will simply enable multiple crypto engines. Referencing an
1858 unsupported engine will prevent haproxy from starting. Note that many engines
1859 will lead to lower HTTPS performance than pure software with recent
1860 processors. The optional command "algo" sets the default algorithms an ENGINE
1861 will supply using the OPENSSL function ENGINE_set_default_string(). A value
Davor Ocelice9ed2812017-12-25 17:49:28 +01001862 of "ALL" uses the engine for all cryptographic operations. If no list of
1863 algo is specified then the value of "ALL" is used. A comma-separated list
Grant Zhang872f9c22017-01-21 01:10:18 +00001864 of different algorithms may be specified, including: RSA, DSA, DH, EC, RAND,
1865 CIPHERS, DIGESTS, PKEY, PKEY_CRYPTO, PKEY_ASN1. This is the same format that
1866 openssl configuration file uses:
1867 https://www.openssl.org/docs/man1.0.2/apps/config.html
1868
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00001869ssl-mode-async
1870 Adds SSL_MODE_ASYNC mode to the SSL context. This enables asynchronous TLS
Emeric Brun3854e012017-05-17 20:42:48 +02001871 I/O operations if asynchronous capable SSL engines are used. The current
Emeric Brunb5e42a82017-06-06 12:35:14 +00001872 implementation supports a maximum of 32 engines. The Openssl ASYNC API
1873 doesn't support moving read/write buffers and is not compliant with
1874 haproxy's buffer management. So the asynchronous mode is disabled on
John Roeslerfb2fce12019-07-10 15:45:51 -05001875 read/write operations (it is only enabled during initial and renegotiation
Emeric Brunb5e42a82017-06-06 12:35:14 +00001876 handshakes).
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00001877
Willy Tarreau33cb0652014-12-23 22:52:37 +01001878tune.buffers.limit <number>
1879 Sets a hard limit on the number of buffers which may be allocated per process.
1880 The default value is zero which means unlimited. The minimum non-zero value
1881 will always be greater than "tune.buffers.reserve" and should ideally always
1882 be about twice as large. Forcing this value can be particularly useful to
1883 limit the amount of memory a process may take, while retaining a sane
Davor Ocelice9ed2812017-12-25 17:49:28 +01001884 behavior. When this limit is reached, sessions which need a buffer wait for
Willy Tarreau33cb0652014-12-23 22:52:37 +01001885 another one to be released by another session. Since buffers are dynamically
1886 allocated and released, the waiting time is very short and not perceptible
1887 provided that limits remain reasonable. In fact sometimes reducing the limit
1888 may even increase performance by increasing the CPU cache's efficiency. Tests
1889 have shown good results on average HTTP traffic with a limit to 1/10 of the
1890 expected global maxconn setting, which also significantly reduces memory
1891 usage. The memory savings come from the fact that a number of connections
1892 will not allocate 2*tune.bufsize. It is best not to touch this value unless
1893 advised to do so by an haproxy core developer.
1894
Willy Tarreau1058ae72014-12-23 22:40:40 +01001895tune.buffers.reserve <number>
1896 Sets the number of buffers which are pre-allocated and reserved for use only
1897 during memory shortage conditions resulting in failed memory allocations. The
1898 minimum value is 2 and is also the default. There is no reason a user would
1899 want to change this value, it's mostly aimed at haproxy core developers.
1900
Willy Tarreau27a674e2009-08-17 07:23:33 +02001901tune.bufsize <number>
1902 Sets the buffer size to this size (in bytes). Lower values allow more
1903 sessions to coexist in the same amount of RAM, and higher values allow some
1904 applications with very large cookies to work. The default value is 16384 and
1905 can be changed at build time. It is strongly recommended not to change this
1906 from the default value, as very low values will break some services such as
1907 statistics, and values larger than default size will increase memory usage,
1908 possibly causing the system to run out of memory. At least the global maxconn
Willy Tarreau45a66cc2017-11-24 11:28:00 +01001909 parameter should be decreased by the same factor as this one is increased. In
1910 addition, use of HTTP/2 mandates that this value must be 16384 or more. If an
1911 HTTP request is larger than (tune.bufsize - tune.maxrewrite), haproxy will
Dmitry Sivachenkof6f4f7b2012-10-21 18:10:25 +04001912 return HTTP 400 (Bad Request) error. Similarly if an HTTP response is larger
Willy Tarreauc77d3642018-12-12 06:19:42 +01001913 than this size, haproxy will return HTTP 502 (Bad Gateway). Note that the
1914 value set using this parameter will automatically be rounded up to the next
1915 multiple of 8 on 32-bit machines and 16 on 64-bit machines.
Willy Tarreau27a674e2009-08-17 07:23:33 +02001916
Willy Tarreau43961d52010-10-04 20:39:20 +02001917tune.chksize <number>
1918 Sets the check buffer size to this size (in bytes). Higher values may help
1919 find string or regex patterns in very large pages, though doing so may imply
1920 more memory and CPU usage. The default value is 16384 and can be changed at
1921 build time. It is not recommended to change this value, but to use better
1922 checks whenever possible.
1923
William Lallemandf3747832012-11-09 12:33:10 +01001924tune.comp.maxlevel <number>
1925 Sets the maximum compression level. The compression level affects CPU
1926 usage during compression. This value affects CPU usage during compression.
1927 Each session using compression initializes the compression algorithm with
1928 this value. The default value is 1.
1929
Willy Tarreauc299e1e2019-02-27 11:35:12 +01001930tune.fail-alloc
1931 If compiled with DEBUG_FAIL_ALLOC, gives the percentage of chances an
1932 allocation attempt fails. Must be between 0 (no failure) and 100 (no
1933 success). This is useful to debug and make sure memory failures are handled
1934 gracefully.
1935
Willy Tarreaubc52bec2020-06-18 08:58:47 +02001936tune.fd.edge-triggered { on | off } [ EXPERIMENTAL ]
1937 Enables ('on') or disables ('off') the edge-triggered polling mode for FDs
1938 that support it. This is currently only support with epoll. It may noticeably
1939 reduce the number of epoll_ctl() calls and slightly improve performance in
1940 certain scenarios. This is still experimental, it may result in frozen
1941 connections if bugs are still present, and is disabled by default.
1942
Willy Tarreaufe20e5b2017-07-27 11:42:14 +02001943tune.h2.header-table-size <number>
1944 Sets the HTTP/2 dynamic header table size. It defaults to 4096 bytes and
1945 cannot be larger than 65536 bytes. A larger value may help certain clients
1946 send more compact requests, depending on their capabilities. This amount of
1947 memory is consumed for each HTTP/2 connection. It is recommended not to
1948 change it.
1949
Willy Tarreaue6baec02017-07-27 11:45:11 +02001950tune.h2.initial-window-size <number>
1951 Sets the HTTP/2 initial window size, which is the number of bytes the client
Davor Ocelice9ed2812017-12-25 17:49:28 +01001952 can upload before waiting for an acknowledgment from haproxy. This setting
1953 only affects payload contents (i.e. the body of POST requests), not headers.
Willy Tarreaue6baec02017-07-27 11:45:11 +02001954 The default value is 65535, which roughly allows up to 5 Mbps of upload
1955 bandwidth per client over a network showing a 100 ms ping time, or 500 Mbps
1956 over a 1-ms local network. It can make sense to increase this value to allow
1957 faster uploads, or to reduce it to increase fairness when dealing with many
1958 clients. It doesn't affect resource usage.
1959
Willy Tarreau5242ef82017-07-27 11:47:28 +02001960tune.h2.max-concurrent-streams <number>
1961 Sets the HTTP/2 maximum number of concurrent streams per connection (ie the
1962 number of outstanding requests on a single connection). The default value is
1963 100. A larger one may slightly improve page load time for complex sites when
1964 visited over high latency networks, but increases the amount of resources a
1965 single client may allocate. A value of zero disables the limit so a single
1966 client may create as many streams as allocatable by haproxy. It is highly
1967 recommended not to change this value.
1968
Willy Tarreaua24b35c2019-02-21 13:24:36 +01001969tune.h2.max-frame-size <number>
1970 Sets the HTTP/2 maximum frame size that haproxy announces it is willing to
1971 receive to its peers. The default value is the largest between 16384 and the
1972 buffer size (tune.bufsize). In any case, haproxy will not announce support
1973 for frame sizes larger than buffers. The main purpose of this setting is to
1974 allow to limit the maximum frame size setting when using large buffers. Too
1975 large frame sizes might have performance impact or cause some peers to
1976 misbehave. It is highly recommended not to change this value.
1977
Willy Tarreau193b8c62012-11-22 00:17:38 +01001978tune.http.cookielen <number>
1979 Sets the maximum length of captured cookies. This is the maximum value that
1980 the "capture cookie xxx len yyy" will be allowed to take, and any upper value
1981 will automatically be truncated to this one. It is important not to set too
1982 high a value because all cookie captures still allocate this size whatever
1983 their configured value (they share a same pool). This value is per request
1984 per response, so the memory allocated is twice this value per connection.
1985 When not specified, the limit is set to 63 characters. It is recommended not
1986 to change this value.
1987
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001988tune.http.logurilen <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001989 Sets the maximum length of request URI in logs. This prevents truncating long
1990 request URIs with valuable query strings in log lines. This is not related
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001991 to syslog limits. If you increase this limit, you may also increase the
Davor Ocelice9ed2812017-12-25 17:49:28 +01001992 'log ... len yyy' parameter. Your syslog daemon may also need specific
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001993 configuration directives too.
1994 The default value is 1024.
1995
Willy Tarreauac1932d2011-10-24 19:14:41 +02001996tune.http.maxhdr <number>
1997 Sets the maximum number of headers in a request. When a request comes with a
1998 number of headers greater than this value (including the first line), it is
1999 rejected with a "400 Bad Request" status code. Similarly, too large responses
2000 are blocked with "502 Bad Gateway". The default value is 101, which is enough
2001 for all usages, considering that the widely deployed Apache server uses the
2002 same limit. It can be useful to push this limit further to temporarily allow
Christopher Faulet50174f32017-06-21 16:31:35 +02002003 a buggy application to work by the time it gets fixed. The accepted range is
2004 1..32767. Keep in mind that each new header consumes 32bits of memory for
2005 each session, so don't push this limit too high.
Willy Tarreauac1932d2011-10-24 19:14:41 +02002006
Willy Tarreau76cc6992020-07-01 18:49:24 +02002007tune.idle-pool.shared { on | off }
2008 Enables ('on') or disables ('off') sharing of idle connection pools between
2009 threads for a same server. The default is to share them between threads in
2010 order to minimize the number of persistent connections to a server, and to
2011 optimize the connection reuse rate. But to help with debugging or when
2012 suspecting a bug in HAProxy around connection reuse, it can be convenient to
2013 forcefully disable this idle pool sharing between multiple threads, and force
2014 this option to "off". The default is on.
2015
Willy Tarreau7e312732014-02-12 16:35:14 +01002016tune.idletimer <timeout>
2017 Sets the duration after which haproxy will consider that an empty buffer is
2018 probably associated with an idle stream. This is used to optimally adjust
2019 some packet sizes while forwarding large and small data alternatively. The
2020 decision to use splice() or to send large buffers in SSL is modulated by this
2021 parameter. The value is in milliseconds between 0 and 65535. A value of zero
2022 means that haproxy will not try to detect idle streams. The default is 1000,
Davor Ocelice9ed2812017-12-25 17:49:28 +01002023 which seems to correctly detect end user pauses (e.g. read a page before
John Roeslerfb2fce12019-07-10 15:45:51 -05002024 clicking). There should be no reason for changing this value. Please check
Willy Tarreau7e312732014-02-12 16:35:14 +01002025 tune.ssl.maxrecord below.
2026
Willy Tarreau7ac908b2019-02-27 12:02:18 +01002027tune.listener.multi-queue { on | off }
2028 Enables ('on') or disables ('off') the listener's multi-queue accept which
2029 spreads the incoming traffic to all threads a "bind" line is allowed to run
2030 on instead of taking them for itself. This provides a smoother traffic
2031 distribution and scales much better, especially in environments where threads
2032 may be unevenly loaded due to external activity (network interrupts colliding
2033 with one thread for example). This option is enabled by default, but it may
2034 be forcefully disabled for troubleshooting or for situations where it is
2035 estimated that the operating system already provides a good enough
2036 distribution and connections are extremely short-lived.
2037
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002038tune.lua.forced-yield <number>
2039 This directive forces the Lua engine to execute a yield each <number> of
Tim Düsterhus4896c442016-11-29 02:15:19 +01002040 instructions executed. This permits interrupting a long script and allows the
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002041 HAProxy scheduler to process other tasks like accepting connections or
2042 forwarding traffic. The default value is 10000 instructions. If HAProxy often
Davor Ocelice9ed2812017-12-25 17:49:28 +01002043 executes some Lua code but more responsiveness is required, this value can be
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002044 lowered. If the Lua code is quite long and its result is absolutely required
2045 to process the data, the <number> can be increased.
2046
Willy Tarreau32f61e22015-03-18 17:54:59 +01002047tune.lua.maxmem
2048 Sets the maximum amount of RAM in megabytes per process usable by Lua. By
2049 default it is zero which means unlimited. It is important to set a limit to
2050 ensure that a bug in a script will not result in the system running out of
2051 memory.
2052
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002053tune.lua.session-timeout <timeout>
2054 This is the execution timeout for the Lua sessions. This is useful for
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02002055 preventing infinite loops or spending too much time in Lua. This timeout
2056 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01002057 not taken in account. The default timeout is 4s.
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002058
2059tune.lua.task-timeout <timeout>
2060 Purpose is the same as "tune.lua.session-timeout", but this timeout is
2061 dedicated to the tasks. By default, this timeout isn't set because a task may
2062 remain alive during of the lifetime of HAProxy. For example, a task used to
2063 check servers.
2064
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02002065tune.lua.service-timeout <timeout>
2066 This is the execution timeout for the Lua services. This is useful for
2067 preventing infinite loops or spending too much time in Lua. This timeout
2068 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01002069 not taken in account. The default timeout is 4s.
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02002070
Willy Tarreaua0250ba2008-01-06 11:22:57 +01002071tune.maxaccept <number>
Willy Tarreau16a21472012-11-19 12:39:59 +01002072 Sets the maximum number of consecutive connections a process may accept in a
2073 row before switching to other work. In single process mode, higher numbers
2074 give better performance at high connection rates. However in multi-process
2075 modes, keeping a bit of fairness between processes generally is better to
2076 increase performance. This value applies individually to each listener, so
2077 that the number of processes a listener is bound to is taken into account.
2078 This value defaults to 64. In multi-process mode, it is divided by twice
2079 the number of processes the listener is bound to. Setting this value to -1
2080 completely disables the limitation. It should normally not be needed to tweak
2081 this value.
Willy Tarreaua0250ba2008-01-06 11:22:57 +01002082
2083tune.maxpollevents <number>
2084 Sets the maximum amount of events that can be processed at once in a call to
2085 the polling system. The default value is adapted to the operating system. It
2086 has been noticed that reducing it below 200 tends to slightly decrease
2087 latency at the expense of network bandwidth, and increasing it above 200
2088 tends to trade latency for slightly increased bandwidth.
2089
Willy Tarreau27a674e2009-08-17 07:23:33 +02002090tune.maxrewrite <number>
2091 Sets the reserved buffer space to this size in bytes. The reserved space is
2092 used for header rewriting or appending. The first reads on sockets will never
2093 fill more than bufsize-maxrewrite. Historically it has defaulted to half of
2094 bufsize, though that does not make much sense since there are rarely large
2095 numbers of headers to add. Setting it too high prevents processing of large
2096 requests or responses. Setting it too low prevents addition of new headers
2097 to already large requests or to POST requests. It is generally wise to set it
2098 to about 1024. It is automatically readjusted to half of bufsize if it is
2099 larger than that. This means you don't have to worry about it when changing
2100 bufsize.
2101
Willy Tarreauf3045d22015-04-29 16:24:50 +02002102tune.pattern.cache-size <number>
2103 Sets the size of the pattern lookup cache to <number> entries. This is an LRU
2104 cache which reminds previous lookups and their results. It is used by ACLs
2105 and maps on slow pattern lookups, namely the ones using the "sub", "reg",
2106 "dir", "dom", "end", "bin" match methods as well as the case-insensitive
2107 strings. It applies to pattern expressions which means that it will be able
2108 to memorize the result of a lookup among all the patterns specified on a
2109 configuration line (including all those loaded from files). It automatically
2110 invalidates entries which are updated using HTTP actions or on the CLI. The
2111 default cache size is set to 10000 entries, which limits its footprint to
Willy Tarreau403bfbb2019-10-23 06:59:31 +02002112 about 5 MB per process/thread on 32-bit systems and 8 MB per process/thread
2113 on 64-bit systems, as caches are thread/process local. There is a very low
Willy Tarreauf3045d22015-04-29 16:24:50 +02002114 risk of collision in this cache, which is in the order of the size of the
2115 cache divided by 2^64. Typically, at 10000 requests per second with the
2116 default cache size of 10000 entries, there's 1% chance that a brute force
2117 attack could cause a single collision after 60 years, or 0.1% after 6 years.
2118 This is considered much lower than the risk of a memory corruption caused by
2119 aging components. If this is not acceptable, the cache can be disabled by
2120 setting this parameter to 0.
2121
Willy Tarreaubd9a0a72011-10-23 21:14:29 +02002122tune.pipesize <number>
2123 Sets the kernel pipe buffer size to this size (in bytes). By default, pipes
2124 are the default size for the system. But sometimes when using TCP splicing,
2125 it can improve performance to increase pipe sizes, especially if it is
2126 suspected that pipes are not filled and that many calls to splice() are
2127 performed. This has an impact on the kernel's memory footprint, so this must
2128 not be changed if impacts are not understood.
2129
Olivier Houchard88698d92019-04-16 19:07:22 +02002130tune.pool-high-fd-ratio <number>
2131 This setting sets the max number of file descriptors (in percentage) used by
2132 haproxy globally against the maximum number of file descriptors haproxy can
2133 use before we start killing idle connections when we can't reuse a connection
2134 and we have to create a new one. The default is 25 (one quarter of the file
2135 descriptor will mean that roughly half of the maximum front connections can
2136 keep an idle connection behind, anything beyond this probably doesn't make
John Roeslerfb2fce12019-07-10 15:45:51 -05002137 much sense in the general case when targeting connection reuse).
Olivier Houchard88698d92019-04-16 19:07:22 +02002138
Willy Tarreau83ca3052020-07-01 18:30:16 +02002139tune.pool-low-fd-ratio <number>
2140 This setting sets the max number of file descriptors (in percentage) used by
2141 haproxy globally against the maximum number of file descriptors haproxy can
2142 use before we stop putting connection into the idle pool for reuse. The
2143 default is 20.
2144
Willy Tarreaue803de22010-01-21 17:43:04 +01002145tune.rcvbuf.client <number>
2146tune.rcvbuf.server <number>
2147 Forces the kernel socket receive buffer size on the client or the server side
2148 to the specified value in bytes. This value applies to all TCP/HTTP frontends
2149 and backends. It should normally never be set, and the default size (0) lets
John Roeslerfb2fce12019-07-10 15:45:51 -05002150 the kernel auto-tune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01002151 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01002152 order to save kernel memory by preventing it from buffering too large amounts
2153 of received data. Lower values will significantly increase CPU usage though.
2154
Willy Tarreaub22fc302015-12-14 12:04:35 +01002155tune.recv_enough <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01002156 HAProxy uses some hints to detect that a short read indicates the end of the
Willy Tarreaub22fc302015-12-14 12:04:35 +01002157 socket buffers. One of them is that a read returns more than <recv_enough>
2158 bytes, which defaults to 10136 (7 segments of 1448 each). This default value
2159 may be changed by this setting to better deal with workloads involving lots
2160 of short messages such as telnet or SSH sessions.
2161
Olivier Houchard1599b802018-05-24 18:59:04 +02002162tune.runqueue-depth <number>
John Roeslerfb2fce12019-07-10 15:45:51 -05002163 Sets the maximum amount of task that can be processed at once when running
Olivier Houchard1599b802018-05-24 18:59:04 +02002164 tasks. The default value is 200. Increasing it may incur latency when
Willy Tarreaue7723bd2020-06-24 11:11:02 +02002165 dealing with I/Os, making it too small can incur extra overhead. When
2166 experimenting with much larger values, it may be useful to also enable
2167 tune.sched.low-latency to limit the maximum latency to the lowest possible.
2168
2169tune.sched.low-latency { on | off }
2170 Enables ('on') or disables ('off') the low-latency task scheduler. By default
2171 haproxy processes tasks from several classes one class at a time as this is
2172 the most efficient. But when running with large values of tune.runqueue-depth
2173 this can have a measurable effect on request or connection latency. When this
2174 low-latency setting is enabled, tasks of lower priority classes will always
2175 be executed before other ones if they exist. This will permit to lower the
2176 maximum latency experienced by new requests or connections in the middle of
2177 massive traffic, at the expense of a higher impact on this large traffic.
2178 For regular usage it is better to leave this off. The default value is off.
Olivier Houchard1599b802018-05-24 18:59:04 +02002179
Willy Tarreaue803de22010-01-21 17:43:04 +01002180tune.sndbuf.client <number>
2181tune.sndbuf.server <number>
2182 Forces the kernel socket send buffer size on the client or the server side to
2183 the specified value in bytes. This value applies to all TCP/HTTP frontends
2184 and backends. It should normally never be set, and the default size (0) lets
John Roeslerfb2fce12019-07-10 15:45:51 -05002185 the kernel auto-tune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01002186 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01002187 order to save kernel memory by preventing it from buffering too large amounts
2188 of received data. Lower values will significantly increase CPU usage though.
2189 Another use case is to prevent write timeouts with extremely slow clients due
2190 to the kernel waiting for a large part of the buffer to be read before
2191 notifying haproxy again.
2192
Willy Tarreau6ec58db2012-11-16 16:32:15 +01002193tune.ssl.cachesize <number>
Emeric Brunaf9619d2012-11-28 18:47:52 +01002194 Sets the size of the global SSL session cache, in a number of blocks. A block
2195 is large enough to contain an encoded session without peer certificate.
2196 An encoded session with peer certificate is stored in multiple blocks
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002197 depending on the size of the peer certificate. A block uses approximately
Emeric Brunaf9619d2012-11-28 18:47:52 +01002198 200 bytes of memory. The default value may be forced at build time, otherwise
Davor Ocelice9ed2812017-12-25 17:49:28 +01002199 defaults to 20000. When the cache is full, the most idle entries are purged
Emeric Brunaf9619d2012-11-28 18:47:52 +01002200 and reassigned. Higher values reduce the occurrence of such a purge, hence
2201 the number of CPU-intensive SSL handshakes by ensuring that all users keep
2202 their session as long as possible. All entries are pre-allocated upon startup
Emeric Brun22890a12012-12-28 14:41:32 +01002203 and are shared between all processes if "nbproc" is greater than 1. Setting
2204 this value to 0 disables the SSL session cache.
Willy Tarreau6ec58db2012-11-16 16:32:15 +01002205
Emeric Brun8dc60392014-05-09 13:52:00 +02002206tune.ssl.force-private-cache
Lukas Tribus27935782018-10-01 02:00:16 +02002207 This option disables SSL session cache sharing between all processes. It
Emeric Brun8dc60392014-05-09 13:52:00 +02002208 should normally not be used since it will force many renegotiations due to
2209 clients hitting a random process. But it may be required on some operating
2210 systems where none of the SSL cache synchronization method may be used. In
2211 this case, adding a first layer of hash-based load balancing before the SSL
2212 layer might limit the impact of the lack of session sharing.
2213
William Lallemand7d42ef52020-07-06 11:41:30 +02002214tune.ssl.keylog { on | off }
2215 This option activates the logging of the TLS keys. It should be used with
2216 care as it will consume more memory per SSL session and could decrease
2217 performances. This is disabled by default.
2218
2219 These sample fetches should be used to generate the SSLKEYLOGFILE that is
2220 required to decipher traffic with wireshark.
2221
2222 https://developer.mozilla.org/en-US/docs/Mozilla/Projects/NSS/Key_Log_Format
2223
2224 The SSLKEYLOG is a series of lines which are formatted this way:
2225
2226 <Label> <space> <ClientRandom> <space> <Secret>
2227
2228 The ClientRandom is provided by the %[ssl_fc_client_random,hex] sample
2229 fetch, the secret and the Label could be find in the array below. You need
2230 to generate a SSLKEYLOGFILE with all the labels in this array.
2231
2232 The following sample fetches are hexadecimal strings and does not need to be
2233 converted.
2234
2235 SSLKEYLOGFILE Label | Sample fetches for the Secrets
2236 --------------------------------|-----------------------------------------
2237 CLIENT_EARLY_TRAFFIC_SECRET | %[ssl_fc_client_early_traffic_secret]
2238 CLIENT_HANDSHAKE_TRAFFIC_SECRET | %[ssl_fc_client_handshake_traffic_secret]
2239 SERVER_HANDSHAKE_TRAFFIC_SECRET | %[ssl_fc_server_handshake_traffic_secret]
2240 CLIENT_TRAFFIC_SECRET_0 | %[ssl_fc_client_traffic_secret_0]
2241 SERVER_TRAFFIC_SECRET_0 | %[ssl_fc_server_traffic_secret_0]
William Lallemandd742b6c2020-07-07 10:14:56 +02002242 EXPORTER_SECRET | %[ssl_fc_exporter_secret]
2243 EARLY_EXPORTER_SECRET | %[ssl_fc_early_exporter_secret]
William Lallemand7d42ef52020-07-06 11:41:30 +02002244
2245 This is only available with OpenSSL 1.1.1, and useful with TLS1.3 session.
2246
2247 If you want to generate the content of a SSLKEYLOGFILE with TLS < 1.3, you
2248 only need this line:
2249
2250 "CLIENT_RANDOM %[ssl_fc_client_random,hex] %[ssl_fc_session_key,hex]"
2251
Emeric Brun4f65bff2012-11-16 15:11:00 +01002252tune.ssl.lifetime <timeout>
2253 Sets how long a cached SSL session may remain valid. This time is expressed
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002254 in seconds and defaults to 300 (5 min). It is important to understand that it
Emeric Brun4f65bff2012-11-16 15:11:00 +01002255 does not guarantee that sessions will last that long, because if the cache is
2256 full, the longest idle sessions will be purged despite their configured
2257 lifetime. The real usefulness of this setting is to prevent sessions from
2258 being used for too long.
2259
Willy Tarreaubfd59462013-02-21 07:46:09 +01002260tune.ssl.maxrecord <number>
2261 Sets the maximum amount of bytes passed to SSL_write() at a time. Default
2262 value 0 means there is no limit. Over SSL/TLS, the client can decipher the
2263 data only once it has received a full record. With large records, it means
2264 that clients might have to download up to 16kB of data before starting to
2265 process them. Limiting the value can improve page load times on browsers
2266 located over high latency or low bandwidth networks. It is suggested to find
2267 optimal values which fit into 1 or 2 TCP segments (generally 1448 bytes over
2268 Ethernet with TCP timestamps enabled, or 1460 when timestamps are disabled),
2269 keeping in mind that SSL/TLS add some overhead. Typical values of 1419 and
2270 2859 gave good results during tests. Use "strace -e trace=write" to find the
Davor Ocelice9ed2812017-12-25 17:49:28 +01002271 best value. HAProxy will automatically switch to this setting after an idle
Willy Tarreau7e312732014-02-12 16:35:14 +01002272 stream has been detected (see tune.idletimer above).
Willy Tarreaubfd59462013-02-21 07:46:09 +01002273
Remi Gacognef46cd6e2014-06-12 14:58:40 +02002274tune.ssl.default-dh-param <number>
2275 Sets the maximum size of the Diffie-Hellman parameters used for generating
2276 the ephemeral/temporary Diffie-Hellman key in case of DHE key exchange. The
2277 final size will try to match the size of the server's RSA (or DSA) key (e.g,
2278 a 2048 bits temporary DH key for a 2048 bits RSA key), but will not exceed
Willy Tarreau3ba77d22020-05-08 09:31:18 +02002279 this maximum value. Default value if 2048. Only 1024 or higher values are
Remi Gacognef46cd6e2014-06-12 14:58:40 +02002280 allowed. Higher values will increase the CPU load, and values greater than
2281 1024 bits are not supported by Java 7 and earlier clients. This value is not
Remi Gacogne47783ef2015-05-29 15:53:22 +02002282 used if static Diffie-Hellman parameters are supplied either directly
2283 in the certificate file or by using the ssl-dh-param-file parameter.
Remi Gacognef46cd6e2014-06-12 14:58:40 +02002284
Christopher Faulet31af49d2015-06-09 17:29:50 +02002285tune.ssl.ssl-ctx-cache-size <number>
2286 Sets the size of the cache used to store generated certificates to <number>
2287 entries. This is a LRU cache. Because generating a SSL certificate
2288 dynamically is expensive, they are cached. The default cache size is set to
2289 1000 entries.
2290
Thierry FOURNIER5bf77322017-02-25 12:45:22 +01002291tune.ssl.capture-cipherlist-size <number>
2292 Sets the maximum size of the buffer used for capturing client-hello cipher
2293 list. If the value is 0 (default value) the capture is disabled, otherwise
2294 a buffer is allocated for each SSL/TLS connection.
2295
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002296tune.vars.global-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01002297tune.vars.proc-max-size <size>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002298tune.vars.reqres-max-size <size>
2299tune.vars.sess-max-size <size>
2300tune.vars.txn-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01002301 These five tunes help to manage the maximum amount of memory used by the
2302 variables system. "global" limits the overall amount of memory available for
2303 all scopes. "proc" limits the memory for the process scope, "sess" limits the
2304 memory for the session scope, "txn" for the transaction scope, and "reqres"
2305 limits the memory for each request or response processing.
2306 Memory accounting is hierarchical, meaning more coarse grained limits include
2307 the finer grained ones: "proc" includes "sess", "sess" includes "txn", and
2308 "txn" includes "reqres".
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002309
Daniel Schneller0b547052016-03-21 20:46:57 +01002310 For example, when "tune.vars.sess-max-size" is limited to 100,
2311 "tune.vars.txn-max-size" and "tune.vars.reqres-max-size" cannot exceed
2312 100 either. If we create a variable "txn.var" that contains 100 bytes,
2313 all available space is consumed.
2314 Notice that exceeding the limits at runtime will not result in an error
2315 message, but values might be cut off or corrupted. So make sure to accurately
2316 plan for the amount of space needed to store all your variables.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002317
William Lallemanda509e4c2012-11-07 16:54:34 +01002318tune.zlib.memlevel <number>
2319 Sets the memLevel parameter in zlib initialization for each session. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002320 defines how much memory should be allocated for the internal compression
William Lallemanda509e4c2012-11-07 16:54:34 +01002321 state. A value of 1 uses minimum memory but is slow and reduces compression
Davor Ocelice9ed2812017-12-25 17:49:28 +01002322 ratio, a value of 9 uses maximum memory for optimal speed. Can be a value
William Lallemanda509e4c2012-11-07 16:54:34 +01002323 between 1 and 9. The default value is 8.
2324
2325tune.zlib.windowsize <number>
2326 Sets the window size (the size of the history buffer) as a parameter of the
2327 zlib initialization for each session. Larger values of this parameter result
Davor Ocelice9ed2812017-12-25 17:49:28 +01002328 in better compression at the expense of memory usage. Can be a value between
2329 8 and 15. The default value is 15.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002330
Willy Tarreauc57f0e22009-05-10 13:12:33 +020023313.3. Debugging
2332--------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02002333
Willy Tarreau6a06a402007-07-15 20:15:28 +02002334quiet
2335 Do not display any message during startup. It is equivalent to the command-
2336 line argument "-q".
2337
Willy Tarreau3eb10b82020-04-15 16:42:39 +02002338zero-warning
2339 When this option is set, haproxy will refuse to start if any warning was
2340 emitted while processing the configuration. It is highly recommended to set
2341 this option on configurations that are not changed often, as it helps detect
2342 subtle mistakes and keep the configuration clean and forward-compatible. Note
2343 that "haproxy -c" will also report errors in such a case. This option is
2344 equivalent to command line argument "-dW".
2345
Emeric Brunf099e792010-09-27 12:05:28 +02002346
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010023473.4. Userlists
2348--------------
2349It is possible to control access to frontend/backend/listen sections or to
2350http stats by allowing only authenticated and authorized users. To do this,
2351it is required to create at least one userlist and to define users.
2352
2353userlist <listname>
Cyril Bonté78caf842010-03-10 22:41:43 +01002354 Creates new userlist with name <listname>. Many independent userlists can be
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002355 used to store authentication & authorization data for independent customers.
2356
2357group <groupname> [users <user>,<user>,(...)]
Cyril Bonté78caf842010-03-10 22:41:43 +01002358 Adds group <groupname> to the current userlist. It is also possible to
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002359 attach users to this group by using a comma separated list of names
2360 proceeded by "users" keyword.
2361
Cyril Bontéf0c60612010-02-06 14:44:47 +01002362user <username> [password|insecure-password <password>]
2363 [groups <group>,<group>,(...)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002364 Adds user <username> to the current userlist. Both secure (encrypted) and
2365 insecure (unencrypted) passwords can be used. Encrypted passwords are
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01002366 evaluated using the crypt(3) function, so depending on the system's
2367 capabilities, different algorithms are supported. For example, modern Glibc
2368 based Linux systems support MD5, SHA-256, SHA-512, and, of course, the
2369 classic DES-based method of encrypting passwords.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002370
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01002371 Attention: Be aware that using encrypted passwords might cause significantly
2372 increased CPU usage, depending on the number of requests, and the algorithm
2373 used. For any of the hashed variants, the password for each request must
2374 be processed through the chosen algorithm, before it can be compared to the
2375 value specified in the config file. Most current algorithms are deliberately
2376 designed to be expensive to compute to achieve resistance against brute
2377 force attacks. They do not simply salt/hash the clear text password once,
2378 but thousands of times. This can quickly become a major factor in haproxy's
2379 overall CPU consumption!
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002380
2381 Example:
Cyril Bontéf0c60612010-02-06 14:44:47 +01002382 userlist L1
2383 group G1 users tiger,scott
2384 group G2 users xdb,scott
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002385
Cyril Bontéf0c60612010-02-06 14:44:47 +01002386 user tiger password $6$k6y3o.eP$JlKBx9za9667qe4(...)xHSwRv6J.C0/D7cV91
2387 user scott insecure-password elgato
2388 user xdb insecure-password hello
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002389
Cyril Bontéf0c60612010-02-06 14:44:47 +01002390 userlist L2
2391 group G1
2392 group G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002393
Cyril Bontéf0c60612010-02-06 14:44:47 +01002394 user tiger password $6$k6y3o.eP$JlKBx(...)xHSwRv6J.C0/D7cV91 groups G1
2395 user scott insecure-password elgato groups G1,G2
2396 user xdb insecure-password hello groups G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002397
2398 Please note that both lists are functionally identical.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002399
Emeric Brunf099e792010-09-27 12:05:28 +02002400
24013.5. Peers
Cyril Bontédc4d9032012-04-08 21:57:39 +02002402----------
Emeric Brun94900952015-06-11 18:25:54 +02002403It is possible to propagate entries of any data-types in stick-tables between
2404several haproxy instances over TCP connections in a multi-master fashion. Each
2405instance pushes its local updates and insertions to remote peers. The pushed
2406values overwrite remote ones without aggregation. Interrupted exchanges are
2407automatically detected and recovered from the last known point.
2408In addition, during a soft restart, the old process connects to the new one
2409using such a TCP connection to push all its entries before the new process
2410tries to connect to other peers. That ensures very fast replication during a
2411reload, it typically takes a fraction of a second even for large tables.
2412Note that Server IDs are used to identify servers remotely, so it is important
2413that configurations look similar or at least that the same IDs are forced on
2414each server on all participants.
Emeric Brunf099e792010-09-27 12:05:28 +02002415
2416peers <peersect>
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002417 Creates a new peer list with name <peersect>. It is an independent section,
Emeric Brunf099e792010-09-27 12:05:28 +02002418 which is referenced by one or more stick-tables.
2419
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002420bind [<address>]:<port_range> [, ...] [param*]
2421 Defines the binding parameters of the local peer of this "peers" section.
2422 Such lines are not supported with "peer" line in the same "peers" section.
2423
Willy Tarreau77e4bd12015-05-01 20:02:17 +02002424disabled
2425 Disables a peers section. It disables both listening and any synchronization
2426 related to this section. This is provided to disable synchronization of stick
2427 tables without having to comment out all "peers" references.
2428
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002429default-bind [param*]
2430 Defines the binding parameters for the local peer, excepted its address.
2431
2432default-server [param*]
2433 Change default options for a server in a "peers" section.
2434
2435 Arguments:
2436 <param*> is a list of parameters for this server. The "default-server"
2437 keyword accepts an important number of options and has a complete
2438 section dedicated to it. Please refer to section 5 for more
2439 details.
2440
2441
2442 See also: "server" and section 5 about server options
2443
Willy Tarreau77e4bd12015-05-01 20:02:17 +02002444enable
2445 This re-enables a disabled peers section which was previously disabled.
2446
Frédéric Lécailleb6f759b2019-11-05 09:57:45 +01002447log <address> [len <length>] [format <format>] [sample <ranges>:<smp_size>]
2448 <facility> [<level> [<minlevel>]]
2449 "peers" sections support the same "log" keyword as for the proxies to
2450 log information about the "peers" listener. See "log" option for proxies for
2451 more details.
2452
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002453peer <peername> <ip>:<port> [param*]
Emeric Brunf099e792010-09-27 12:05:28 +02002454 Defines a peer inside a peers section.
2455 If <peername> is set to the local peer name (by default hostname, or forced
Dragan Dosen13cd54c2020-06-18 18:24:05 +02002456 using "-L" command line option or "localpeer" global configuration setting),
2457 haproxy will listen for incoming remote peer connection on <ip>:<port>.
2458 Otherwise, <ip>:<port> defines where to connect to in order to join the
2459 remote peer, and <peername> is used at the protocol level to identify and
2460 validate the remote peer on the server side.
Emeric Brunf099e792010-09-27 12:05:28 +02002461
2462 During a soft restart, local peer <ip>:<port> is used by the old instance to
2463 connect the new one and initiate a complete replication (teaching process).
2464
2465 It is strongly recommended to have the exact same peers declaration on all
Dragan Dosen13cd54c2020-06-18 18:24:05 +02002466 peers and to only rely on the "-L" command line argument or the "localpeer"
2467 global configuration setting to change the local peer name. This makes it
2468 easier to maintain coherent configuration files across all peers.
Emeric Brunf099e792010-09-27 12:05:28 +02002469
William Lallemandb2f07452015-05-12 14:27:13 +02002470 You may want to reference some environment variables in the address
2471 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01002472
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002473 Note: "peer" keyword may transparently be replaced by "server" keyword (see
2474 "server" keyword explanation below).
2475
2476server <peername> [<ip>:<port>] [param*]
Michael Prokop4438c602019-05-24 10:25:45 +02002477 As previously mentioned, "peer" keyword may be replaced by "server" keyword
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002478 with a support for all "server" parameters found in 5.2 paragraph.
2479 If the underlying peer is local, <ip>:<port> parameters must not be present.
2480 These parameters must be provided on a "bind" line (see "bind" keyword
2481 of this "peers" section).
2482 Some of these parameters are irrelevant for "peers" sections.
2483
2484
Cyril Bontédc4d9032012-04-08 21:57:39 +02002485 Example:
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002486 # The old way.
Emeric Brunf099e792010-09-27 12:05:28 +02002487 peers mypeers
Willy Tarreauf7b30a92010-12-06 22:59:17 +01002488 peer haproxy1 192.168.0.1:1024
2489 peer haproxy2 192.168.0.2:1024
2490 peer haproxy3 10.2.0.1:1024
Emeric Brunf099e792010-09-27 12:05:28 +02002491
2492 backend mybackend
2493 mode tcp
2494 balance roundrobin
2495 stick-table type ip size 20k peers mypeers
2496 stick on src
2497
Willy Tarreauf7b30a92010-12-06 22:59:17 +01002498 server srv1 192.168.0.30:80
2499 server srv2 192.168.0.31:80
Emeric Brunf099e792010-09-27 12:05:28 +02002500
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002501 Example:
2502 peers mypeers
2503 bind 127.0.0.11:10001 ssl crt mycerts/pem
2504 default-server ssl verify none
2505 server hostA 127.0.0.10:10000
2506 server hostB #local peer
Emeric Brunf099e792010-09-27 12:05:28 +02002507
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01002508
2509table <tablename> type {ip | integer | string [len <length>] | binary [len <length>]}
2510 size <size> [expire <expire>] [nopurge] [store <data_type>]*
2511
2512 Configure a stickiness table for the current section. This line is parsed
2513 exactly the same way as the "stick-table" keyword in others section, except
John Roeslerfb2fce12019-07-10 15:45:51 -05002514 for the "peers" argument which is not required here and with an additional
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01002515 mandatory first parameter to designate the stick-table. Contrary to others
2516 sections, there may be several "table" lines in "peers" sections (see also
2517 "stick-table" keyword).
2518
2519 Also be aware of the fact that "peers" sections have their own stick-table
2520 namespaces to avoid collisions between stick-table names identical in
2521 different "peers" section. This is internally handled prepending the "peers"
2522 sections names to the name of the stick-tables followed by a '/' character.
2523 If somewhere else in the configuration file you have to refer to such
2524 stick-tables declared in "peers" sections you must use the prefixed version
2525 of the stick-table name as follows:
2526
2527 peers mypeers
2528 peer A ...
2529 peer B ...
2530 table t1 ...
2531
2532 frontend fe1
2533 tcp-request content track-sc0 src table mypeers/t1
2534
2535 This is also this prefixed version of the stick-table names which must be
2536 used to refer to stick-tables through the CLI.
2537
2538 About "peers" protocol, as only "peers" belonging to the same section may
2539 communicate with each others, there is no need to do such a distinction.
2540 Several "peers" sections may declare stick-tables with the same name.
2541 This is shorter version of the stick-table name which is sent over the network.
2542 There is only a '/' character as prefix to avoid stick-table name collisions between
2543 stick-tables declared as backends and stick-table declared in "peers" sections
2544 as follows in this weird but supported configuration:
2545
2546 peers mypeers
2547 peer A ...
2548 peer B ...
2549 table t1 type string size 10m store gpc0
2550
2551 backend t1
2552 stick-table type string size 10m store gpc0 peers mypeers
2553
Daniel Corbett67a82712020-07-06 23:01:19 -04002554 Here "t1" table declared in "mypeers" section has "mypeers/t1" as global name.
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01002555 "t1" table declared as a backend as "t1" as global name. But at peer protocol
2556 level the former table is named "/t1", the latter is again named "t1".
2557
Simon Horman51a1cf62015-02-03 13:00:44 +090025583.6. Mailers
2559------------
2560It is possible to send email alerts when the state of servers changes.
2561If configured email alerts are sent to each mailer that is configured
2562in a mailers section. Email is sent to mailers using SMTP.
2563
Pieter Baauw386a1272015-08-16 15:26:24 +02002564mailers <mailersect>
Simon Horman51a1cf62015-02-03 13:00:44 +09002565 Creates a new mailer list with the name <mailersect>. It is an
2566 independent section which is referenced by one or more proxies.
2567
2568mailer <mailername> <ip>:<port>
2569 Defines a mailer inside a mailers section.
2570
2571 Example:
2572 mailers mymailers
2573 mailer smtp1 192.168.0.1:587
2574 mailer smtp2 192.168.0.2:587
2575
2576 backend mybackend
2577 mode tcp
2578 balance roundrobin
2579
2580 email-alert mailers mymailers
2581 email-alert from test1@horms.org
2582 email-alert to test2@horms.org
2583
2584 server srv1 192.168.0.30:80
2585 server srv2 192.168.0.31:80
2586
Pieter Baauw235fcfc2016-02-13 15:33:40 +01002587timeout mail <time>
2588 Defines the time available for a mail/connection to be made and send to
2589 the mail-server. If not defined the default value is 10 seconds. To allow
2590 for at least two SYN-ACK packets to be send during initial TCP handshake it
2591 is advised to keep this value above 4 seconds.
2592
2593 Example:
2594 mailers mymailers
2595 timeout mail 20s
2596 mailer smtp1 192.168.0.1:587
Simon Horman51a1cf62015-02-03 13:00:44 +09002597
William Lallemandc9515522019-06-12 16:32:11 +020025983.7. Programs
2599-------------
2600In master-worker mode, it is possible to launch external binaries with the
2601master, these processes are called programs. These programs are launched and
2602managed the same way as the workers.
2603
2604During a reload of HAProxy, those processes are dealing with the same
2605sequence as a worker:
2606
2607 - the master is re-executed
2608 - the master sends a SIGUSR1 signal to the program
2609 - if "option start-on-reload" is not disabled, the master launches a new
2610 instance of the program
2611
2612During a stop, or restart, a SIGTERM is sent to the programs.
2613
2614program <name>
2615 This is a new program section, this section will create an instance <name>
2616 which is visible in "show proc" on the master CLI. (See "9.4. Master CLI" in
2617 the management guide).
2618
2619command <command> [arguments*]
2620 Define the command to start with optional arguments. The command is looked
2621 up in the current PATH if it does not include an absolute path. This is a
2622 mandatory option of the program section. Arguments containing spaces must
2623 be enclosed in quotes or double quotes or be prefixed by a backslash.
2624
Andrew Heberle97236962019-07-12 11:50:26 +08002625user <user name>
2626 Changes the executed command user ID to the <user name> from /etc/passwd.
2627 See also "group".
2628
2629group <group name>
2630 Changes the executed command group ID to the <group name> from /etc/group.
2631 See also "user".
2632
William Lallemandc9515522019-06-12 16:32:11 +02002633option start-on-reload
2634no option start-on-reload
2635 Start (or not) a new instance of the program upon a reload of the master.
2636 The default is to start a new instance. This option may only be used in a
2637 program section.
2638
2639
Christopher Faulet76edc0f2020-01-13 15:52:01 +010026403.8. HTTP-errors
2641----------------
2642
2643It is possible to globally declare several groups of HTTP errors, to be
2644imported afterwards in any proxy section. Same group may be referenced at
2645several places and can be fully or partially imported.
2646
2647http-errors <name>
2648 Create a new http-errors group with the name <name>. It is an independent
2649 section that may be referenced by one or more proxies using its name.
2650
2651errorfile <code> <file>
2652 Associate a file contents to an HTTP error code
2653
2654 Arguments :
2655 <code> is the HTTP status code. Currently, HAProxy is capable of
Christopher Faulet612f2ea2020-05-27 09:57:28 +02002656 generating codes 200, 400, 401, 403, 404, 405, 407, 408, 410,
2657 425, 429, 500, 502, 503, and 504.
Christopher Faulet76edc0f2020-01-13 15:52:01 +01002658
2659 <file> designates a file containing the full HTTP response. It is
2660 recommended to follow the common practice of appending ".http" to
2661 the filename so that people do not confuse the response with HTML
2662 error pages, and to use absolute paths, since files are read
2663 before any chroot is performed.
2664
2665 Please referrers to "errorfile" keyword in section 4 for details.
2666
2667 Example:
2668 http-errors website-1
2669 errorfile 400 /etc/haproxy/errorfiles/site1/400.http
2670 errorfile 404 /etc/haproxy/errorfiles/site1/404.http
2671 errorfile 408 /dev/null # work around Chrome pre-connect bug
2672
2673 http-errors website-2
2674 errorfile 400 /etc/haproxy/errorfiles/site2/400.http
2675 errorfile 404 /etc/haproxy/errorfiles/site2/404.http
2676 errorfile 408 /dev/null # work around Chrome pre-connect bug
2677
Emeric Brun99c453d2020-05-25 15:01:04 +020026783.9. Rings
2679----------
2680
2681It is possible to globally declare ring-buffers, to be used as target for log
2682servers or traces.
2683
2684ring <ringname>
2685 Creates a new ring-buffer with name <ringname>.
2686
2687description <text>
Daniel Corbett67a82712020-07-06 23:01:19 -04002688 The description is an optional description string of the ring. It will
Emeric Brun99c453d2020-05-25 15:01:04 +02002689 appear on CLI. By default, <name> is reused to fill this field.
2690
2691format <format>
2692 Format used to store events into the ring buffer.
2693
2694 Arguments:
2695 <format> is the log format used when generating syslog messages. It may be
2696 one of the following :
2697
2698 iso A message containing only the ISO date, followed by the text.
2699 The PID, process name and system name are omitted. This is
2700 designed to be used with a local log server.
2701
2702 raw A message containing only the text. The level, PID, date, time,
2703 process name and system name are omitted. This is designed to be
2704 used in containers or during development, where the severity
2705 only depends on the file descriptor used (stdout/stderr). This
2706 is the default.
2707
2708 rfc3164 The RFC3164 syslog message format. This is the default.
2709 (https://tools.ietf.org/html/rfc3164)
2710
2711 rfc5424 The RFC5424 syslog message format.
2712 (https://tools.ietf.org/html/rfc5424)
2713
2714 short A message containing only a level between angle brackets such as
2715 '<3>', followed by the text. The PID, date, time, process name
2716 and system name are omitted. This is designed to be used with a
2717 local log server. This format is compatible with what the systemd
2718 logger consumes.
2719
Emeric Brun54648852020-07-06 15:54:06 +02002720 priority A message containing only a level plus syslog facility between angle
2721 brackets such as '<63>', followed by the text. The PID, date, time,
2722 process name and system name are omitted. This is designed to be used
2723 with a local log server.
2724
Emeric Brun99c453d2020-05-25 15:01:04 +02002725 timed A message containing only a level between angle brackets such as
2726 '<3>', followed by ISO date and by the text. The PID, process
2727 name and system name are omitted. This is designed to be
2728 used with a local log server.
2729
2730maxlen <length>
2731 The maximum length of an event message stored into the ring,
2732 including formatted header. If an event message is longer than
2733 <length>, it will be truncated to this length.
2734
Emeric Brun494c5052020-05-28 11:13:15 +02002735server <name> <address> [param*]
2736 Used to configure a syslog tcp server to forward messages from ring buffer.
2737 This supports for all "server" parameters found in 5.2 paragraph. Some of
2738 these parameters are irrelevant for "ring" sections. Important point: there
2739 is little reason to add more than one server to a ring, because all servers
2740 will receive the exact same copy of the ring contents, and as such the ring
2741 will progress at the speed of the slowest server. If one server does not
2742 respond, it will prevent old messages from being purged and may block new
2743 messages from being inserted into the ring. The proper way to send messages
2744 to multiple servers is to use one distinct ring per log server, not to
Emeric Brun97556472020-05-30 01:42:45 +02002745 attach multiple servers to the same ring. Note that specific server directive
2746 "log-proto" is used to set the protocol used to send messages.
Emeric Brun494c5052020-05-28 11:13:15 +02002747
Emeric Brun99c453d2020-05-25 15:01:04 +02002748size <size>
2749 This is the optional size in bytes for the ring-buffer. Default value is
2750 set to BUFSIZE.
2751
Emeric Brun494c5052020-05-28 11:13:15 +02002752timeout connect <timeout>
2753 Set the maximum time to wait for a connection attempt to a server to succeed.
2754
2755 Arguments :
2756 <timeout> is the timeout value specified in milliseconds by default, but
2757 can be in any other unit if the number is suffixed by the unit,
2758 as explained at the top of this document.
2759
2760timeout server <timeout>
2761 Set the maximum time for pending data staying into output buffer.
2762
2763 Arguments :
2764 <timeout> is the timeout value specified in milliseconds by default, but
2765 can be in any other unit if the number is suffixed by the unit,
2766 as explained at the top of this document.
2767
Emeric Brun99c453d2020-05-25 15:01:04 +02002768 Example:
2769 global
2770 log ring@myring local7
2771
2772 ring myring
2773 description "My local buffer"
2774 format rfc3164
2775 maxlen 1200
2776 size 32764
Emeric Brun494c5052020-05-28 11:13:15 +02002777 timeout connect 5s
2778 timeout server 10s
Emeric Brun97556472020-05-30 01:42:45 +02002779 server mysyslogsrv 127.0.0.1:6514 log-proto octet-count
Emeric Brun99c453d2020-05-25 15:01:04 +02002780
Emeric Brun12941c82020-07-07 14:19:42 +020027813.10. Log forwarding
2782-------------------
2783
2784It is possible to declare one or multiple log forwarding section,
2785haproxy will forward all received log messages to a log servers list.
2786
2787log-forward <name>
2788 Creates a new log forwarder proxy identified as <name>.
2789
Emeric Bruncbb7bf72020-10-05 14:39:35 +02002790backlog <conns>
2791 Give hints to the system about the approximate listen backlog desired size
2792 on connections accept.
2793
2794bind <addr> [param*]
2795 Used to configure a stream log listener to receive messages to forward.
Emeric Brunda46c1c2020-10-08 08:39:02 +02002796 This supports the "bind" parameters found in 5.1 paragraph including
2797 those about ssl but some statements such as "alpn" may be irrelevant for
2798 syslog protocol over TCP.
2799 Those listeners support both "Octet Counting" and "Non-Transparent-Framing"
Emeric Bruncbb7bf72020-10-05 14:39:35 +02002800 modes as defined in rfc-6587.
2801
Willy Tarreau76aaa7f2020-09-16 15:07:22 +02002802dgram-bind <addr> [param*]
Emeric Bruncbb7bf72020-10-05 14:39:35 +02002803 Used to configure a datagram log listener to receive messages to forward.
2804 Addresses must be in IPv4 or IPv6 form,followed by a port. This supports
2805 for some of the "bind" parameters found in 5.1 paragraph among which
2806 "interface", "namespace" or "transparent", the other ones being
Willy Tarreau26ff5da2020-09-16 15:22:19 +02002807 silently ignored as irrelevant for UDP/syslog case.
Emeric Brun12941c82020-07-07 14:19:42 +02002808
2809log global
2810log <address> [len <length>] [format <format>] [sample <ranges>:<smp_size>]
2811 <facility> [<level> [<minlevel>]]
2812 Used to configure target log servers. See more details on proxies
2813 documentation.
2814 If no format specified, haproxy tries to keep the incoming log format.
2815 Configured facility is ignored, except if incoming message does not
2816 present a facility but one is mandatory on the outgoing format.
2817 If there is no timestamp available in the input format, but the field
2818 exists in output format, haproxy will use the local date.
2819
2820 Example:
2821 global
2822 log stderr format iso local7
2823
2824 ring myring
2825 description "My local buffer"
2826 format rfc5424
2827 maxlen 1200
2828 size 32764
2829 timeout connect 5s
2830 timeout server 10s
2831 # syslog tcp server
2832 server mysyslogsrv 127.0.0.1:514 log-proto octet-count
2833
2834 log-forward sylog-loadb
Emeric Bruncbb7bf72020-10-05 14:39:35 +02002835 dgram-bind 127.0.0.1:1514
2836 bind 127.0.0.1:1514
Emeric Brun12941c82020-07-07 14:19:42 +02002837 # all messages on stderr
2838 log global
2839 # all messages on local tcp syslog server
2840 log ring@myring local0
2841 # load balance messages on 4 udp syslog servers
2842 log 127.0.0.1:10001 sample 1:4 local0
2843 log 127.0.0.1:10002 sample 2:4 local0
2844 log 127.0.0.1:10003 sample 3:4 local0
2845 log 127.0.0.1:10004 sample 4:4 local0
Christopher Faulet76edc0f2020-01-13 15:52:01 +01002846
Emeric Bruncbb7bf72020-10-05 14:39:35 +02002847maxconn <conns>
2848 Fix the maximum number of concurrent connections on a log forwarder.
2849 10 is the default.
2850
2851timeout client <timeout>
2852 Set the maximum inactivity time on the client side.
2853
Willy Tarreauc57f0e22009-05-10 13:12:33 +020028544. Proxies
Willy Tarreau6a06a402007-07-15 20:15:28 +02002855----------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002856
Willy Tarreau6a06a402007-07-15 20:15:28 +02002857Proxy configuration can be located in a set of sections :
William Lallemand6e62fb62015-04-28 16:55:23 +02002858 - defaults [<name>]
Willy Tarreau6a06a402007-07-15 20:15:28 +02002859 - frontend <name>
2860 - backend <name>
2861 - listen <name>
2862
2863A "defaults" section sets default parameters for all other sections following
2864its declaration. Those default parameters are reset by the next "defaults"
2865section. See below for the list of parameters which can be set in a "defaults"
Willy Tarreau0ba27502007-12-24 16:55:16 +01002866section. The name is optional but its use is encouraged for better readability.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002867
2868A "frontend" section describes a set of listening sockets accepting client
2869connections.
2870
2871A "backend" section describes a set of servers to which the proxy will connect
2872to forward incoming connections.
2873
2874A "listen" section defines a complete proxy with its frontend and backend
2875parts combined in one section. It is generally useful for TCP-only traffic.
2876
Willy Tarreau0ba27502007-12-24 16:55:16 +01002877All proxy names must be formed from upper and lower case letters, digits,
2878'-' (dash), '_' (underscore) , '.' (dot) and ':' (colon). ACL names are
2879case-sensitive, which means that "www" and "WWW" are two different proxies.
2880
2881Historically, all proxy names could overlap, it just caused troubles in the
2882logs. Since the introduction of content switching, it is mandatory that two
2883proxies with overlapping capabilities (frontend/backend) have different names.
2884However, it is still permitted that a frontend and a backend share the same
2885name, as this configuration seems to be commonly encountered.
2886
2887Right now, two major proxy modes are supported : "tcp", also known as layer 4,
2888and "http", also known as layer 7. In layer 4 mode, HAProxy simply forwards
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002889bidirectional traffic between two sides. In layer 7 mode, HAProxy analyzes the
Willy Tarreau0ba27502007-12-24 16:55:16 +01002890protocol, and can interact with it by allowing, blocking, switching, adding,
2891modifying, or removing arbitrary contents in requests or responses, based on
2892arbitrary criteria.
2893
Willy Tarreau70dffda2014-01-30 03:07:23 +01002894In HTTP mode, the processing applied to requests and responses flowing over
2895a connection depends in the combination of the frontend's HTTP options and
Julien Pivotto21ad3152019-12-10 13:11:17 +01002896the backend's. HAProxy supports 3 connection modes :
Willy Tarreau70dffda2014-01-30 03:07:23 +01002897
2898 - KAL : keep alive ("option http-keep-alive") which is the default mode : all
2899 requests and responses are processed, and connections remain open but idle
2900 between responses and new requests.
2901
Willy Tarreau70dffda2014-01-30 03:07:23 +01002902 - SCL: server close ("option http-server-close") : the server-facing
2903 connection is closed after the end of the response is received, but the
2904 client-facing connection remains open.
2905
Christopher Faulet315b39c2018-09-21 16:26:19 +02002906 - CLO: close ("option httpclose"): the connection is closed after the end of
2907 the response and "Connection: close" appended in both directions.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002908
2909The effective mode that will be applied to a connection passing through a
2910frontend and a backend can be determined by both proxy modes according to the
2911following matrix, but in short, the modes are symmetric, keep-alive is the
Christopher Faulet315b39c2018-09-21 16:26:19 +02002912weakest option and close is the strongest.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002913
Christopher Faulet315b39c2018-09-21 16:26:19 +02002914 Backend mode
Willy Tarreau70dffda2014-01-30 03:07:23 +01002915
Christopher Faulet315b39c2018-09-21 16:26:19 +02002916 | KAL | SCL | CLO
2917 ----+-----+-----+----
2918 KAL | KAL | SCL | CLO
2919 ----+-----+-----+----
Christopher Faulet315b39c2018-09-21 16:26:19 +02002920 mode SCL | SCL | SCL | CLO
2921 ----+-----+-----+----
2922 CLO | CLO | CLO | CLO
Willy Tarreau70dffda2014-01-30 03:07:23 +01002923
Willy Tarreau0ba27502007-12-24 16:55:16 +01002924
Willy Tarreau70dffda2014-01-30 03:07:23 +01002925
Willy Tarreauc57f0e22009-05-10 13:12:33 +020029264.1. Proxy keywords matrix
2927--------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002928
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002929The following list of keywords is supported. Most of them may only be used in a
2930limited set of section types. Some of them are marked as "deprecated" because
2931they are inherited from an old syntax which may be confusing or functionally
2932limited, and there are new recommended keywords to replace them. Keywords
Davor Ocelice9ed2812017-12-25 17:49:28 +01002933marked with "(*)" can be optionally inverted using the "no" prefix, e.g. "no
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002934option contstats". This makes sense when the option has been enabled by default
Willy Tarreau3842f002009-06-14 11:39:52 +02002935and must be disabled for a specific instance. Such options may also be prefixed
2936with "default" in order to restore default settings regardless of what has been
2937specified in a previous "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002938
Willy Tarreau6a06a402007-07-15 20:15:28 +02002939
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002940 keyword defaults frontend listen backend
2941------------------------------------+----------+----------+---------+---------
2942acl - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002943backlog X X X -
2944balance X - X X
2945bind - X X -
2946bind-process X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002947capture cookie - X X -
2948capture request header - X X -
2949capture response header - X X -
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09002950clitcpka-cnt X X X -
2951clitcpka-idle X X X -
2952clitcpka-intvl X X X -
William Lallemand82fe75c2012-10-23 10:25:10 +02002953compression X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002954cookie X - X X
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02002955declare capture - X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002956default-server X - X X
2957default_backend X X X -
2958description - X X X
2959disabled X X X X
2960dispatch - - X X
Simon Horman51a1cf62015-02-03 13:00:44 +09002961email-alert from X X X X
Simon Horman64e34162015-02-06 11:11:57 +09002962email-alert level X X X X
Simon Horman51a1cf62015-02-03 13:00:44 +09002963email-alert mailers X X X X
2964email-alert myhostname X X X X
2965email-alert to X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002966enabled X X X X
2967errorfile X X X X
Christopher Faulet76edc0f2020-01-13 15:52:01 +01002968errorfiles X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002969errorloc X X X X
2970errorloc302 X X X X
2971-- keyword -------------------------- defaults - frontend - listen -- backend -
2972errorloc303 X X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01002973force-persist - - X X
Christopher Fauletc3fe5332016-04-07 15:30:10 +02002974filter - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002975fullconn X - X X
2976grace X X X X
2977hash-type X - X X
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01002978http-after-response - X X X
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02002979http-check comment X - X X
Christopher Faulete5870d82020-04-15 11:32:03 +02002980http-check connect X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002981http-check disable-on-404 X - X X
Christopher Faulete5870d82020-04-15 11:32:03 +02002982http-check expect X - X X
Peter Gervai8912ae62020-06-11 18:26:36 +02002983http-check send X - X X
Willy Tarreau7ab6aff2010-10-12 06:30:16 +02002984http-check send-state X - X X
Christopher Faulete5870d82020-04-15 11:32:03 +02002985http-check set-var X - X X
2986http-check unset-var X - X X
Christopher Faulet3b967c12020-05-15 15:47:44 +02002987http-error X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002988http-request - X X X
Willy Tarreaue365c0b2013-06-11 16:06:12 +02002989http-response - X X X
Willy Tarreau30631952015-08-06 15:05:24 +02002990http-reuse X - X X
Baptiste Assmann2c42ef52013-10-09 21:57:02 +02002991http-send-name-header - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002992id - X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01002993ignore-persist - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02002994load-server-state-from-file X - X X
William Lallemand0f99e342011-10-12 17:50:54 +02002995log (*) X X X X
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01002996log-format X X X -
Dragan Dosen7ad31542015-09-28 17:16:47 +02002997log-format-sd X X X -
Willy Tarreau094af4e2015-01-07 15:03:42 +01002998log-tag X X X X
Willy Tarreauc35362a2014-04-25 13:58:37 +02002999max-keep-alive-queue X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003000maxconn X X X -
3001mode X X X X
3002monitor fail - X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003003monitor-uri X X X -
3004option abortonclose (*) X - X X
3005option accept-invalid-http-request (*) X X X -
3006option accept-invalid-http-response (*) X - X X
3007option allbackups (*) X - X X
3008option checkcache (*) X - X X
3009option clitcpka (*) X X X -
3010option contstats (*) X X X -
Christopher Faulet89aed322020-06-02 17:33:56 +02003011option disable-h2-upgrade (*) X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003012option dontlog-normal (*) X X X -
3013option dontlognull (*) X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003014-- keyword -------------------------- defaults - frontend - listen -- backend -
3015option forwardfor X X X X
Christopher Faulet98fbe952019-07-22 16:18:24 +02003016option h1-case-adjust-bogus-client (*) X X X -
3017option h1-case-adjust-bogus-server (*) X - X X
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02003018option http-buffer-request (*) X X X X
Willy Tarreau82649f92015-05-01 22:40:51 +02003019option http-ignore-probes (*) X X X -
Willy Tarreau16bfb022010-01-16 19:48:41 +01003020option http-keep-alive (*) X X X X
Willy Tarreau96e31212011-05-30 18:10:30 +02003021option http-no-delay (*) X X X X
Christopher Faulet98db9762018-09-21 10:25:19 +02003022option http-pretend-keepalive (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003023option http-server-close (*) X X X X
3024option http-use-proxy-header (*) X X X -
3025option httpchk X - X X
3026option httpclose (*) X X X X
Freddy Spierenburge88b7732019-03-25 14:35:17 +01003027option httplog X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003028option http_proxy (*) X X X X
Jamie Gloudon801a0a32012-08-25 00:18:33 -04003029option independent-streams (*) X X X X
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02003030option ldap-check X - X X
Simon Horman98637e52014-06-20 12:30:16 +09003031option external-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003032option log-health-checks (*) X - X X
3033option log-separate-errors (*) X X X -
3034option logasap (*) X X X -
3035option mysql-check X - X X
3036option nolinger (*) X X X X
3037option originalto X X X X
3038option persist (*) X - X X
Baptiste Assmann809e22a2015-10-12 20:22:55 +02003039option pgsql-check X - X X
3040option prefer-last-server (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003041option redispatch (*) X - X X
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02003042option redis-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003043option smtpchk X - X X
3044option socket-stats (*) X X X -
3045option splice-auto (*) X X X X
3046option splice-request (*) X X X X
3047option splice-response (*) X X X X
Christopher Fauletba7bc162016-11-07 21:07:38 +01003048option spop-check - - - X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003049option srvtcpka (*) X - X X
3050option ssl-hello-chk X - X X
3051-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreaued179852013-12-16 01:07:00 +01003052option tcp-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003053option tcp-smart-accept (*) X X X -
3054option tcp-smart-connect (*) X - X X
3055option tcpka X X X X
3056option tcplog X X X X
3057option transparent (*) X - X X
Simon Horman98637e52014-06-20 12:30:16 +09003058external-check command X - X X
3059external-check path X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003060persist rdp-cookie X - X X
3061rate-limit sessions X X X -
3062redirect - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003063-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003064retries X - X X
Olivier Houcharda254a372019-04-05 15:30:12 +02003065retry-on X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003066server - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02003067server-state-file-name X - X X
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02003068server-template - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003069source X - X X
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09003070srvtcpka-cnt X - X X
3071srvtcpka-idle X - X X
3072srvtcpka-intvl X - X X
Baptiste Assmann5a549212015-10-12 20:30:24 +02003073stats admin - X X X
3074stats auth X X X X
3075stats enable X X X X
3076stats hide-version X X X X
3077stats http-request - X X X
3078stats realm X X X X
3079stats refresh X X X X
3080stats scope X X X X
3081stats show-desc X X X X
3082stats show-legends X X X X
3083stats show-node X X X X
3084stats uri X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003085-- keyword -------------------------- defaults - frontend - listen -- backend -
3086stick match - - X X
3087stick on - - X X
3088stick store-request - - X X
Willy Tarreaud8dc99f2011-07-01 11:33:25 +02003089stick store-response - - X X
Adam Spiers68af3c12017-04-06 16:31:39 +01003090stick-table - X X X
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02003091tcp-check comment X - X X
Christopher Faulet404f9192020-04-09 23:13:54 +02003092tcp-check connect X - X X
3093tcp-check expect X - X X
3094tcp-check send X - X X
Christopher Fauletb50b3e62020-05-05 18:43:43 +02003095tcp-check send-lf X - X X
Christopher Faulet404f9192020-04-09 23:13:54 +02003096tcp-check send-binary X - X X
Christopher Fauletb50b3e62020-05-05 18:43:43 +02003097tcp-check send-binary-lf X - X X
Christopher Faulet404f9192020-04-09 23:13:54 +02003098tcp-check set-var X - X X
3099tcp-check unset-var X - X X
Willy Tarreaue9656522010-08-17 15:40:09 +02003100tcp-request connection - X X -
3101tcp-request content - X X X
Willy Tarreaua56235c2010-09-14 11:31:36 +02003102tcp-request inspect-delay - X X X
Willy Tarreau4f614292016-10-21 17:49:36 +02003103tcp-request session - X X -
Emeric Brun0a3b67f2010-09-24 15:34:53 +02003104tcp-response content - - X X
3105tcp-response inspect-delay - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003106timeout check X - X X
3107timeout client X X X -
Willy Tarreau05cdd962014-05-10 14:30:07 +02003108timeout client-fin X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003109timeout connect X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003110timeout http-keep-alive X X X X
3111timeout http-request X X X X
3112timeout queue X - X X
3113timeout server X - X X
Willy Tarreau05cdd962014-05-10 14:30:07 +02003114timeout server-fin X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003115timeout tarpit X X X X
Willy Tarreauce887fd2012-05-12 12:50:00 +02003116timeout tunnel X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003117transparent (deprecated) X - X X
William Lallemanda73203e2012-03-12 12:48:57 +01003118unique-id-format X X X -
3119unique-id-header X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003120use_backend - X X -
Christopher Fauletb30b3102019-09-12 23:03:09 +02003121use-fcgi-app - - X X
Willy Tarreau4a5cade2012-04-05 21:09:48 +02003122use-server - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003123------------------------------------+----------+----------+---------+---------
3124 keyword defaults frontend listen backend
Willy Tarreau6a06a402007-07-15 20:15:28 +02003125
Willy Tarreau0ba27502007-12-24 16:55:16 +01003126
Willy Tarreauc57f0e22009-05-10 13:12:33 +020031274.2. Alphabetically sorted keywords reference
3128---------------------------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01003129
3130This section provides a description of each keyword and its usage.
3131
3132
3133acl <aclname> <criterion> [flags] [operator] <value> ...
3134 Declare or complete an access list.
3135 May be used in sections : defaults | frontend | listen | backend
3136 no | yes | yes | yes
3137 Example:
3138 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
3139 acl invalid_src src_port 0:1023
3140 acl local_dst hdr(host) -i localhost
3141
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003142 See section 7 about ACL usage.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003143
3144
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01003145backlog <conns>
3146 Give hints to the system about the approximate listen backlog desired size
3147 May be used in sections : defaults | frontend | listen | backend
3148 yes | yes | yes | no
3149 Arguments :
3150 <conns> is the number of pending connections. Depending on the operating
3151 system, it may represent the number of already acknowledged
Cyril Bontédc4d9032012-04-08 21:57:39 +02003152 connections, of non-acknowledged ones, or both.
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01003153
3154 In order to protect against SYN flood attacks, one solution is to increase
3155 the system's SYN backlog size. Depending on the system, sometimes it is just
3156 tunable via a system parameter, sometimes it is not adjustable at all, and
3157 sometimes the system relies on hints given by the application at the time of
3158 the listen() syscall. By default, HAProxy passes the frontend's maxconn value
3159 to the listen() syscall. On systems which can make use of this value, it can
3160 sometimes be useful to be able to specify a different value, hence this
3161 backlog parameter.
3162
3163 On Linux 2.4, the parameter is ignored by the system. On Linux 2.6, it is
3164 used as a hint and the system accepts up to the smallest greater power of
3165 two, and never more than some limits (usually 32768).
3166
3167 See also : "maxconn" and the target operating system's tuning guide.
3168
3169
Willy Tarreau0ba27502007-12-24 16:55:16 +01003170balance <algorithm> [ <arguments> ]
Willy Tarreau226071e2014-04-10 11:55:45 +02003171balance url_param <param> [check_post]
Willy Tarreau0ba27502007-12-24 16:55:16 +01003172 Define the load balancing algorithm to be used in a backend.
3173 May be used in sections : defaults | frontend | listen | backend
3174 yes | no | yes | yes
3175 Arguments :
3176 <algorithm> is the algorithm used to select a server when doing load
3177 balancing. This only applies when no persistence information
3178 is available, or when a connection is redispatched to another
3179 server. <algorithm> may be one of the following :
3180
3181 roundrobin Each server is used in turns, according to their weights.
3182 This is the smoothest and fairest algorithm when the server's
3183 processing time remains equally distributed. This algorithm
3184 is dynamic, which means that server weights may be adjusted
Willy Tarreau9757a382009-10-03 12:56:50 +02003185 on the fly for slow starts for instance. It is limited by
Godbacha34bdc02013-07-22 07:44:53 +08003186 design to 4095 active servers per backend. Note that in some
Willy Tarreau9757a382009-10-03 12:56:50 +02003187 large farms, when a server becomes up after having been down
3188 for a very short time, it may sometimes take a few hundreds
3189 requests for it to be re-integrated into the farm and start
3190 receiving traffic. This is normal, though very rare. It is
3191 indicated here in case you would have the chance to observe
3192 it, so that you don't worry.
3193
3194 static-rr Each server is used in turns, according to their weights.
3195 This algorithm is as similar to roundrobin except that it is
3196 static, which means that changing a server's weight on the
3197 fly will have no effect. On the other hand, it has no design
3198 limitation on the number of servers, and when a server goes
3199 up, it is always immediately reintroduced into the farm, once
3200 the full map is recomputed. It also uses slightly less CPU to
3201 run (around -1%).
Willy Tarreau0ba27502007-12-24 16:55:16 +01003202
Willy Tarreau2d2a7f82008-03-17 12:07:56 +01003203 leastconn The server with the lowest number of connections receives the
3204 connection. Round-robin is performed within groups of servers
3205 of the same load to ensure that all servers will be used. Use
3206 of this algorithm is recommended where very long sessions are
3207 expected, such as LDAP, SQL, TSE, etc... but is not very well
3208 suited for protocols using short sessions such as HTTP. This
3209 algorithm is dynamic, which means that server weights may be
Willy Tarreau8c855f62020-10-22 17:41:45 +02003210 adjusted on the fly for slow starts for instance. It will
3211 also consider the number of queued connections in addition to
3212 the established ones in order to minimize queuing.
Willy Tarreau2d2a7f82008-03-17 12:07:56 +01003213
Willy Tarreauf09c6602012-02-13 17:12:08 +01003214 first The first server with available connection slots receives the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03003215 connection. The servers are chosen from the lowest numeric
Willy Tarreauf09c6602012-02-13 17:12:08 +01003216 identifier to the highest (see server parameter "id"), which
3217 defaults to the server's position in the farm. Once a server
Willy Tarreau64559c52012-04-07 09:08:45 +02003218 reaches its maxconn value, the next server is used. It does
Willy Tarreauf09c6602012-02-13 17:12:08 +01003219 not make sense to use this algorithm without setting maxconn.
3220 The purpose of this algorithm is to always use the smallest
3221 number of servers so that extra servers can be powered off
3222 during non-intensive hours. This algorithm ignores the server
3223 weight, and brings more benefit to long session such as RDP
Willy Tarreau64559c52012-04-07 09:08:45 +02003224 or IMAP than HTTP, though it can be useful there too. In
3225 order to use this algorithm efficiently, it is recommended
3226 that a cloud controller regularly checks server usage to turn
3227 them off when unused, and regularly checks backend queue to
3228 turn new servers on when the queue inflates. Alternatively,
3229 using "http-check send-state" may inform servers on the load.
Willy Tarreauf09c6602012-02-13 17:12:08 +01003230
Willy Tarreau0ba27502007-12-24 16:55:16 +01003231 source The source IP address is hashed and divided by the total
3232 weight of the running servers to designate which server will
3233 receive the request. This ensures that the same client IP
3234 address will always reach the same server as long as no
3235 server goes down or up. If the hash result changes due to the
3236 number of running servers changing, many clients will be
3237 directed to a different server. This algorithm is generally
3238 used in TCP mode where no cookie may be inserted. It may also
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003239 be used on the Internet to provide a best-effort stickiness
Willy Tarreau0ba27502007-12-24 16:55:16 +01003240 to clients which refuse session cookies. This algorithm is
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003241 static by default, which means that changing a server's
3242 weight on the fly will have no effect, but this can be
3243 changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003244
Oskar Stolc8dc41842012-05-19 10:19:54 +01003245 uri This algorithm hashes either the left part of the URI (before
3246 the question mark) or the whole URI (if the "whole" parameter
3247 is present) and divides the hash value by the total weight of
3248 the running servers. The result designates which server will
3249 receive the request. This ensures that the same URI will
3250 always be directed to the same server as long as no server
3251 goes up or down. This is used with proxy caches and
3252 anti-virus proxies in order to maximize the cache hit rate.
3253 Note that this algorithm may only be used in an HTTP backend.
3254 This algorithm is static by default, which means that
3255 changing a server's weight on the fly will have no effect,
3256 but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003257
Oskar Stolc8dc41842012-05-19 10:19:54 +01003258 This algorithm supports two optional parameters "len" and
Marek Majkowski9c30fc12008-04-27 23:25:55 +02003259 "depth", both followed by a positive integer number. These
3260 options may be helpful when it is needed to balance servers
3261 based on the beginning of the URI only. The "len" parameter
3262 indicates that the algorithm should only consider that many
3263 characters at the beginning of the URI to compute the hash.
3264 Note that having "len" set to 1 rarely makes sense since most
3265 URIs start with a leading "/".
3266
3267 The "depth" parameter indicates the maximum directory depth
3268 to be used to compute the hash. One level is counted for each
3269 slash in the request. If both parameters are specified, the
3270 evaluation stops when either is reached.
3271
Willy Tarreau57a37412020-09-23 08:56:29 +02003272 A "path-only" parameter indicates that the hashing key starts
3273 at the first '/' of the path. This can be used to ignore the
3274 authority part of absolute URIs, and to make sure that HTTP/1
3275 and HTTP/2 URIs will provide the same hash.
3276
Willy Tarreau0ba27502007-12-24 16:55:16 +01003277 url_param The URL parameter specified in argument will be looked up in
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003278 the query string of each HTTP GET request.
3279
3280 If the modifier "check_post" is used, then an HTTP POST
Cyril Bontédc4d9032012-04-08 21:57:39 +02003281 request entity will be searched for the parameter argument,
3282 when it is not found in a query string after a question mark
Willy Tarreau226071e2014-04-10 11:55:45 +02003283 ('?') in the URL. The message body will only start to be
3284 analyzed once either the advertised amount of data has been
3285 received or the request buffer is full. In the unlikely event
3286 that chunked encoding is used, only the first chunk is
Cyril Bontédc4d9032012-04-08 21:57:39 +02003287 scanned. Parameter values separated by a chunk boundary, may
Willy Tarreau226071e2014-04-10 11:55:45 +02003288 be randomly balanced if at all. This keyword used to support
3289 an optional <max_wait> parameter which is now ignored.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003290
3291 If the parameter is found followed by an equal sign ('=') and
3292 a value, then the value is hashed and divided by the total
3293 weight of the running servers. The result designates which
3294 server will receive the request.
3295
3296 This is used to track user identifiers in requests and ensure
3297 that a same user ID will always be sent to the same server as
3298 long as no server goes up or down. If no value is found or if
3299 the parameter is not found, then a round robin algorithm is
3300 applied. Note that this algorithm may only be used in an HTTP
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003301 backend. This algorithm is static by default, which means
3302 that changing a server's weight on the fly will have no
3303 effect, but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003304
Cyril Bontédc4d9032012-04-08 21:57:39 +02003305 hdr(<name>) The HTTP header <name> will be looked up in each HTTP
3306 request. Just as with the equivalent ACL 'hdr()' function,
3307 the header name in parenthesis is not case sensitive. If the
3308 header is absent or if it does not contain any value, the
3309 roundrobin algorithm is applied instead.
Benoitaffb4812009-03-25 13:02:10 +01003310
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003311 An optional 'use_domain_only' parameter is available, for
Benoitaffb4812009-03-25 13:02:10 +01003312 reducing the hash algorithm to the main domain part with some
3313 specific headers such as 'Host'. For instance, in the Host
3314 value "haproxy.1wt.eu", only "1wt" will be considered.
3315
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003316 This algorithm is static by default, which means that
3317 changing a server's weight on the fly will have no effect,
3318 but this can be changed using "hash-type".
3319
Willy Tarreau21c741a2019-01-14 18:14:27 +01003320 random
3321 random(<draws>)
3322 A random number will be used as the key for the consistent
Willy Tarreau760e81d2018-05-03 07:20:40 +02003323 hashing function. This means that the servers' weights are
3324 respected, dynamic weight changes immediately take effect, as
3325 well as new server additions. Random load balancing can be
3326 useful with large farms or when servers are frequently added
Willy Tarreau21c741a2019-01-14 18:14:27 +01003327 or removed as it may avoid the hammering effect that could
3328 result from roundrobin or leastconn in this situation. The
3329 hash-balance-factor directive can be used to further improve
3330 fairness of the load balancing, especially in situations
3331 where servers show highly variable response times. When an
3332 argument <draws> is present, it must be an integer value one
3333 or greater, indicating the number of draws before selecting
3334 the least loaded of these servers. It was indeed demonstrated
3335 that picking the least loaded of two servers is enough to
3336 significantly improve the fairness of the algorithm, by
3337 always avoiding to pick the most loaded server within a farm
3338 and getting rid of any bias that could be induced by the
3339 unfair distribution of the consistent list. Higher values N
3340 will take away N-1 of the highest loaded servers at the
3341 expense of performance. With very high values, the algorithm
3342 will converge towards the leastconn's result but much slower.
3343 The default value is 2, which generally shows very good
3344 distribution and performance. This algorithm is also known as
3345 the Power of Two Random Choices and is described here :
3346 http://www.eecs.harvard.edu/~michaelm/postscripts/handbook2001.pdf
Willy Tarreau760e81d2018-05-03 07:20:40 +02003347
Emeric Brun736aa232009-06-30 17:56:00 +02003348 rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02003349 rdp-cookie(<name>)
Emeric Brun736aa232009-06-30 17:56:00 +02003350 The RDP cookie <name> (or "mstshash" if omitted) will be
3351 looked up and hashed for each incoming TCP request. Just as
3352 with the equivalent ACL 'req_rdp_cookie()' function, the name
3353 is not case-sensitive. This mechanism is useful as a degraded
3354 persistence mode, as it makes it possible to always send the
3355 same user (or the same session ID) to the same server. If the
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003356 cookie is not found, the normal roundrobin algorithm is
Emeric Brun736aa232009-06-30 17:56:00 +02003357 used instead.
3358
3359 Note that for this to work, the frontend must ensure that an
3360 RDP cookie is already present in the request buffer. For this
3361 you must use 'tcp-request content accept' rule combined with
3362 a 'req_rdp_cookie_cnt' ACL.
3363
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003364 This algorithm is static by default, which means that
3365 changing a server's weight on the fly will have no effect,
3366 but this can be changed using "hash-type".
3367
Cyril Bontédc4d9032012-04-08 21:57:39 +02003368 See also the rdp_cookie pattern fetch function.
Simon Hormanab814e02011-06-24 14:50:20 +09003369
Willy Tarreau0ba27502007-12-24 16:55:16 +01003370 <arguments> is an optional list of arguments which may be needed by some
Marek Majkowski9c30fc12008-04-27 23:25:55 +02003371 algorithms. Right now, only "url_param" and "uri" support an
3372 optional argument.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003373
Willy Tarreau3cd9af22009-03-15 14:06:41 +01003374 The load balancing algorithm of a backend is set to roundrobin when no other
3375 algorithm, mode nor option have been set. The algorithm may only be set once
3376 for each backend.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003377
Lukas Tribus80512b12018-10-27 20:07:40 +02003378 With authentication schemes that require the same connection like NTLM, URI
John Roeslerfb2fce12019-07-10 15:45:51 -05003379 based algorithms must not be used, as they would cause subsequent requests
Lukas Tribus80512b12018-10-27 20:07:40 +02003380 to be routed to different backend servers, breaking the invalid assumptions
3381 NTLM relies on.
3382
Willy Tarreau0ba27502007-12-24 16:55:16 +01003383 Examples :
3384 balance roundrobin
3385 balance url_param userid
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003386 balance url_param session_id check_post 64
Benoitaffb4812009-03-25 13:02:10 +01003387 balance hdr(User-Agent)
3388 balance hdr(host)
3389 balance hdr(Host) use_domain_only
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003390
3391 Note: the following caveats and limitations on using the "check_post"
3392 extension with "url_param" must be considered :
3393
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003394 - all POST requests are eligible for consideration, because there is no way
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003395 to determine if the parameters will be found in the body or entity which
3396 may contain binary data. Therefore another method may be required to
3397 restrict consideration of POST requests that have no URL parameters in
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02003398 the body. (see acl http_end)
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003399
3400 - using a <max_wait> value larger than the request buffer size does not
3401 make sense and is useless. The buffer size is set at build time, and
3402 defaults to 16 kB.
3403
3404 - Content-Encoding is not supported, the parameter search will probably
3405 fail; and load balancing will fall back to Round Robin.
3406
3407 - Expect: 100-continue is not supported, load balancing will fall back to
3408 Round Robin.
3409
Lukas Tribus23953682017-04-28 13:24:30 +00003410 - Transfer-Encoding (RFC7230 3.3.1) is only supported in the first chunk.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003411 If the entire parameter value is not present in the first chunk, the
3412 selection of server is undefined (actually, defined by how little
3413 actually appeared in the first chunk).
3414
3415 - This feature does not support generation of a 100, 411 or 501 response.
3416
3417 - In some cases, requesting "check_post" MAY attempt to scan the entire
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003418 contents of a message body. Scanning normally terminates when linear
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003419 white space or control characters are found, indicating the end of what
3420 might be a URL parameter list. This is probably not a concern with SGML
3421 type message bodies.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003422
Willy Tarreau294d0f02015-08-10 19:40:12 +02003423 See also : "dispatch", "cookie", "transparent", "hash-type" and "http_proxy".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003424
3425
Willy Tarreaub6205fd2012-09-24 12:27:33 +02003426bind [<address>]:<port_range> [, ...] [param*]
3427bind /<path> [, ...] [param*]
Willy Tarreau0ba27502007-12-24 16:55:16 +01003428 Define one or several listening addresses and/or ports in a frontend.
3429 May be used in sections : defaults | frontend | listen | backend
3430 no | yes | yes | no
3431 Arguments :
Willy Tarreaub1e52e82008-01-13 14:49:51 +01003432 <address> is optional and can be a host name, an IPv4 address, an IPv6
3433 address, or '*'. It designates the address the frontend will
3434 listen on. If unset, all IPv4 addresses of the system will be
3435 listened on. The same will apply for '*' or the system's
David du Colombier9c938da2011-03-17 10:40:27 +01003436 special address "0.0.0.0". The IPv6 equivalent is '::'.
Willy Tarreau24709282013-03-10 21:32:12 +01003437 Optionally, an address family prefix may be used before the
3438 address to force the family regardless of the address format,
3439 which can be useful to specify a path to a unix socket with
3440 no slash ('/'). Currently supported prefixes are :
3441 - 'ipv4@' -> address is always IPv4
3442 - 'ipv6@' -> address is always IPv6
Emeric Brun3835c0d2020-07-07 09:46:09 +02003443 - 'udp@' -> address is resolved as IPv4 or IPv6 and
Emeric Brun12941c82020-07-07 14:19:42 +02003444 protocol UDP is used. Currently those listeners are
3445 supported only in log-forward sections.
Emeric Brun3835c0d2020-07-07 09:46:09 +02003446 - 'udp4@' -> address is always IPv4 and protocol UDP
Emeric Brun12941c82020-07-07 14:19:42 +02003447 is used. Currently those listeners are supported
3448 only in log-forward sections.
Emeric Brun3835c0d2020-07-07 09:46:09 +02003449 - 'udp6@' -> address is always IPv6 and protocol UDP
Emeric Brun12941c82020-07-07 14:19:42 +02003450 is used. Currently those listeners are supported
3451 only in log-forward sections.
Willy Tarreau24709282013-03-10 21:32:12 +01003452 - 'unix@' -> address is a path to a local unix socket
Willy Tarreau70f72e02014-07-08 00:37:50 +02003453 - 'abns@' -> address is in abstract namespace (Linux only).
3454 Note: since abstract sockets are not "rebindable", they
3455 do not cope well with multi-process mode during
3456 soft-restart, so it is better to avoid them if
3457 nbproc is greater than 1. The effect is that if the
3458 new process fails to start, only one of the old ones
3459 will be able to rebind to the socket.
Willy Tarreau40aa0702013-03-10 23:51:38 +01003460 - 'fd@<n>' -> use file descriptor <n> inherited from the
3461 parent. The fd must be bound and may or may not already
3462 be listening.
William Lallemand2fe7dd02018-09-11 16:51:29 +02003463 - 'sockpair@<n>'-> like fd@ but you must use the fd of a
3464 connected unix socket or of a socketpair. The bind waits
3465 to receive a FD over the unix socket and uses it as if it
3466 was the FD of an accept(). Should be used carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02003467 You may want to reference some environment variables in the
3468 address parameter, see section 2.3 about environment
3469 variables.
Willy Tarreaub1e52e82008-01-13 14:49:51 +01003470
Willy Tarreauc5011ca2010-03-22 11:53:56 +01003471 <port_range> is either a unique TCP port, or a port range for which the
3472 proxy will accept connections for the IP address specified
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003473 above. The port is mandatory for TCP listeners. Note that in
3474 the case of an IPv6 address, the port is always the number
3475 after the last colon (':'). A range can either be :
Willy Tarreauc5011ca2010-03-22 11:53:56 +01003476 - a numerical port (ex: '80')
3477 - a dash-delimited ports range explicitly stating the lower
3478 and upper bounds (ex: '2000-2100') which are included in
3479 the range.
3480
3481 Particular care must be taken against port ranges, because
3482 every <address:port> couple consumes one socket (= a file
3483 descriptor), so it's easy to consume lots of descriptors
3484 with a simple range, and to run out of sockets. Also, each
3485 <address:port> couple must be used only once among all
3486 instances running on a same system. Please note that binding
3487 to ports lower than 1024 generally require particular
Jamie Gloudon801a0a32012-08-25 00:18:33 -04003488 privileges to start the program, which are independent of
Willy Tarreauc5011ca2010-03-22 11:53:56 +01003489 the 'uid' parameter.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003490
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003491 <path> is a UNIX socket path beginning with a slash ('/'). This is
Davor Ocelice9ed2812017-12-25 17:49:28 +01003492 alternative to the TCP listening port. HAProxy will then
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003493 receive UNIX connections on the socket located at this place.
3494 The path must begin with a slash and by default is absolute.
3495 It can be relative to the prefix defined by "unix-bind" in
3496 the global section. Note that the total length of the prefix
3497 followed by the socket path cannot exceed some system limits
3498 for UNIX sockets, which commonly are set to 107 characters.
3499
Willy Tarreaub6205fd2012-09-24 12:27:33 +02003500 <param*> is a list of parameters common to all sockets declared on the
3501 same line. These numerous parameters depend on OS and build
3502 options and have a complete section dedicated to them. Please
3503 refer to section 5 to for more details.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02003504
Willy Tarreau0ba27502007-12-24 16:55:16 +01003505 It is possible to specify a list of address:port combinations delimited by
3506 commas. The frontend will then listen on all of these addresses. There is no
3507 fixed limit to the number of addresses and ports which can be listened on in
3508 a frontend, as well as there is no limit to the number of "bind" statements
3509 in a frontend.
3510
3511 Example :
3512 listen http_proxy
3513 bind :80,:443
3514 bind 10.0.0.1:10080,10.0.0.1:10443
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003515 bind /var/run/ssl-frontend.sock user root mode 600 accept-proxy
Willy Tarreau0ba27502007-12-24 16:55:16 +01003516
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02003517 listen http_https_proxy
3518 bind :80
Cyril Bonté0d44fc62012-10-09 22:45:33 +02003519 bind :443 ssl crt /etc/haproxy/site.pem
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02003520
Willy Tarreau24709282013-03-10 21:32:12 +01003521 listen http_https_proxy_explicit
3522 bind ipv6@:80
3523 bind ipv4@public_ssl:443 ssl crt /etc/haproxy/site.pem
3524 bind unix@ssl-frontend.sock user root mode 600 accept-proxy
3525
Willy Tarreaudad36a32013-03-11 01:20:04 +01003526 listen external_bind_app1
William Lallemandb2f07452015-05-12 14:27:13 +02003527 bind "fd@${FD_APP1}"
Willy Tarreaudad36a32013-03-11 01:20:04 +01003528
Willy Tarreau55dcaf62015-09-27 15:03:15 +02003529 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
3530 sun_path length is used for the address length. Some other programs
3531 such as socat use the string length only by default. Pass the option
3532 ",unix-tightsocklen=0" to any abstract socket definition in socat to
3533 make it compatible with HAProxy's.
3534
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003535 See also : "source", "option forwardfor", "unix-bind" and the PROXY protocol
Willy Tarreaub6205fd2012-09-24 12:27:33 +02003536 documentation, and section 5 about bind options.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003537
3538
Christopher Fauletff4121f2017-11-22 16:38:49 +01003539bind-process [ all | odd | even | <process_num>[-[<process_num>]] ] ...
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003540 Limit visibility of an instance to a certain set of processes numbers.
3541 May be used in sections : defaults | frontend | listen | backend
3542 yes | yes | yes | yes
3543 Arguments :
3544 all All process will see this instance. This is the default. It
3545 may be used to override a default value.
3546
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003547 odd This instance will be enabled on processes 1,3,5,...63. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003548 option may be combined with other numbers.
3549
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003550 even This instance will be enabled on processes 2,4,6,...64. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003551 option may be combined with other numbers. Do not use it
3552 with less than 2 processes otherwise some instances might be
3553 missing from all processes.
3554
Christopher Fauletff4121f2017-11-22 16:38:49 +01003555 process_num The instance will be enabled on this process number or range,
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003556 whose values must all be between 1 and 32 or 64 depending on
Christopher Fauletff4121f2017-11-22 16:38:49 +01003557 the machine's word size. Ranges can be partially defined. The
3558 higher bound can be omitted. In such case, it is replaced by
3559 the corresponding maximum value. If a proxy is bound to
3560 process numbers greater than the configured global.nbproc, it
3561 will either be forced to process #1 if a single process was
Willy Tarreau102df612014-05-07 23:56:38 +02003562 specified, or to all processes otherwise.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003563
3564 This keyword limits binding of certain instances to certain processes. This
3565 is useful in order not to have too many processes listening to the same
3566 ports. For instance, on a dual-core machine, it might make sense to set
3567 'nbproc 2' in the global section, then distributes the listeners among 'odd'
3568 and 'even' instances.
3569
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003570 At the moment, it is not possible to reference more than 32 or 64 processes
3571 using this keyword, but this should be more than enough for most setups.
3572 Please note that 'all' really means all processes regardless of the machine's
3573 word size, and is not limited to the first 32 or 64.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003574
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02003575 Each "bind" line may further be limited to a subset of the proxy's processes,
3576 please consult the "process" bind keyword in section 5.1.
3577
Willy Tarreaub369a042014-09-16 13:21:03 +02003578 When a frontend has no explicit "bind-process" line, it tries to bind to all
3579 the processes referenced by its "bind" lines. That means that frontends can
3580 easily adapt to their listeners' processes.
3581
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003582 If some backends are referenced by frontends bound to other processes, the
3583 backend automatically inherits the frontend's processes.
3584
3585 Example :
3586 listen app_ip1
3587 bind 10.0.0.1:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02003588 bind-process odd
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003589
3590 listen app_ip2
3591 bind 10.0.0.2:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02003592 bind-process even
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003593
3594 listen management
3595 bind 10.0.0.3:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02003596 bind-process 1 2 3 4
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003597
Willy Tarreau110ecc12012-11-15 17:50:01 +01003598 listen management
3599 bind 10.0.0.4:80
3600 bind-process 1-4
3601
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02003602 See also : "nbproc" in global section, and "process" in section 5.1.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003603
3604
Willy Tarreau0ba27502007-12-24 16:55:16 +01003605capture cookie <name> len <length>
3606 Capture and log a cookie in the request and in the response.
3607 May be used in sections : defaults | frontend | listen | backend
3608 no | yes | yes | no
3609 Arguments :
3610 <name> is the beginning of the name of the cookie to capture. In order
3611 to match the exact name, simply suffix the name with an equal
3612 sign ('='). The full name will appear in the logs, which is
3613 useful with application servers which adjust both the cookie name
Davor Ocelice9ed2812017-12-25 17:49:28 +01003614 and value (e.g. ASPSESSIONXXX).
Willy Tarreau0ba27502007-12-24 16:55:16 +01003615
3616 <length> is the maximum number of characters to report in the logs, which
3617 include the cookie name, the equal sign and the value, all in the
3618 standard "name=value" form. The string will be truncated on the
3619 right if it exceeds <length>.
3620
3621 Only the first cookie is captured. Both the "cookie" request headers and the
3622 "set-cookie" response headers are monitored. This is particularly useful to
3623 check for application bugs causing session crossing or stealing between
3624 users, because generally the user's cookies can only change on a login page.
3625
3626 When the cookie was not presented by the client, the associated log column
3627 will report "-". When a request does not cause a cookie to be assigned by the
3628 server, a "-" is reported in the response column.
3629
3630 The capture is performed in the frontend only because it is necessary that
3631 the log format does not change for a given frontend depending on the
3632 backends. This may change in the future. Note that there can be only one
Willy Tarreau193b8c62012-11-22 00:17:38 +01003633 "capture cookie" statement in a frontend. The maximum capture length is set
3634 by the global "tune.http.cookielen" setting and defaults to 63 characters. It
3635 is not possible to specify a capture in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003636
3637 Example:
3638 capture cookie ASPSESSION len 32
3639
3640 See also : "capture request header", "capture response header" as well as
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003641 section 8 about logging.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003642
3643
3644capture request header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01003645 Capture and log the last occurrence of the specified request header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003646 May be used in sections : defaults | frontend | listen | backend
3647 no | yes | yes | no
3648 Arguments :
3649 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003650 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01003651 appear in the requests, with the first letter of each word in
3652 upper case. The header name will not appear in the logs, only the
3653 value is reported, but the position in the logs is respected.
3654
3655 <length> is the maximum number of characters to extract from the value and
3656 report in the logs. The string will be truncated on the right if
3657 it exceeds <length>.
3658
Willy Tarreau4460d032012-11-21 23:37:37 +01003659 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01003660 value will be added to the logs between braces ('{}'). If multiple headers
3661 are captured, they will be delimited by a vertical bar ('|') and will appear
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003662 in the same order they were declared in the configuration. Non-existent
3663 headers will be logged just as an empty string. Common uses for request
3664 header captures include the "Host" field in virtual hosting environments, the
3665 "Content-length" when uploads are supported, "User-agent" to quickly
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003666 differentiate between real users and robots, and "X-Forwarded-For" in proxied
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003667 environments to find where the request came from.
3668
3669 Note that when capturing headers such as "User-agent", some spaces may be
3670 logged, making the log analysis more difficult. Thus be careful about what
3671 you log if you know your log parser is not smart enough to rely on the
3672 braces.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003673
Willy Tarreau0900abb2012-11-22 00:21:46 +01003674 There is no limit to the number of captured request headers nor to their
3675 length, though it is wise to keep them low to limit memory usage per session.
3676 In order to keep log format consistent for a same frontend, header captures
3677 can only be declared in a frontend. It is not possible to specify a capture
3678 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003679
3680 Example:
3681 capture request header Host len 15
3682 capture request header X-Forwarded-For len 15
Cyril Bontéd1b0f7c2015-10-26 22:37:39 +01003683 capture request header Referer len 15
Willy Tarreau0ba27502007-12-24 16:55:16 +01003684
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003685 See also : "capture cookie", "capture response header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01003686 about logging.
3687
3688
3689capture response header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01003690 Capture and log the last occurrence of the specified response header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003691 May be used in sections : defaults | frontend | listen | backend
3692 no | yes | yes | no
3693 Arguments :
3694 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003695 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01003696 appear in the response, with the first letter of each word in
3697 upper case. The header name will not appear in the logs, only the
3698 value is reported, but the position in the logs is respected.
3699
3700 <length> is the maximum number of characters to extract from the value and
3701 report in the logs. The string will be truncated on the right if
3702 it exceeds <length>.
3703
Willy Tarreau4460d032012-11-21 23:37:37 +01003704 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01003705 result will be added to the logs between braces ('{}') after the captured
3706 request headers. If multiple headers are captured, they will be delimited by
3707 a vertical bar ('|') and will appear in the same order they were declared in
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003708 the configuration. Non-existent headers will be logged just as an empty
3709 string. Common uses for response header captures include the "Content-length"
3710 header which indicates how many bytes are expected to be returned, the
3711 "Location" header to track redirections.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003712
Willy Tarreau0900abb2012-11-22 00:21:46 +01003713 There is no limit to the number of captured response headers nor to their
3714 length, though it is wise to keep them low to limit memory usage per session.
3715 In order to keep log format consistent for a same frontend, header captures
3716 can only be declared in a frontend. It is not possible to specify a capture
3717 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003718
3719 Example:
3720 capture response header Content-length len 9
3721 capture response header Location len 15
3722
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003723 See also : "capture cookie", "capture request header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01003724 about logging.
3725
3726
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09003727clitcpka-cnt <count>
3728 Sets the maximum number of keepalive probes TCP should send before dropping
3729 the connection on the client side.
3730 May be used in sections : defaults | frontend | listen | backend
3731 yes | yes | yes | no
3732 Arguments :
3733 <count> is the maximum number of keepalive probes.
3734
3735 This keyword corresponds to the socket option TCP_KEEPCNT. If this keyword
3736 is not specified, system-wide TCP parameter (tcp_keepalive_probes) is used.
Willy Tarreau52543212020-07-09 05:58:51 +02003737 The availability of this setting depends on the operating system. It is
3738 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09003739
3740 See also : "option clitcpka", "clitcpka-idle", "clitcpka-intvl".
3741
3742
3743clitcpka-idle <timeout>
3744 Sets the time the connection needs to remain idle before TCP starts sending
3745 keepalive probes, if enabled the sending of TCP keepalive packets on the
3746 client side.
3747 May be used in sections : defaults | frontend | listen | backend
3748 yes | yes | yes | no
3749 Arguments :
3750 <timeout> is the time the connection needs to remain idle before TCP starts
3751 sending keepalive probes. It is specified in seconds by default,
3752 but can be in any other unit if the number is suffixed by the
3753 unit, as explained at the top of this document.
3754
3755 This keyword corresponds to the socket option TCP_KEEPIDLE. If this keyword
3756 is not specified, system-wide TCP parameter (tcp_keepalive_time) is used.
Willy Tarreau52543212020-07-09 05:58:51 +02003757 The availability of this setting depends on the operating system. It is
3758 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09003759
3760 See also : "option clitcpka", "clitcpka-cnt", "clitcpka-intvl".
3761
3762
3763clitcpka-intvl <timeout>
3764 Sets the time between individual keepalive probes on the client side.
3765 May be used in sections : defaults | frontend | listen | backend
3766 yes | yes | yes | no
3767 Arguments :
3768 <timeout> is the time between individual keepalive probes. It is specified
3769 in seconds by default, but can be in any other unit if the number
3770 is suffixed by the unit, as explained at the top of this
3771 document.
3772
3773 This keyword corresponds to the socket option TCP_KEEPINTVL. If this keyword
3774 is not specified, system-wide TCP parameter (tcp_keepalive_intvl) is used.
Willy Tarreau52543212020-07-09 05:58:51 +02003775 The availability of this setting depends on the operating system. It is
3776 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09003777
3778 See also : "option clitcpka", "clitcpka-cnt", "clitcpka-idle".
3779
3780
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003781compression algo <algorithm> ...
3782compression type <mime type> ...
Willy Tarreau70737d12012-10-27 00:34:28 +02003783compression offload
William Lallemand82fe75c2012-10-23 10:25:10 +02003784 Enable HTTP compression.
3785 May be used in sections : defaults | frontend | listen | backend
3786 yes | yes | yes | yes
3787 Arguments :
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003788 algo is followed by the list of supported compression algorithms.
3789 type is followed by the list of MIME types that will be compressed.
3790 offload makes haproxy work as a compression offloader only (see notes).
3791
3792 The currently supported algorithms are :
Willy Tarreauc91840a2015-03-28 17:00:39 +01003793 identity this is mostly for debugging, and it was useful for developing
3794 the compression feature. Identity does not apply any change on
3795 data.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003796
Willy Tarreauc91840a2015-03-28 17:00:39 +01003797 gzip applies gzip compression. This setting is only available when
Baptiste Assmannf085d632015-12-21 17:57:32 +01003798 support for zlib or libslz was built in.
Willy Tarreauc91840a2015-03-28 17:00:39 +01003799
3800 deflate same as "gzip", but with deflate algorithm and zlib format.
3801 Note that this algorithm has ambiguous support on many
3802 browsers and no support at all from recent ones. It is
3803 strongly recommended not to use it for anything else than
3804 experimentation. This setting is only available when support
Baptiste Assmannf085d632015-12-21 17:57:32 +01003805 for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003806
Willy Tarreauc91840a2015-03-28 17:00:39 +01003807 raw-deflate same as "deflate" without the zlib wrapper, and used as an
3808 alternative when the browser wants "deflate". All major
3809 browsers understand it and despite violating the standards,
3810 it is known to work better than "deflate", at least on MSIE
3811 and some versions of Safari. Do not use it in conjunction
3812 with "deflate", use either one or the other since both react
3813 to the same Accept-Encoding token. This setting is only
Baptiste Assmannf085d632015-12-21 17:57:32 +01003814 available when support for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003815
Dmitry Sivachenko87c208b2012-11-22 20:03:26 +04003816 Compression will be activated depending on the Accept-Encoding request
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003817 header. With identity, it does not take care of that header.
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04003818 If backend servers support HTTP compression, these directives
3819 will be no-op: haproxy will see the compressed response and will not
3820 compress again. If backend servers do not support HTTP compression and
3821 there is Accept-Encoding header in request, haproxy will compress the
3822 matching response.
Willy Tarreau70737d12012-10-27 00:34:28 +02003823
3824 The "offload" setting makes haproxy remove the Accept-Encoding header to
3825 prevent backend servers from compressing responses. It is strongly
3826 recommended not to do this because this means that all the compression work
3827 will be done on the single point where haproxy is located. However in some
3828 deployment scenarios, haproxy may be installed in front of a buggy gateway
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04003829 with broken HTTP compression implementation which can't be turned off.
3830 In that case haproxy can be used to prevent that gateway from emitting
3831 invalid payloads. In this case, simply removing the header in the
3832 configuration does not work because it applies before the header is parsed,
3833 so that prevents haproxy from compressing. The "offload" setting should
Willy Tarreauffea9fd2014-07-12 16:37:02 +02003834 then be used for such scenarios. Note: for now, the "offload" setting is
3835 ignored when set in a defaults section.
William Lallemand82fe75c2012-10-23 10:25:10 +02003836
William Lallemand05097442012-11-20 12:14:28 +01003837 Compression is disabled when:
Baptiste Assmann650d53d2013-01-05 15:44:44 +01003838 * the request does not advertise a supported compression algorithm in the
3839 "Accept-Encoding" header
3840 * the response message is not HTTP/1.1
Tim Duesterhusbb48c9a2019-01-30 23:46:04 +01003841 * HTTP status code is not one of 200, 201, 202, or 203
Baptiste Assmann650d53d2013-01-05 15:44:44 +01003842 * response contain neither a "Content-Length" header nor a
3843 "Transfer-Encoding" whose last value is "chunked"
3844 * response contains a "Content-Type" header whose first value starts with
3845 "multipart"
3846 * the response contains the "no-transform" value in the "Cache-control"
3847 header
3848 * User-Agent matches "Mozilla/4" unless it is MSIE 6 with XP SP2, or MSIE 7
3849 and later
3850 * The response contains a "Content-Encoding" header, indicating that the
3851 response is already compressed (see compression offload)
Tim Duesterhusbb48c9a2019-01-30 23:46:04 +01003852 * The response contains an invalid "ETag" header or multiple ETag headers
William Lallemand05097442012-11-20 12:14:28 +01003853
Tim Duesterhusb229f012019-01-29 16:38:56 +01003854 Note: The compression does not emit the Warning header.
William Lallemand05097442012-11-20 12:14:28 +01003855
William Lallemand82fe75c2012-10-23 10:25:10 +02003856 Examples :
3857 compression algo gzip
3858 compression type text/html text/plain
Willy Tarreau0ba27502007-12-24 16:55:16 +01003859
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003860
Willy Tarreau55165fe2009-05-10 12:02:55 +02003861cookie <name> [ rewrite | insert | prefix ] [ indirect ] [ nocache ]
Willy Tarreau4992dd22012-05-31 21:02:17 +02003862 [ postonly ] [ preserve ] [ httponly ] [ secure ]
3863 [ domain <domain> ]* [ maxidle <idle> ] [ maxlife <life> ]
Christopher Faulet2f533902020-01-21 11:06:48 +01003864 [ dynamic ] [ attr <value> ]*
Willy Tarreau0ba27502007-12-24 16:55:16 +01003865 Enable cookie-based persistence in a backend.
3866 May be used in sections : defaults | frontend | listen | backend
3867 yes | no | yes | yes
3868 Arguments :
3869 <name> is the name of the cookie which will be monitored, modified or
3870 inserted in order to bring persistence. This cookie is sent to
3871 the client via a "Set-Cookie" header in the response, and is
3872 brought back by the client in a "Cookie" header in all requests.
3873 Special care should be taken to choose a name which does not
3874 conflict with any likely application cookie. Also, if the same
Davor Ocelice9ed2812017-12-25 17:49:28 +01003875 backends are subject to be used by the same clients (e.g.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003876 HTTP/HTTPS), care should be taken to use different cookie names
3877 between all backends if persistence between them is not desired.
3878
3879 rewrite This keyword indicates that the cookie will be provided by the
3880 server and that haproxy will have to modify its value to set the
3881 server's identifier in it. This mode is handy when the management
3882 of complex combinations of "Set-cookie" and "Cache-control"
3883 headers is left to the application. The application can then
3884 decide whether or not it is appropriate to emit a persistence
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003885 cookie. Since all responses should be monitored, this mode
3886 doesn't work in HTTP tunnel mode. Unless the application
Davor Ocelice9ed2812017-12-25 17:49:28 +01003887 behavior is very complex and/or broken, it is advised not to
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003888 start with this mode for new deployments. This keyword is
3889 incompatible with "insert" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003890
3891 insert This keyword indicates that the persistence cookie will have to
Willy Tarreaua79094d2010-08-31 22:54:15 +02003892 be inserted by haproxy in server responses if the client did not
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003893
Willy Tarreaua79094d2010-08-31 22:54:15 +02003894 already have a cookie that would have permitted it to access this
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003895 server. When used without the "preserve" option, if the server
Michael Prokop4438c602019-05-24 10:25:45 +02003896 emits a cookie with the same name, it will be removed before
Davor Ocelice9ed2812017-12-25 17:49:28 +01003897 processing. For this reason, this mode can be used to upgrade
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003898 existing configurations running in the "rewrite" mode. The cookie
3899 will only be a session cookie and will not be stored on the
3900 client's disk. By default, unless the "indirect" option is added,
3901 the server will see the cookies emitted by the client. Due to
3902 caching effects, it is generally wise to add the "nocache" or
3903 "postonly" keywords (see below). The "insert" keyword is not
3904 compatible with "rewrite" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003905
3906 prefix This keyword indicates that instead of relying on a dedicated
3907 cookie for the persistence, an existing one will be completed.
3908 This may be needed in some specific environments where the client
3909 does not support more than one single cookie and the application
3910 already needs it. In this case, whenever the server sets a cookie
3911 named <name>, it will be prefixed with the server's identifier
3912 and a delimiter. The prefix will be removed from all client
3913 requests so that the server still finds the cookie it emitted.
3914 Since all requests and responses are subject to being modified,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003915 this mode doesn't work with tunnel mode. The "prefix" keyword is
Willy Tarreau37229df2011-10-17 12:24:55 +02003916 not compatible with "rewrite" and "insert". Note: it is highly
3917 recommended not to use "indirect" with "prefix", otherwise server
3918 cookie updates would not be sent to clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003919
Willy Tarreaua79094d2010-08-31 22:54:15 +02003920 indirect When this option is specified, no cookie will be emitted to a
3921 client which already has a valid one for the server which has
3922 processed the request. If the server sets such a cookie itself,
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003923 it will be removed, unless the "preserve" option is also set. In
3924 "insert" mode, this will additionally remove cookies from the
3925 requests transmitted to the server, making the persistence
3926 mechanism totally transparent from an application point of view.
Willy Tarreau37229df2011-10-17 12:24:55 +02003927 Note: it is highly recommended not to use "indirect" with
3928 "prefix", otherwise server cookie updates would not be sent to
3929 clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003930
3931 nocache This option is recommended in conjunction with the insert mode
3932 when there is a cache between the client and HAProxy, as it
3933 ensures that a cacheable response will be tagged non-cacheable if
3934 a cookie needs to be inserted. This is important because if all
3935 persistence cookies are added on a cacheable home page for
3936 instance, then all customers will then fetch the page from an
3937 outer cache and will all share the same persistence cookie,
3938 leading to one server receiving much more traffic than others.
3939 See also the "insert" and "postonly" options.
3940
3941 postonly This option ensures that cookie insertion will only be performed
3942 on responses to POST requests. It is an alternative to the
3943 "nocache" option, because POST responses are not cacheable, so
3944 this ensures that the persistence cookie will never get cached.
3945 Since most sites do not need any sort of persistence before the
3946 first POST which generally is a login request, this is a very
3947 efficient method to optimize caching without risking to find a
3948 persistence cookie in the cache.
3949 See also the "insert" and "nocache" options.
3950
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003951 preserve This option may only be used with "insert" and/or "indirect". It
3952 allows the server to emit the persistence cookie itself. In this
3953 case, if a cookie is found in the response, haproxy will leave it
3954 untouched. This is useful in order to end persistence after a
3955 logout request for instance. For this, the server just has to
Davor Ocelice9ed2812017-12-25 17:49:28 +01003956 emit a cookie with an invalid value (e.g. empty) or with a date in
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003957 the past. By combining this mechanism with the "disable-on-404"
3958 check option, it is possible to perform a completely graceful
3959 shutdown because users will definitely leave the server after
3960 they logout.
3961
Willy Tarreau4992dd22012-05-31 21:02:17 +02003962 httponly This option tells haproxy to add an "HttpOnly" cookie attribute
3963 when a cookie is inserted. This attribute is used so that a
3964 user agent doesn't share the cookie with non-HTTP components.
3965 Please check RFC6265 for more information on this attribute.
3966
3967 secure This option tells haproxy to add a "Secure" cookie attribute when
3968 a cookie is inserted. This attribute is used so that a user agent
3969 never emits this cookie over non-secure channels, which means
3970 that a cookie learned with this flag will be presented only over
3971 SSL/TLS connections. Please check RFC6265 for more information on
3972 this attribute.
3973
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02003974 domain This option allows to specify the domain at which a cookie is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003975 inserted. It requires exactly one parameter: a valid domain
Willy Tarreau68a897b2009-12-03 23:28:34 +01003976 name. If the domain begins with a dot, the browser is allowed to
3977 use it for any host ending with that name. It is also possible to
3978 specify several domain names by invoking this option multiple
3979 times. Some browsers might have small limits on the number of
3980 domains, so be careful when doing that. For the record, sending
3981 10 domains to MSIE 6 or Firefox 2 works as expected.
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02003982
Willy Tarreau996a92c2010-10-13 19:30:47 +02003983 maxidle This option allows inserted cookies to be ignored after some idle
3984 time. It only works with insert-mode cookies. When a cookie is
3985 sent to the client, the date this cookie was emitted is sent too.
3986 Upon further presentations of this cookie, if the date is older
3987 than the delay indicated by the parameter (in seconds), it will
3988 be ignored. Otherwise, it will be refreshed if needed when the
3989 response is sent to the client. This is particularly useful to
3990 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01003991 too long on the same server (e.g. after a farm size change). When
Willy Tarreau996a92c2010-10-13 19:30:47 +02003992 this option is set and a cookie has no date, it is always
3993 accepted, but gets refreshed in the response. This maintains the
3994 ability for admins to access their sites. Cookies that have a
3995 date in the future further than 24 hours are ignored. Doing so
3996 lets admins fix timezone issues without risking kicking users off
3997 the site.
3998
3999 maxlife This option allows inserted cookies to be ignored after some life
4000 time, whether they're in use or not. It only works with insert
4001 mode cookies. When a cookie is first sent to the client, the date
4002 this cookie was emitted is sent too. Upon further presentations
4003 of this cookie, if the date is older than the delay indicated by
4004 the parameter (in seconds), it will be ignored. If the cookie in
4005 the request has no date, it is accepted and a date will be set.
4006 Cookies that have a date in the future further than 24 hours are
4007 ignored. Doing so lets admins fix timezone issues without risking
4008 kicking users off the site. Contrary to maxidle, this value is
4009 not refreshed, only the first visit date counts. Both maxidle and
4010 maxlife may be used at the time. This is particularly useful to
4011 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01004012 too long on the same server (e.g. after a farm size change). This
Willy Tarreau996a92c2010-10-13 19:30:47 +02004013 is stronger than the maxidle method in that it forces a
4014 redispatch after some absolute delay.
4015
Olivier Houchard4e694042017-03-14 20:01:29 +01004016 dynamic Activate dynamic cookies. When used, a session cookie is
4017 dynamically created for each server, based on the IP and port
4018 of the server, and a secret key, specified in the
4019 "dynamic-cookie-key" backend directive.
4020 The cookie will be regenerated each time the IP address change,
4021 and is only generated for IPv4/IPv6.
4022
Christopher Faulet2f533902020-01-21 11:06:48 +01004023 attr This option tells haproxy to add an extra attribute when a
4024 cookie is inserted. The attribute value can contain any
4025 characters except control ones or ";". This option may be
4026 repeated.
4027
Willy Tarreau0ba27502007-12-24 16:55:16 +01004028 There can be only one persistence cookie per HTTP backend, and it can be
4029 declared in a defaults section. The value of the cookie will be the value
4030 indicated after the "cookie" keyword in a "server" statement. If no cookie
4031 is declared for a given server, the cookie is not set.
Willy Tarreau6a06a402007-07-15 20:15:28 +02004032
Willy Tarreau0ba27502007-12-24 16:55:16 +01004033 Examples :
4034 cookie JSESSIONID prefix
4035 cookie SRV insert indirect nocache
4036 cookie SRV insert postonly indirect
Willy Tarreau996a92c2010-10-13 19:30:47 +02004037 cookie SRV insert indirect nocache maxidle 30m maxlife 8h
Willy Tarreau0ba27502007-12-24 16:55:16 +01004038
Willy Tarreau294d0f02015-08-10 19:40:12 +02004039 See also : "balance source", "capture cookie", "server" and "ignore-persist".
Willy Tarreau0ba27502007-12-24 16:55:16 +01004040
Willy Tarreau983e01e2010-01-11 18:42:06 +01004041
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02004042declare capture [ request | response ] len <length>
4043 Declares a capture slot.
4044 May be used in sections : defaults | frontend | listen | backend
4045 no | yes | yes | no
4046 Arguments:
4047 <length> is the length allowed for the capture.
4048
4049 This declaration is only available in the frontend or listen section, but the
4050 reserved slot can be used in the backends. The "request" keyword allocates a
4051 capture slot for use in the request, and "response" allocates a capture slot
4052 for use in the response.
4053
4054 See also: "capture-req", "capture-res" (sample converters),
Baptiste Assmann5ac425c2015-10-21 23:13:46 +02004055 "capture.req.hdr", "capture.res.hdr" (sample fetches),
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02004056 "http-request capture" and "http-response capture".
4057
4058
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01004059default-server [param*]
4060 Change default options for a server in a backend
4061 May be used in sections : defaults | frontend | listen | backend
4062 yes | no | yes | yes
4063 Arguments:
Willy Tarreau983e01e2010-01-11 18:42:06 +01004064 <param*> is a list of parameters for this server. The "default-server"
4065 keyword accepts an important number of options and has a complete
4066 section dedicated to it. Please refer to section 5 for more
4067 details.
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01004068
Willy Tarreau983e01e2010-01-11 18:42:06 +01004069 Example :
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01004070 default-server inter 1000 weight 13
4071
4072 See also: "server" and section 5 about server options
Willy Tarreau0ba27502007-12-24 16:55:16 +01004073
Willy Tarreau983e01e2010-01-11 18:42:06 +01004074
Willy Tarreau0ba27502007-12-24 16:55:16 +01004075default_backend <backend>
4076 Specify the backend to use when no "use_backend" rule has been matched.
4077 May be used in sections : defaults | frontend | listen | backend
4078 yes | yes | yes | no
4079 Arguments :
4080 <backend> is the name of the backend to use.
4081
4082 When doing content-switching between frontend and backends using the
4083 "use_backend" keyword, it is often useful to indicate which backend will be
4084 used when no rule has matched. It generally is the dynamic backend which
4085 will catch all undetermined requests.
4086
Willy Tarreau0ba27502007-12-24 16:55:16 +01004087 Example :
4088
4089 use_backend dynamic if url_dyn
4090 use_backend static if url_css url_img extension_img
4091 default_backend dynamic
4092
Willy Tarreau98d04852015-05-26 12:18:29 +02004093 See also : "use_backend"
Willy Tarreau2769aa02007-12-27 18:26:09 +01004094
Willy Tarreau0ba27502007-12-24 16:55:16 +01004095
Baptiste Assmann27f51342013-10-09 06:51:49 +02004096description <string>
4097 Describe a listen, frontend or backend.
4098 May be used in sections : defaults | frontend | listen | backend
4099 no | yes | yes | yes
4100 Arguments : string
4101
4102 Allows to add a sentence to describe the related object in the HAProxy HTML
4103 stats page. The description will be printed on the right of the object name
4104 it describes.
4105 No need to backslash spaces in the <string> arguments.
4106
4107
Willy Tarreau0ba27502007-12-24 16:55:16 +01004108disabled
4109 Disable a proxy, frontend or backend.
4110 May be used in sections : defaults | frontend | listen | backend
4111 yes | yes | yes | yes
4112 Arguments : none
4113
4114 The "disabled" keyword is used to disable an instance, mainly in order to
4115 liberate a listening port or to temporarily disable a service. The instance
4116 will still be created and its configuration will be checked, but it will be
4117 created in the "stopped" state and will appear as such in the statistics. It
4118 will not receive any traffic nor will it send any health-checks or logs. It
4119 is possible to disable many instances at once by adding the "disabled"
4120 keyword in a "defaults" section.
4121
4122 See also : "enabled"
4123
4124
Willy Tarreau5ce94572010-06-07 14:35:41 +02004125dispatch <address>:<port>
4126 Set a default server address
4127 May be used in sections : defaults | frontend | listen | backend
4128 no | no | yes | yes
Cyril Bonté108cf6e2012-04-21 23:30:29 +02004129 Arguments :
Willy Tarreau5ce94572010-06-07 14:35:41 +02004130
4131 <address> is the IPv4 address of the default server. Alternatively, a
4132 resolvable hostname is supported, but this name will be resolved
4133 during start-up.
4134
4135 <ports> is a mandatory port specification. All connections will be sent
4136 to this port, and it is not permitted to use port offsets as is
4137 possible with normal servers.
4138
Willy Tarreau787aed52011-04-15 06:45:37 +02004139 The "dispatch" keyword designates a default server for use when no other
Willy Tarreau5ce94572010-06-07 14:35:41 +02004140 server can take the connection. In the past it was used to forward non
4141 persistent connections to an auxiliary load balancer. Due to its simple
4142 syntax, it has also been used for simple TCP relays. It is recommended not to
4143 use it for more clarity, and to use the "server" directive instead.
4144
4145 See also : "server"
4146
Olivier Houchard4e694042017-03-14 20:01:29 +01004147
4148dynamic-cookie-key <string>
4149 Set the dynamic cookie secret key for a backend.
4150 May be used in sections : defaults | frontend | listen | backend
4151 yes | no | yes | yes
4152 Arguments : The secret key to be used.
4153
4154 When dynamic cookies are enabled (see the "dynamic" directive for cookie),
Davor Ocelice9ed2812017-12-25 17:49:28 +01004155 a dynamic cookie is created for each server (unless one is explicitly
Olivier Houchard4e694042017-03-14 20:01:29 +01004156 specified on the "server" line), using a hash of the IP address of the
4157 server, the TCP port, and the secret key.
Davor Ocelice9ed2812017-12-25 17:49:28 +01004158 That way, we can ensure session persistence across multiple load-balancers,
Olivier Houchard4e694042017-03-14 20:01:29 +01004159 even if servers are dynamically added or removed.
Willy Tarreau5ce94572010-06-07 14:35:41 +02004160
Willy Tarreau0ba27502007-12-24 16:55:16 +01004161enabled
4162 Enable a proxy, frontend or backend.
4163 May be used in sections : defaults | frontend | listen | backend
4164 yes | yes | yes | yes
4165 Arguments : none
4166
4167 The "enabled" keyword is used to explicitly enable an instance, when the
4168 defaults has been set to "disabled". This is very rarely used.
4169
4170 See also : "disabled"
4171
4172
4173errorfile <code> <file>
4174 Return a file contents instead of errors generated by HAProxy
4175 May be used in sections : defaults | frontend | listen | backend
4176 yes | yes | yes | yes
4177 Arguments :
4178 <code> is the HTTP status code. Currently, HAProxy is capable of
Christopher Faulet612f2ea2020-05-27 09:57:28 +02004179 generating codes 200, 400, 401, 403, 404, 405, 407, 408, 410,
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02004180 413, 425, 429, 500, 502, 503, and 504.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004181
4182 <file> designates a file containing the full HTTP response. It is
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01004183 recommended to follow the common practice of appending ".http" to
Willy Tarreau0ba27502007-12-24 16:55:16 +01004184 the filename so that people do not confuse the response with HTML
Willy Tarreau59140a22009-02-22 12:02:30 +01004185 error pages, and to use absolute paths, since files are read
4186 before any chroot is performed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004187
4188 It is important to understand that this keyword is not meant to rewrite
4189 errors returned by the server, but errors detected and returned by HAProxy.
4190 This is why the list of supported errors is limited to a small set.
4191
Willy Tarreauae94d4d2011-05-11 16:28:49 +02004192 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
4193
Christopher Faulet70170672020-05-18 17:42:48 +02004194 The files are parsed when HAProxy starts and must be valid according to the
4195 HTTP specification. They should not exceed the configured buffer size
4196 (BUFSIZE), which generally is 16 kB, otherwise an internal error will be
4197 returned. It is also wise not to put any reference to local contents
4198 (e.g. images) in order to avoid loops between the client and HAProxy when all
4199 servers are down, causing an error to be returned instead of an
4200 image. Finally, The response cannot exceed (tune.bufsize - tune.maxrewrite)
4201 so that "http-after-response" rules still have room to operate (see
4202 "tune.maxrewrite").
Willy Tarreau59140a22009-02-22 12:02:30 +01004203
Willy Tarreau0ba27502007-12-24 16:55:16 +01004204 The files are read at the same time as the configuration and kept in memory.
4205 For this reason, the errors continue to be returned even when the process is
4206 chrooted, and no file change is considered while the process is running. A
Willy Tarreauc27debf2008-01-06 08:57:02 +01004207 simple method for developing those files consists in associating them to the
Willy Tarreau0ba27502007-12-24 16:55:16 +01004208 403 status code and interrogating a blocked URL.
4209
Christopher Faulet3b967c12020-05-15 15:47:44 +02004210 See also : "http-error", "errorloc", "errorloc302", "errorloc303"
Willy Tarreau0ba27502007-12-24 16:55:16 +01004211
Willy Tarreau59140a22009-02-22 12:02:30 +01004212 Example :
4213 errorfile 400 /etc/haproxy/errorfiles/400badreq.http
Willy Tarreau989222a2016-01-15 10:26:26 +01004214 errorfile 408 /dev/null # work around Chrome pre-connect bug
Willy Tarreau59140a22009-02-22 12:02:30 +01004215 errorfile 403 /etc/haproxy/errorfiles/403forbid.http
4216 errorfile 503 /etc/haproxy/errorfiles/503sorry.http
4217
Willy Tarreau2769aa02007-12-27 18:26:09 +01004218
Christopher Faulet76edc0f2020-01-13 15:52:01 +01004219errorfiles <name> [<code> ...]
4220 Import, fully or partially, the error files defined in the <name> http-errors
4221 section.
4222 May be used in sections : defaults | frontend | listen | backend
4223 yes | yes | yes | yes
4224 Arguments :
4225 <name> is the name of an existing http-errors section.
4226
4227 <code> is a HTTP status code. Several status code may be listed.
Christopher Faulet612f2ea2020-05-27 09:57:28 +02004228 Currently, HAProxy is capable of generating codes 200, 400, 401,
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02004229 403, 404, 405, 407, 408, 410, 413, 425, 429, 500, 502, 503, and 504.
Christopher Faulet76edc0f2020-01-13 15:52:01 +01004230
4231 Errors defined in the http-errors section with the name <name> are imported
4232 in the current proxy. If no status code is specified, all error files of the
4233 http-errors section are imported. Otherwise, only error files associated to
4234 the listed status code are imported. Those error files override the already
4235 defined custom errors for the proxy. And they may be overridden by following
Daniel Corbett67a82712020-07-06 23:01:19 -04004236 ones. Functionally, it is exactly the same as declaring all error files by
Christopher Faulet76edc0f2020-01-13 15:52:01 +01004237 hand using "errorfile" directives.
4238
Christopher Faulet3b967c12020-05-15 15:47:44 +02004239 See also : "http-error", "errorfile", "errorloc", "errorloc302" ,
4240 "errorloc303" and section 3.8 about http-errors.
Christopher Faulet76edc0f2020-01-13 15:52:01 +01004241
4242 Example :
4243 errorfiles generic
4244 errorfiles site-1 403 404
4245
4246
Willy Tarreau2769aa02007-12-27 18:26:09 +01004247errorloc <code> <url>
4248errorloc302 <code> <url>
4249 Return an HTTP redirection to a URL instead of errors generated by HAProxy
4250 May be used in sections : defaults | frontend | listen | backend
4251 yes | yes | yes | yes
4252 Arguments :
4253 <code> is the HTTP status code. Currently, HAProxy is capable of
Christopher Faulet612f2ea2020-05-27 09:57:28 +02004254 generating codes 200, 400, 401, 403, 404, 405, 407, 408, 410,
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02004255 413, 425, 429, 500, 502, 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01004256
4257 <url> it is the exact contents of the "Location" header. It may contain
4258 either a relative URI to an error page hosted on the same site,
4259 or an absolute URI designating an error page on another site.
4260 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01004261 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01004262
4263 It is important to understand that this keyword is not meant to rewrite
4264 errors returned by the server, but errors detected and returned by HAProxy.
4265 This is why the list of supported errors is limited to a small set.
4266
Willy Tarreauae94d4d2011-05-11 16:28:49 +02004267 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
4268
Willy Tarreau2769aa02007-12-27 18:26:09 +01004269 Note that both keyword return the HTTP 302 status code, which tells the
4270 client to fetch the designated URL using the same HTTP method. This can be
4271 quite problematic in case of non-GET methods such as POST, because the URL
4272 sent to the client might not be allowed for something other than GET. To
Willy Tarreau989222a2016-01-15 10:26:26 +01004273 work around this problem, please use "errorloc303" which send the HTTP 303
Willy Tarreau2769aa02007-12-27 18:26:09 +01004274 status code, indicating to the client that the URL must be fetched with a GET
4275 request.
4276
Christopher Faulet3b967c12020-05-15 15:47:44 +02004277 See also : "http-error", "errorfile", "errorloc303"
Willy Tarreau2769aa02007-12-27 18:26:09 +01004278
4279
4280errorloc303 <code> <url>
4281 Return an HTTP redirection to a URL instead of errors generated by HAProxy
4282 May be used in sections : defaults | frontend | listen | backend
4283 yes | yes | yes | yes
4284 Arguments :
4285 <code> is the HTTP status code. Currently, HAProxy is capable of
Christopher Faulet612f2ea2020-05-27 09:57:28 +02004286 generating codes 200, 400, 401, 403, 404, 405, 407, 408, 410,
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02004287 413, 425, 429, 500, 502, 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01004288
4289 <url> it is the exact contents of the "Location" header. It may contain
4290 either a relative URI to an error page hosted on the same site,
4291 or an absolute URI designating an error page on another site.
4292 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01004293 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01004294
4295 It is important to understand that this keyword is not meant to rewrite
4296 errors returned by the server, but errors detected and returned by HAProxy.
4297 This is why the list of supported errors is limited to a small set.
4298
Willy Tarreauae94d4d2011-05-11 16:28:49 +02004299 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
4300
Willy Tarreau2769aa02007-12-27 18:26:09 +01004301 Note that both keyword return the HTTP 303 status code, which tells the
4302 client to fetch the designated URL using the same HTTP GET method. This
4303 solves the usual problems associated with "errorloc" and the 302 code. It is
4304 possible that some very old browsers designed before HTTP/1.1 do not support
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01004305 it, but no such problem has been reported till now.
Willy Tarreau2769aa02007-12-27 18:26:09 +01004306
Christopher Faulet3b967c12020-05-15 15:47:44 +02004307 See also : "http-error", "errorfile", "errorloc", "errorloc302"
Willy Tarreau2769aa02007-12-27 18:26:09 +01004308
4309
Simon Horman51a1cf62015-02-03 13:00:44 +09004310email-alert from <emailaddr>
4311 Declare the from email address to be used in both the envelope and header
Davor Ocelice9ed2812017-12-25 17:49:28 +01004312 of email alerts. This is the address that email alerts are sent from.
Simon Horman51a1cf62015-02-03 13:00:44 +09004313 May be used in sections: defaults | frontend | listen | backend
4314 yes | yes | yes | yes
4315
4316 Arguments :
4317
4318 <emailaddr> is the from email address to use when sending email alerts
4319
4320 Also requires "email-alert mailers" and "email-alert to" to be set
4321 and if so sending email alerts is enabled for the proxy.
4322
Simon Horman64e34162015-02-06 11:11:57 +09004323 See also : "email-alert level", "email-alert mailers",
Cyril Bonté307ee1e2015-09-28 23:16:06 +02004324 "email-alert myhostname", "email-alert to", section 3.6 about
4325 mailers.
Simon Horman64e34162015-02-06 11:11:57 +09004326
4327
4328email-alert level <level>
4329 Declare the maximum log level of messages for which email alerts will be
4330 sent. This acts as a filter on the sending of email alerts.
4331 May be used in sections: defaults | frontend | listen | backend
4332 yes | yes | yes | yes
4333
4334 Arguments :
4335
4336 <level> One of the 8 syslog levels:
4337 emerg alert crit err warning notice info debug
4338 The above syslog levels are ordered from lowest to highest.
4339
4340 By default level is alert
4341
4342 Also requires "email-alert from", "email-alert mailers" and
4343 "email-alert to" to be set and if so sending email alerts is enabled
4344 for the proxy.
4345
Simon Horman1421e212015-04-30 13:10:35 +09004346 Alerts are sent when :
4347
4348 * An un-paused server is marked as down and <level> is alert or lower
4349 * A paused server is marked as down and <level> is notice or lower
4350 * A server is marked as up or enters the drain state and <level>
4351 is notice or lower
4352 * "option log-health-checks" is enabled, <level> is info or lower,
4353 and a health check status update occurs
4354
Simon Horman64e34162015-02-06 11:11:57 +09004355 See also : "email-alert from", "email-alert mailers",
4356 "email-alert myhostname", "email-alert to",
Simon Horman51a1cf62015-02-03 13:00:44 +09004357 section 3.6 about mailers.
4358
4359
4360email-alert mailers <mailersect>
4361 Declare the mailers to be used when sending email alerts
4362 May be used in sections: defaults | frontend | listen | backend
4363 yes | yes | yes | yes
4364
4365 Arguments :
4366
4367 <mailersect> is the name of the mailers section to send email alerts.
4368
4369 Also requires "email-alert from" and "email-alert to" to be set
4370 and if so sending email alerts is enabled for the proxy.
4371
Simon Horman64e34162015-02-06 11:11:57 +09004372 See also : "email-alert from", "email-alert level", "email-alert myhostname",
4373 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09004374
4375
4376email-alert myhostname <hostname>
4377 Declare the to hostname address to be used when communicating with
4378 mailers.
4379 May be used in sections: defaults | frontend | listen | backend
4380 yes | yes | yes | yes
4381
4382 Arguments :
4383
Baptiste Assmann738bad92015-12-21 15:27:53 +01004384 <hostname> is the hostname to use when communicating with mailers
Simon Horman51a1cf62015-02-03 13:00:44 +09004385
4386 By default the systems hostname is used.
4387
4388 Also requires "email-alert from", "email-alert mailers" and
4389 "email-alert to" to be set and if so sending email alerts is enabled
4390 for the proxy.
4391
Simon Horman64e34162015-02-06 11:11:57 +09004392 See also : "email-alert from", "email-alert level", "email-alert mailers",
4393 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09004394
4395
4396email-alert to <emailaddr>
Davor Ocelice9ed2812017-12-25 17:49:28 +01004397 Declare both the recipient address in the envelope and to address in the
Simon Horman51a1cf62015-02-03 13:00:44 +09004398 header of email alerts. This is the address that email alerts are sent to.
4399 May be used in sections: defaults | frontend | listen | backend
4400 yes | yes | yes | yes
4401
4402 Arguments :
4403
4404 <emailaddr> is the to email address to use when sending email alerts
4405
4406 Also requires "email-alert mailers" and "email-alert to" to be set
4407 and if so sending email alerts is enabled for the proxy.
4408
Simon Horman64e34162015-02-06 11:11:57 +09004409 See also : "email-alert from", "email-alert level", "email-alert mailers",
Simon Horman51a1cf62015-02-03 13:00:44 +09004410 "email-alert myhostname", section 3.6 about mailers.
4411
4412
Willy Tarreau4de91492010-01-22 19:10:05 +01004413force-persist { if | unless } <condition>
4414 Declare a condition to force persistence on down servers
4415 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01004416 no | no | yes | yes
Willy Tarreau4de91492010-01-22 19:10:05 +01004417
4418 By default, requests are not dispatched to down servers. It is possible to
4419 force this using "option persist", but it is unconditional and redispatches
4420 to a valid server if "option redispatch" is set. That leaves with very little
4421 possibilities to force some requests to reach a server which is artificially
4422 marked down for maintenance operations.
4423
4424 The "force-persist" statement allows one to declare various ACL-based
4425 conditions which, when met, will cause a request to ignore the down status of
4426 a server and still try to connect to it. That makes it possible to start a
4427 server, still replying an error to the health checks, and run a specially
4428 configured browser to test the service. Among the handy methods, one could
4429 use a specific source IP address, or a specific cookie. The cookie also has
4430 the advantage that it can easily be added/removed on the browser from a test
4431 page. Once the service is validated, it is then possible to open the service
4432 to the world by returning a valid response to health checks.
4433
4434 The forced persistence is enabled when an "if" condition is met, or unless an
4435 "unless" condition is met. The final redispatch is always disabled when this
4436 is used.
4437
Cyril Bonté0d4bf012010-04-25 23:21:46 +02004438 See also : "option redispatch", "ignore-persist", "persist",
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +02004439 and section 7 about ACL usage.
Willy Tarreau4de91492010-01-22 19:10:05 +01004440
Christopher Fauletc3fe5332016-04-07 15:30:10 +02004441
4442filter <name> [param*]
4443 Add the filter <name> in the filter list attached to the proxy.
4444 May be used in sections : defaults | frontend | listen | backend
4445 no | yes | yes | yes
4446 Arguments :
4447 <name> is the name of the filter. Officially supported filters are
4448 referenced in section 9.
4449
Tim Düsterhus4896c442016-11-29 02:15:19 +01004450 <param*> is a list of parameters accepted by the filter <name>. The
Christopher Fauletc3fe5332016-04-07 15:30:10 +02004451 parsing of these parameters are the responsibility of the
Tim Düsterhus4896c442016-11-29 02:15:19 +01004452 filter. Please refer to the documentation of the corresponding
4453 filter (section 9) for all details on the supported parameters.
Christopher Fauletc3fe5332016-04-07 15:30:10 +02004454
4455 Multiple occurrences of the filter line can be used for the same proxy. The
4456 same filter can be referenced many times if needed.
4457
4458 Example:
4459 listen
4460 bind *:80
4461
4462 filter trace name BEFORE-HTTP-COMP
4463 filter compression
4464 filter trace name AFTER-HTTP-COMP
4465
4466 compression algo gzip
4467 compression offload
4468
4469 server srv1 192.168.0.1:80
4470
4471 See also : section 9.
4472
Willy Tarreau4de91492010-01-22 19:10:05 +01004473
Willy Tarreau2769aa02007-12-27 18:26:09 +01004474fullconn <conns>
4475 Specify at what backend load the servers will reach their maxconn
4476 May be used in sections : defaults | frontend | listen | backend
4477 yes | no | yes | yes
4478 Arguments :
4479 <conns> is the number of connections on the backend which will make the
4480 servers use the maximal number of connections.
4481
Willy Tarreau198a7442008-01-17 12:05:32 +01004482 When a server has a "maxconn" parameter specified, it means that its number
Willy Tarreau2769aa02007-12-27 18:26:09 +01004483 of concurrent connections will never go higher. Additionally, if it has a
Willy Tarreau198a7442008-01-17 12:05:32 +01004484 "minconn" parameter, it indicates a dynamic limit following the backend's
Willy Tarreau2769aa02007-12-27 18:26:09 +01004485 load. The server will then always accept at least <minconn> connections,
4486 never more than <maxconn>, and the limit will be on the ramp between both
4487 values when the backend has less than <conns> concurrent connections. This
4488 makes it possible to limit the load on the servers during normal loads, but
4489 push it further for important loads without overloading the servers during
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004490 exceptional loads.
Willy Tarreau2769aa02007-12-27 18:26:09 +01004491
Willy Tarreaufbb78422011-06-05 15:38:35 +02004492 Since it's hard to get this value right, haproxy automatically sets it to
4493 10% of the sum of the maxconns of all frontends that may branch to this
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01004494 backend (based on "use_backend" and "default_backend" rules). That way it's
4495 safe to leave it unset. However, "use_backend" involving dynamic names are
4496 not counted since there is no way to know if they could match or not.
Willy Tarreaufbb78422011-06-05 15:38:35 +02004497
Willy Tarreau2769aa02007-12-27 18:26:09 +01004498 Example :
4499 # The servers will accept between 100 and 1000 concurrent connections each
4500 # and the maximum of 1000 will be reached when the backend reaches 10000
4501 # connections.
4502 backend dynamic
4503 fullconn 10000
4504 server srv1 dyn1:80 minconn 100 maxconn 1000
4505 server srv2 dyn2:80 minconn 100 maxconn 1000
4506
4507 See also : "maxconn", "server"
4508
4509
Willy Tarreauab0a5192020-10-09 19:07:01 +02004510grace <time> (deprecated)
Willy Tarreau2769aa02007-12-27 18:26:09 +01004511 Maintain a proxy operational for some time after a soft stop
4512 May be used in sections : defaults | frontend | listen | backend
Cyril Bonté99ed3272010-01-24 23:29:44 +01004513 yes | yes | yes | yes
Willy Tarreau2769aa02007-12-27 18:26:09 +01004514 Arguments :
4515 <time> is the time (by default in milliseconds) for which the instance
4516 will remain operational with the frontend sockets still listening
4517 when a soft-stop is received via the SIGUSR1 signal.
4518
4519 This may be used to ensure that the services disappear in a certain order.
4520 This was designed so that frontends which are dedicated to monitoring by an
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004521 external equipment fail immediately while other ones remain up for the time
Willy Tarreau2769aa02007-12-27 18:26:09 +01004522 needed by the equipment to detect the failure.
4523
4524 Note that currently, there is very little benefit in using this parameter,
4525 and it may in fact complicate the soft-reconfiguration process more than
4526 simplify it.
4527
Willy Tarreau0ba27502007-12-24 16:55:16 +01004528
Andrew Rodland17be45e2016-10-25 17:04:12 -04004529hash-balance-factor <factor>
4530 Specify the balancing factor for bounded-load consistent hashing
4531 May be used in sections : defaults | frontend | listen | backend
4532 yes | no | no | yes
4533 Arguments :
4534 <factor> is the control for the maximum number of concurrent requests to
4535 send to a server, expressed as a percentage of the average number
Frédéric Lécaille93d33162019-03-06 09:35:59 +01004536 of concurrent requests across all of the active servers.
Andrew Rodland17be45e2016-10-25 17:04:12 -04004537
4538 Specifying a "hash-balance-factor" for a server with "hash-type consistent"
4539 enables an algorithm that prevents any one server from getting too many
4540 requests at once, even if some hash buckets receive many more requests than
4541 others. Setting <factor> to 0 (the default) disables the feature. Otherwise,
4542 <factor> is a percentage greater than 100. For example, if <factor> is 150,
4543 then no server will be allowed to have a load more than 1.5 times the average.
4544 If server weights are used, they will be respected.
4545
4546 If the first-choice server is disqualified, the algorithm will choose another
4547 server based on the request hash, until a server with additional capacity is
4548 found. A higher <factor> allows more imbalance between the servers, while a
4549 lower <factor> means that more servers will be checked on average, affecting
4550 performance. Reasonable values are from 125 to 200.
4551
Willy Tarreau760e81d2018-05-03 07:20:40 +02004552 This setting is also used by "balance random" which internally relies on the
4553 consistent hashing mechanism.
4554
Andrew Rodland17be45e2016-10-25 17:04:12 -04004555 See also : "balance" and "hash-type".
4556
4557
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004558hash-type <method> <function> <modifier>
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004559 Specify a method to use for mapping hashes to servers
4560 May be used in sections : defaults | frontend | listen | backend
4561 yes | no | yes | yes
4562 Arguments :
Bhaskar98634f02013-10-29 23:30:51 -04004563 <method> is the method used to select a server from the hash computed by
4564 the <function> :
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004565
Bhaskar98634f02013-10-29 23:30:51 -04004566 map-based the hash table is a static array containing all alive servers.
4567 The hashes will be very smooth, will consider weights, but
4568 will be static in that weight changes while a server is up
4569 will be ignored. This means that there will be no slow start.
4570 Also, since a server is selected by its position in the array,
4571 most mappings are changed when the server count changes. This
4572 means that when a server goes up or down, or when a server is
4573 added to a farm, most connections will be redistributed to
4574 different servers. This can be inconvenient with caches for
4575 instance.
Willy Tarreau798a39c2010-11-24 15:04:29 +01004576
Bhaskar98634f02013-10-29 23:30:51 -04004577 consistent the hash table is a tree filled with many occurrences of each
4578 server. The hash key is looked up in the tree and the closest
4579 server is chosen. This hash is dynamic, it supports changing
4580 weights while the servers are up, so it is compatible with the
4581 slow start feature. It has the advantage that when a server
4582 goes up or down, only its associations are moved. When a
4583 server is added to the farm, only a few part of the mappings
4584 are redistributed, making it an ideal method for caches.
4585 However, due to its principle, the distribution will never be
4586 very smooth and it may sometimes be necessary to adjust a
4587 server's weight or its ID to get a more balanced distribution.
4588 In order to get the same distribution on multiple load
4589 balancers, it is important that all servers have the exact
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004590 same IDs. Note: consistent hash uses sdbm and avalanche if no
4591 hash function is specified.
Bhaskar98634f02013-10-29 23:30:51 -04004592
4593 <function> is the hash function to be used :
4594
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03004595 sdbm this function was created initially for sdbm (a public-domain
Bhaskar98634f02013-10-29 23:30:51 -04004596 reimplementation of ndbm) database library. It was found to do
4597 well in scrambling bits, causing better distribution of the keys
4598 and fewer splits. It also happens to be a good general hashing
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004599 function with good distribution, unless the total server weight
4600 is a multiple of 64, in which case applying the avalanche
4601 modifier may help.
Bhaskar98634f02013-10-29 23:30:51 -04004602
4603 djb2 this function was first proposed by Dan Bernstein many years ago
4604 on comp.lang.c. Studies have shown that for certain workload this
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004605 function provides a better distribution than sdbm. It generally
4606 works well with text-based inputs though it can perform extremely
4607 poorly with numeric-only input or when the total server weight is
4608 a multiple of 33, unless the avalanche modifier is also used.
4609
Willy Tarreaua0f42712013-11-14 14:30:35 +01004610 wt6 this function was designed for haproxy while testing other
4611 functions in the past. It is not as smooth as the other ones, but
4612 is much less sensible to the input data set or to the number of
4613 servers. It can make sense as an alternative to sdbm+avalanche or
4614 djb2+avalanche for consistent hashing or when hashing on numeric
4615 data such as a source IP address or a visitor identifier in a URL
4616 parameter.
4617
Willy Tarreau324f07f2015-01-20 19:44:50 +01004618 crc32 this is the most common CRC32 implementation as used in Ethernet,
4619 gzip, PNG, etc. It is slower than the other ones but may provide
4620 a better distribution or less predictable results especially when
4621 used on strings.
4622
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004623 <modifier> indicates an optional method applied after hashing the key :
4624
4625 avalanche This directive indicates that the result from the hash
4626 function above should not be used in its raw form but that
4627 a 4-byte full avalanche hash must be applied first. The
4628 purpose of this step is to mix the resulting bits from the
4629 previous hash in order to avoid any undesired effect when
4630 the input contains some limited values or when the number of
4631 servers is a multiple of one of the hash's components (64
4632 for SDBM, 33 for DJB2). Enabling avalanche tends to make the
4633 result less predictable, but it's also not as smooth as when
4634 using the original function. Some testing might be needed
4635 with some workloads. This hash is one of the many proposed
4636 by Bob Jenkins.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004637
Bhaskar98634f02013-10-29 23:30:51 -04004638 The default hash type is "map-based" and is recommended for most usages. The
4639 default function is "sdbm", the selection of a function should be based on
4640 the range of the values being hashed.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004641
Andrew Rodland17be45e2016-10-25 17:04:12 -04004642 See also : "balance", "hash-balance-factor", "server"
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004643
4644
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01004645http-after-response <action> <options...> [ { if | unless } <condition> ]
4646 Access control for all Layer 7 responses (server, applet/service and internal
4647 ones).
4648
4649 May be used in sections: defaults | frontend | listen | backend
4650 no | yes | yes | yes
4651
4652 The http-after-response statement defines a set of rules which apply to layer
4653 7 processing. The rules are evaluated in their declaration order when they
4654 are met in a frontend, listen or backend section. Any rule may optionally be
4655 followed by an ACL-based condition, in which case it will only be evaluated
4656 if the condition is true. Since these rules apply on responses, the backend
4657 rules are applied first, followed by the frontend's rules.
4658
4659 Unlike http-response rules, these ones are applied on all responses, the
4660 server ones but also to all responses generated by HAProxy. These rules are
4661 evaluated at the end of the responses analysis, before the data forwarding.
4662
4663 The first keyword is the rule's action. The supported actions are described
4664 below.
4665
4666 There is no limit to the number of http-after-response statements per
4667 instance.
4668
4669 Example:
4670 http-after-response set-header Strict-Transport-Security "max-age=31536000"
4671 http-after-response set-header Cache-Control "no-store,no-cache,private"
4672 http-after-response set-header Pragma "no-cache"
4673
4674http-after-response add-header <name> <fmt> [ { if | unless } <condition> ]
4675
4676 This appends an HTTP header field whose name is specified in <name> and whose
4677 value is defined by <fmt> which follows the log-format rules (see Custom Log
4678 Format in section 8.2.4). This may be used to send a cookie to a client for
4679 example, or to pass some internal information.
4680 This rule is not final, so it is possible to add other similar rules.
4681 Note that header addition is performed immediately, so one rule might reuse
4682 the resulting header from a previous rule.
4683
4684http-after-response allow [ { if | unless } <condition> ]
4685
4686 This stops the evaluation of the rules and lets the response pass the check.
4687 No further "http-after-response" rules are evaluated.
4688
4689http-after-response del-header <name> [ { if | unless } <condition> ]
4690
4691 This removes all HTTP header fields whose name is specified in <name>.
4692
4693http-after-response replace-header <name> <regex-match> <replace-fmt>
4694 [ { if | unless } <condition> ]
4695
4696 This works like "http-response replace-header".
4697
4698 Example:
4699 http-after-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2
4700
4701 # applied to:
4702 Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT
4703
4704 # outputs:
4705 Set-Cookie: C=1;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT
4706
4707 # assuming the backend IP is 192.168.1.20.
4708
4709http-after-response replace-value <name> <regex-match> <replace-fmt>
4710 [ { if | unless } <condition> ]
4711
4712 This works like "http-response replace-value".
4713
4714 Example:
4715 http-after-response replace-value Cache-control ^public$ private
4716
4717 # applied to:
4718 Cache-Control: max-age=3600, public
4719
4720 # outputs:
4721 Cache-Control: max-age=3600, private
4722
4723http-after-response set-header <name> <fmt> [ { if | unless } <condition> ]
4724
4725 This does the same as "add-header" except that the header name is first
4726 removed if it existed. This is useful when passing security information to
4727 the server, where the header must not be manipulated by external users.
4728
4729http-after-response set-status <status> [reason <str>]
4730 [ { if | unless } <condition> ]
4731
4732 This replaces the response status code with <status> which must be an integer
4733 between 100 and 999. Optionally, a custom reason text can be provided defined
4734 by <str>, or the default reason for the specified code will be used as a
4735 fallback.
4736
4737 Example:
4738 # return "431 Request Header Fields Too Large"
4739 http-response set-status 431
4740 # return "503 Slow Down", custom reason
4741 http-response set-status 503 reason "Slow Down"
4742
4743http-after-response set-var(<var-name>) <expr> [ { if | unless } <condition> ]
4744
4745 This is used to set the contents of a variable. The variable is declared
4746 inline.
4747
4748 Arguments:
4749 <var-name> The name of the variable starts with an indication about its
4750 scope. The scopes allowed are:
4751 "proc" : the variable is shared with the whole process
4752 "sess" : the variable is shared with the whole session
4753 "txn" : the variable is shared with the transaction
4754 (request and response)
4755 "req" : the variable is shared only during request
4756 processing
4757 "res" : the variable is shared only during response
4758 processing
4759 This prefix is followed by a name. The separator is a '.'.
4760 The name may only contain characters 'a-z', 'A-Z', '0-9', '.'
4761 and '_'.
4762
4763 <expr> Is a standard HAProxy expression formed by a sample-fetch
4764 followed by some converters.
4765
4766 Example:
4767 http-after-response set-var(sess.last_redir) res.hdr(location)
4768
4769http-after-response strict-mode { on | off }
4770
4771 This enables or disables the strict rewriting mode for following rules. It
4772 does not affect rules declared before it and it is only applicable on rules
4773 performing a rewrite on the responses. When the strict mode is enabled, any
4774 rewrite failure triggers an internal error. Otherwise, such errors are
4775 silently ignored. The purpose of the strict rewriting mode is to make some
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05004776 rewrites optional while others must be performed to continue the response
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01004777 processing.
4778
4779 By default, the strict rewriting mode is enabled. Its value is also reset
4780 when a ruleset evaluation ends. So, for instance, if you change the mode on
Daniel Corbett67a82712020-07-06 23:01:19 -04004781 the backend, the default mode is restored when HAProxy starts the frontend
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01004782 rules evaluation.
4783
4784http-after-response unset-var(<var-name>) [ { if | unless } <condition> ]
4785
4786 This is used to unset a variable. See "http-after-response set-var" for
4787 details about <var-name>.
4788
4789 Example:
4790 http-after-response unset-var(sess.last_redir)
4791
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02004792
4793http-check comment <string>
4794 Defines a comment for the following the http-check rule, reported in logs if
4795 it fails.
4796 May be used in sections : defaults | frontend | listen | backend
4797 yes | no | yes | yes
4798
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02004799 Arguments :
4800 <string> is the comment message to add in logs if the following http-check
4801 rule fails.
4802
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02004803 It only works for connect, send and expect rules. It is useful to make
4804 user-friendly error reporting.
4805
Daniel Corbett67a82712020-07-06 23:01:19 -04004806 See also : "option httpchk", "http-check connect", "http-check send" and
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02004807 "http-check expect".
4808
4809
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02004810http-check connect [default] [port <expr>] [addr <ip>] [send-proxy]
4811 [via-socks4] [ssl] [sni <sni>] [alpn <alpn>] [linger]
Christopher Fauletedc6ed92020-04-23 16:27:59 +02004812 [proto <name>] [comment <msg>]
Christopher Faulete5870d82020-04-15 11:32:03 +02004813 Opens a new connection to perform an HTTP health check
4814 May be used in sections : defaults | frontend | listen | backend
4815 yes | no | yes | yes
4816
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02004817 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02004818 comment <msg> defines a message to report if the rule evaluation fails.
4819
Christopher Faulete5870d82020-04-15 11:32:03 +02004820 default Use default options of the server line to do the health
Daniel Corbett67a82712020-07-06 23:01:19 -04004821 checks. The server options are used only if not redefined.
Christopher Faulete5870d82020-04-15 11:32:03 +02004822
4823 port <expr> if not set, check port or server port is used.
4824 It tells HAProxy where to open the connection to.
4825 <port> must be a valid TCP port source integer, from 1 to
4826 65535 or an sample-fetch expression.
4827
4828 addr <ip> defines the IP address to do the health check.
4829
4830 send-proxy send a PROXY protocol string
4831
4832 via-socks4 enables outgoing health checks using upstream socks4 proxy.
4833
4834 ssl opens a ciphered connection
4835
4836 sni <sni> specifies the SNI to use to do health checks over SSL.
4837
4838 alpn <alpn> defines which protocols to advertise with ALPN. The protocol
4839 list consists in a comma-delimited list of protocol names,
4840 for instance: "h2,http/1.1". If it is not set, the server ALPN
4841 is used.
4842
Christopher Fauletedc6ed92020-04-23 16:27:59 +02004843 proto <name> forces the multiplexer's protocol to use for this connection.
4844 It must be an HTTP mux protocol and it must be usable on the
4845 backend side. The list of available protocols is reported in
4846 haproxy -vv.
4847
Christopher Faulete5870d82020-04-15 11:32:03 +02004848 linger cleanly close the connection instead of using a single RST.
4849
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02004850 Just like tcp-check health checks, it is possible to configure the connection
4851 to use to perform HTTP health check. This directive should also be used to
4852 describe a scenario involving several request/response exchanges, possibly on
4853 different ports or with different servers.
4854
4855 When there are no TCP port configured on the server line neither server port
4856 directive, then the first step of the http-check sequence must be to specify
4857 the port with a "http-check connect".
4858
4859 In an http-check ruleset a 'connect' is required, it is also mandatory to start
4860 the ruleset with a 'connect' rule. Purpose is to ensure admin know what they
4861 do.
4862
4863 When a connect must start the ruleset, if may still be preceded by set-var,
4864 unset-var or comment rules.
4865
4866 Examples :
Christopher Faulete5870d82020-04-15 11:32:03 +02004867 # check HTTP and HTTPs services on a server.
4868 # first open port 80 thanks to server line port directive, then
4869 # tcp-check opens port 443, ciphered and run a request on it:
4870 option httpchk
4871
4872 http-check connect
Christopher Fauleta5c14ef2020-04-29 14:19:13 +02004873 http-check send meth GET uri / ver HTTP/1.1 hdr host haproxy.1wt.eu
Christopher Faulet8021a5f2020-04-24 13:53:12 +02004874 http-check expect status 200-399
Christopher Faulete5870d82020-04-15 11:32:03 +02004875 http-check connect port 443 ssl sni haproxy.1wt.eu
Christopher Fauleta5c14ef2020-04-29 14:19:13 +02004876 http-check send meth GET uri / ver HTTP/1.1 hdr host haproxy.1wt.eu
Christopher Faulet8021a5f2020-04-24 13:53:12 +02004877 http-check expect status 200-399
Christopher Faulete5870d82020-04-15 11:32:03 +02004878
4879 server www 10.0.0.1 check port 80
4880
4881 See also : "option httpchk", "http-check send", "http-check expect"
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01004882
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02004883
Willy Tarreau0ba27502007-12-24 16:55:16 +01004884http-check disable-on-404
4885 Enable a maintenance mode upon HTTP/404 response to health-checks
4886 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau2769aa02007-12-27 18:26:09 +01004887 yes | no | yes | yes
Willy Tarreau0ba27502007-12-24 16:55:16 +01004888 Arguments : none
4889
4890 When this option is set, a server which returns an HTTP code 404 will be
4891 excluded from further load-balancing, but will still receive persistent
4892 connections. This provides a very convenient method for Web administrators
4893 to perform a graceful shutdown of their servers. It is also important to note
4894 that a server which is detected as failed while it was in this mode will not
4895 generate an alert, just a notice. If the server responds 2xx or 3xx again, it
4896 will immediately be reinserted into the farm. The status on the stats page
4897 reports "NOLB" for a server in this mode. It is important to note that this
Willy Tarreaubd741542010-03-16 18:46:54 +01004898 option only works in conjunction with the "httpchk" option. If this option
4899 is used with "http-check expect", then it has precedence over it so that 404
4900 responses will still be considered as soft-stop.
4901
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02004902 See also : "option httpchk" and "http-check expect".
Willy Tarreaubd741542010-03-16 18:46:54 +01004903
4904
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02004905http-check expect [min-recv <int>] [comment <msg>]
Christopher Faulete5870d82020-04-15 11:32:03 +02004906 [ok-status <st>] [error-status <st>] [tout-status <st>]
4907 [on-success <fmt>] [on-error <fmt>] [status-code <expr>]
4908 [!] <match> <pattern>
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004909 Make HTTP health checks consider response contents or specific status codes
Willy Tarreaubd741542010-03-16 18:46:54 +01004910 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau1ee51a62011-08-19 20:04:17 +02004911 yes | no | yes | yes
Christopher Faulete5870d82020-04-15 11:32:03 +02004912
Willy Tarreaubd741542010-03-16 18:46:54 +01004913 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02004914 comment <msg> defines a message to report if the rule evaluation fails.
4915
Christopher Faulete5870d82020-04-15 11:32:03 +02004916 min-recv is optional and can define the minimum amount of data required to
4917 evaluate the current expect rule. If the number of received bytes
4918 is under this limit, the check will wait for more data. This
4919 option can be used to resolve some ambiguous matching rules or to
4920 avoid executing costly regex matches on content known to be still
4921 incomplete. If an exact string is used, the minimum between the
4922 string length and this parameter is used. This parameter is
4923 ignored if it is set to -1. If the expect rule does not match,
4924 the check will wait for more data. If set to 0, the evaluation
4925 result is always conclusive.
4926
4927 ok-status <st> is optional and can be used to set the check status if
4928 the expect rule is successfully evaluated and if it is
4929 the last rule in the tcp-check ruleset. "L7OK", "L7OKC",
Christopher Fauletd888f0f2020-05-07 07:40:17 +02004930 "L6OK" and "L4OK" are supported :
4931 - L7OK : check passed on layer 7
4932 - L7OKC : check conditionally passed on layer 7, for
4933 example 404 with disable-on-404
4934 - L6OK : check passed on layer 6
4935 - L4OK : check passed on layer 4
4936 By default "L7OK" is used.
Christopher Faulete5870d82020-04-15 11:32:03 +02004937
4938 error-status <st> is optional and can be used to set the check status if
4939 an error occurred during the expect rule evaluation.
Christopher Fauletd888f0f2020-05-07 07:40:17 +02004940 "L7RSP", "L7STS", "L6RSP" and "L4CON" are supported :
4941 - L7RSP : layer 7 invalid response - protocol error
4942 - L7STS : layer 7 response error, for example HTTP 5xx
4943 - L6RSP : layer 6 invalid response - protocol error
4944 - L4CON : layer 1-4 connection problem
4945 By default "L7RSP" is used.
Christopher Faulete5870d82020-04-15 11:32:03 +02004946
4947 tout-status <st> is optional and can be used to set the check status if
4948 a timeout occurred during the expect rule evaluation.
Christopher Fauletd888f0f2020-05-07 07:40:17 +02004949 "L7TOUT", "L6TOUT", and "L4TOUT" are supported :
4950 - L7TOUT : layer 7 (HTTP/SMTP) timeout
4951 - L6TOUT : layer 6 (SSL) timeout
4952 - L4TOUT : layer 1-4 timeout
Christopher Faulete5870d82020-04-15 11:32:03 +02004953 By default "L7TOUT" is used.
4954
4955 on-success <fmt> is optional and can be used to customize the
4956 informational message reported in logs if the expect
4957 rule is successfully evaluated and if it is the last rule
4958 in the tcp-check ruleset. <fmt> is a log-format string.
4959
4960 on-error <fmt> is optional and can be used to customize the
4961 informational message reported in logs if an error
4962 occurred during the expect rule evaluation. <fmt> is a
4963 log-format string.
4964
Willy Tarreaubd741542010-03-16 18:46:54 +01004965 <match> is a keyword indicating how to look for a specific pattern in the
Christopher Fauletb5594262020-05-05 20:23:13 +02004966 response. The keyword may be one of "status", "rstatus", "hdr",
4967 "fhdr", "string", or "rstring". The keyword may be preceded by an
Willy Tarreaubd741542010-03-16 18:46:54 +01004968 exclamation mark ("!") to negate the match. Spaces are allowed
4969 between the exclamation mark and the keyword. See below for more
4970 details on the supported keywords.
4971
Christopher Faulet39708192020-05-05 10:47:36 +02004972 <pattern> is the pattern to look for. It may be a string, a regular
4973 expression or a more complex pattern with several arguments. If
4974 the string pattern contains spaces, they must be escaped with the
4975 usual backslash ('\').
Willy Tarreaubd741542010-03-16 18:46:54 +01004976
4977 By default, "option httpchk" considers that response statuses 2xx and 3xx
4978 are valid, and that others are invalid. When "http-check expect" is used,
4979 it defines what is considered valid or invalid. Only one "http-check"
4980 statement is supported in a backend. If a server fails to respond or times
4981 out, the check obviously fails. The available matches are :
4982
Christopher Faulet8021a5f2020-04-24 13:53:12 +02004983 status <codes> : test the status codes found parsing <codes> string. it
4984 must be a comma-separated list of status codes or range
4985 codes. A health check response will be considered as
4986 valid if the response's status code matches any status
4987 code or is inside any range of the list. If the "status"
4988 keyword is prefixed with "!", then the response will be
4989 considered invalid if the status code matches.
Willy Tarreaubd741542010-03-16 18:46:54 +01004990
4991 rstatus <regex> : test a regular expression for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004992 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01004993 response's status code matches the expression. If the
4994 "rstatus" keyword is prefixed with "!", then the response
4995 will be considered invalid if the status code matches.
4996 This is mostly used to check for multiple codes.
4997
Christopher Fauletb5594262020-05-05 20:23:13 +02004998 hdr { name | name-lf } [ -m <meth> ] <name>
4999 [ { value | value-lf } [ -m <meth> ] <value> :
Christopher Faulet39708192020-05-05 10:47:36 +02005000 test the specified header pattern on the HTTP response
5001 headers. The name pattern is mandatory but the value
5002 pattern is optional. If not specified, only the header
5003 presence is verified. <meth> is the matching method,
5004 applied on the header name or the header value. Supported
5005 matching methods are "str" (exact match), "beg" (prefix
5006 match), "end" (suffix match), "sub" (substring match) or
5007 "reg" (regex match). If not specified, exact matching
Christopher Fauletb5594262020-05-05 20:23:13 +02005008 method is used. If the "name-lf" parameter is used,
5009 <name> is evaluated as a log-format string. If "value-lf"
5010 parameter is used, <value> is evaluated as a log-format
5011 string. These parameters cannot be used with the regex
5012 matching method. Finally, the header value is considered
5013 as comma-separated list. Note that matchings are case
5014 insensitive on the header names.
5015
5016 fhdr { name | name-lf } [ -m <meth> ] <name>
5017 [ { value | value-lf } [ -m <meth> ] <value> :
5018 test the specified full header pattern on the HTTP
5019 response headers. It does exactly the same than "hdr"
5020 keyword, except the full header value is tested, commas
5021 are not considered as delimiters.
Christopher Faulet39708192020-05-05 10:47:36 +02005022
Willy Tarreaubd741542010-03-16 18:46:54 +01005023 string <string> : test the exact string match in the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04005024 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01005025 response's body contains this exact string. If the
5026 "string" keyword is prefixed with "!", then the response
5027 will be considered invalid if the body contains this
5028 string. This can be used to look for a mandatory word at
5029 the end of a dynamic page, or to detect a failure when a
Davor Ocelice9ed2812017-12-25 17:49:28 +01005030 specific error appears on the check page (e.g. a stack
Willy Tarreaubd741542010-03-16 18:46:54 +01005031 trace).
5032
5033 rstring <regex> : test a regular expression on the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04005034 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01005035 response's body matches this expression. If the "rstring"
5036 keyword is prefixed with "!", then the response will be
5037 considered invalid if the body matches the expression.
5038 This can be used to look for a mandatory word at the end
5039 of a dynamic page, or to detect a failure when a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01005040 error appears on the check page (e.g. a stack trace).
Willy Tarreaubd741542010-03-16 18:46:54 +01005041
Christopher Fauletaaab0832020-05-05 15:54:22 +02005042 string-lf <fmt> : test a log-format string match in the HTTP response body.
5043 A health check response will be considered valid if the
5044 response's body contains the string resulting of the
5045 evaluation of <fmt>, which follows the log-format rules.
5046 If prefixed with "!", then the response will be
5047 considered invalid if the body contains the string.
5048
Willy Tarreaubd741542010-03-16 18:46:54 +01005049 It is important to note that the responses will be limited to a certain size
5050 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
5051 Thus, too large responses may not contain the mandatory pattern when using
5052 "string" or "rstring". If a large response is absolutely required, it is
5053 possible to change the default max size by setting the global variable.
5054 However, it is worth keeping in mind that parsing very large responses can
5055 waste some CPU cycles, especially when regular expressions are used, and that
5056 it is always better to focus the checks on smaller resources.
5057
Christopher Faulete5870d82020-04-15 11:32:03 +02005058 In an http-check ruleset, the last expect rule may be implicit. If no expect
5059 rule is specified after the last "http-check send", an implicit expect rule
5060 is defined to match on 2xx or 3xx status codes. It means this rule is also
5061 defined if there is no "http-check" rule at all, when only "option httpchk"
5062 is set.
Cyril Bonté32602d22015-01-30 00:07:07 +01005063
Willy Tarreaubd741542010-03-16 18:46:54 +01005064 Last, if "http-check expect" is combined with "http-check disable-on-404",
5065 then this last one has precedence when the server responds with 404.
5066
5067 Examples :
5068 # only accept status 200 as valid
Christopher Faulet8021a5f2020-04-24 13:53:12 +02005069 http-check expect status 200,201,300-310
Willy Tarreaubd741542010-03-16 18:46:54 +01005070
Christopher Faulet39708192020-05-05 10:47:36 +02005071 # be sure a sessid coookie is set
5072 http-check expect header name "set-cookie" value -m beg "sessid="
5073
Willy Tarreaubd741542010-03-16 18:46:54 +01005074 # consider SQL errors as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01005075 http-check expect ! string SQL\ Error
Willy Tarreaubd741542010-03-16 18:46:54 +01005076
5077 # consider status 5xx only as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01005078 http-check expect ! rstatus ^5
Willy Tarreaubd741542010-03-16 18:46:54 +01005079
5080 # check that we have a correct hexadecimal tag before /html
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03005081 http-check expect rstring <!--tag:[0-9a-f]*--></html>
Willy Tarreau0ba27502007-12-24 16:55:16 +01005082
Christopher Faulete5870d82020-04-15 11:32:03 +02005083 See also : "option httpchk", "http-check connect", "http-check disable-on-404"
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005084 and "http-check send".
Willy Tarreau2769aa02007-12-27 18:26:09 +01005085
5086
Christopher Faulet7c95f5f2020-05-06 15:06:34 +02005087http-check send [meth <method>] [{ uri <uri> | uri-lf <fmt> }>] [ver <version>]
Christopher Faulet574e7bd2020-05-06 15:38:58 +02005088 [hdr <name> <fmt>]* [{ body <string> | body-lf <fmt> }]
5089 [comment <msg>]
Christopher Faulet8acb1282020-04-09 08:44:06 +02005090 Add a possible list of headers and/or a body to the request sent during HTTP
5091 health checks.
5092 May be used in sections : defaults | frontend | listen | backend
5093 yes | no | yes | yes
5094 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005095 comment <msg> defines a message to report if the rule evaluation fails.
5096
Christopher Faulete5870d82020-04-15 11:32:03 +02005097 meth <method> is the optional HTTP method used with the requests. When not
5098 set, the "OPTIONS" method is used, as it generally requires
5099 low server processing and is easy to filter out from the
5100 logs. Any method may be used, though it is not recommended
5101 to invent non-standard ones.
5102
Christopher Faulet7c95f5f2020-05-06 15:06:34 +02005103 uri <uri> is optional and set the URI referenced in the HTTP requests
5104 to the string <uri>. It defaults to "/" which is accessible
5105 by default on almost any server, but may be changed to any
5106 other URI. Query strings are permitted.
5107
5108 uri-lf <fmt> is optional and set the URI referenced in the HTTP requests
5109 using the log-format string <fmt>. It defaults to "/" which
5110 is accessible by default on almost any server, but may be
5111 changed to any other URI. Query strings are permitted.
Christopher Faulet8acb1282020-04-09 08:44:06 +02005112
Christopher Faulet907701b2020-04-28 09:37:00 +02005113 ver <version> is the optional HTTP version string. It defaults to
Christopher Faulete5870d82020-04-15 11:32:03 +02005114 "HTTP/1.0" but some servers might behave incorrectly in HTTP
Daniel Corbett67a82712020-07-06 23:01:19 -04005115 1.0, so turning it to HTTP/1.1 may sometimes help. Note that
Christopher Faulete5870d82020-04-15 11:32:03 +02005116 the Host field is mandatory in HTTP/1.1, use "hdr" argument
5117 to add it.
5118
5119 hdr <name> <fmt> adds the HTTP header field whose name is specified in
5120 <name> and whose value is defined by <fmt>, which follows
5121 to the log-format rules.
5122
5123 body <string> add the body defined by <string> to the request sent during
5124 HTTP health checks. If defined, the "Content-Length" header
5125 is thus automatically added to the request.
Christopher Faulet8acb1282020-04-09 08:44:06 +02005126
Christopher Faulet574e7bd2020-05-06 15:38:58 +02005127 body-lf <fmt> add the body defined by the log-format string <fmt> to the
5128 request sent during HTTP health checks. If defined, the
5129 "Content-Length" header is thus automatically added to the
5130 request.
5131
Christopher Faulet8acb1282020-04-09 08:44:06 +02005132 In addition to the request line defined by the "option httpchk" directive,
5133 this one is the valid way to add some headers and optionally a body to the
5134 request sent during HTTP health checks. If a body is defined, the associate
Christopher Faulet9df910c2020-04-29 14:20:47 +02005135 "Content-Length" header is automatically added. Thus, this header or
5136 "Transfer-encoding" header should not be present in the request provided by
5137 "http-check send". If so, it will be ignored. The old trick consisting to add
5138 headers after the version string on the "option httpchk" line is now
Christopher Faulet8acb1282020-04-09 08:44:06 +02005139 deprecated. Note also the "Connection: close" header is still added if a
Daniel Corbett67a82712020-07-06 23:01:19 -04005140 "http-check expect" directive is defined independently of this directive, just
Christopher Faulet8acb1282020-04-09 08:44:06 +02005141 like the state header if the directive "http-check send-state" is defined.
5142
Christopher Faulete5870d82020-04-15 11:32:03 +02005143 Also "http-check send" doesn't support HTTP keep-alive. Keep in mind that it
5144 will automatically append a "Connection: close" header, meaning that this
Christopher Faulet9df910c2020-04-29 14:20:47 +02005145 header should not be present in the request provided by "http-check send". If
5146 so, it will be ignored.
5147
5148 Note that the Host header and the request authority, when both defined, are
5149 automatically synchronized. It means when the HTTP request is sent, when a
5150 Host is inserted in the request, the request authority is accordingly
5151 updated. Thus, don't be surprised if the Host header value overwrites the
5152 configured request authority.
5153
5154 Note also for now, no Host header is automatically added in HTTP/1.1 or above
5155 requests. You should add it explicitly.
Christopher Faulete5870d82020-04-15 11:32:03 +02005156
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005157 See also : "option httpchk", "http-check send-state" and "http-check expect".
Christopher Faulet8acb1282020-04-09 08:44:06 +02005158
5159
Willy Tarreauef781042010-01-27 11:53:01 +01005160http-check send-state
5161 Enable emission of a state header with HTTP health checks
5162 May be used in sections : defaults | frontend | listen | backend
5163 yes | no | yes | yes
5164 Arguments : none
5165
5166 When this option is set, haproxy will systematically send a special header
5167 "X-Haproxy-Server-State" with a list of parameters indicating to each server
5168 how they are seen by haproxy. This can be used for instance when a server is
5169 manipulated without access to haproxy and the operator needs to know whether
5170 haproxy still sees it up or not, or if the server is the last one in a farm.
5171
5172 The header is composed of fields delimited by semi-colons, the first of which
5173 is a word ("UP", "DOWN", "NOLB"), possibly followed by a number of valid
5174 checks on the total number before transition, just as appears in the stats
5175 interface. Next headers are in the form "<variable>=<value>", indicating in
5176 no specific order some values available in the stats interface :
Joseph Lynch514061c2015-01-15 17:52:59 -08005177 - a variable "address", containing the address of the backend server.
5178 This corresponds to the <address> field in the server declaration. For
5179 unix domain sockets, it will read "unix".
5180
5181 - a variable "port", containing the port of the backend server. This
5182 corresponds to the <port> field in the server declaration. For unix
5183 domain sockets, it will read "unix".
5184
Willy Tarreauef781042010-01-27 11:53:01 +01005185 - a variable "name", containing the name of the backend followed by a slash
5186 ("/") then the name of the server. This can be used when a server is
5187 checked in multiple backends.
5188
5189 - a variable "node" containing the name of the haproxy node, as set in the
5190 global "node" variable, otherwise the system's hostname if unspecified.
5191
5192 - a variable "weight" indicating the weight of the server, a slash ("/")
5193 and the total weight of the farm (just counting usable servers). This
5194 helps to know if other servers are available to handle the load when this
5195 one fails.
5196
5197 - a variable "scur" indicating the current number of concurrent connections
5198 on the server, followed by a slash ("/") then the total number of
5199 connections on all servers of the same backend.
5200
5201 - a variable "qcur" indicating the current number of requests in the
5202 server's queue.
5203
5204 Example of a header received by the application server :
5205 >>> X-Haproxy-Server-State: UP 2/3; name=bck/srv2; node=lb1; weight=1/2; \
5206 scur=13/22; qcur=0
5207
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005208 See also : "option httpchk", "http-check disable-on-404" and
5209 "http-check send".
Willy Tarreauef781042010-01-27 11:53:01 +01005210
Christopher Faulete5870d82020-04-15 11:32:03 +02005211
5212http-check set-var(<var-name>) <expr>
Christopher Faulete5870d82020-04-15 11:32:03 +02005213 This operation sets the content of a variable. The variable is declared inline.
Christopher Faulete5870d82020-04-15 11:32:03 +02005214 May be used in sections: defaults | frontend | listen | backend
5215 yes | no | yes | yes
5216
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005217 Arguments :
Christopher Faulete5870d82020-04-15 11:32:03 +02005218 <var-name> The name of the variable starts with an indication about its
5219 scope. The scopes allowed for http-check are:
5220 "proc" : the variable is shared with the whole process.
5221 "sess" : the variable is shared with the tcp-check session.
5222 "check": the variable is declared for the lifetime of the tcp-check.
5223 This prefix is followed by a name. The separator is a '.'.
5224 The name may only contain characters 'a-z', 'A-Z', '0-9', '.',
5225 and '-'.
5226
5227 <expr> Is a sample-fetch expression potentially followed by converters.
5228
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005229 Examples :
5230 http-check set-var(check.port) int(1234)
Christopher Faulete5870d82020-04-15 11:32:03 +02005231
5232
5233http-check unset-var(<var-name>)
Christopher Faulete5870d82020-04-15 11:32:03 +02005234 Free a reference to a variable within its scope.
Christopher Faulete5870d82020-04-15 11:32:03 +02005235 May be used in sections: defaults | frontend | listen | backend
5236 yes | no | yes | yes
5237
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005238 Arguments :
Christopher Faulete5870d82020-04-15 11:32:03 +02005239 <var-name> The name of the variable starts with an indication about its
5240 scope. The scopes allowed for http-check are:
5241 "proc" : the variable is shared with the whole process.
5242 "sess" : the variable is shared with the tcp-check session.
5243 "check": the variable is declared for the lifetime of the tcp-check.
5244 This prefix is followed by a name. The separator is a '.'.
5245 The name may only contain characters 'a-z', 'A-Z', '0-9', '.',
5246 and '-'.
5247
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005248 Examples :
5249 http-check unset-var(check.port)
Christopher Faulete5870d82020-04-15 11:32:03 +02005250
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005251
Christopher Faulet3b967c12020-05-15 15:47:44 +02005252http-error status <code> [content-type <type>]
5253 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
5254 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
5255 [ hdr <name> <fmt> ]*
5256 Defines a custom error message to use instead of errors generated by HAProxy.
5257 May be used in sections : defaults | frontend | listen | backend
5258 yes | yes | yes | yes
5259 Arguments :
Ilya Shipitsin11057a32020-06-21 21:18:27 +05005260 status <code> is the HTTP status code. It must be specified.
Christopher Faulet3b967c12020-05-15 15:47:44 +02005261 Currently, HAProxy is capable of generating codes
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02005262 200, 400, 401, 403, 404, 405, 407, 408, 410, 413, 425,
5263 429, 500, 502, 503, and 504.
Christopher Faulet3b967c12020-05-15 15:47:44 +02005264
5265 content-type <type> is the response content type, for instance
5266 "text/plain". This parameter is ignored and should be
5267 omitted when an errorfile is configured or when the
5268 payload is empty. Otherwise, it must be defined.
5269
5270 default-errorfiles Reset the previously defined error message for current
5271 proxy for the status <code>. If used on a backend, the
5272 frontend error message is used, if defined. If used on
5273 a frontend, the default error message is used.
5274
5275 errorfile <file> designates a file containing the full HTTP response.
5276 It is recommended to follow the common practice of
5277 appending ".http" to the filename so that people do
5278 not confuse the response with HTML error pages, and to
5279 use absolute paths, since files are read before any
5280 chroot is performed.
5281
5282 errorfiles <name> designates the http-errors section to use to import
5283 the error message with the status code <code>. If no
5284 such message is found, the proxy's error messages are
5285 considered.
5286
5287 file <file> specifies the file to use as response payload. If the
5288 file is not empty, its content-type must be set as
5289 argument to "content-type", otherwise, any
5290 "content-type" argument is ignored. <file> is
5291 considered as a raw string.
5292
5293 string <str> specifies the raw string to use as response payload.
5294 The content-type must always be set as argument to
5295 "content-type".
5296
5297 lf-file <file> specifies the file to use as response payload. If the
5298 file is not empty, its content-type must be set as
5299 argument to "content-type", otherwise, any
5300 "content-type" argument is ignored. <file> is
5301 evaluated as a log-format string.
5302
5303 lf-string <str> specifies the log-format string to use as response
5304 payload. The content-type must always be set as
5305 argument to "content-type".
5306
5307 hdr <name> <fmt> adds to the response the HTTP header field whose name
5308 is specified in <name> and whose value is defined by
5309 <fmt>, which follows to the log-format rules.
5310 This parameter is ignored if an errorfile is used.
5311
5312 This directive may be used instead of "errorfile", to define a custom error
5313 message. As "errorfile" directive, it is used for errors detected and
5314 returned by HAProxy. If an errorfile is defined, it is parsed when HAProxy
5315 starts and must be valid according to the HTTP standards. The generated
5316 response must not exceed the configured buffer size (BUFFSIZE), otherwise an
5317 internal error will be returned. Finally, if you consider to use some
5318 http-after-response rules to rewrite these errors, the reserved buffer space
5319 should be available (see "tune.maxrewrite").
5320
5321 The files are read at the same time as the configuration and kept in memory.
5322 For this reason, the errors continue to be returned even when the process is
5323 chrooted, and no file change is considered while the process is running.
5324
5325 See also : "errorfile", "errorfiles", "errorloc", "errorloc302",
5326 "errorloc303" and section 3.8 about http-errors.
5327
5328
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005329http-request <action> [options...] [ { if | unless } <condition> ]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01005330 Access control for Layer 7 requests
5331
5332 May be used in sections: defaults | frontend | listen | backend
5333 no | yes | yes | yes
5334
Willy Tarreau20b0de52012-12-24 15:45:22 +01005335 The http-request statement defines a set of rules which apply to layer 7
5336 processing. The rules are evaluated in their declaration order when they are
5337 met in a frontend, listen or backend section. Any rule may optionally be
5338 followed by an ACL-based condition, in which case it will only be evaluated
5339 if the condition is true.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01005340
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005341 The first keyword is the rule's action. The supported actions are described
5342 below.
Willy Tarreau20b0de52012-12-24 15:45:22 +01005343
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005344 There is no limit to the number of http-request statements per instance.
Willy Tarreau20b0de52012-12-24 15:45:22 +01005345
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005346 Example:
5347 acl nagios src 192.168.129.3
5348 acl local_net src 192.168.0.0/16
5349 acl auth_ok http_auth(L1)
Willy Tarreau20b0de52012-12-24 15:45:22 +01005350
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005351 http-request allow if nagios
5352 http-request allow if local_net auth_ok
5353 http-request auth realm Gimme if local_net auth_ok
5354 http-request deny
Willy Tarreau81499eb2012-12-27 12:19:02 +01005355
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005356 Example:
5357 acl key req.hdr(X-Add-Acl-Key) -m found
5358 acl add path /addacl
5359 acl del path /delacl
Willy Tarreau20b0de52012-12-24 15:45:22 +01005360
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005361 acl myhost hdr(Host) -f myhost.lst
Willy Tarreau20b0de52012-12-24 15:45:22 +01005362
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005363 http-request add-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key add
5364 http-request del-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key del
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02005365
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005366 Example:
5367 acl value req.hdr(X-Value) -m found
5368 acl setmap path /setmap
5369 acl delmap path /delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06005370
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005371 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06005372
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005373 http-request set-map(map.lst) %[src] %[req.hdr(X-Value)] if setmap value
5374 http-request del-map(map.lst) %[src] if delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06005375
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005376 See also : "stats http-request", section 3.4 about userlists and section 7
5377 about ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06005378
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005379http-request add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005380
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005381 This is used to add a new entry into an ACL. The ACL must be loaded from a
5382 file (even a dummy empty file). The file name of the ACL to be updated is
5383 passed between parentheses. It takes one argument: <key fmt>, which follows
5384 log-format rules, to collect content of the new entry. It performs a lookup
5385 in the ACL before insertion, to avoid duplicated (or more) values. This
5386 lookup is done by a linear search and can be expensive with large lists!
5387 It is the equivalent of the "add acl" command from the stats socket, but can
5388 be triggered by an HTTP request.
Sasha Pachev218f0642014-06-16 12:05:59 -06005389
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005390http-request add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005391
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005392 This appends an HTTP header field whose name is specified in <name> and
5393 whose value is defined by <fmt> which follows the log-format rules (see
5394 Custom Log Format in section 8.2.4). This is particularly useful to pass
5395 connection-specific information to the server (e.g. the client's SSL
5396 certificate), or to combine several headers into one. This rule is not
5397 final, so it is possible to add other similar rules. Note that header
5398 addition is performed immediately, so one rule might reuse the resulting
5399 header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06005400
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005401http-request allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005402
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005403 This stops the evaluation of the rules and lets the request pass the check.
5404 No further "http-request" rules are evaluated.
Sasha Pachev218f0642014-06-16 12:05:59 -06005405
Sasha Pachev218f0642014-06-16 12:05:59 -06005406
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005407http-request auth [realm <realm>] [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005408
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005409 This stops the evaluation of the rules and immediately responds with an
5410 HTTP 401 or 407 error code to invite the user to present a valid user name
5411 and password. No further "http-request" rules are evaluated. An optional
5412 "realm" parameter is supported, it sets the authentication realm that is
5413 returned with the response (typically the application's name).
Sasha Pachev218f0642014-06-16 12:05:59 -06005414
Christopher Faulet612f2ea2020-05-27 09:57:28 +02005415 The corresponding proxy's error message is used. It may be customized using
5416 an "errorfile" or an "http-error" directive. For 401 responses, all
5417 occurrences of the WWW-Authenticate header are removed and replaced by a new
5418 one with a basic authentication challenge for realm "<realm>". For 407
5419 responses, the same is done on the Proxy-Authenticate header. If the error
5420 message must not be altered, consider to use "http-request return" rule
5421 instead.
5422
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005423 Example:
5424 acl auth_ok http_auth_group(L1) G1
5425 http-request auth unless auth_ok
Sasha Pachev218f0642014-06-16 12:05:59 -06005426
Jarno Huuskonen251a6b72019-01-04 14:05:02 +02005427http-request cache-use <name> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005428
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02005429 See section 6.2 about cache setup.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005430
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005431http-request capture <sample> [ len <length> | id <id> ]
5432 [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005433
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005434 This captures sample expression <sample> from the request buffer, and
5435 converts it to a string of at most <len> characters. The resulting string is
5436 stored into the next request "capture" slot, so it will possibly appear next
5437 to some captured HTTP headers. It will then automatically appear in the logs,
5438 and it will be possible to extract it using sample fetch rules to feed it
5439 into headers or anything. The length should be limited given that this size
5440 will be allocated for each capture during the whole session life.
5441 Please check section 7.3 (Fetching samples) and "capture request header" for
5442 more information.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005443
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005444 If the keyword "id" is used instead of "len", the action tries to store the
5445 captured string in a previously declared capture slot. This is useful to run
5446 captures in backends. The slot id can be declared by a previous directive
Baptiste Assmann19a69b32020-01-16 14:34:22 +01005447 "http-request capture" or with the "declare capture" keyword.
5448
5449 When using this action in a backend, double check that the relevant
5450 frontend(s) have the required capture slots otherwise, this rule will be
5451 ignored at run time. This can't be detected at configuration parsing time
5452 due to HAProxy's ability to dynamically resolve backend name at runtime.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005453
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005454http-request del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005455
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005456 This is used to delete an entry from an ACL. The ACL must be loaded from a
5457 file (even a dummy empty file). The file name of the ACL to be updated is
5458 passed between parentheses. It takes one argument: <key fmt>, which follows
5459 log-format rules, to collect content of the entry to delete.
5460 It is the equivalent of the "del acl" command from the stats socket, but can
5461 be triggered by an HTTP request.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005462
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005463http-request del-header <name> [ { if | unless } <condition> ]
Willy Tarreauf4c43c12013-06-11 17:01:13 +02005464
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005465 This removes all HTTP header fields whose name is specified in <name>.
Willy Tarreau9a355ec2013-06-11 17:45:46 +02005466
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005467http-request del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau42cf39e2013-06-11 18:51:32 +02005468
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005469 This is used to delete an entry from a MAP. The MAP must be loaded from a
5470 file (even a dummy empty file). The file name of the MAP to be updated is
5471 passed between parentheses. It takes one argument: <key fmt>, which follows
5472 log-format rules, to collect content of the entry to delete.
5473 It takes one argument: "file name" It is the equivalent of the "del map"
5474 command from the stats socket, but can be triggered by an HTTP request.
Willy Tarreau51347ed2013-06-11 19:34:13 +02005475
Christopher Faulet5cb513a2020-05-13 17:56:56 +02005476http-request deny [deny_status <status>] [ { if | unless } <condition> ]
5477http-request deny [ { status | deny_status } <code>] [content-type <type>]
5478 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
5479 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
5480 [ hdr <name> <fmt> ]*
5481 [ { if | unless } <condition> ]
Patrick Hemmer268a7072018-05-11 12:52:31 -04005482
Christopher Faulet5cb513a2020-05-13 17:56:56 +02005483 This stops the evaluation of the rules and immediately rejects the request.
5484 By default an HTTP 403 error is returned. But the response may be customized
5485 using same syntax than "http-request return" rules. Thus, see "http-request
Ilya Shipitsin11057a32020-06-21 21:18:27 +05005486 return" for details. For compatibility purpose, when no argument is defined,
Christopher Faulet5cb513a2020-05-13 17:56:56 +02005487 or only "deny_status", the argument "default-errorfiles" is implied. It means
5488 "http-request deny [deny_status <status>]" is an alias of
5489 "http-request deny [status <status>] default-errorfiles".
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005490 No further "http-request" rules are evaluated.
Christopher Faulet5cb513a2020-05-13 17:56:56 +02005491 See also "http-request return".
Patrick Hemmer268a7072018-05-11 12:52:31 -04005492
Olivier Houchard602bf7d2019-05-10 13:59:15 +02005493http-request disable-l7-retry [ { if | unless } <condition> ]
5494 This disables any attempt to retry the request if it fails for any other
5495 reason than a connection failure. This can be useful for example to make
5496 sure POST requests aren't retried on failure.
5497
Baptiste Assmann333939c2019-01-21 08:34:50 +01005498http-request do-resolve(<var>,<resolvers>,[ipv4,ipv6]) <expr> :
5499
5500 This action performs a DNS resolution of the output of <expr> and stores
5501 the result in the variable <var>. It uses the DNS resolvers section
5502 pointed by <resolvers>.
5503 It is possible to choose a resolution preference using the optional
5504 arguments 'ipv4' or 'ipv6'.
5505 When performing the DNS resolution, the client side connection is on
5506 pause waiting till the end of the resolution.
5507 If an IP address can be found, it is stored into <var>. If any kind of
5508 error occurs, then <var> is not set.
5509 One can use this action to discover a server IP address at run time and
5510 based on information found in the request (IE a Host header).
5511 If this action is used to find the server's IP address (using the
5512 "set-dst" action), then the server IP address in the backend must be set
5513 to 0.0.0.0.
5514
5515 Example:
5516 resolvers mydns
5517 nameserver local 127.0.0.53:53
5518 nameserver google 8.8.8.8:53
5519 timeout retry 1s
5520 hold valid 10s
5521 hold nx 3s
5522 hold other 3s
5523 hold obsolete 0s
5524 accepted_payload_size 8192
5525
5526 frontend fe
5527 bind 10.42.0.1:80
5528 http-request do-resolve(txn.myip,mydns,ipv4) hdr(Host),lower
5529 http-request capture var(txn.myip) len 40
5530
5531 # return 503 when the variable is not set,
5532 # which mean DNS resolution error
5533 use_backend b_503 unless { var(txn.myip) -m found }
5534
5535 default_backend be
5536
5537 backend b_503
5538 # dummy backend used to return 503.
5539 # one can use the errorfile directive to send a nice
5540 # 503 error page to end users
5541
5542 backend be
5543 # rule to prevent HAProxy from reconnecting to services
5544 # on the local network (forged DNS name used to scan the network)
5545 http-request deny if { var(txn.myip) -m ip 127.0.0.0/8 10.0.0.0/8 }
5546 http-request set-dst var(txn.myip)
5547 server clear 0.0.0.0:0
5548
5549 NOTE: Don't forget to set the "protection" rules to ensure HAProxy won't
5550 be used to scan the network or worst won't loop over itself...
5551
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01005552http-request early-hint <name> <fmt> [ { if | unless } <condition> ]
5553
5554 This is used to build an HTTP 103 Early Hints response prior to any other one.
5555 This appends an HTTP header field to this response whose name is specified in
5556 <name> and whose value is defined by <fmt> which follows the log-format rules
5557 (see Custom Log Format in section 8.2.4). This is particularly useful to pass
Frédéric Lécaille3aac1062018-11-13 09:42:13 +01005558 to the client some Link headers to preload resources required to render the
5559 HTML documents.
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01005560
5561 See RFC 8297 for more information.
5562
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005563http-request redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005564
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005565 This performs an HTTP redirection based on a redirect rule. This is exactly
5566 the same as the "redirect" statement except that it inserts a redirect rule
5567 which can be processed in the middle of other "http-request" rules and that
5568 these rules use the "log-format" strings. See the "redirect" keyword for the
5569 rule's syntax.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005570
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005571http-request reject [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005572
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005573 This stops the evaluation of the rules and immediately closes the connection
5574 without sending any response. It acts similarly to the
5575 "tcp-request content reject" rules. It can be useful to force an immediate
5576 connection closure on HTTP/2 connections.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005577
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005578http-request replace-header <name> <match-regex> <replace-fmt>
5579 [ { if | unless } <condition> ]
Willy Tarreaua9083d02015-05-08 15:27:59 +02005580
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05005581 This matches the value of all occurrences of header field <name> against
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005582 <match-regex>. Matching is performed case-sensitively. Matching values are
5583 completely replaced by <replace-fmt>. Format characters are allowed in
5584 <replace-fmt> and work like <fmt> arguments in "http-request add-header".
5585 Standard back-references using the backslash ('\') followed by a number are
5586 supported.
Thierry FOURNIER82bf70d2015-05-26 17:58:29 +02005587
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005588 This action acts on whole header lines, regardless of the number of values
5589 they may contain. Thus it is well-suited to process headers naturally
5590 containing commas in their value, such as If-Modified-Since. Headers that
5591 contain a comma-separated list of values, such as Accept, should be processed
5592 using "http-request replace-value".
William Lallemand86d0df02017-11-24 21:36:45 +01005593
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005594 Example:
5595 http-request replace-header Cookie foo=([^;]*);(.*) foo=\1;ip=%bi;\2
5596
5597 # applied to:
5598 Cookie: foo=foobar; expires=Tue, 14-Jun-2016 01:40:45 GMT;
5599
5600 # outputs:
5601 Cookie: foo=foobar;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT;
5602
5603 # assuming the backend IP is 192.168.1.20
Willy Tarreau09448f72014-06-25 18:12:15 +02005604
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005605 http-request replace-header User-Agent curl foo
5606
5607 # applied to:
5608 User-Agent: curl/7.47.0
Willy Tarreau09448f72014-06-25 18:12:15 +02005609
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005610 # outputs:
5611 User-Agent: foo
Willy Tarreau09448f72014-06-25 18:12:15 +02005612
Willy Tarreau262c3f12019-12-17 06:52:51 +01005613http-request replace-path <match-regex> <replace-fmt>
5614 [ { if | unless } <condition> ]
5615
5616 This works like "replace-header" except that it works on the request's path
5617 component instead of a header. The path component starts at the first '/'
Christopher Faulet82c83322020-09-02 14:16:59 +02005618 after an optional scheme+authority and ends before the question mark. Thus,
5619 the replacement does not modify the scheme, the authority and the
5620 query-string.
Willy Tarreau262c3f12019-12-17 06:52:51 +01005621
5622 It is worth noting that regular expressions may be more expensive to evaluate
5623 than certain ACLs, so rare replacements may benefit from a condition to avoid
5624 performing the evaluation at all if it does not match.
5625
5626 Example:
5627 # prefix /foo : turn /bar?q=1 into /foo/bar?q=1 :
5628 http-request replace-path (.*) /foo\1
5629
Willy Tarreau262c3f12019-12-17 06:52:51 +01005630 # strip /foo : turn /foo/bar?q=1 into /bar?q=1
5631 http-request replace-path /foo/(.*) /\1
5632 # or more efficient if only some requests match :
5633 http-request replace-path /foo/(.*) /\1 if { url_beg /foo/ }
5634
Christopher Faulet312294f2020-09-02 17:17:44 +02005635http-request replace-pathq <match-regex> <replace-fmt>
5636 [ { if | unless } <condition> ]
5637
5638 This does the same as "http-request replace-path" except that the path
5639 contains the query-string if any is present. Thus, the path and the
5640 query-string are replaced.
5641
5642 Example:
5643 # suffix /foo : turn /bar?q=1 into /bar/foo?q=1 :
5644 http-request replace-pathq ([^?]*)(\?(.*))? \1/foo\2
5645
Willy Tarreau33810222019-06-12 17:44:02 +02005646http-request replace-uri <match-regex> <replace-fmt>
5647 [ { if | unless } <condition> ]
5648
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005649 This works like "replace-header" except that it works on the request's URI part
5650 instead of a header. The URI part may contain an optional scheme, authority or
5651 query string. These are considered to be part of the value that is matched
5652 against.
5653
5654 It is worth noting that regular expressions may be more expensive to evaluate
5655 than certain ACLs, so rare replacements may benefit from a condition to avoid
5656 performing the evaluation at all if it does not match.
Willy Tarreau33810222019-06-12 17:44:02 +02005657
Willy Tarreau62b59132019-12-17 06:51:20 +01005658 IMPORTANT NOTE: historically in HTTP/1.x, the vast majority of requests sent
5659 by browsers use the "origin form", which differs from the "absolute form" in
5660 that they do not contain a scheme nor authority in the URI portion. Mostly
5661 only requests sent to proxies, those forged by hand and some emitted by
5662 certain applications use the absolute form. As such, "replace-uri" usually
5663 works fine most of the time in HTTP/1.x with rules starting with a "/". But
5664 with HTTP/2, clients are encouraged to send absolute URIs only, which look
5665 like the ones HTTP/1 clients use to talk to proxies. Such partial replace-uri
5666 rules may then fail in HTTP/2 when they work in HTTP/1. Either the rules need
Willy Tarreau262c3f12019-12-17 06:52:51 +01005667 to be adapted to optionally match a scheme and authority, or replace-path
5668 should be used.
Willy Tarreau33810222019-06-12 17:44:02 +02005669
Willy Tarreau62b59132019-12-17 06:51:20 +01005670 Example:
5671 # rewrite all "http" absolute requests to "https":
5672 http-request replace-uri ^http://(.*) https://\1
Willy Tarreau33810222019-06-12 17:44:02 +02005673
Willy Tarreau62b59132019-12-17 06:51:20 +01005674 # prefix /foo : turn /bar?q=1 into /foo/bar?q=1 :
5675 http-request replace-uri ([^/:]*://[^/]*)?(.*) \1/foo\2
Willy Tarreau33810222019-06-12 17:44:02 +02005676
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005677http-request replace-value <name> <match-regex> <replace-fmt>
5678 [ { if | unless } <condition> ]
Willy Tarreau09448f72014-06-25 18:12:15 +02005679
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005680 This works like "replace-header" except that it matches the regex against
5681 every comma-delimited value of the header field <name> instead of the
5682 entire header. This is suited for all headers which are allowed to carry
5683 more than one value. An example could be the Accept header.
Willy Tarreau09448f72014-06-25 18:12:15 +02005684
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005685 Example:
5686 http-request replace-value X-Forwarded-For ^192\.168\.(.*)$ 172.16.\1
Thierry FOURNIER236657b2015-08-19 08:25:14 +02005687
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005688 # applied to:
5689 X-Forwarded-For: 192.168.10.1, 192.168.13.24, 10.0.0.37
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02005690
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005691 # outputs:
5692 X-Forwarded-For: 172.16.10.1, 172.16.13.24, 10.0.0.37
Frédéric Lécaille6778b272018-01-29 15:22:53 +01005693
Christopher Faulet24231ab2020-01-24 17:44:23 +01005694http-request return [status <code>] [content-type <type>]
5695 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
5696 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
Christopher Faulet4a2c1422020-01-31 17:36:01 +01005697 [ hdr <name> <fmt> ]*
Christopher Faulet24231ab2020-01-24 17:44:23 +01005698 [ { if | unless } <condition> ]
5699
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05005700 This stops the evaluation of the rules and immediately returns a response. The
Christopher Faulet24231ab2020-01-24 17:44:23 +01005701 default status code used for the response is 200. It can be optionally
5702 specified as an arguments to "status". The response content-type may also be
Daniel Corbett67a82712020-07-06 23:01:19 -04005703 specified as an argument to "content-type". Finally the response itself may
Sébastien Grossab877122020-10-08 10:06:03 +02005704 be defined. It can be a full HTTP response specifying the errorfile to use,
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05005705 or the response payload specifying the file or the string to use. These rules
Christopher Faulet24231ab2020-01-24 17:44:23 +01005706 are followed to create the response :
5707
5708 * If neither the errorfile nor the payload to use is defined, a dummy
5709 response is returned. Only the "status" argument is considered. It can be
5710 any code in the range [200, 599]. The "content-type" argument, if any, is
5711 ignored.
5712
5713 * If "default-errorfiles" argument is set, the proxy's errorfiles are
5714 considered. If the "status" argument is defined, it must be one of the
Daniel Corbett67a82712020-07-06 23:01:19 -04005715 status code handled by haproxy (200, 400, 403, 404, 405, 408, 410, 413,
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02005716 425, 429, 500, 502, 503, and 504). The "content-type" argument, if any,
5717 is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01005718
5719 * If a specific errorfile is defined, with an "errorfile" argument, the
5720 corresponding file, containing a full HTTP response, is returned. Only the
5721 "status" argument is considered. It must be one of the status code handled
Daniel Corbett67a82712020-07-06 23:01:19 -04005722 by haproxy (200, 400, 403, 404, 405, 408, 410, 413, 425, 429, 500, 502, 503,
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02005723 and 504). The "content-type" argument, if any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01005724
5725 * If an http-errors section is defined, with an "errorfiles" argument, the
5726 corresponding file in the specified http-errors section, containing a full
5727 HTTP response, is returned. Only the "status" argument is considered. It
Daniel Corbett67a82712020-07-06 23:01:19 -04005728 must be one of the status code handled by haproxy (200, 400, 403, 404, 405,
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02005729 408, 410, 413, 425, 429, 500, 502, 503, and 504). The "content-type"
5730 argument, if any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01005731
5732 * If a "file" or a "lf-file" argument is specified, the file's content is
5733 used as the response payload. If the file is not empty, its content-type
5734 must be set as argument to "content-type". Otherwise, any "content-type"
5735 argument is ignored. With a "lf-file" argument, the file's content is
5736 evaluated as a log-format string. With a "file" argument, it is considered
5737 as a raw content.
5738
5739 * If a "string" or "lf-string" argument is specified, the defined string is
5740 used as the response payload. The content-type must always be set as
5741 argument to "content-type". With a "lf-string" argument, the string is
5742 evaluated as a log-format string. With a "string" argument, it is
5743 considered as a raw string.
5744
Sébastien Grossab877122020-10-08 10:06:03 +02005745 When the response is not based on an errorfile, it is possible to append HTTP
Christopher Faulet4a2c1422020-01-31 17:36:01 +01005746 header fields to the response using "hdr" arguments. Otherwise, all "hdr"
5747 arguments are ignored. For each one, the header name is specified in <name>
5748 and its value is defined by <fmt> which follows the log-format rules.
5749
Christopher Faulet24231ab2020-01-24 17:44:23 +01005750 Note that the generated response must be smaller than a buffer. And to avoid
5751 any warning, when an errorfile or a raw file is loaded, the buffer space
Sébastien Grossab877122020-10-08 10:06:03 +02005752 reserved for the headers rewriting should also be free.
Christopher Faulet24231ab2020-01-24 17:44:23 +01005753
5754 No further "http-request" rules are evaluated.
5755
5756 Example:
Daniel Corbett67a82712020-07-06 23:01:19 -04005757 http-request return errorfile /etc/haproxy/errorfiles/200.http \
Christopher Faulet24231ab2020-01-24 17:44:23 +01005758 if { path /ping }
5759
5760 http-request return content-type image/x-icon file /var/www/favicon.ico \
5761 if { path /favicon.ico }
5762
5763 http-request return status 403 content-type text/plain \
5764 lf-string "Access denied. IP %[src] is blacklisted." \
5765 if { src -f /etc/haproxy/blacklist.lst }
5766
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005767http-request sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
5768http-request sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02005769
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005770 This actions increments the GPC0 or GPC1 counter according with the sticky
5771 counter designated by <sc-id>. If an error occurs, this action silently fails
5772 and the actions evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02005773
Cédric Dufour0d7712d2019-11-06 18:38:53 +01005774http-request sc-set-gpt0(<sc-id>) { <int> | <expr> }
5775 [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02005776
Cédric Dufour0d7712d2019-11-06 18:38:53 +01005777 This action sets the 32-bit unsigned GPT0 tag according to the sticky counter
5778 designated by <sc-id> and the value of <int>/<expr>. The expected result is a
5779 boolean. If an error occurs, this action silently fails and the actions
5780 evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02005781
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005782http-request set-dst <expr> [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02005783
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005784 This is used to set the destination IP address to the value of specified
5785 expression. Useful when a proxy in front of HAProxy rewrites destination IP,
5786 but provides the correct IP in a HTTP header; or you want to mask the IP for
5787 privacy. If you want to connect to the new address/port, use '0.0.0.0:0' as a
5788 server address in the backend.
Christopher Faulet85d79c92016-11-09 16:54:56 +01005789
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005790 Arguments:
5791 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
5792 by some converters.
Christopher Faulet85d79c92016-11-09 16:54:56 +01005793
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005794 Example:
5795 http-request set-dst hdr(x-dst)
5796 http-request set-dst dst,ipmask(24)
Christopher Faulet85d79c92016-11-09 16:54:56 +01005797
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005798 When possible, set-dst preserves the original destination port as long as the
5799 address family allows it, otherwise the destination port is set to 0.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02005800
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005801http-request set-dst-port <expr> [ { if | unless } <condition> ]
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02005802
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005803 This is used to set the destination port address to the value of specified
5804 expression. If you want to connect to the new address/port, use '0.0.0.0:0'
5805 as a server address in the backend.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02005806
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005807 Arguments:
5808 <expr> Is a standard HAProxy expression formed by a sample-fetch
5809 followed by some converters.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02005810
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005811 Example:
5812 http-request set-dst-port hdr(x-port)
5813 http-request set-dst-port int(4000)
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02005814
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005815 When possible, set-dst-port preserves the original destination address as
5816 long as the address family supports a port, otherwise it forces the
5817 destination address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +02005818
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005819http-request set-header <name> <fmt> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02005820
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005821 This does the same as "http-request add-header" except that the header name
5822 is first removed if it existed. This is useful when passing security
5823 information to the server, where the header must not be manipulated by
5824 external users. Note that the new value is computed before the removal so it
5825 is possible to concatenate a value to an existing header.
William Lallemand44be6402016-05-25 01:51:35 +02005826
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005827 Example:
5828 http-request set-header X-Haproxy-Current-Date %T
5829 http-request set-header X-SSL %[ssl_fc]
5830 http-request set-header X-SSL-Session_ID %[ssl_fc_session_id,hex]
5831 http-request set-header X-SSL-Client-Verify %[ssl_c_verify]
5832 http-request set-header X-SSL-Client-DN %{+Q}[ssl_c_s_dn]
5833 http-request set-header X-SSL-Client-CN %{+Q}[ssl_c_s_dn(cn)]
5834 http-request set-header X-SSL-Issuer %{+Q}[ssl_c_i_dn]
5835 http-request set-header X-SSL-Client-NotBefore %{+Q}[ssl_c_notbefore]
5836 http-request set-header X-SSL-Client-NotAfter %{+Q}[ssl_c_notafter]
William Lallemand44be6402016-05-25 01:51:35 +02005837
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005838http-request set-log-level <level> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02005839
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005840 This is used to change the log level of the current request when a certain
5841 condition is met. Valid levels are the 8 syslog levels (see the "log"
5842 keyword) plus the special level "silent" which disables logging for this
5843 request. This rule is not final so the last matching rule wins. This rule
5844 can be useful to disable health checks coming from another equipment.
William Lallemand13e9b0c2016-05-25 02:34:07 +02005845
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005846http-request set-map(<file-name>) <key fmt> <value fmt>
5847 [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02005848
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005849 This is used to add a new entry into a MAP. The MAP must be loaded from a
5850 file (even a dummy empty file). The file name of the MAP to be updated is
5851 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
5852 log-format rules, used to collect MAP key, and <value fmt>, which follows
5853 log-format rules, used to collect content for the new entry.
5854 It performs a lookup in the MAP before insertion, to avoid duplicated (or
5855 more) values. This lookup is done by a linear search and can be expensive
5856 with large lists! It is the equivalent of the "set map" command from the
5857 stats socket, but can be triggered by an HTTP request.
William Lallemand13e9b0c2016-05-25 02:34:07 +02005858
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005859http-request set-mark <mark> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02005860
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005861 This is used to set the Netfilter MARK on all packets sent to the client to
5862 the value passed in <mark> on platforms which support it. This value is an
5863 unsigned 32 bit value which can be matched by netfilter and by the routing
5864 table. It can be expressed both in decimal or hexadecimal format (prefixed by
5865 "0x"). This can be useful to force certain packets to take a different route
5866 (for example a cheaper network path for bulk downloads). This works on Linux
5867 kernels 2.6.32 and above and requires admin privileges.
Willy Tarreau00005ce2016-10-21 15:07:45 +02005868
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005869http-request set-method <fmt> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02005870
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005871 This rewrites the request method with the result of the evaluation of format
5872 string <fmt>. There should be very few valid reasons for having to do so as
5873 this is more likely to break something than to fix it.
William Lallemand13e9b0c2016-05-25 02:34:07 +02005874
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005875http-request set-nice <nice> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02005876
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005877 This sets the "nice" factor of the current request being processed. It only
5878 has effect against the other requests being processed at the same time.
5879 The default value is 0, unless altered by the "nice" setting on the "bind"
5880 line. The accepted range is -1024..1024. The higher the value, the nicest
5881 the request will be. Lower values will make the request more important than
5882 other ones. This can be useful to improve the speed of some requests, or
5883 lower the priority of non-important requests. Using this setting without
5884 prior experimentation can cause some major slowdown.
William Lallemand13e9b0c2016-05-25 02:34:07 +02005885
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005886http-request set-path <fmt> [ { if | unless } <condition> ]
Willy Tarreau00005ce2016-10-21 15:07:45 +02005887
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005888 This rewrites the request path with the result of the evaluation of format
5889 string <fmt>. The query string, if any, is left intact. If a scheme and
5890 authority is found before the path, they are left intact as well. If the
5891 request doesn't have a path ("*"), this one is replaced with the format.
5892 This can be used to prepend a directory component in front of a path for
5893 example. See also "http-request set-query" and "http-request set-uri".
Willy Tarreau2d392c22015-08-24 01:43:45 +02005894
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005895 Example :
5896 # prepend the host name before the path
5897 http-request set-path /%[hdr(host)]%[path]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02005898
Christopher Faulet312294f2020-09-02 17:17:44 +02005899http-request set-pathq <fmt> [ { if | unless } <condition> ]
5900
5901 This does the same as "http-request set-path" except that the query-string is
5902 also rewritten. It may be used to remove the query-string, including the
5903 question mark (it is not possible using "http-request set-query").
5904
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005905http-request set-priority-class <expr> [ { if | unless } <condition> ]
Olivier Houchardccaa7de2017-10-02 11:51:03 +02005906
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005907 This is used to set the queue priority class of the current request.
5908 The value must be a sample expression which converts to an integer in the
5909 range -2047..2047. Results outside this range will be truncated.
5910 The priority class determines the order in which queued requests are
5911 processed. Lower values have higher priority.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02005912
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005913http-request set-priority-offset <expr> [ { if | unless } <condition> ]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02005914
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005915 This is used to set the queue priority timestamp offset of the current
5916 request. The value must be a sample expression which converts to an integer
5917 in the range -524287..524287. Results outside this range will be truncated.
5918 When a request is queued, it is ordered first by the priority class, then by
5919 the current timestamp adjusted by the given offset in milliseconds. Lower
5920 values have higher priority.
5921 Note that the resulting timestamp is is only tracked with enough precision
5922 for 524,287ms (8m44s287ms). If the request is queued long enough to where the
5923 adjusted timestamp exceeds this value, it will be misidentified as highest
5924 priority. Thus it is important to set "timeout queue" to a value, where when
5925 combined with the offset, does not exceed this limit.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02005926
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005927http-request set-query <fmt> [ { if | unless } <condition> ]
Willy Tarreau20b0de52012-12-24 15:45:22 +01005928
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005929 This rewrites the request's query string which appears after the first
5930 question mark ("?") with the result of the evaluation of format string <fmt>.
5931 The part prior to the question mark is left intact. If the request doesn't
5932 contain a question mark and the new value is not empty, then one is added at
5933 the end of the URI, followed by the new value. If a question mark was
5934 present, it will never be removed even if the value is empty. This can be
5935 used to add or remove parameters from the query string.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08005936
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005937 See also "http-request set-query" and "http-request set-uri".
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01005938
5939 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005940 # replace "%3D" with "=" in the query string
5941 http-request set-query %[query,regsub(%3D,=,g)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01005942
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005943http-request set-src <expr> [ { if | unless } <condition> ]
5944 This is used to set the source IP address to the value of specified
5945 expression. Useful when a proxy in front of HAProxy rewrites source IP, but
5946 provides the correct IP in a HTTP header; or you want to mask source IP for
Olivier Doucet56e31202020-04-21 09:32:56 +02005947 privacy. All subsequent calls to "src" fetch will return this value
5948 (see example).
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005949
5950 Arguments :
5951 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
5952 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01005953
Olivier Doucet56e31202020-04-21 09:32:56 +02005954 See also "option forwardfor".
5955
Cyril Bonté78caf842010-03-10 22:41:43 +01005956 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005957 http-request set-src hdr(x-forwarded-for)
5958 http-request set-src src,ipmask(24)
5959
Olivier Doucet56e31202020-04-21 09:32:56 +02005960 # After the masking this will track connections
5961 # based on the IP address with the last byte zeroed out.
5962 http-request track-sc0 src
5963
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005964 When possible, set-src preserves the original source port as long as the
5965 address family allows it, otherwise the source port is set to 0.
5966
5967http-request set-src-port <expr> [ { if | unless } <condition> ]
5968
5969 This is used to set the source port address to the value of specified
5970 expression.
5971
5972 Arguments:
5973 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
5974 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01005975
Willy Tarreau20b0de52012-12-24 15:45:22 +01005976 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005977 http-request set-src-port hdr(x-port)
5978 http-request set-src-port int(4000)
5979
5980 When possible, set-src-port preserves the original source address as long as
5981 the address family supports a port, otherwise it forces the source address to
5982 IPv4 "0.0.0.0" before rewriting the port.
5983
5984http-request set-tos <tos> [ { if | unless } <condition> ]
5985
5986 This is used to set the TOS or DSCP field value of packets sent to the client
5987 to the value passed in <tos> on platforms which support this. This value
5988 represents the whole 8 bits of the IP TOS field, and can be expressed both in
5989 decimal or hexadecimal format (prefixed by "0x"). Note that only the 6 higher
5990 bits are used in DSCP or TOS, and the two lower bits are always 0. This can
5991 be used to adjust some routing behavior on border routers based on some
5992 information from the request.
5993
5994 See RFC 2474, 2597, 3260 and 4594 for more information.
5995
5996http-request set-uri <fmt> [ { if | unless } <condition> ]
5997
5998 This rewrites the request URI with the result of the evaluation of format
5999 string <fmt>. The scheme, authority, path and query string are all replaced
6000 at once. This can be used to rewrite hosts in front of proxies, or to
6001 perform complex modifications to the URI such as moving parts between the
6002 path and the query string.
6003 See also "http-request set-path" and "http-request set-query".
6004
6005http-request set-var(<var-name>) <expr> [ { if | unless } <condition> ]
6006
6007 This is used to set the contents of a variable. The variable is declared
6008 inline.
6009
6010 Arguments:
6011 <var-name> The name of the variable starts with an indication about its
6012 scope. The scopes allowed are:
6013 "proc" : the variable is shared with the whole process
6014 "sess" : the variable is shared with the whole session
6015 "txn" : the variable is shared with the transaction
6016 (request and response)
6017 "req" : the variable is shared only during request
6018 processing
6019 "res" : the variable is shared only during response
6020 processing
6021 This prefix is followed by a name. The separator is a '.'.
6022 The name may only contain characters 'a-z', 'A-Z', '0-9'
6023 and '_'.
6024
6025 <expr> Is a standard HAProxy expression formed by a sample-fetch
6026 followed by some converters.
Willy Tarreau20b0de52012-12-24 15:45:22 +01006027
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006028 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006029 http-request set-var(req.my_var) req.fhdr(user-agent),lower
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006030
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006031http-request send-spoe-group <engine-name> <group-name>
6032 [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006033
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006034 This action is used to trigger sending of a group of SPOE messages. To do so,
6035 the SPOE engine used to send messages must be defined, as well as the SPOE
6036 group to send. Of course, the SPOE engine must refer to an existing SPOE
6037 filter. If not engine name is provided on the SPOE filter line, the SPOE
6038 agent name must be used.
6039
6040 Arguments:
6041 <engine-name> The SPOE engine name.
6042
6043 <group-name> The SPOE group name as specified in the engine
6044 configuration.
6045
6046http-request silent-drop [ { if | unless } <condition> ]
6047
6048 This stops the evaluation of the rules and makes the client-facing connection
6049 suddenly disappear using a system-dependent way that tries to prevent the
6050 client from being notified. The effect it then that the client still sees an
6051 established connection while there's none on HAProxy. The purpose is to
6052 achieve a comparable effect to "tarpit" except that it doesn't use any local
6053 resource at all on the machine running HAProxy. It can resist much higher
6054 loads than "tarpit", and slow down stronger attackers. It is important to
6055 understand the impact of using this mechanism. All stateful equipment placed
6056 between the client and HAProxy (firewalls, proxies, load balancers) will also
6057 keep the established connection for a long time and may suffer from this
6058 action.
6059 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
6060 option is used to block the emission of a TCP reset. On other systems, the
6061 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
6062 router, though it's still delivered to local networks. Do not use it unless
6063 you fully understand how it works.
6064
Christopher Faulet46f95542019-12-20 10:07:22 +01006065http-request strict-mode { on | off }
6066
6067 This enables or disables the strict rewriting mode for following rules. It
6068 does not affect rules declared before it and it is only applicable on rules
6069 performing a rewrite on the requests. When the strict mode is enabled, any
6070 rewrite failure triggers an internal error. Otherwise, such errors are
6071 silently ignored. The purpose of the strict rewriting mode is to make some
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05006072 rewrites optional while others must be performed to continue the request
Christopher Faulet46f95542019-12-20 10:07:22 +01006073 processing.
6074
Christopher Faulet1aea50e2020-01-17 16:03:53 +01006075 By default, the strict rewriting mode is enabled. Its value is also reset
Christopher Faulet46f95542019-12-20 10:07:22 +01006076 when a ruleset evaluation ends. So, for instance, if you change the mode on
6077 the frontend, the default mode is restored when HAProxy starts the backend
6078 rules evaluation.
6079
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006080http-request tarpit [deny_status <status>] [ { if | unless } <condition> ]
6081http-request tarpit [ { status | deny_status } <code>] [content-type <type>]
6082 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
6083 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
6084 [ hdr <name> <fmt> ]*
6085 [ { if | unless } <condition> ]
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006086
6087 This stops the evaluation of the rules and immediately blocks the request
6088 without responding for a delay specified by "timeout tarpit" or
6089 "timeout connect" if the former is not set. After that delay, if the client
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006090 is still connected, a response is returned so that the client does not
6091 suspect it has been tarpitted. Logs will report the flags "PT". The goal of
6092 the tarpit rule is to slow down robots during an attack when they're limited
6093 on the number of concurrent requests. It can be very efficient against very
6094 dumb robots, and will significantly reduce the load on firewalls compared to
6095 a "deny" rule. But when facing "correctly" developed robots, it can make
6096 things worse by forcing haproxy and the front firewall to support insane
6097 number of concurrent connections. By default an HTTP error 500 is returned.
6098 But the response may be customized using same syntax than
6099 "http-request return" rules. Thus, see "http-request return" for details.
Ilya Shipitsin11057a32020-06-21 21:18:27 +05006100 For compatibility purpose, when no argument is defined, or only "deny_status",
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006101 the argument "default-errorfiles" is implied. It means
6102 "http-request tarpit [deny_status <status>]" is an alias of
6103 "http-request tarpit [status <status>] default-errorfiles".
6104 No further "http-request" rules are evaluated.
6105 See also "http-request return" and "http-request silent-drop".
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006106
6107http-request track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
6108http-request track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
6109http-request track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
6110
6111 This enables tracking of sticky counters from current request. These rules do
6112 not stop evaluation and do not change default action. The number of counters
6113 that may be simultaneously tracked by the same connection is set in
6114 MAX_SESS_STKCTR at build time (reported in haproxy -vv) which defaults to 3,
Matteo Contrini1857b8c2020-10-16 17:35:54 +02006115 so the track-sc number is between 0 and (MAX_SESS_STKCTR-1). The first
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006116 "track-sc0" rule executed enables tracking of the counters of the specified
6117 table as the first set. The first "track-sc1" rule executed enables tracking
6118 of the counters of the specified table as the second set. The first
6119 "track-sc2" rule executed enables tracking of the counters of the specified
6120 table as the third set. It is a recommended practice to use the first set of
6121 counters for the per-frontend counters and the second set for the per-backend
6122 ones. But this is just a guideline, all may be used everywhere.
6123
6124 Arguments :
6125 <key> is mandatory, and is a sample expression rule as described in
6126 section 7.3. It describes what elements of the incoming request or
6127 connection will be analyzed, extracted, combined, and used to
6128 select which table entry to update the counters.
6129
6130 <table> is an optional table to be used instead of the default one, which
6131 is the stick-table declared in the current proxy. All the counters
6132 for the matches and updates for the key will then be performed in
6133 that table until the session ends.
6134
6135 Once a "track-sc*" rule is executed, the key is looked up in the table and if
6136 it is not found, an entry is allocated for it. Then a pointer to that entry
6137 is kept during all the session's life, and this entry's counters are updated
6138 as often as possible, every time the session's counters are updated, and also
6139 systematically when the session ends. Counters are only updated for events
6140 that happen after the tracking has been started. As an exception, connection
6141 counters and request counters are systematically updated so that they reflect
6142 useful information.
6143
6144 If the entry tracks concurrent connection counters, one connection is counted
6145 for as long as the entry is tracked, and the entry will not expire during
6146 that time. Tracking counters also provides a performance advantage over just
6147 checking the keys, because only one table lookup is performed for all ACL
6148 checks that make use of it.
6149
6150http-request unset-var(<var-name>) [ { if | unless } <condition> ]
6151
6152 This is used to unset a variable. See above for details about <var-name>.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006153
6154 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006155 http-request unset-var(req.my_var)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006156
Christopher Faulet579d83b2019-11-22 15:34:17 +01006157http-request use-service <service-name> [ { if | unless } <condition> ]
6158
6159 This directive executes the configured HTTP service to reply to the request
6160 and stops the evaluation of the rules. An HTTP service may choose to reply by
6161 sending any valid HTTP response or it may immediately close the connection
6162 without sending any response. Outside natives services, for instance the
6163 Prometheus exporter, it is possible to write your own services in Lua. No
6164 further "http-request" rules are evaluated.
6165
6166 Arguments :
6167 <service-name> is mandatory. It is the service to call
6168
6169 Example:
6170 http-request use-service prometheus-exporter if { path /metrics }
6171
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006172http-request wait-for-handshake [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006173
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006174 This will delay the processing of the request until the SSL handshake
6175 happened. This is mostly useful to delay processing early data until we're
6176 sure they are valid.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006177
Willy Tarreauef781042010-01-27 11:53:01 +01006178
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006179http-response <action> <options...> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02006180 Access control for Layer 7 responses
6181
6182 May be used in sections: defaults | frontend | listen | backend
6183 no | yes | yes | yes
6184
6185 The http-response statement defines a set of rules which apply to layer 7
6186 processing. The rules are evaluated in their declaration order when they are
6187 met in a frontend, listen or backend section. Any rule may optionally be
6188 followed by an ACL-based condition, in which case it will only be evaluated
6189 if the condition is true. Since these rules apply on responses, the backend
6190 rules are applied first, followed by the frontend's rules.
6191
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006192 The first keyword is the rule's action. The supported actions are described
6193 below.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02006194
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006195 There is no limit to the number of http-response statements per instance.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02006196
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006197 Example:
6198 acl key_acl res.hdr(X-Acl-Key) -m found
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02006199
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006200 acl myhost hdr(Host) -f myhost.lst
Sasha Pachev218f0642014-06-16 12:05:59 -06006201
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006202 http-response add-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
6203 http-response del-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
Sasha Pachev218f0642014-06-16 12:05:59 -06006204
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006205 Example:
6206 acl value res.hdr(X-Value) -m found
Sasha Pachev218f0642014-06-16 12:05:59 -06006207
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006208 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06006209
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006210 http-response set-map(map.lst) %[src] %[res.hdr(X-Value)] if value
6211 http-response del-map(map.lst) %[src] if ! value
Sasha Pachev218f0642014-06-16 12:05:59 -06006212
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006213 See also : "http-request", section 3.4 about userlists and section 7 about
6214 ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06006215
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006216http-response add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006217
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006218 This is used to add a new entry into an ACL. The ACL must be loaded from a
6219 file (even a dummy empty file). The file name of the ACL to be updated is
6220 passed between parentheses. It takes one argument: <key fmt>, which follows
6221 log-format rules, to collect content of the new entry. It performs a lookup
6222 in the ACL before insertion, to avoid duplicated (or more) values.
6223 This lookup is done by a linear search and can be expensive with large lists!
6224 It is the equivalent of the "add acl" command from the stats socket, but can
6225 be triggered by an HTTP response.
Sasha Pachev218f0642014-06-16 12:05:59 -06006226
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006227http-response add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006228
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006229 This appends an HTTP header field whose name is specified in <name> and whose
6230 value is defined by <fmt> which follows the log-format rules (see Custom Log
6231 Format in section 8.2.4). This may be used to send a cookie to a client for
6232 example, or to pass some internal information.
6233 This rule is not final, so it is possible to add other similar rules.
6234 Note that header addition is performed immediately, so one rule might reuse
6235 the resulting header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06006236
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006237http-response allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006238
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006239 This stops the evaluation of the rules and lets the response pass the check.
6240 No further "http-response" rules are evaluated for the current section.
Sasha Pachev218f0642014-06-16 12:05:59 -06006241
Jarno Huuskonen251a6b72019-01-04 14:05:02 +02006242http-response cache-store <name> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006243
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02006244 See section 6.2 about cache setup.
Sasha Pachev218f0642014-06-16 12:05:59 -06006245
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006246http-response capture <sample> id <id> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006247
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006248 This captures sample expression <sample> from the response buffer, and
6249 converts it to a string. The resulting string is stored into the next request
6250 "capture" slot, so it will possibly appear next to some captured HTTP
6251 headers. It will then automatically appear in the logs, and it will be
6252 possible to extract it using sample fetch rules to feed it into headers or
6253 anything. Please check section 7.3 (Fetching samples) and
6254 "capture response header" for more information.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02006255
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006256 The keyword "id" is the id of the capture slot which is used for storing the
6257 string. The capture slot must be defined in an associated frontend.
6258 This is useful to run captures in backends. The slot id can be declared by a
6259 previous directive "http-response capture" or with the "declare capture"
6260 keyword.
Baptiste Assmann19a69b32020-01-16 14:34:22 +01006261
6262 When using this action in a backend, double check that the relevant
6263 frontend(s) have the required capture slots otherwise, this rule will be
6264 ignored at run time. This can't be detected at configuration parsing time
6265 due to HAProxy's ability to dynamically resolve backend name at runtime.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02006266
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006267http-response del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02006268
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006269 This is used to delete an entry from an ACL. The ACL must be loaded from a
6270 file (even a dummy empty file). The file name of the ACL to be updated is
6271 passed between parentheses. It takes one argument: <key fmt>, which follows
6272 log-format rules, to collect content of the entry to delete.
6273 It is the equivalent of the "del acl" command from the stats socket, but can
6274 be triggered by an HTTP response.
Willy Tarreauf4c43c12013-06-11 17:01:13 +02006275
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006276http-response del-header <name> [ { if | unless } <condition> ]
Willy Tarreau9a355ec2013-06-11 17:45:46 +02006277
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006278 This removes all HTTP header fields whose name is specified in <name>.
Willy Tarreau42cf39e2013-06-11 18:51:32 +02006279
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006280http-response del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau51347ed2013-06-11 19:34:13 +02006281
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006282 This is used to delete an entry from a MAP. The MAP must be loaded from a
6283 file (even a dummy empty file). The file name of the MAP to be updated is
6284 passed between parentheses. It takes one argument: <key fmt>, which follows
6285 log-format rules, to collect content of the entry to delete.
6286 It takes one argument: "file name" It is the equivalent of the "del map"
6287 command from the stats socket, but can be triggered by an HTTP response.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006288
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006289http-response deny [deny_status <status>] [ { if | unless } <condition> ]
6290http-response deny [ { status | deny_status } <code>] [content-type <type>]
6291 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
6292 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
6293 [ hdr <name> <fmt> ]*
6294 [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006295
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006296 This stops the evaluation of the rules and immediately rejects the response.
6297 By default an HTTP 502 error is returned. But the response may be customized
6298 using same syntax than "http-response return" rules. Thus, see
Ilya Shipitsin11057a32020-06-21 21:18:27 +05006299 "http-response return" for details. For compatibility purpose, when no
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006300 argument is defined, or only "deny_status", the argument "default-errorfiles"
6301 is implied. It means "http-response deny [deny_status <status>]" is an alias
6302 of "http-response deny [status <status>] default-errorfiles".
Christopher Faulet040c8cd2020-01-13 16:43:45 +01006303 No further "http-response" rules are evaluated.
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006304 See also "http-response return".
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006305
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006306http-response redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006307
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006308 This performs an HTTP redirection based on a redirect rule.
6309 This supports a format string similarly to "http-request redirect" rules,
6310 with the exception that only the "location" type of redirect is possible on
6311 the response. See the "redirect" keyword for the rule's syntax. When a
6312 redirect rule is applied during a response, connections to the server are
6313 closed so that no data can be forwarded from the server to the client.
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02006314
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006315http-response replace-header <name> <regex-match> <replace-fmt>
6316 [ { if | unless } <condition> ]
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02006317
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006318 This works like "http-request replace-header" except that it works on the
6319 server's response instead of the client's request.
William Lallemand86d0df02017-11-24 21:36:45 +01006320
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006321 Example:
6322 http-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2
Willy Tarreau51d861a2015-05-22 17:30:48 +02006323
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006324 # applied to:
6325 Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006326
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006327 # outputs:
6328 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 +02006329
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006330 # assuming the backend IP is 192.168.1.20.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006331
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006332http-response replace-value <name> <regex-match> <replace-fmt>
6333 [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006334
Tim Duesterhus6bd909b2020-01-17 15:53:18 +01006335 This works like "http-request replace-value" except that it works on the
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006336 server's response instead of the client's request.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006337
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006338 Example:
6339 http-response replace-value Cache-control ^public$ private
Christopher Faulet85d79c92016-11-09 16:54:56 +01006340
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006341 # applied to:
6342 Cache-Control: max-age=3600, public
Christopher Faulet85d79c92016-11-09 16:54:56 +01006343
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006344 # outputs:
6345 Cache-Control: max-age=3600, private
Christopher Faulet85d79c92016-11-09 16:54:56 +01006346
Christopher Faulet24231ab2020-01-24 17:44:23 +01006347http-response return [status <code>] [content-type <type>]
6348 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
6349 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
Christopher Faulet4a2c1422020-01-31 17:36:01 +01006350 [ hdr <name> <value> ]*
Christopher Faulet24231ab2020-01-24 17:44:23 +01006351 [ { if | unless } <condition> ]
6352
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05006353 This stops the evaluation of the rules and immediately returns a response. The
Christopher Faulet24231ab2020-01-24 17:44:23 +01006354 default status code used for the response is 200. It can be optionally
6355 specified as an arguments to "status". The response content-type may also be
Daniel Corbett67a82712020-07-06 23:01:19 -04006356 specified as an argument to "content-type". Finally the response itself may
Christopher Faulet24231ab2020-01-24 17:44:23 +01006357 be defined. If can be a full HTTP response specifying the errorfile to use,
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05006358 or the response payload specifying the file or the string to use. These rules
Christopher Faulet24231ab2020-01-24 17:44:23 +01006359 are followed to create the response :
6360
6361 * If neither the errorfile nor the payload to use is defined, a dummy
6362 response is returned. Only the "status" argument is considered. It can be
6363 any code in the range [200, 599]. The "content-type" argument, if any, is
6364 ignored.
6365
6366 * If "default-errorfiles" argument is set, the proxy's errorfiles are
6367 considered. If the "status" argument is defined, it must be one of the
Daniel Corbett67a82712020-07-06 23:01:19 -04006368 status code handled by haproxy (200, 400, 403, 404, 405, 408, 410, 413,
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02006369 425, 429, 500, 502, 503, and 504). The "content-type" argument, if any,
6370 is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006371
6372 * If a specific errorfile is defined, with an "errorfile" argument, the
6373 corresponding file, containing a full HTTP response, is returned. Only the
6374 "status" argument is considered. It must be one of the status code handled
Daniel Corbett67a82712020-07-06 23:01:19 -04006375 by haproxy (200, 400, 403, 404, 405, 408, 410, 413, 425, 429, 500, 502, 503,
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02006376 and 504). The "content-type" argument, if any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006377
6378 * If an http-errors section is defined, with an "errorfiles" argument, the
6379 corresponding file in the specified http-errors section, containing a full
6380 HTTP response, is returned. Only the "status" argument is considered. It
Daniel Corbett67a82712020-07-06 23:01:19 -04006381 must be one of the status code handled by haproxy (200, 400, 403, 404, 405,
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02006382 408, 410, 413, 425, 429, 500, 502, 503, and 504). The "content-type"
6383 argument, if any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006384
6385 * If a "file" or a "lf-file" argument is specified, the file's content is
6386 used as the response payload. If the file is not empty, its content-type
6387 must be set as argument to "content-type". Otherwise, any "content-type"
6388 argument is ignored. With a "lf-file" argument, the file's content is
6389 evaluated as a log-format string. With a "file" argument, it is considered
6390 as a raw content.
6391
6392 * If a "string" or "lf-string" argument is specified, the defined string is
6393 used as the response payload. The content-type must always be set as
6394 argument to "content-type". With a "lf-string" argument, the string is
6395 evaluated as a log-format string. With a "string" argument, it is
6396 considered as a raw string.
6397
Christopher Faulet4a2c1422020-01-31 17:36:01 +01006398 When the response is not based an errorfile, it is possible to appends HTTP
6399 header fields to the response using "hdr" arguments. Otherwise, all "hdr"
6400 arguments are ignored. For each one, the header name is specified in <name>
6401 and its value is defined by <fmt> which follows the log-format rules.
6402
Christopher Faulet24231ab2020-01-24 17:44:23 +01006403 Note that the generated response must be smaller than a buffer. And to avoid
6404 any warning, when an errorfile or a raw file is loaded, the buffer space
Ilya Shipitsin11057a32020-06-21 21:18:27 +05006405 reserved to the headers rewriting should also be free.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006406
6407 No further "http-response" rules are evaluated.
6408
6409 Example:
Daniel Corbett67a82712020-07-06 23:01:19 -04006410 http-response return errorfile /etc/haproxy/errorfiles/200.http \
Christopher Faulet24231ab2020-01-24 17:44:23 +01006411 if { status eq 404 }
6412
6413 http-response return content-type text/plain \
6414 string "This is the end !" \
6415 if { status eq 500 }
6416
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006417http-response sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
6418http-response sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Ruoshan Huange4edc6b2016-07-14 15:07:45 +08006419
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006420 This action increments the GPC0 or GPC1 counter according with the sticky
6421 counter designated by <sc-id>. If an error occurs, this action silently fails
6422 and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +02006423
Cédric Dufour0d7712d2019-11-06 18:38:53 +01006424http-response sc-set-gpt0(<sc-id>) { <int> | <expr> }
6425 [ { if | unless } <condition> ]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02006426
Cédric Dufour0d7712d2019-11-06 18:38:53 +01006427 This action sets the 32-bit unsigned GPT0 tag according to the sticky counter
6428 designated by <sc-id> and the value of <int>/<expr>. The expected result is a
6429 boolean. If an error occurs, this action silently fails and the actions
6430 evaluation continues.
Frédéric Lécaille6778b272018-01-29 15:22:53 +01006431
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006432http-response send-spoe-group [ { if | unless } <condition> ]
Willy Tarreau2d392c22015-08-24 01:43:45 +02006433
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006434 This action is used to trigger sending of a group of SPOE messages. To do so,
6435 the SPOE engine used to send messages must be defined, as well as the SPOE
6436 group to send. Of course, the SPOE engine must refer to an existing SPOE
6437 filter. If not engine name is provided on the SPOE filter line, the SPOE
6438 agent name must be used.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006439
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006440 Arguments:
6441 <engine-name> The SPOE engine name.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006442
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006443 <group-name> The SPOE group name as specified in the engine
6444 configuration.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006445
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006446http-response set-header <name> <fmt> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02006447
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006448 This does the same as "add-header" except that the header name is first
6449 removed if it existed. This is useful when passing security information to
6450 the server, where the header must not be manipulated by external users.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02006451
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006452http-response set-log-level <level> [ { if | unless } <condition> ]
6453
6454 This is used to change the log level of the current request when a certain
6455 condition is met. Valid levels are the 8 syslog levels (see the "log"
6456 keyword) plus the special level "silent" which disables logging for this
6457 request. This rule is not final so the last matching rule wins. This rule can
6458 be useful to disable health checks coming from another equipment.
6459
6460http-response set-map(<file-name>) <key fmt> <value fmt>
6461
6462 This is used to add a new entry into a MAP. The MAP must be loaded from a
6463 file (even a dummy empty file). The file name of the MAP to be updated is
6464 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
6465 log-format rules, used to collect MAP key, and <value fmt>, which follows
6466 log-format rules, used to collect content for the new entry. It performs a
6467 lookup in the MAP before insertion, to avoid duplicated (or more) values.
6468 This lookup is done by a linear search and can be expensive with large lists!
6469 It is the equivalent of the "set map" command from the stats socket, but can
6470 be triggered by an HTTP response.
6471
6472http-response set-mark <mark> [ { if | unless } <condition> ]
6473
6474 This is used to set the Netfilter MARK on all packets sent to the client to
6475 the value passed in <mark> on platforms which support it. This value is an
6476 unsigned 32 bit value which can be matched by netfilter and by the routing
6477 table. It can be expressed both in decimal or hexadecimal format (prefixed
6478 by "0x"). This can be useful to force certain packets to take a different
6479 route (for example a cheaper network path for bulk downloads). This works on
6480 Linux kernels 2.6.32 and above and requires admin privileges.
6481
6482http-response set-nice <nice> [ { if | unless } <condition> ]
6483
6484 This sets the "nice" factor of the current request being processed.
6485 It only has effect against the other requests being processed at the same
6486 time. The default value is 0, unless altered by the "nice" setting on the
6487 "bind" line. The accepted range is -1024..1024. The higher the value, the
6488 nicest the request will be. Lower values will make the request more important
6489 than other ones. This can be useful to improve the speed of some requests, or
6490 lower the priority of non-important requests. Using this setting without
6491 prior experimentation can cause some major slowdown.
6492
6493http-response set-status <status> [reason <str>]
6494 [ { if | unless } <condition> ]
6495
6496 This replaces the response status code with <status> which must be an integer
6497 between 100 and 999. Optionally, a custom reason text can be provided defined
6498 by <str>, or the default reason for the specified code will be used as a
6499 fallback.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08006500
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006501 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006502 # return "431 Request Header Fields Too Large"
6503 http-response set-status 431
6504 # return "503 Slow Down", custom reason
6505 http-response set-status 503 reason "Slow Down".
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006506
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006507http-response set-tos <tos> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006508
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006509 This is used to set the TOS or DSCP field value of packets sent to the client
6510 to the value passed in <tos> on platforms which support this.
6511 This value represents the whole 8 bits of the IP TOS field, and can be
6512 expressed both in decimal or hexadecimal format (prefixed by "0x"). Note that
6513 only the 6 higher bits are used in DSCP or TOS, and the two lower bits are
6514 always 0. This can be used to adjust some routing behavior on border routers
6515 based on some information from the request.
6516
6517 See RFC 2474, 2597, 3260 and 4594 for more information.
6518
6519http-response set-var(<var-name>) <expr> [ { if | unless } <condition> ]
6520
6521 This is used to set the contents of a variable. The variable is declared
6522 inline.
6523
6524 Arguments:
6525 <var-name> The name of the variable starts with an indication about its
6526 scope. The scopes allowed are:
6527 "proc" : the variable is shared with the whole process
6528 "sess" : the variable is shared with the whole session
6529 "txn" : the variable is shared with the transaction
6530 (request and response)
6531 "req" : the variable is shared only during request
6532 processing
6533 "res" : the variable is shared only during response
6534 processing
6535 This prefix is followed by a name. The separator is a '.'.
6536 The name may only contain characters 'a-z', 'A-Z', '0-9', '.'
6537 and '_'.
6538
6539 <expr> Is a standard HAProxy expression formed by a sample-fetch
6540 followed by some converters.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006541
6542 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006543 http-response set-var(sess.last_redir) res.hdr(location)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006544
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006545http-response silent-drop [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006546
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006547 This stops the evaluation of the rules and makes the client-facing connection
6548 suddenly disappear using a system-dependent way that tries to prevent the
6549 client from being notified. The effect it then that the client still sees an
6550 established connection while there's none on HAProxy. The purpose is to
6551 achieve a comparable effect to "tarpit" except that it doesn't use any local
6552 resource at all on the machine running HAProxy. It can resist much higher
6553 loads than "tarpit", and slow down stronger attackers. It is important to
6554 understand the impact of using this mechanism. All stateful equipment placed
6555 between the client and HAProxy (firewalls, proxies, load balancers) will also
6556 keep the established connection for a long time and may suffer from this
6557 action.
6558 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
6559 option is used to block the emission of a TCP reset. On other systems, the
6560 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
6561 router, though it's still delivered to local networks. Do not use it unless
6562 you fully understand how it works.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006563
Christopher Faulet46f95542019-12-20 10:07:22 +01006564http-response strict-mode { on | off }
6565
6566 This enables or disables the strict rewriting mode for following rules. It
6567 does not affect rules declared before it and it is only applicable on rules
6568 performing a rewrite on the responses. When the strict mode is enabled, any
6569 rewrite failure triggers an internal error. Otherwise, such errors are
6570 silently ignored. The purpose of the strict rewriting mode is to make some
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05006571 rewrites optional while others must be performed to continue the response
Christopher Faulet46f95542019-12-20 10:07:22 +01006572 processing.
6573
Christopher Faulet1aea50e2020-01-17 16:03:53 +01006574 By default, the strict rewriting mode is enabled. Its value is also reset
Christopher Faulet46f95542019-12-20 10:07:22 +01006575 when a ruleset evaluation ends. So, for instance, if you change the mode on
Daniel Corbett67a82712020-07-06 23:01:19 -04006576 the backend, the default mode is restored when HAProxy starts the frontend
Christopher Faulet46f95542019-12-20 10:07:22 +01006577 rules evaluation.
6578
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006579http-response track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
6580http-response track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
6581http-response track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02006582
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006583 This enables tracking of sticky counters from current response. Please refer
6584 to "http-request track-sc" for a complete description. The only difference
6585 from "http-request track-sc" is the <key> sample expression can only make use
6586 of samples in response (e.g. res.*, status etc.) and samples below Layer 6
6587 (e.g. SSL-related samples, see section 7.3.4). If the sample is not
6588 supported, haproxy will fail and warn while parsing the config.
6589
6590http-response unset-var(<var-name>) [ { if | unless } <condition> ]
6591
6592 This is used to unset a variable. See "http-response set-var" for details
6593 about <var-name>.
6594
6595 Example:
6596 http-response unset-var(sess.last_redir)
6597
Baptiste Assmann5ecb77f2013-10-06 23:24:13 +02006598
Willy Tarreau30631952015-08-06 15:05:24 +02006599http-reuse { never | safe | aggressive | always }
6600 Declare how idle HTTP connections may be shared between requests
6601
6602 May be used in sections: defaults | frontend | listen | backend
6603 yes | no | yes | yes
6604
6605 By default, a connection established between haproxy and the backend server
Olivier Houchard86006a52018-12-14 19:37:49 +01006606 which is considered safe for reuse is moved back to the server's idle
6607 connections pool so that any other request can make use of it. This is the
6608 "safe" strategy below.
Willy Tarreau30631952015-08-06 15:05:24 +02006609
6610 The argument indicates the desired connection reuse strategy :
6611
Olivier Houchard86006a52018-12-14 19:37:49 +01006612 - "never" : idle connections are never shared between sessions. This mode
6613 may be enforced to cancel a different strategy inherited from
6614 a defaults section or for troubleshooting. For example, if an
6615 old bogus application considers that multiple requests over
6616 the same connection come from the same client and it is not
6617 possible to fix the application, it may be desirable to
6618 disable connection sharing in a single backend. An example of
6619 such an application could be an old haproxy using cookie
6620 insertion in tunnel mode and not checking any request past the
6621 first one.
Willy Tarreau30631952015-08-06 15:05:24 +02006622
Olivier Houchard86006a52018-12-14 19:37:49 +01006623 - "safe" : this is the default and the recommended strategy. The first
6624 request of a session is always sent over its own connection,
6625 and only subsequent requests may be dispatched over other
6626 existing connections. This ensures that in case the server
6627 closes the connection when the request is being sent, the
6628 browser can decide to silently retry it. Since it is exactly
6629 equivalent to regular keep-alive, there should be no side
Amaury Denoyelle27179652020-10-14 18:17:12 +02006630 effects. There is also a special handling for the connections
6631 using protocols subject to Head-of-line blocking (backend with
6632 h2 or fcgi). In this case, when at least one stream is
6633 processed, the used connection is reserved to handle streams
6634 of the same session. When no more streams are processed, the
6635 connection is released and can be reused.
Willy Tarreau30631952015-08-06 15:05:24 +02006636
6637 - "aggressive" : this mode may be useful in webservices environments where
6638 all servers are not necessarily known and where it would be
6639 appreciable to deliver most first requests over existing
6640 connections. In this case, first requests are only delivered
6641 over existing connections that have been reused at least once,
6642 proving that the server correctly supports connection reuse.
6643 It should only be used when it's sure that the client can
6644 retry a failed request once in a while and where the benefit
Michael Prokop4438c602019-05-24 10:25:45 +02006645 of aggressive connection reuse significantly outweighs the
Willy Tarreau30631952015-08-06 15:05:24 +02006646 downsides of rare connection failures.
6647
6648 - "always" : this mode is only recommended when the path to the server is
6649 known for never breaking existing connections quickly after
6650 releasing them. It allows the first request of a session to be
6651 sent to an existing connection. This can provide a significant
6652 performance increase over the "safe" strategy when the backend
6653 is a cache farm, since such components tend to show a
Davor Ocelice9ed2812017-12-25 17:49:28 +01006654 consistent behavior and will benefit from the connection
Willy Tarreau30631952015-08-06 15:05:24 +02006655 sharing. It is recommended that the "http-keep-alive" timeout
6656 remains low in this mode so that no dead connections remain
6657 usable. In most cases, this will lead to the same performance
6658 gains as "aggressive" but with more risks. It should only be
6659 used when it improves the situation over "aggressive".
6660
6661 When http connection sharing is enabled, a great care is taken to respect the
Davor Ocelice9ed2812017-12-25 17:49:28 +01006662 connection properties and compatibility. Specifically :
6663 - connections made with "usesrc" followed by a client-dependent value
6664 ("client", "clientip", "hdr_ip") are marked private and never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02006665
Amaury Denoyelle7239c242020-10-15 16:41:09 +02006666 - connections sent to a server with a variable value as TLS SNI extension
6667 are marked private and are never shared. This is not the case if the SNI
6668 is guaranteed to be a constant, as for example using a literal string;
Willy Tarreau30631952015-08-06 15:05:24 +02006669
Lukas Tribusfd9b68c2018-10-27 20:06:59 +02006670 - connections with certain bogus authentication schemes (relying on the
6671 connection) like NTLM are detected, marked private and are never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02006672
Lukas Tribuse8adfeb2019-11-06 11:50:25 +01006673 A connection pool is involved and configurable with "pool-max-conn".
Willy Tarreau30631952015-08-06 15:05:24 +02006674
6675 Note: connection reuse improves the accuracy of the "server maxconn" setting,
6676 because almost no new connection will be established while idle connections
6677 remain available. This is particularly true with the "always" strategy.
6678
6679 See also : "option http-keep-alive", "server maxconn"
6680
6681
Mark Lamourinec2247f02012-01-04 13:02:01 -05006682http-send-name-header [<header>]
6683 Add the server name to a request. Use the header string given by <header>
Mark Lamourinec2247f02012-01-04 13:02:01 -05006684 May be used in sections: defaults | frontend | listen | backend
6685 yes | no | yes | yes
Mark Lamourinec2247f02012-01-04 13:02:01 -05006686 Arguments :
Mark Lamourinec2247f02012-01-04 13:02:01 -05006687 <header> The header string to use to send the server name
6688
Willy Tarreau81bef7e2019-10-07 14:58:02 +02006689 The "http-send-name-header" statement causes the header field named <header>
6690 to be set to the name of the target server at the moment the request is about
6691 to be sent on the wire. Any existing occurrences of this header are removed.
6692 Upon retries and redispatches, the header field is updated to always reflect
6693 the server being attempted to connect to. Given that this header is modified
6694 very late in the connection setup, it may have unexpected effects on already
6695 modified headers. For example using it with transport-level header such as
6696 connection, content-length, transfer-encoding and so on will likely result in
6697 invalid requests being sent to the server. Additionally it has been reported
6698 that this directive is currently being used as a way to overwrite the Host
6699 header field in outgoing requests; while this trick has been known to work
6700 as a side effect of the feature for some time, it is not officially supported
6701 and might possibly not work anymore in a future version depending on the
6702 technical difficulties this feature induces. A long-term solution instead
6703 consists in fixing the application which required this trick so that it binds
6704 to the correct host name.
Mark Lamourinec2247f02012-01-04 13:02:01 -05006705
6706 See also : "server"
6707
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01006708id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +02006709 Set a persistent ID to a proxy.
6710 May be used in sections : defaults | frontend | listen | backend
6711 no | yes | yes | yes
6712 Arguments : none
6713
6714 Set a persistent ID for the proxy. This ID must be unique and positive.
6715 An unused ID will automatically be assigned if unset. The first assigned
6716 value will be 1. This ID is currently only returned in statistics.
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01006717
6718
Cyril Bonté0d4bf012010-04-25 23:21:46 +02006719ignore-persist { if | unless } <condition>
6720 Declare a condition to ignore persistence
6721 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01006722 no | no | yes | yes
Cyril Bonté0d4bf012010-04-25 23:21:46 +02006723
6724 By default, when cookie persistence is enabled, every requests containing
6725 the cookie are unconditionally persistent (assuming the target server is up
6726 and running).
6727
6728 The "ignore-persist" statement allows one to declare various ACL-based
6729 conditions which, when met, will cause a request to ignore persistence.
6730 This is sometimes useful to load balance requests for static files, which
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03006731 often don't require persistence. This can also be used to fully disable
Cyril Bonté0d4bf012010-04-25 23:21:46 +02006732 persistence for a specific User-Agent (for example, some web crawler bots).
6733
Cyril Bonté0d4bf012010-04-25 23:21:46 +02006734 The persistence is ignored when an "if" condition is met, or unless an
6735 "unless" condition is met.
6736
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03006737 Example:
6738 acl url_static path_beg /static /images /img /css
6739 acl url_static path_end .gif .png .jpg .css .js
6740 ignore-persist if url_static
6741
Cyril Bonté0d4bf012010-04-25 23:21:46 +02006742 See also : "force-persist", "cookie", and section 7 about ACL usage.
6743
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02006744load-server-state-from-file { global | local | none }
6745 Allow seamless reload of HAProxy
6746 May be used in sections: defaults | frontend | listen | backend
6747 yes | no | yes | yes
6748
6749 This directive points HAProxy to a file where server state from previous
6750 running process has been saved. That way, when starting up, before handling
6751 traffic, the new process can apply old states to servers exactly has if no
Davor Ocelice9ed2812017-12-25 17:49:28 +01006752 reload occurred. The purpose of the "load-server-state-from-file" directive is
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02006753 to tell haproxy which file to use. For now, only 2 arguments to either prevent
6754 loading state or load states from a file containing all backends and servers.
6755 The state file can be generated by running the command "show servers state"
6756 over the stats socket and redirect output.
6757
Davor Ocelice9ed2812017-12-25 17:49:28 +01006758 The format of the file is versioned and is very specific. To understand it,
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02006759 please read the documentation of the "show servers state" command (chapter
Willy Tarreau1af20c72017-06-23 16:01:14 +02006760 9.3 of Management Guide).
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02006761
6762 Arguments:
6763 global load the content of the file pointed by the global directive
6764 named "server-state-file".
6765
6766 local load the content of the file pointed by the directive
6767 "server-state-file-name" if set. If not set, then the backend
6768 name is used as a file name.
6769
6770 none don't load any stat for this backend
6771
6772 Notes:
Willy Tarreaue5a60682016-11-09 14:54:53 +01006773 - server's IP address is preserved across reloads by default, but the
6774 order can be changed thanks to the server's "init-addr" setting. This
6775 means that an IP address change performed on the CLI at run time will
Davor Ocelice9ed2812017-12-25 17:49:28 +01006776 be preserved, and that any change to the local resolver (e.g. /etc/hosts)
Willy Tarreaue5a60682016-11-09 14:54:53 +01006777 will possibly not have any effect if the state file is in use.
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02006778
6779 - server's weight is applied from previous running process unless it has
6780 has changed between previous and new configuration files.
6781
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02006782 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02006783
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02006784 global
6785 stats socket /tmp/socket
6786 server-state-file /tmp/server_state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02006787
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02006788 defaults
6789 load-server-state-from-file global
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02006790
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02006791 backend bk
6792 server s1 127.0.0.1:22 check weight 11
6793 server s2 127.0.0.1:22 check weight 12
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02006794
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02006795
6796 Then one can run :
6797
6798 socat /tmp/socket - <<< "show servers state" > /tmp/server_state
6799
6800 Content of the file /tmp/server_state would be like this:
6801
6802 1
6803 # <field names skipped for the doc example>
6804 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
6805 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
6806
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02006807 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02006808
6809 global
6810 stats socket /tmp/socket
6811 server-state-base /etc/haproxy/states
6812
6813 defaults
6814 load-server-state-from-file local
6815
6816 backend bk
6817 server s1 127.0.0.1:22 check weight 11
6818 server s2 127.0.0.1:22 check weight 12
6819
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02006820
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02006821 Then one can run :
6822
6823 socat /tmp/socket - <<< "show servers state bk" > /etc/haproxy/states/bk
6824
6825 Content of the file /etc/haproxy/states/bk would be like this:
6826
6827 1
6828 # <field names skipped for the doc example>
6829 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
6830 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
6831
6832 See also: "server-state-file", "server-state-file-name", and
6833 "show servers state"
6834
Cyril Bonté0d4bf012010-04-25 23:21:46 +02006835
Willy Tarreau2769aa02007-12-27 18:26:09 +01006836log global
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02006837log <address> [len <length>] [format <format>] [sample <ranges>:<smp_size>]
6838 <facility> [<level> [<minlevel>]]
William Lallemand0f99e342011-10-12 17:50:54 +02006839no log
Willy Tarreau2769aa02007-12-27 18:26:09 +01006840 Enable per-instance logging of events and traffic.
6841 May be used in sections : defaults | frontend | listen | backend
6842 yes | yes | yes | yes
William Lallemand0f99e342011-10-12 17:50:54 +02006843
6844 Prefix :
6845 no should be used when the logger list must be flushed. For example,
6846 if you don't want to inherit from the default logger list. This
6847 prefix does not allow arguments.
6848
Willy Tarreau2769aa02007-12-27 18:26:09 +01006849 Arguments :
6850 global should be used when the instance's logging parameters are the
6851 same as the global ones. This is the most common usage. "global"
6852 replaces <address>, <facility> and <level> with those of the log
6853 entries found in the "global" section. Only one "log global"
6854 statement may be used per instance, and this form takes no other
6855 parameter.
6856
6857 <address> indicates where to send the logs. It takes the same format as
6858 for the "global" section's logs, and can be one of :
6859
6860 - An IPv4 address optionally followed by a colon (':') and a UDP
6861 port. If no port is specified, 514 is used by default (the
6862 standard syslog port).
6863
David du Colombier24bb5f52011-03-17 10:40:23 +01006864 - An IPv6 address followed by a colon (':') and optionally a UDP
6865 port. If no port is specified, 514 is used by default (the
6866 standard syslog port).
6867
Willy Tarreau2769aa02007-12-27 18:26:09 +01006868 - A filesystem path to a UNIX domain socket, keeping in mind
6869 considerations for chroot (be sure the path is accessible
6870 inside the chroot) and uid/gid (be sure the path is
Davor Ocelice9ed2812017-12-25 17:49:28 +01006871 appropriately writable).
Willy Tarreau2769aa02007-12-27 18:26:09 +01006872
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01006873 - A file descriptor number in the form "fd@<number>", which may
6874 point to a pipe, terminal, or socket. In this case unbuffered
6875 logs are used and one writev() call per log is performed. This
6876 is a bit expensive but acceptable for most workloads. Messages
6877 sent this way will not be truncated but may be dropped, in
6878 which case the DroppedLogs counter will be incremented. The
6879 writev() call is atomic even on pipes for messages up to
6880 PIPE_BUF size, which POSIX recommends to be at least 512 and
6881 which is 4096 bytes on most modern operating systems. Any
6882 larger message may be interleaved with messages from other
6883 processes. Exceptionally for debugging purposes the file
6884 descriptor may also be directed to a file, but doing so will
6885 significantly slow haproxy down as non-blocking calls will be
6886 ignored. Also there will be no way to purge nor rotate this
6887 file without restarting the process. Note that the configured
6888 syslog format is preserved, so the output is suitable for use
Willy Tarreauc1b06452018-11-12 11:57:56 +01006889 with a TCP syslog server. See also the "short" and "raw"
6890 formats below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01006891
6892 - "stdout" / "stderr", which are respectively aliases for "fd@1"
6893 and "fd@2", see above.
6894
Willy Tarreauc046d162019-08-30 15:24:59 +02006895 - A ring buffer in the form "ring@<name>", which will correspond
6896 to an in-memory ring buffer accessible over the CLI using the
6897 "show events" command, which will also list existing rings and
6898 their sizes. Such buffers are lost on reload or restart but
6899 when used as a complement this can help troubleshooting by
6900 having the logs instantly available.
6901
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01006902 You may want to reference some environment variables in the
6903 address parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01006904
Willy Tarreau18324f52014-06-27 18:10:07 +02006905 <length> is an optional maximum line length. Log lines larger than this
6906 value will be truncated before being sent. The reason is that
6907 syslog servers act differently on log line length. All servers
6908 support the default value of 1024, but some servers simply drop
6909 larger lines while others do log them. If a server supports long
6910 lines, it may make sense to set this value here in order to avoid
6911 truncating long lines. Similarly, if a server drops long lines,
6912 it is preferable to truncate them before sending them. Accepted
6913 values are 80 to 65535 inclusive. The default value of 1024 is
6914 generally fine for all standard usages. Some specific cases of
Davor Ocelice9ed2812017-12-25 17:49:28 +01006915 long captures or JSON-formatted logs may require larger values.
Willy Tarreau18324f52014-06-27 18:10:07 +02006916
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02006917 <ranges> A list of comma-separated ranges to identify the logs to sample.
6918 This is used to balance the load of the logs to send to the log
6919 server. The limits of the ranges cannot be null. They are numbered
6920 from 1. The size or period (in number of logs) of the sample must
6921 be set with <sample_size> parameter.
6922
6923 <sample_size>
6924 The size of the sample in number of logs to consider when balancing
6925 their logging loads. It is used to balance the load of the logs to
6926 send to the syslog server. This size must be greater or equal to the
6927 maximum of the high limits of the ranges.
6928 (see also <ranges> parameter).
6929
Willy Tarreauadb345d2018-11-12 07:56:13 +01006930 <format> is the log format used when generating syslog messages. It may be
6931 one of the following :
6932
6933 rfc3164 The RFC3164 syslog message format. This is the default.
6934 (https://tools.ietf.org/html/rfc3164)
6935
6936 rfc5424 The RFC5424 syslog message format.
6937 (https://tools.ietf.org/html/rfc5424)
6938
Emeric Brun54648852020-07-06 15:54:06 +02006939 priority A message containing only a level plus syslog facility between
6940 angle brackets such as '<63>', followed by the text. The PID,
6941 date, time, process name and system name are omitted. This is
6942 designed to be used with a local log server.
6943
Willy Tarreaue8746a02018-11-12 08:45:00 +01006944 short A message containing only a level between angle brackets such as
6945 '<3>', followed by the text. The PID, date, time, process name
6946 and system name are omitted. This is designed to be used with a
6947 local log server. This format is compatible with what the
6948 systemd logger consumes.
6949
Emeric Brun54648852020-07-06 15:54:06 +02006950 timed A message containing only a level between angle brackets such as
6951 '<3>', followed by ISO date and by the text. The PID, process
6952 name and system name are omitted. This is designed to be
6953 used with a local log server.
6954
6955 iso A message containing only the ISO date, followed by the text.
6956 The PID, process name and system name are omitted. This is
6957 designed to be used with a local log server.
6958
Willy Tarreauc1b06452018-11-12 11:57:56 +01006959 raw A message containing only the text. The level, PID, date, time,
6960 process name and system name are omitted. This is designed to
6961 be used in containers or during development, where the severity
6962 only depends on the file descriptor used (stdout/stderr).
6963
Willy Tarreau2769aa02007-12-27 18:26:09 +01006964 <facility> must be one of the 24 standard syslog facilities :
6965
Willy Tarreaue8746a02018-11-12 08:45:00 +01006966 kern user mail daemon auth syslog lpr news
6967 uucp cron auth2 ftp ntp audit alert cron2
6968 local0 local1 local2 local3 local4 local5 local6 local7
6969
Willy Tarreauc1b06452018-11-12 11:57:56 +01006970 Note that the facility is ignored for the "short" and "raw"
6971 formats, but still required as a positional field. It is
6972 recommended to use "daemon" in this case to make it clear that
6973 it's only supposed to be used locally.
Willy Tarreau2769aa02007-12-27 18:26:09 +01006974
6975 <level> is optional and can be specified to filter outgoing messages. By
6976 default, all messages are sent. If a level is specified, only
6977 messages with a severity at least as important as this level
Willy Tarreauf7edefa2009-05-10 17:20:05 +02006978 will be sent. An optional minimum level can be specified. If it
6979 is set, logs emitted with a more severe level than this one will
6980 be capped to this level. This is used to avoid sending "emerg"
6981 messages on all terminals on some default syslog configurations.
6982 Eight levels are known :
Willy Tarreau2769aa02007-12-27 18:26:09 +01006983
6984 emerg alert crit err warning notice info debug
6985
William Lallemand0f99e342011-10-12 17:50:54 +02006986 It is important to keep in mind that it is the frontend which decides what to
6987 log from a connection, and that in case of content switching, the log entries
6988 from the backend will be ignored. Connections are logged at level "info".
Willy Tarreaucc6c8912009-02-22 10:53:55 +01006989
6990 However, backend log declaration define how and where servers status changes
6991 will be logged. Level "notice" will be used to indicate a server going up,
6992 "warning" will be used for termination signals and definitive service
6993 termination, and "alert" will be used for when a server goes down.
6994
6995 Note : According to RFC3164, messages are truncated to 1024 bytes before
6996 being emitted.
Willy Tarreau2769aa02007-12-27 18:26:09 +01006997
6998 Example :
6999 log global
Willy Tarreauc1b06452018-11-12 11:57:56 +01007000 log stdout format short daemon # send log to systemd
7001 log stdout format raw daemon # send everything to stdout
7002 log stderr format raw daemon notice # send important events to stderr
Willy Tarreauf7edefa2009-05-10 17:20:05 +02007003 log 127.0.0.1:514 local0 notice # only send important events
7004 log 127.0.0.1:514 local0 notice notice # same but limit output level
William Lallemandb2f07452015-05-12 14:27:13 +02007005 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
Willy Tarreaudad36a32013-03-11 01:20:04 +01007006
Willy Tarreau2769aa02007-12-27 18:26:09 +01007007
William Lallemand48940402012-01-30 16:47:22 +01007008log-format <string>
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01007009 Specifies the log format string to use for traffic logs
7010 May be used in sections: defaults | frontend | listen | backend
7011 yes | yes | yes | no
William Lallemand48940402012-01-30 16:47:22 +01007012
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01007013 This directive specifies the log format string that will be used for all logs
7014 resulting from traffic passing through the frontend using this line. If the
7015 directive is used in a defaults section, all subsequent frontends will use
7016 the same log format. Please see section 8.2.4 which covers the log format
7017 string in depth.
William Lallemand48940402012-01-30 16:47:22 +01007018
Guillaume de Lafond29f45602017-03-31 19:52:15 +02007019 "log-format" directive overrides previous "option tcplog", "log-format" and
7020 "option httplog" directives.
7021
Dragan Dosen7ad31542015-09-28 17:16:47 +02007022log-format-sd <string>
7023 Specifies the RFC5424 structured-data log format string
7024 May be used in sections: defaults | frontend | listen | backend
7025 yes | yes | yes | no
7026
7027 This directive specifies the RFC5424 structured-data log format string that
7028 will be used for all logs resulting from traffic passing through the frontend
7029 using this line. If the directive is used in a defaults section, all
7030 subsequent frontends will use the same log format. Please see section 8.2.4
7031 which covers the log format string in depth.
7032
7033 See https://tools.ietf.org/html/rfc5424#section-6.3 for more information
7034 about the RFC5424 structured-data part.
7035
7036 Note : This log format string will be used only for loggers that have set
7037 log format to "rfc5424".
7038
7039 Example :
7040 log-format-sd [exampleSDID@1234\ bytes=\"%B\"\ status=\"%ST\"]
7041
7042
Willy Tarreau094af4e2015-01-07 15:03:42 +01007043log-tag <string>
7044 Specifies the log tag to use for all outgoing logs
7045 May be used in sections: defaults | frontend | listen | backend
7046 yes | yes | yes | yes
7047
7048 Sets the tag field in the syslog header to this string. It defaults to the
7049 log-tag set in the global section, otherwise the program name as launched
7050 from the command line, which usually is "haproxy". Sometimes it can be useful
7051 to differentiate between multiple processes running on the same host, or to
7052 differentiate customer instances running in the same process. In the backend,
7053 logs about servers up/down will use this tag. As a hint, it can be convenient
7054 to set a log-tag related to a hosted customer in a defaults section then put
7055 all the frontends and backends for that customer, then start another customer
7056 in a new defaults section. See also the global "log-tag" directive.
Willy Tarreau2769aa02007-12-27 18:26:09 +01007057
Willy Tarreauc35362a2014-04-25 13:58:37 +02007058max-keep-alive-queue <value>
7059 Set the maximum server queue size for maintaining keep-alive connections
7060 May be used in sections: defaults | frontend | listen | backend
7061 yes | no | yes | yes
7062
7063 HTTP keep-alive tries to reuse the same server connection whenever possible,
7064 but sometimes it can be counter-productive, for example if a server has a lot
7065 of connections while other ones are idle. This is especially true for static
7066 servers.
7067
7068 The purpose of this setting is to set a threshold on the number of queued
7069 connections at which haproxy stops trying to reuse the same server and prefers
7070 to find another one. The default value, -1, means there is no limit. A value
7071 of zero means that keep-alive requests will never be queued. For very close
7072 servers which can be reached with a low latency and which are not sensible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01007073 breaking keep-alive, a low value is recommended (e.g. local static server can
Willy Tarreauc35362a2014-04-25 13:58:37 +02007074 use a value of 10 or less). For remote servers suffering from a high latency,
7075 higher values might be needed to cover for the latency and/or the cost of
7076 picking a different server.
7077
7078 Note that this has no impact on responses which are maintained to the same
7079 server consecutively to a 401 response. They will still go to the same server
7080 even if they have to be queued.
7081
7082 See also : "option http-server-close", "option prefer-last-server", server
7083 "maxconn" and cookie persistence.
7084
Olivier Houcharda4d4fdf2018-12-14 19:27:06 +01007085max-session-srv-conns <nb>
7086 Set the maximum number of outgoing connections we can keep idling for a given
7087 client session. The default is 5 (it precisely equals MAX_SRV_LIST which is
7088 defined at build time).
Willy Tarreauc35362a2014-04-25 13:58:37 +02007089
Willy Tarreau2769aa02007-12-27 18:26:09 +01007090maxconn <conns>
7091 Fix the maximum number of concurrent connections on a frontend
7092 May be used in sections : defaults | frontend | listen | backend
7093 yes | yes | yes | no
7094 Arguments :
7095 <conns> is the maximum number of concurrent connections the frontend will
7096 accept to serve. Excess connections will be queued by the system
7097 in the socket's listen queue and will be served once a connection
7098 closes.
7099
7100 If the system supports it, it can be useful on big sites to raise this limit
7101 very high so that haproxy manages connection queues, instead of leaving the
7102 clients with unanswered connection attempts. This value should not exceed the
7103 global maxconn. Also, keep in mind that a connection contains two buffers
Baptiste Assmann79fb45d2016-03-06 23:34:31 +01007104 of tune.bufsize (16kB by default) each, as well as some other data resulting
7105 in about 33 kB of RAM being consumed per established connection. That means
7106 that a medium system equipped with 1GB of RAM can withstand around
7107 20000-25000 concurrent connections if properly tuned.
Willy Tarreau2769aa02007-12-27 18:26:09 +01007108
7109 Also, when <conns> is set to large values, it is possible that the servers
7110 are not sized to accept such loads, and for this reason it is generally wise
7111 to assign them some reasonable connection limits.
7112
Willy Tarreauc8d5b952019-02-27 17:25:52 +01007113 When this value is set to zero, which is the default, the global "maxconn"
7114 value is used.
Vincent Bernat6341be52012-06-27 17:18:30 +02007115
Willy Tarreau2769aa02007-12-27 18:26:09 +01007116 See also : "server", global section's "maxconn", "fullconn"
7117
7118
Willy Tarreau77e0dae2020-10-14 15:44:27 +02007119mode { tcp|http }
Willy Tarreau2769aa02007-12-27 18:26:09 +01007120 Set the running mode or protocol of the instance
7121 May be used in sections : defaults | frontend | listen | backend
7122 yes | yes | yes | yes
7123 Arguments :
7124 tcp The instance will work in pure TCP mode. A full-duplex connection
7125 will be established between clients and servers, and no layer 7
7126 examination will be performed. This is the default mode. It
7127 should be used for SSL, SSH, SMTP, ...
7128
7129 http The instance will work in HTTP mode. The client request will be
7130 analyzed in depth before connecting to any server. Any request
7131 which is not RFC-compliant will be rejected. Layer 7 filtering,
7132 processing and switching will be possible. This is the mode which
7133 brings HAProxy most of its value.
7134
Cyril Bonté108cf6e2012-04-21 23:30:29 +02007135 When doing content switching, it is mandatory that the frontend and the
7136 backend are in the same mode (generally HTTP), otherwise the configuration
7137 will be refused.
Willy Tarreau2769aa02007-12-27 18:26:09 +01007138
Cyril Bonté108cf6e2012-04-21 23:30:29 +02007139 Example :
Willy Tarreau2769aa02007-12-27 18:26:09 +01007140 defaults http_instances
7141 mode http
7142
Willy Tarreau0ba27502007-12-24 16:55:16 +01007143
Cyril Bontéf0c60612010-02-06 14:44:47 +01007144monitor fail { if | unless } <condition>
Willy Tarreau2769aa02007-12-27 18:26:09 +01007145 Add a condition to report a failure to a monitor HTTP request.
Willy Tarreau0ba27502007-12-24 16:55:16 +01007146 May be used in sections : defaults | frontend | listen | backend
7147 no | yes | yes | no
Willy Tarreau0ba27502007-12-24 16:55:16 +01007148 Arguments :
7149 if <cond> the monitor request will fail if the condition is satisfied,
7150 and will succeed otherwise. The condition should describe a
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007151 combined test which must induce a failure if all conditions
Willy Tarreau0ba27502007-12-24 16:55:16 +01007152 are met, for instance a low number of servers both in a
7153 backend and its backup.
7154
7155 unless <cond> the monitor request will succeed only if the condition is
7156 satisfied, and will fail otherwise. Such a condition may be
7157 based on a test on the presence of a minimum number of active
7158 servers in a list of backends.
7159
7160 This statement adds a condition which can force the response to a monitor
7161 request to report a failure. By default, when an external component queries
7162 the URI dedicated to monitoring, a 200 response is returned. When one of the
7163 conditions above is met, haproxy will return 503 instead of 200. This is
7164 very useful to report a site failure to an external component which may base
7165 routing advertisements between multiple sites on the availability reported by
7166 haproxy. In this case, one would rely on an ACL involving the "nbsrv"
Willy Tarreauae94d4d2011-05-11 16:28:49 +02007167 criterion. Note that "monitor fail" only works in HTTP mode. Both status
7168 messages may be tweaked using "errorfile" or "errorloc" if needed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01007169
7170 Example:
7171 frontend www
Willy Tarreau2769aa02007-12-27 18:26:09 +01007172 mode http
Willy Tarreau0ba27502007-12-24 16:55:16 +01007173 acl site_dead nbsrv(dynamic) lt 2
7174 acl site_dead nbsrv(static) lt 2
7175 monitor-uri /site_alive
7176 monitor fail if site_dead
7177
Willy Tarreau9e9919d2020-10-14 15:55:23 +02007178 See also : "monitor-uri", "errorfile", "errorloc"
Willy Tarreau2769aa02007-12-27 18:26:09 +01007179
7180
Willy Tarreau2769aa02007-12-27 18:26:09 +01007181monitor-uri <uri>
7182 Intercept a URI used by external components' monitor requests
7183 May be used in sections : defaults | frontend | listen | backend
7184 yes | yes | yes | no
7185 Arguments :
7186 <uri> is the exact URI which we want to intercept to return HAProxy's
7187 health status instead of forwarding the request.
7188
7189 When an HTTP request referencing <uri> will be received on a frontend,
7190 HAProxy will not forward it nor log it, but instead will return either
7191 "HTTP/1.0 200 OK" or "HTTP/1.0 503 Service unavailable", depending on failure
7192 conditions defined with "monitor fail". This is normally enough for any
7193 front-end HTTP probe to detect that the service is UP and running without
7194 forwarding the request to a backend server. Note that the HTTP method, the
7195 version and all headers are ignored, but the request must at least be valid
7196 at the HTTP level. This keyword may only be used with an HTTP-mode frontend.
7197
Willy Tarreau721d8e02017-12-01 18:25:08 +01007198 Monitor requests are processed very early, just after the request is parsed
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02007199 and even before any "http-request". The only rulesets applied before are the
7200 tcp-request ones. They cannot be logged either, and it is the intended
7201 purpose. They are only used to report HAProxy's health to an upper component,
7202 nothing more. However, it is possible to add any number of conditions using
7203 "monitor fail" and ACLs so that the result can be adjusted to whatever check
7204 can be imagined (most often the number of available servers in a backend).
Willy Tarreau2769aa02007-12-27 18:26:09 +01007205
Christopher Faulet6072beb2020-02-18 15:34:58 +01007206 Note: if <uri> starts by a slash ('/'), the matching is performed against the
7207 request's path instead of the request's uri. It is a workaround to let
7208 the HTTP/2 requests match the monitor-uri. Indeed, in HTTP/2, clients
7209 are encouraged to send absolute URIs only.
7210
Willy Tarreau2769aa02007-12-27 18:26:09 +01007211 Example :
7212 # Use /haproxy_test to report haproxy's status
7213 frontend www
7214 mode http
7215 monitor-uri /haproxy_test
7216
Willy Tarreau9e9919d2020-10-14 15:55:23 +02007217 See also : "monitor fail"
Willy Tarreau2769aa02007-12-27 18:26:09 +01007218
Willy Tarreau0ba27502007-12-24 16:55:16 +01007219
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007220option abortonclose
7221no option abortonclose
7222 Enable or disable early dropping of aborted requests pending in queues.
7223 May be used in sections : defaults | frontend | listen | backend
7224 yes | no | yes | yes
7225 Arguments : none
7226
7227 In presence of very high loads, the servers will take some time to respond.
7228 The per-instance connection queue will inflate, and the response time will
7229 increase respective to the size of the queue times the average per-session
7230 response time. When clients will wait for more than a few seconds, they will
Willy Tarreau198a7442008-01-17 12:05:32 +01007231 often hit the "STOP" button on their browser, leaving a useless request in
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007232 the queue, and slowing down other users, and the servers as well, because the
7233 request will eventually be served, then aborted at the first error
7234 encountered while delivering the response.
7235
7236 As there is no way to distinguish between a full STOP and a simple output
7237 close on the client side, HTTP agents should be conservative and consider
7238 that the client might only have closed its output channel while waiting for
7239 the response. However, this introduces risks of congestion when lots of users
7240 do the same, and is completely useless nowadays because probably no client at
7241 all will close the session while waiting for the response. Some HTTP agents
Davor Ocelice9ed2812017-12-25 17:49:28 +01007242 support this behavior (Squid, Apache, HAProxy), and others do not (TUX, most
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007243 hardware-based load balancers). So the probability for a closed input channel
Willy Tarreau198a7442008-01-17 12:05:32 +01007244 to represent a user hitting the "STOP" button is close to 100%, and the risk
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007245 of being the single component to break rare but valid traffic is extremely
7246 low, which adds to the temptation to be able to abort a session early while
7247 still not served and not pollute the servers.
7248
Davor Ocelice9ed2812017-12-25 17:49:28 +01007249 In HAProxy, the user can choose the desired behavior using the option
7250 "abortonclose". By default (without the option) the behavior is HTTP
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007251 compliant and aborted requests will be served. But when the option is
7252 specified, a session with an incoming channel closed will be aborted while
7253 it is still possible, either pending in the queue for a connection slot, or
7254 during the connection establishment if the server has not yet acknowledged
7255 the connection request. This considerably reduces the queue size and the load
7256 on saturated servers when users are tempted to click on STOP, which in turn
Willy Tarreaud72758d2010-01-12 10:42:19 +01007257 reduces the response time for other users.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007258
7259 If this option has been enabled in a "defaults" section, it can be disabled
7260 in a specific instance by prepending the "no" keyword before it.
7261
7262 See also : "timeout queue" and server's "maxconn" and "maxqueue" parameters
7263
7264
Willy Tarreau4076a152009-04-02 15:18:36 +02007265option accept-invalid-http-request
7266no option accept-invalid-http-request
7267 Enable or disable relaxing of HTTP request parsing
7268 May be used in sections : defaults | frontend | listen | backend
7269 yes | yes | yes | no
7270 Arguments : none
7271
Willy Tarreau91852eb2015-05-01 13:26:00 +02007272 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02007273 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01007274 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02007275 forbidden characters are essentially used to build attacks exploiting server
7276 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
7277 server will emit invalid header names for whatever reason (configuration,
7278 implementation) and the issue will not be immediately fixed. In such a case,
7279 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau422246e2012-01-07 23:54:13 +01007280 even if that does not make sense, by specifying this option. Similarly, the
7281 list of characters allowed to appear in a URI is well defined by RFC3986, and
7282 chars 0-31, 32 (space), 34 ('"'), 60 ('<'), 62 ('>'), 92 ('\'), 94 ('^'), 96
7283 ('`'), 123 ('{'), 124 ('|'), 125 ('}'), 127 (delete) and anything above are
Davor Ocelice9ed2812017-12-25 17:49:28 +01007284 not allowed at all. HAProxy always blocks a number of them (0..32, 127). The
Willy Tarreau91852eb2015-05-01 13:26:00 +02007285 remaining ones are blocked by default unless this option is enabled. This
Willy Tarreau13317662015-05-01 13:47:08 +02007286 option also relaxes the test on the HTTP version, it allows HTTP/0.9 requests
7287 to pass through (no version specified) and multiple digits for both the major
7288 and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02007289
7290 This option should never be enabled by default as it hides application bugs
7291 and open security breaches. It should only be deployed after a problem has
7292 been confirmed.
7293
7294 When this option is enabled, erroneous header names will still be accepted in
7295 requests, but the complete request will be captured in order to permit later
Willy Tarreau422246e2012-01-07 23:54:13 +01007296 analysis using the "show errors" request on the UNIX stats socket. Similarly,
7297 requests containing invalid chars in the URI part will be logged. Doing this
Willy Tarreau4076a152009-04-02 15:18:36 +02007298 also helps confirming that the issue has been solved.
7299
7300 If this option has been enabled in a "defaults" section, it can be disabled
7301 in a specific instance by prepending the "no" keyword before it.
7302
7303 See also : "option accept-invalid-http-response" and "show errors" on the
7304 stats socket.
7305
7306
7307option accept-invalid-http-response
7308no option accept-invalid-http-response
7309 Enable or disable relaxing of HTTP response parsing
7310 May be used in sections : defaults | frontend | listen | backend
7311 yes | no | yes | yes
7312 Arguments : none
7313
Willy Tarreau91852eb2015-05-01 13:26:00 +02007314 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02007315 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01007316 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02007317 forbidden characters are essentially used to build attacks exploiting server
7318 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
7319 server will emit invalid header names for whatever reason (configuration,
7320 implementation) and the issue will not be immediately fixed. In such a case,
7321 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau91852eb2015-05-01 13:26:00 +02007322 even if that does not make sense, by specifying this option. This option also
7323 relaxes the test on the HTTP version format, it allows multiple digits for
7324 both the major and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02007325
7326 This option should never be enabled by default as it hides application bugs
7327 and open security breaches. It should only be deployed after a problem has
7328 been confirmed.
7329
7330 When this option is enabled, erroneous header names will still be accepted in
7331 responses, but the complete response will be captured in order to permit
7332 later analysis using the "show errors" request on the UNIX stats socket.
7333 Doing this also helps confirming that the issue has been solved.
7334
7335 If this option has been enabled in a "defaults" section, it can be disabled
7336 in a specific instance by prepending the "no" keyword before it.
7337
7338 See also : "option accept-invalid-http-request" and "show errors" on the
7339 stats socket.
7340
7341
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007342option allbackups
7343no option allbackups
7344 Use either all backup servers at a time or only the first one
7345 May be used in sections : defaults | frontend | listen | backend
7346 yes | no | yes | yes
7347 Arguments : none
7348
7349 By default, the first operational backup server gets all traffic when normal
7350 servers are all down. Sometimes, it may be preferred to use multiple backups
7351 at once, because one will not be enough. When "option allbackups" is enabled,
7352 the load balancing will be performed among all backup servers when all normal
7353 ones are unavailable. The same load balancing algorithm will be used and the
7354 servers' weights will be respected. Thus, there will not be any priority
7355 order between the backup servers anymore.
7356
7357 This option is mostly used with static server farms dedicated to return a
7358 "sorry" page when an application is completely offline.
7359
7360 If this option has been enabled in a "defaults" section, it can be disabled
7361 in a specific instance by prepending the "no" keyword before it.
7362
7363
7364option checkcache
7365no option checkcache
Godbach7056a352013-12-11 20:01:07 +08007366 Analyze all server responses and block responses with cacheable cookies
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007367 May be used in sections : defaults | frontend | listen | backend
7368 yes | no | yes | yes
7369 Arguments : none
7370
7371 Some high-level frameworks set application cookies everywhere and do not
7372 always let enough control to the developer to manage how the responses should
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007373 be cached. When a session cookie is returned on a cacheable object, there is a
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007374 high risk of session crossing or stealing between users traversing the same
7375 caches. In some situations, it is better to block the response than to let
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02007376 some sensitive session information go in the wild.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007377
7378 The option "checkcache" enables deep inspection of all server responses for
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007379 strict compliance with HTTP specification in terms of cacheability. It
Willy Tarreau198a7442008-01-17 12:05:32 +01007380 carefully checks "Cache-control", "Pragma" and "Set-cookie" headers in server
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007381 response to check if there's a risk of caching a cookie on a client-side
7382 proxy. When this option is enabled, the only responses which can be delivered
Willy Tarreau198a7442008-01-17 12:05:32 +01007383 to the client are :
Davor Ocelice9ed2812017-12-25 17:49:28 +01007384 - all those without "Set-Cookie" header;
Willy Tarreauc55ddce2017-12-21 11:41:38 +01007385 - all those with a return code other than 200, 203, 204, 206, 300, 301,
7386 404, 405, 410, 414, 501, provided that the server has not set a
Davor Ocelice9ed2812017-12-25 17:49:28 +01007387 "Cache-control: public" header field;
Willy Tarreau24ea0bc2017-12-21 11:32:55 +01007388 - all those that result from a request using a method other than GET, HEAD,
7389 OPTIONS, TRACE, provided that the server has not set a 'Cache-Control:
Davor Ocelice9ed2812017-12-25 17:49:28 +01007390 public' header field;
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007391 - those with a 'Pragma: no-cache' header
7392 - those with a 'Cache-control: private' header
7393 - those with a 'Cache-control: no-store' header
7394 - those with a 'Cache-control: max-age=0' header
7395 - those with a 'Cache-control: s-maxage=0' header
7396 - those with a 'Cache-control: no-cache' header
7397 - those with a 'Cache-control: no-cache="set-cookie"' header
7398 - those with a 'Cache-control: no-cache="set-cookie,' header
7399 (allowing other fields after set-cookie)
7400
7401 If a response doesn't respect these requirements, then it will be blocked
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02007402 just as if it was from an "http-response deny" rule, with an "HTTP 502 bad
7403 gateway". The session state shows "PH--" meaning that the proxy blocked the
7404 response during headers processing. Additionally, an alert will be sent in
7405 the logs so that admins are informed that there's something to be fixed.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007406
7407 Due to the high impact on the application, the application should be tested
7408 in depth with the option enabled before going to production. It is also a
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01007409 good practice to always activate it during tests, even if it is not used in
Davor Ocelice9ed2812017-12-25 17:49:28 +01007410 production, as it will report potentially dangerous application behaviors.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007411
7412 If this option has been enabled in a "defaults" section, it can be disabled
7413 in a specific instance by prepending the "no" keyword before it.
7414
7415
7416option clitcpka
7417no option clitcpka
7418 Enable or disable the sending of TCP keepalive packets on the client side
7419 May be used in sections : defaults | frontend | listen | backend
7420 yes | yes | yes | no
7421 Arguments : none
7422
7423 When there is a firewall or any session-aware component between a client and
7424 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01007425 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007426 components decides to expire a session which has remained idle for too long.
7427
7428 Enabling socket-level TCP keep-alives makes the system regularly send packets
7429 to the other end of the connection, leaving it active. The delay between
7430 keep-alive probes is controlled by the system only and depends both on the
7431 operating system and its tuning parameters.
7432
7433 It is important to understand that keep-alive packets are neither emitted nor
7434 received at the application level. It is only the network stacks which sees
7435 them. For this reason, even if one side of the proxy already uses keep-alives
7436 to maintain its connection alive, those keep-alive packets will not be
7437 forwarded to the other side of the proxy.
7438
7439 Please note that this has nothing to do with HTTP keep-alive.
7440
7441 Using option "clitcpka" enables the emission of TCP keep-alive probes on the
7442 client side of a connection, which should help when session expirations are
7443 noticed between HAProxy and a client.
7444
7445 If this option has been enabled in a "defaults" section, it can be disabled
7446 in a specific instance by prepending the "no" keyword before it.
7447
7448 See also : "option srvtcpka", "option tcpka"
7449
7450
Willy Tarreau0ba27502007-12-24 16:55:16 +01007451option contstats
7452 Enable continuous traffic statistics updates
7453 May be used in sections : defaults | frontend | listen | backend
7454 yes | yes | yes | no
7455 Arguments : none
7456
7457 By default, counters used for statistics calculation are incremented
7458 only when a session finishes. It works quite well when serving small
7459 objects, but with big ones (for example large images or archives) or
7460 with A/V streaming, a graph generated from haproxy counters looks like
Willy Tarreaudef0d222016-11-08 22:03:00 +01007461 a hedgehog. With this option enabled counters get incremented frequently
7462 along the session, typically every 5 seconds, which is often enough to
7463 produce clean graphs. Recounting touches a hotpath directly so it is not
7464 not enabled by default, as it can cause a lot of wakeups for very large
7465 session counts and cause a small performance drop.
Willy Tarreau0ba27502007-12-24 16:55:16 +01007466
Christopher Faulet89aed322020-06-02 17:33:56 +02007467option disable-h2-upgrade
7468no option disable-h2-upgrade
7469 Enable or disable the implicit HTTP/2 upgrade from an HTTP/1.x client
7470 connection.
7471 May be used in sections : defaults | frontend | listen | backend
7472 yes | yes | yes | no
7473 Arguments : none
7474
7475 By default, HAProxy is able to implicitly upgrade an HTTP/1.x client
7476 connection to an HTTP/2 connection if the first request it receives from a
7477 given HTTP connection matches the HTTP/2 connection preface (i.e. the string
7478 "PRI * HTTP/2.0\r\n\r\nSM\r\n\r\n"). This way, it is possible to support
7479 HTTP/1.x and HTTP/2 clients on a non-SSL connections. This option must be used to
7480 disable the implicit upgrade. Note this implicit upgrade is only supported
7481 for HTTP proxies, thus this option too. Note also it is possible to force the
7482 HTTP/2 on clear connections by specifying "proto h2" on the bind line.
7483
7484 If this option has been enabled in a "defaults" section, it can be disabled
7485 in a specific instance by prepending the "no" keyword before it.
Willy Tarreau0ba27502007-12-24 16:55:16 +01007486
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02007487option dontlog-normal
7488no option dontlog-normal
7489 Enable or disable logging of normal, successful connections
7490 May be used in sections : defaults | frontend | listen | backend
7491 yes | yes | yes | no
7492 Arguments : none
7493
7494 There are large sites dealing with several thousand connections per second
7495 and for which logging is a major pain. Some of them are even forced to turn
7496 logs off and cannot debug production issues. Setting this option ensures that
7497 normal connections, those which experience no error, no timeout, no retry nor
7498 redispatch, will not be logged. This leaves disk space for anomalies. In HTTP
7499 mode, the response status code is checked and return codes 5xx will still be
7500 logged.
7501
7502 It is strongly discouraged to use this option as most of the time, the key to
7503 complex issues is in the normal logs which will not be logged here. If you
7504 need to separate logs, see the "log-separate-errors" option instead.
7505
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007506 See also : "log", "dontlognull", "log-separate-errors" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02007507 logging.
7508
7509
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007510option dontlognull
7511no option dontlognull
7512 Enable or disable logging of null connections
7513 May be used in sections : defaults | frontend | listen | backend
7514 yes | yes | yes | no
7515 Arguments : none
7516
7517 In certain environments, there are components which will regularly connect to
7518 various systems to ensure that they are still alive. It can be the case from
7519 another load balancer as well as from monitoring systems. By default, even a
7520 simple port probe or scan will produce a log. If those connections pollute
7521 the logs too much, it is possible to enable option "dontlognull" to indicate
7522 that a connection on which no data has been transferred will not be logged,
Willy Tarreau0f228a02015-05-01 15:37:53 +02007523 which typically corresponds to those probes. Note that errors will still be
7524 returned to the client and accounted for in the stats. If this is not what is
7525 desired, option http-ignore-probes can be used instead.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007526
7527 It is generally recommended not to use this option in uncontrolled
Davor Ocelice9ed2812017-12-25 17:49:28 +01007528 environments (e.g. internet), otherwise scans and other malicious activities
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007529 would not be logged.
7530
7531 If this option has been enabled in a "defaults" section, it can be disabled
7532 in a specific instance by prepending the "no" keyword before it.
7533
Willy Tarreau9e9919d2020-10-14 15:55:23 +02007534 See also : "log", "http-ignore-probes", "monitor-uri", and
Willy Tarreau0f228a02015-05-01 15:37:53 +02007535 section 8 about logging.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007536
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007537
Willy Tarreau87cf5142011-08-19 22:57:24 +02007538option forwardfor [ except <network> ] [ header <name> ] [ if-none ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01007539 Enable insertion of the X-Forwarded-For header to requests sent to servers
7540 May be used in sections : defaults | frontend | listen | backend
7541 yes | yes | yes | yes
7542 Arguments :
7543 <network> is an optional argument used to disable this option for sources
7544 matching <network>
Ross Westaf72a1d2008-08-03 10:51:45 +02007545 <name> an optional argument to specify a different "X-Forwarded-For"
Willy Tarreaud72758d2010-01-12 10:42:19 +01007546 header name.
Willy Tarreauc27debf2008-01-06 08:57:02 +01007547
7548 Since HAProxy works in reverse-proxy mode, the servers see its IP address as
7549 their client address. This is sometimes annoying when the client's IP address
7550 is expected in server logs. To solve this problem, the well-known HTTP header
7551 "X-Forwarded-For" may be added by HAProxy to all requests sent to the server.
7552 This header contains a value representing the client's IP address. Since this
7553 header is always appended at the end of the existing header list, the server
7554 must be configured to always use the last occurrence of this header only. See
Ross Westaf72a1d2008-08-03 10:51:45 +02007555 the server's manual to find how to enable use of this standard header. Note
7556 that only the last occurrence of the header must be used, since it is really
7557 possible that the client has already brought one.
7558
Willy Tarreaud72758d2010-01-12 10:42:19 +01007559 The keyword "header" may be used to supply a different header name to replace
Ross Westaf72a1d2008-08-03 10:51:45 +02007560 the default "X-Forwarded-For". This can be useful where you might already
Davor Ocelice9ed2812017-12-25 17:49:28 +01007561 have a "X-Forwarded-For" header from a different application (e.g. stunnel),
Willy Tarreaud72758d2010-01-12 10:42:19 +01007562 and you need preserve it. Also if your backend server doesn't use the
Davor Ocelice9ed2812017-12-25 17:49:28 +01007563 "X-Forwarded-For" header and requires different one (e.g. Zeus Web Servers
Ross Westaf72a1d2008-08-03 10:51:45 +02007564 require "X-Cluster-Client-IP").
Willy Tarreauc27debf2008-01-06 08:57:02 +01007565
7566 Sometimes, a same HAProxy instance may be shared between a direct client
7567 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
7568 used to decrypt HTTPS traffic). It is possible to disable the addition of the
7569 header for a known source address or network by adding the "except" keyword
7570 followed by the network address. In this case, any source IP matching the
7571 network will not cause an addition of this header. Most common uses are with
7572 private networks or 127.0.0.1.
7573
Willy Tarreau87cf5142011-08-19 22:57:24 +02007574 Alternatively, the keyword "if-none" states that the header will only be
7575 added if it is not present. This should only be used in perfectly trusted
7576 environment, as this might cause a security issue if headers reaching haproxy
7577 are under the control of the end-user.
7578
Willy Tarreauc27debf2008-01-06 08:57:02 +01007579 This option may be specified either in the frontend or in the backend. If at
Ross Westaf72a1d2008-08-03 10:51:45 +02007580 least one of them uses it, the header will be added. Note that the backend's
7581 setting of the header subargument takes precedence over the frontend's if
Willy Tarreau87cf5142011-08-19 22:57:24 +02007582 both are defined. In the case of the "if-none" argument, if at least one of
7583 the frontend or the backend does not specify it, it wants the addition to be
7584 mandatory, so it wins.
Willy Tarreauc27debf2008-01-06 08:57:02 +01007585
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007586 Example :
Willy Tarreauc27debf2008-01-06 08:57:02 +01007587 # Public HTTP address also used by stunnel on the same machine
7588 frontend www
7589 mode http
7590 option forwardfor except 127.0.0.1 # stunnel already adds the header
7591
Ross Westaf72a1d2008-08-03 10:51:45 +02007592 # Those servers want the IP Address in X-Client
7593 backend www
7594 mode http
7595 option forwardfor header X-Client
7596
Willy Tarreau87cf5142011-08-19 22:57:24 +02007597 See also : "option httpclose", "option http-server-close",
Christopher Faulet315b39c2018-09-21 16:26:19 +02007598 "option http-keep-alive"
Willy Tarreauc27debf2008-01-06 08:57:02 +01007599
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02007600
Christopher Faulet98fbe952019-07-22 16:18:24 +02007601option h1-case-adjust-bogus-client
7602no option h1-case-adjust-bogus-client
7603 Enable or disable the case adjustment of HTTP/1 headers sent to bogus clients
7604 May be used in sections : defaults | frontend | listen | backend
7605 yes | yes | yes | no
7606 Arguments : none
7607
7608 There is no standard case for header names because, as stated in RFC7230,
7609 they are case-insensitive. So applications must handle them in a case-
7610 insensitive manner. But some bogus applications violate the standards and
7611 erroneously rely on the cases most commonly used by browsers. This problem
7612 becomes critical with HTTP/2 because all header names must be exchanged in
7613 lower case, and HAProxy follows the same convention. All header names are
7614 sent in lower case to clients and servers, regardless of the HTTP version.
7615
7616 When HAProxy receives an HTTP/1 response, its header names are converted to
7617 lower case and manipulated and sent this way to the clients. If a client is
7618 known to violate the HTTP standards and to fail to process a response coming
7619 from HAProxy, it is possible to transform the lower case header names to a
7620 different format when the response is formatted and sent to the client, by
7621 enabling this option and specifying the list of headers to be reformatted
7622 using the global directives "h1-case-adjust" or "h1-case-adjust-file". This
7623 must only be a temporary workaround for the time it takes the client to be
7624 fixed, because clients which require such workarounds might be vulnerable to
7625 content smuggling attacks and must absolutely be fixed.
7626
7627 Please note that this option will not affect standards-compliant clients.
7628
7629 If this option has been enabled in a "defaults" section, it can be disabled
7630 in a specific instance by prepending the "no" keyword before it.
7631
7632 See also: "option h1-case-adjust-bogus-server", "h1-case-adjust",
7633 "h1-case-adjust-file".
7634
7635
7636option h1-case-adjust-bogus-server
7637no option h1-case-adjust-bogus-server
7638 Enable or disable the case adjustment of HTTP/1 headers sent to bogus servers
7639 May be used in sections : defaults | frontend | listen | backend
7640 yes | no | yes | yes
7641 Arguments : none
7642
7643 There is no standard case for header names because, as stated in RFC7230,
7644 they are case-insensitive. So applications must handle them in a case-
7645 insensitive manner. But some bogus applications violate the standards and
7646 erroneously rely on the cases most commonly used by browsers. This problem
7647 becomes critical with HTTP/2 because all header names must be exchanged in
7648 lower case, and HAProxy follows the same convention. All header names are
7649 sent in lower case to clients and servers, regardless of the HTTP version.
7650
7651 When HAProxy receives an HTTP/1 request, its header names are converted to
7652 lower case and manipulated and sent this way to the servers. If a server is
7653 known to violate the HTTP standards and to fail to process a request coming
7654 from HAProxy, it is possible to transform the lower case header names to a
7655 different format when the request is formatted and sent to the server, by
7656 enabling this option and specifying the list of headers to be reformatted
7657 using the global directives "h1-case-adjust" or "h1-case-adjust-file". This
7658 must only be a temporary workaround for the time it takes the server to be
7659 fixed, because servers which require such workarounds might be vulnerable to
7660 content smuggling attacks and must absolutely be fixed.
7661
7662 Please note that this option will not affect standards-compliant servers.
7663
7664 If this option has been enabled in a "defaults" section, it can be disabled
7665 in a specific instance by prepending the "no" keyword before it.
7666
7667 See also: "option h1-case-adjust-bogus-client", "h1-case-adjust",
7668 "h1-case-adjust-file".
7669
7670
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02007671option http-buffer-request
7672no option http-buffer-request
7673 Enable or disable waiting for whole HTTP request body before proceeding
7674 May be used in sections : defaults | frontend | listen | backend
7675 yes | yes | yes | yes
7676 Arguments : none
7677
7678 It is sometimes desirable to wait for the body of an HTTP request before
7679 taking a decision. This is what is being done by "balance url_param" for
7680 example. The first use case is to buffer requests from slow clients before
7681 connecting to the server. Another use case consists in taking the routing
7682 decision based on the request body's contents. This option placed in a
7683 frontend or backend forces the HTTP processing to wait until either the whole
Christopher Faulet6db8a2e2019-11-19 16:27:25 +01007684 body is received or the request buffer is full. It can have undesired side
7685 effects with some applications abusing HTTP by expecting unbuffered
7686 transmissions between the frontend and the backend, so this should definitely
7687 not be used by default.
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02007688
Baptiste Assmanneccdf432015-10-28 13:49:01 +01007689 See also : "option http-no-delay", "timeout http-request"
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02007690
7691
Willy Tarreau0f228a02015-05-01 15:37:53 +02007692option http-ignore-probes
7693no option http-ignore-probes
7694 Enable or disable logging of null connections and request timeouts
7695 May be used in sections : defaults | frontend | listen | backend
7696 yes | yes | yes | no
7697 Arguments : none
7698
7699 Recently some browsers started to implement a "pre-connect" feature
7700 consisting in speculatively connecting to some recently visited web sites
7701 just in case the user would like to visit them. This results in many
7702 connections being established to web sites, which end up in 408 Request
7703 Timeout if the timeout strikes first, or 400 Bad Request when the browser
7704 decides to close them first. These ones pollute the log and feed the error
7705 counters. There was already "option dontlognull" but it's insufficient in
7706 this case. Instead, this option does the following things :
7707 - prevent any 400/408 message from being sent to the client if nothing
Davor Ocelice9ed2812017-12-25 17:49:28 +01007708 was received over a connection before it was closed;
7709 - prevent any log from being emitted in this situation;
Willy Tarreau0f228a02015-05-01 15:37:53 +02007710 - prevent any error counter from being incremented
7711
7712 That way the empty connection is silently ignored. Note that it is better
7713 not to use this unless it is clear that it is needed, because it will hide
7714 real problems. The most common reason for not receiving a request and seeing
7715 a 408 is due to an MTU inconsistency between the client and an intermediary
7716 element such as a VPN, which blocks too large packets. These issues are
7717 generally seen with POST requests as well as GET with large cookies. The logs
7718 are often the only way to detect them.
7719
7720 If this option has been enabled in a "defaults" section, it can be disabled
7721 in a specific instance by prepending the "no" keyword before it.
7722
7723 See also : "log", "dontlognull", "errorfile", and section 8 about logging.
7724
7725
Willy Tarreau16bfb022010-01-16 19:48:41 +01007726option http-keep-alive
7727no option http-keep-alive
7728 Enable or disable HTTP keep-alive from client to server
7729 May be used in sections : defaults | frontend | listen | backend
7730 yes | yes | yes | yes
7731 Arguments : none
7732
Willy Tarreau70dffda2014-01-30 03:07:23 +01007733 By default HAProxy operates in keep-alive mode with regards to persistent
7734 connections: for each connection it processes each request and response, and
Christopher Faulet315b39c2018-09-21 16:26:19 +02007735 leaves the connection idle on both sides between the end of a response and
7736 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02007737 as "option http-server-close" or "option httpclose". This option allows to
7738 set back the keep-alive mode, which can be useful when another mode was used
7739 in a defaults section.
Willy Tarreau70dffda2014-01-30 03:07:23 +01007740
7741 Setting "option http-keep-alive" enables HTTP keep-alive mode on the client-
7742 and server- sides. This provides the lowest latency on the client side (slow
Willy Tarreau16bfb022010-01-16 19:48:41 +01007743 network) and the fastest session reuse on the server side at the expense
7744 of maintaining idle connections to the servers. In general, it is possible
7745 with this option to achieve approximately twice the request rate that the
7746 "http-server-close" option achieves on small objects. There are mainly two
7747 situations where this option may be useful :
7748
7749 - when the server is non-HTTP compliant and authenticates the connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01007750 instead of requests (e.g. NTLM authentication)
Willy Tarreau16bfb022010-01-16 19:48:41 +01007751
7752 - when the cost of establishing the connection to the server is significant
7753 compared to the cost of retrieving the associated object from the server.
7754
7755 This last case can happen when the server is a fast static server of cache.
7756 In this case, the server will need to be properly tuned to support high enough
7757 connection counts because connections will last until the client sends another
7758 request.
7759
7760 If the client request has to go to another backend or another server due to
7761 content switching or the load balancing algorithm, the idle connection will
Willy Tarreau9420b122013-12-15 18:58:25 +01007762 immediately be closed and a new one re-opened. Option "prefer-last-server" is
7763 available to try optimize server selection so that if the server currently
7764 attached to an idle connection is usable, it will be used.
Willy Tarreau16bfb022010-01-16 19:48:41 +01007765
Willy Tarreau16bfb022010-01-16 19:48:41 +01007766 At the moment, logs will not indicate whether requests came from the same
7767 session or not. The accept date reported in the logs corresponds to the end
7768 of the previous request, and the request time corresponds to the time spent
7769 waiting for a new request. The keep-alive request time is still bound to the
7770 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
7771 not set.
7772
Christopher Faulet159e6672019-07-16 15:09:52 +02007773 This option disables and replaces any previous "option httpclose" or "option
7774 http-server-close". When backend and frontend options differ, all of these 4
7775 options have precedence over "option http-keep-alive".
Willy Tarreau16bfb022010-01-16 19:48:41 +01007776
Christopher Faulet315b39c2018-09-21 16:26:19 +02007777 See also : "option httpclose",, "option http-server-close",
Willy Tarreau9420b122013-12-15 18:58:25 +01007778 "option prefer-last-server", "option http-pretend-keepalive",
Frédéric Lécaille93d33162019-03-06 09:35:59 +01007779 and "1.1. The HTTP transaction model".
Willy Tarreau16bfb022010-01-16 19:48:41 +01007780
7781
Willy Tarreau96e31212011-05-30 18:10:30 +02007782option http-no-delay
7783no option http-no-delay
7784 Instruct the system to favor low interactive delays over performance in HTTP
7785 May be used in sections : defaults | frontend | listen | backend
7786 yes | yes | yes | yes
7787 Arguments : none
7788
7789 In HTTP, each payload is unidirectional and has no notion of interactivity.
7790 Any agent is expected to queue data somewhat for a reasonably low delay.
7791 There are some very rare server-to-server applications that abuse the HTTP
7792 protocol and expect the payload phase to be highly interactive, with many
7793 interleaved data chunks in both directions within a single request. This is
7794 absolutely not supported by the HTTP specification and will not work across
7795 most proxies or servers. When such applications attempt to do this through
7796 haproxy, it works but they will experience high delays due to the network
7797 optimizations which favor performance by instructing the system to wait for
7798 enough data to be available in order to only send full packets. Typical
7799 delays are around 200 ms per round trip. Note that this only happens with
7800 abnormal uses. Normal uses such as CONNECT requests nor WebSockets are not
7801 affected.
7802
7803 When "option http-no-delay" is present in either the frontend or the backend
7804 used by a connection, all such optimizations will be disabled in order to
7805 make the exchanges as fast as possible. Of course this offers no guarantee on
7806 the functionality, as it may break at any other place. But if it works via
7807 HAProxy, it will work as fast as possible. This option should never be used
7808 by default, and should never be used at all unless such a buggy application
7809 is discovered. The impact of using this option is an increase of bandwidth
7810 usage and CPU usage, which may significantly lower performance in high
7811 latency environments.
7812
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02007813 See also : "option http-buffer-request"
7814
Willy Tarreau96e31212011-05-30 18:10:30 +02007815
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02007816option http-pretend-keepalive
7817no option http-pretend-keepalive
7818 Define whether haproxy will announce keepalive to the server or not
7819 May be used in sections : defaults | frontend | listen | backend
Christopher Faulet98db9762018-09-21 10:25:19 +02007820 yes | no | yes | yes
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02007821 Arguments : none
7822
Christopher Faulet315b39c2018-09-21 16:26:19 +02007823 When running with "option http-server-close" or "option httpclose", haproxy
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02007824 adds a "Connection: close" header to the request forwarded to the server.
7825 Unfortunately, when some servers see this header, they automatically refrain
7826 from using the chunked encoding for responses of unknown length, while this
7827 is totally unrelated. The immediate effect is that this prevents haproxy from
7828 maintaining the client connection alive. A second effect is that a client or
7829 a cache could receive an incomplete response without being aware of it, and
7830 consider the response complete.
7831
7832 By setting "option http-pretend-keepalive", haproxy will make the server
7833 believe it will keep the connection alive. The server will then not fall back
7834 to the abnormal undesired above. When haproxy gets the whole response, it
7835 will close the connection with the server just as it would do with the
Christopher Faulet315b39c2018-09-21 16:26:19 +02007836 "option httpclose". That way the client gets a normal response and the
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02007837 connection is correctly closed on the server side.
7838
7839 It is recommended not to enable this option by default, because most servers
7840 will more efficiently close the connection themselves after the last packet,
7841 and release its buffers slightly earlier. Also, the added packet on the
7842 network could slightly reduce the overall peak performance. However it is
7843 worth noting that when this option is enabled, haproxy will have slightly
7844 less work to do. So if haproxy is the bottleneck on the whole architecture,
7845 enabling this option might save a few CPU cycles.
7846
Christopher Faulet98db9762018-09-21 10:25:19 +02007847 This option may be set in backend and listen sections. Using it in a frontend
7848 section will be ignored and a warning will be reported during startup. It is
7849 a backend related option, so there is no real reason to set it on a
7850 frontend. This option may be combined with "option httpclose", which will
7851 cause keepalive to be announced to the server and close to be announced to
7852 the client. This practice is discouraged though.
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02007853
7854 If this option has been enabled in a "defaults" section, it can be disabled
7855 in a specific instance by prepending the "no" keyword before it.
7856
Christopher Faulet315b39c2018-09-21 16:26:19 +02007857 See also : "option httpclose", "option http-server-close", and
Willy Tarreau16bfb022010-01-16 19:48:41 +01007858 "option http-keep-alive"
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02007859
Willy Tarreauc27debf2008-01-06 08:57:02 +01007860
Willy Tarreaub608feb2010-01-02 22:47:18 +01007861option http-server-close
7862no option http-server-close
7863 Enable or disable HTTP connection closing on the server side
7864 May be used in sections : defaults | frontend | listen | backend
7865 yes | yes | yes | yes
7866 Arguments : none
7867
Willy Tarreau70dffda2014-01-30 03:07:23 +01007868 By default HAProxy operates in keep-alive mode with regards to persistent
7869 connections: for each connection it processes each request and response, and
7870 leaves the connection idle on both sides between the end of a response and
7871 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02007872 as "option http-server-close" or "option httpclose". Setting "option
7873 http-server-close" enables HTTP connection-close mode on the server side
7874 while keeping the ability to support HTTP keep-alive and pipelining on the
7875 client side. This provides the lowest latency on the client side (slow
7876 network) and the fastest session reuse on the server side to save server
7877 resources, similarly to "option httpclose". It also permits non-keepalive
7878 capable servers to be served in keep-alive mode to the clients if they
7879 conform to the requirements of RFC7230. Please note that some servers do not
7880 always conform to those requirements when they see "Connection: close" in the
7881 request. The effect will be that keep-alive will never be used. A workaround
7882 consists in enabling "option http-pretend-keepalive".
Willy Tarreaub608feb2010-01-02 22:47:18 +01007883
7884 At the moment, logs will not indicate whether requests came from the same
7885 session or not. The accept date reported in the logs corresponds to the end
7886 of the previous request, and the request time corresponds to the time spent
7887 waiting for a new request. The keep-alive request time is still bound to the
Willy Tarreaub16a5742010-01-10 14:46:16 +01007888 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
7889 not set.
Willy Tarreaub608feb2010-01-02 22:47:18 +01007890
7891 This option may be set both in a frontend and in a backend. It is enabled if
7892 at least one of the frontend or backend holding a connection has it enabled.
Christopher Faulet159e6672019-07-16 15:09:52 +02007893 It disables and replaces any previous "option httpclose" or "option
7894 http-keep-alive". Please check section 4 ("Proxies") to see how this option
7895 combines with others when frontend and backend options differ.
Willy Tarreaub608feb2010-01-02 22:47:18 +01007896
7897 If this option has been enabled in a "defaults" section, it can be disabled
7898 in a specific instance by prepending the "no" keyword before it.
7899
Christopher Faulet315b39c2018-09-21 16:26:19 +02007900 See also : "option httpclose", "option http-pretend-keepalive",
7901 "option http-keep-alive", and "1.1. The HTTP transaction model".
Willy Tarreaub608feb2010-01-02 22:47:18 +01007902
Willy Tarreau88d349d2010-01-25 12:15:43 +01007903option http-use-proxy-header
Cyril Bontéf0c60612010-02-06 14:44:47 +01007904no option http-use-proxy-header
Willy Tarreau88d349d2010-01-25 12:15:43 +01007905 Make use of non-standard Proxy-Connection header instead of Connection
7906 May be used in sections : defaults | frontend | listen | backend
7907 yes | yes | yes | no
7908 Arguments : none
7909
Lukas Tribus23953682017-04-28 13:24:30 +00007910 While RFC7230 explicitly states that HTTP/1.1 agents must use the
Willy Tarreau88d349d2010-01-25 12:15:43 +01007911 Connection header to indicate their wish of persistent or non-persistent
7912 connections, both browsers and proxies ignore this header for proxied
7913 connections and make use of the undocumented, non-standard Proxy-Connection
7914 header instead. The issue begins when trying to put a load balancer between
7915 browsers and such proxies, because there will be a difference between what
7916 haproxy understands and what the client and the proxy agree on.
7917
7918 By setting this option in a frontend, haproxy can automatically switch to use
7919 that non-standard header if it sees proxied requests. A proxied request is
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01007920 defined here as one where the URI begins with neither a '/' nor a '*'. This
7921 is incompatible with the HTTP tunnel mode. Note that this option can only be
7922 specified in a frontend and will affect the request along its whole life.
Willy Tarreau88d349d2010-01-25 12:15:43 +01007923
Willy Tarreau844a7e72010-01-31 21:46:18 +01007924 Also, when this option is set, a request which requires authentication will
7925 automatically switch to use proxy authentication headers if it is itself a
7926 proxied request. That makes it possible to check or enforce authentication in
7927 front of an existing proxy.
7928
Willy Tarreau88d349d2010-01-25 12:15:43 +01007929 This option should normally never be used, except in front of a proxy.
7930
Christopher Faulet315b39c2018-09-21 16:26:19 +02007931 See also : "option httpclose", and "option http-server-close".
Willy Tarreau88d349d2010-01-25 12:15:43 +01007932
Willy Tarreaud63335a2010-02-26 12:56:52 +01007933option httpchk
7934option httpchk <uri>
7935option httpchk <method> <uri>
7936option httpchk <method> <uri> <version>
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02007937 Enables HTTP protocol to check on the servers health
Willy Tarreaud63335a2010-02-26 12:56:52 +01007938 May be used in sections : defaults | frontend | listen | backend
7939 yes | no | yes | yes
7940 Arguments :
7941 <method> is the optional HTTP method used with the requests. When not set,
7942 the "OPTIONS" method is used, as it generally requires low server
7943 processing and is easy to filter out from the logs. Any method
7944 may be used, though it is not recommended to invent non-standard
7945 ones.
7946
7947 <uri> is the URI referenced in the HTTP requests. It defaults to " / "
7948 which is accessible by default on almost any server, but may be
7949 changed to any other URI. Query strings are permitted.
7950
7951 <version> is the optional HTTP version string. It defaults to "HTTP/1.0"
7952 but some servers might behave incorrectly in HTTP 1.0, so turning
7953 it to HTTP/1.1 may sometimes help. Note that the Host field is
Christopher Faulet8acb1282020-04-09 08:44:06 +02007954 mandatory in HTTP/1.1, use "http-check send" directive to add it.
Willy Tarreaud63335a2010-02-26 12:56:52 +01007955
7956 By default, server health checks only consist in trying to establish a TCP
7957 connection. When "option httpchk" is specified, a complete HTTP request is
7958 sent once the TCP connection is established, and responses 2xx and 3xx are
7959 considered valid, while all other ones indicate a server failure, including
7960 the lack of any response.
7961
Christopher Faulete5870d82020-04-15 11:32:03 +02007962 Combined with "http-check" directives, it is possible to customize the
7963 request sent during the HTTP health checks or the matching rules on the
7964 response. It is also possible to configure a send/expect sequence, just like
7965 with the directive "tcp-check" for TCP health checks.
7966
7967 The server configuration is used by default to open connections to perform
7968 HTTP health checks. By it is also possible to overwrite server parameters
7969 using "http-check connect" rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +01007970
Christopher Faulete5870d82020-04-15 11:32:03 +02007971 "httpchk" option does not necessarily require an HTTP backend, it also works
7972 with plain TCP backends. This is particularly useful to check simple scripts
Christopher Faulet14cd3162020-04-16 14:50:06 +02007973 bound to some dedicated ports using the inetd daemon. However, it will always
Daniel Corbett67a82712020-07-06 23:01:19 -04007974 internally relies on an HTX multiplexer. Thus, it means the request
Christopher Faulet14cd3162020-04-16 14:50:06 +02007975 formatting and the response parsing will be strict.
Willy Tarreaud63335a2010-02-26 12:56:52 +01007976
Christopher Faulet8acb1282020-04-09 08:44:06 +02007977 Note : For a while, there was no way to add headers or body in the request
7978 used for HTTP health checks. So a workaround was to hide it at the end
7979 of the version string with a "\r\n" after the version. It is now
7980 deprecated. The directive "http-check send" must be used instead.
7981
Willy Tarreaud63335a2010-02-26 12:56:52 +01007982 Examples :
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02007983 # Relay HTTPS traffic to Apache instance and check service availability
7984 # using HTTP request "OPTIONS * HTTP/1.1" on port 80.
7985 backend https_relay
7986 mode tcp
7987 option httpchk OPTIONS * HTTP/1.1
7988 http-check send hdr Host www
7989 server apache1 192.168.1.1:443 check port 80
Willy Tarreaud63335a2010-02-26 12:56:52 +01007990
Simon Hormanafc47ee2013-11-25 10:46:35 +09007991 See also : "option ssl-hello-chk", "option smtpchk", "option mysql-check",
7992 "option pgsql-check", "http-check" and the "check", "port" and
7993 "inter" server options.
Willy Tarreaud63335a2010-02-26 12:56:52 +01007994
7995
Willy Tarreauc27debf2008-01-06 08:57:02 +01007996option httpclose
7997no option httpclose
Christopher Faulet315b39c2018-09-21 16:26:19 +02007998 Enable or disable HTTP connection closing
Willy Tarreauc27debf2008-01-06 08:57:02 +01007999 May be used in sections : defaults | frontend | listen | backend
8000 yes | yes | yes | yes
8001 Arguments : none
8002
Willy Tarreau70dffda2014-01-30 03:07:23 +01008003 By default HAProxy operates in keep-alive mode with regards to persistent
8004 connections: for each connection it processes each request and response, and
8005 leaves the connection idle on both sides between the end of a response and
8006 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02008007 as "option http-server-close" or "option httpclose".
Willy Tarreau70dffda2014-01-30 03:07:23 +01008008
Christopher Faulet315b39c2018-09-21 16:26:19 +02008009 If "option httpclose" is set, HAProxy will close connections with the server
8010 and the client as soon as the request and the response are received. It will
John Roeslerfb2fce12019-07-10 15:45:51 -05008011 also check if a "Connection: close" header is already set in each direction,
Christopher Faulet315b39c2018-09-21 16:26:19 +02008012 and will add one if missing. Any "Connection" header different from "close"
8013 will also be removed.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008014
Christopher Faulet315b39c2018-09-21 16:26:19 +02008015 This option may also be combined with "option http-pretend-keepalive", which
8016 will disable sending of the "Connection: close" header, but will still cause
8017 the connection to be closed once the whole response is received.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008018
8019 This option may be set both in a frontend and in a backend. It is enabled if
8020 at least one of the frontend or backend holding a connection has it enabled.
Christopher Faulet159e6672019-07-16 15:09:52 +02008021 It disables and replaces any previous "option http-server-close" or "option
8022 http-keep-alive". Please check section 4 ("Proxies") to see how this option
8023 combines with others when frontend and backend options differ.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008024
8025 If this option has been enabled in a "defaults" section, it can be disabled
8026 in a specific instance by prepending the "no" keyword before it.
8027
Christopher Faulet315b39c2018-09-21 16:26:19 +02008028 See also : "option http-server-close" and "1.1. The HTTP transaction model".
Willy Tarreauc27debf2008-01-06 08:57:02 +01008029
8030
Emeric Brun3a058f32009-06-30 18:26:00 +02008031option httplog [ clf ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01008032 Enable logging of HTTP request, session state and timers
8033 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01008034 yes | yes | yes | no
Emeric Brun3a058f32009-06-30 18:26:00 +02008035 Arguments :
8036 clf if the "clf" argument is added, then the output format will be
8037 the CLF format instead of HAProxy's default HTTP format. You can
8038 use this when you need to feed HAProxy's logs through a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01008039 log analyzer which only support the CLF format and which is not
Emeric Brun3a058f32009-06-30 18:26:00 +02008040 extensible.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008041
8042 By default, the log output format is very poor, as it only contains the
8043 source and destination addresses, and the instance name. By specifying
8044 "option httplog", each log line turns into a much richer format including,
8045 but not limited to, the HTTP request, the connection timers, the session
8046 status, the connections numbers, the captured headers and cookies, the
8047 frontend, backend and server name, and of course the source address and
8048 ports.
8049
PiBa-NLbd556bf2014-12-11 21:31:54 +01008050 Specifying only "option httplog" will automatically clear the 'clf' mode
8051 if it was set by default.
Emeric Brun3a058f32009-06-30 18:26:00 +02008052
Guillaume de Lafond29f45602017-03-31 19:52:15 +02008053 "option httplog" overrides any previous "log-format" directive.
8054
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008055 See also : section 8 about logging.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008056
Willy Tarreau55165fe2009-05-10 12:02:55 +02008057
8058option http_proxy
8059no option http_proxy
8060 Enable or disable plain HTTP proxy mode
8061 May be used in sections : defaults | frontend | listen | backend
8062 yes | yes | yes | yes
8063 Arguments : none
8064
8065 It sometimes happens that people need a pure HTTP proxy which understands
8066 basic proxy requests without caching nor any fancy feature. In this case,
8067 it may be worth setting up an HAProxy instance with the "option http_proxy"
8068 set. In this mode, no server is declared, and the connection is forwarded to
8069 the IP address and port found in the URL after the "http://" scheme.
8070
8071 No host address resolution is performed, so this only works when pure IP
8072 addresses are passed. Since this option's usage perimeter is rather limited,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01008073 it will probably be used only by experts who know they need exactly it. This
8074 is incompatible with the HTTP tunnel mode.
Willy Tarreau55165fe2009-05-10 12:02:55 +02008075
8076 If this option has been enabled in a "defaults" section, it can be disabled
8077 in a specific instance by prepending the "no" keyword before it.
8078
8079 Example :
8080 # this backend understands HTTP proxy requests and forwards them directly.
8081 backend direct_forward
8082 option httpclose
8083 option http_proxy
8084
8085 See also : "option httpclose"
8086
Willy Tarreau211ad242009-10-03 21:45:07 +02008087
Jamie Gloudon801a0a32012-08-25 00:18:33 -04008088option independent-streams
8089no option independent-streams
8090 Enable or disable independent timeout processing for both directions
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02008091 May be used in sections : defaults | frontend | listen | backend
8092 yes | yes | yes | yes
8093 Arguments : none
8094
8095 By default, when data is sent over a socket, both the write timeout and the
8096 read timeout for that socket are refreshed, because we consider that there is
8097 activity on that socket, and we have no other means of guessing if we should
8098 receive data or not.
8099
Davor Ocelice9ed2812017-12-25 17:49:28 +01008100 While this default behavior is desirable for almost all applications, there
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02008101 exists a situation where it is desirable to disable it, and only refresh the
8102 read timeout if there are incoming data. This happens on sessions with large
8103 timeouts and low amounts of exchanged data such as telnet session. If the
8104 server suddenly disappears, the output data accumulates in the system's
8105 socket buffers, both timeouts are correctly refreshed, and there is no way
8106 to know the server does not receive them, so we don't timeout. However, when
8107 the underlying protocol always echoes sent data, it would be enough by itself
8108 to detect the issue using the read timeout. Note that this problem does not
8109 happen with more verbose protocols because data won't accumulate long in the
8110 socket buffers.
8111
8112 When this option is set on the frontend, it will disable read timeout updates
8113 on data sent to the client. There probably is little use of this case. When
8114 the option is set on the backend, it will disable read timeout updates on
8115 data sent to the server. Doing so will typically break large HTTP posts from
8116 slow lines, so use it with caution.
8117
Willy Tarreauce887fd2012-05-12 12:50:00 +02008118 See also : "timeout client", "timeout server" and "timeout tunnel"
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02008119
8120
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02008121option ldap-check
8122 Use LDAPv3 health checks for server testing
8123 May be used in sections : defaults | frontend | listen | backend
8124 yes | no | yes | yes
8125 Arguments : none
8126
8127 It is possible to test that the server correctly talks LDAPv3 instead of just
8128 testing that it accepts the TCP connection. When this option is set, an
8129 LDAPv3 anonymous simple bind message is sent to the server, and the response
8130 is analyzed to find an LDAPv3 bind response message.
8131
8132 The server is considered valid only when the LDAP response contains success
8133 resultCode (http://tools.ietf.org/html/rfc4511#section-4.1.9).
8134
8135 Logging of bind requests is server dependent see your documentation how to
8136 configure it.
8137
8138 Example :
8139 option ldap-check
8140
8141 See also : "option httpchk"
8142
8143
Simon Horman98637e52014-06-20 12:30:16 +09008144option external-check
8145 Use external processes for server health checks
8146 May be used in sections : defaults | frontend | listen | backend
8147 yes | no | yes | yes
8148
8149 It is possible to test the health of a server using an external command.
8150 This is achieved by running the executable set using "external-check
8151 command".
8152
8153 Requires the "external-check" global to be set.
8154
8155 See also : "external-check", "external-check command", "external-check path"
8156
8157
Willy Tarreau211ad242009-10-03 21:45:07 +02008158option log-health-checks
8159no option log-health-checks
Willy Tarreaubef1b322014-05-13 21:01:39 +02008160 Enable or disable logging of health checks status updates
Willy Tarreau211ad242009-10-03 21:45:07 +02008161 May be used in sections : defaults | frontend | listen | backend
8162 yes | no | yes | yes
8163 Arguments : none
8164
Willy Tarreaubef1b322014-05-13 21:01:39 +02008165 By default, failed health check are logged if server is UP and successful
8166 health checks are logged if server is DOWN, so the amount of additional
8167 information is limited.
Willy Tarreau211ad242009-10-03 21:45:07 +02008168
Willy Tarreaubef1b322014-05-13 21:01:39 +02008169 When this option is enabled, any change of the health check status or to
8170 the server's health will be logged, so that it becomes possible to know
8171 that a server was failing occasional checks before crashing, or exactly when
8172 it failed to respond a valid HTTP status, then when the port started to
8173 reject connections, then when the server stopped responding at all.
8174
Davor Ocelice9ed2812017-12-25 17:49:28 +01008175 Note that status changes not caused by health checks (e.g. enable/disable on
Willy Tarreaubef1b322014-05-13 21:01:39 +02008176 the CLI) are intentionally not logged by this option.
Willy Tarreau211ad242009-10-03 21:45:07 +02008177
Willy Tarreaubef1b322014-05-13 21:01:39 +02008178 See also: "option httpchk", "option ldap-check", "option mysql-check",
8179 "option pgsql-check", "option redis-check", "option smtpchk",
8180 "option tcp-check", "log" and section 8 about logging.
Willy Tarreau211ad242009-10-03 21:45:07 +02008181
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02008182
8183option log-separate-errors
8184no option log-separate-errors
8185 Change log level for non-completely successful connections
8186 May be used in sections : defaults | frontend | listen | backend
8187 yes | yes | yes | no
8188 Arguments : none
8189
8190 Sometimes looking for errors in logs is not easy. This option makes haproxy
8191 raise the level of logs containing potentially interesting information such
8192 as errors, timeouts, retries, redispatches, or HTTP status codes 5xx. The
8193 level changes from "info" to "err". This makes it possible to log them
8194 separately to a different file with most syslog daemons. Be careful not to
8195 remove them from the original file, otherwise you would lose ordering which
8196 provides very important information.
8197
8198 Using this option, large sites dealing with several thousand connections per
8199 second may log normal traffic to a rotating buffer and only archive smaller
8200 error logs.
8201
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008202 See also : "log", "dontlognull", "dontlog-normal" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02008203 logging.
8204
Willy Tarreauc27debf2008-01-06 08:57:02 +01008205
8206option logasap
8207no option logasap
Jerome Magnin95fb57b2020-04-23 19:01:17 +02008208 Enable or disable early logging.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008209 May be used in sections : defaults | frontend | listen | backend
8210 yes | yes | yes | no
8211 Arguments : none
8212
Jerome Magnin95fb57b2020-04-23 19:01:17 +02008213 By default, logs are emitted when all the log format variables and sample
8214 fetches used in the definition of the log-format string return a value, or
8215 when the session is terminated. This allows the built in log-format strings
8216 to account for the transfer time, or the number of bytes in log messages.
8217
8218 When handling long lived connections such as large file transfers or RDP,
8219 it may take a while for the request or connection to appear in the logs.
8220 Using "option logasap", the log message is created as soon as the server
8221 connection is established in mode tcp, or as soon as the server sends the
8222 complete headers in mode http. Missing information in the logs will be the
Ilya Shipitsin4329a9a2020-05-05 21:17:10 +05008223 total number of bytes which will only indicate the amount of data transferred
Jerome Magnin95fb57b2020-04-23 19:01:17 +02008224 before the message was created and the total time which will not take the
8225 remainder of the connection life or transfer time into account. For the case
8226 of HTTP, it is good practice to capture the Content-Length response header
8227 so that the logs at least indicate how many bytes are expected to be
Ilya Shipitsin4329a9a2020-05-05 21:17:10 +05008228 transferred.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008229
Willy Tarreaucc6c8912009-02-22 10:53:55 +01008230 Examples :
8231 listen http_proxy 0.0.0.0:80
8232 mode http
8233 option httplog
8234 option logasap
8235 log 192.168.2.200 local3
8236
8237 >>> Feb 6 12:14:14 localhost \
8238 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
8239 static/srv1 9/10/7/14/+30 200 +243 - - ---- 3/1/1/1/0 1/0 \
8240 "GET /image.iso HTTP/1.0"
8241
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008242 See also : "option httplog", "capture response header", and section 8 about
Willy Tarreauc27debf2008-01-06 08:57:02 +01008243 logging.
8244
8245
Christopher Faulet62f79fe2020-05-18 18:13:03 +02008246option mysql-check [ user <username> [ { post-41 | pre-41 } ] ]
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02008247 Use MySQL health checks for server testing
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01008248 May be used in sections : defaults | frontend | listen | backend
8249 yes | no | yes | yes
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02008250 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008251 <username> This is the username which will be used when connecting to MySQL
8252 server.
Christopher Faulet62f79fe2020-05-18 18:13:03 +02008253 post-41 Send post v4.1 client compatible checks (the default)
8254 pre-41 Send pre v4.1 client compatible checks
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02008255
8256 If you specify a username, the check consists of sending two MySQL packet,
8257 one Client Authentication packet, and one QUIT packet, to correctly close
Davor Ocelice9ed2812017-12-25 17:49:28 +01008258 MySQL session. We then parse the MySQL Handshake Initialization packet and/or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02008259 Error packet. It is a basic but useful test which does not produce error nor
8260 aborted connect on the server. However, it requires adding an authorization
8261 in the MySQL table, like this :
8262
8263 USE mysql;
8264 INSERT INTO user (Host,User) values ('<ip_of_haproxy>','<username>');
8265 FLUSH PRIVILEGES;
8266
8267 If you don't specify a username (it is deprecated and not recommended), the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008268 check only consists in parsing the Mysql Handshake Initialization packet or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02008269 Error packet, we don't send anything in this mode. It was reported that it
8270 can generate lockout if check is too frequent and/or if there is not enough
8271 traffic. In fact, you need in this case to check MySQL "max_connect_errors"
8272 value as if a connection is established successfully within fewer than MySQL
8273 "max_connect_errors" attempts after a previous connection was interrupted,
8274 the error count for the host is cleared to zero. If HAProxy's server get
8275 blocked, the "FLUSH HOSTS" statement is the only way to unblock it.
8276
8277 Remember that this does not check database presence nor database consistency.
8278 To do this, you can use an external check with xinetd for example.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01008279
Hervé COMMOWICK212f7782011-06-10 14:05:59 +02008280 The check requires MySQL >=3.22, for older version, please use TCP check.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01008281
8282 Most often, an incoming MySQL server needs to see the client's IP address for
8283 various purposes, including IP privilege matching and connection logging.
8284 When possible, it is often wise to masquerade the client's IP address when
8285 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02008286 which requires the transparent proxy feature to be compiled in, and the MySQL
8287 server to route the client via the machine hosting haproxy.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01008288
8289 See also: "option httpchk"
8290
8291
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008292option nolinger
8293no option nolinger
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01008294 Enable or disable immediate session resource cleaning after close
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008295 May be used in sections: defaults | frontend | listen | backend
8296 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008297 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008298
Davor Ocelice9ed2812017-12-25 17:49:28 +01008299 When clients or servers abort connections in a dirty way (e.g. they are
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008300 physically disconnected), the session timeouts triggers and the session is
8301 closed. But it will remain in FIN_WAIT1 state for some time in the system,
8302 using some resources and possibly limiting the ability to establish newer
8303 connections.
8304
8305 When this happens, it is possible to activate "option nolinger" which forces
8306 the system to immediately remove any socket's pending data on close. Thus,
Willy Tarreau4a321032020-10-16 04:55:19 +02008307 a TCP RST is emitted, any pending data are truncated, and the session is
8308 instantly purged from the system's tables. The generally visible effect for
8309 a client is that responses are truncated if the close happens with a last
8310 block of data (e.g. on a redirect or error response). On the server side,
8311 it may help release the source ports immediately when forwarding a client
8312 aborts in tunnels. In both cases, TCP resets are emitted and given that
8313 the session is instantly destroyed, there will be no retransmit. On a lossy
8314 network this can increase problems, especially when there is a firewall on
8315 the lossy side, because the firewall might see and process the reset (hence
8316 purge its session) and block any further traffic for this session,, including
8317 retransmits from the other side. So if the other side doesn't receive it,
8318 it will never receive any RST again, and the firewall might log many blocked
8319 packets.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008320
Willy Tarreau4a321032020-10-16 04:55:19 +02008321 For all these reasons, it is strongly recommended NOT to use this option,
8322 unless absolutely needed as a last resort. In most situations, using the
8323 "client-fin" or "server-fin" timeouts achieves similar results with a more
8324 reliable behavior. On Linux it's also possible to use the "tcp-ut" bind or
8325 server setting.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008326
8327 This option may be used both on frontends and backends, depending on the side
8328 where it is required. Use it on the frontend for clients, and on the backend
Willy Tarreau4a321032020-10-16 04:55:19 +02008329 for servers. While this option is technically supported in "defaults"
8330 sections, it must really not be used there as it risks to accidently
8331 propagate to sections that must no use it and to cause problems there.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008332
8333 If this option has been enabled in a "defaults" section, it can be disabled
8334 in a specific instance by prepending the "no" keyword before it.
8335
Willy Tarreau4a321032020-10-16 04:55:19 +02008336 See also: "timeout client-fin", "timeout server-fin", "tcp-ut" bind or server
8337 keywords.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008338
Willy Tarreau55165fe2009-05-10 12:02:55 +02008339option originalto [ except <network> ] [ header <name> ]
8340 Enable insertion of the X-Original-To header to requests sent to servers
8341 May be used in sections : defaults | frontend | listen | backend
8342 yes | yes | yes | yes
8343 Arguments :
8344 <network> is an optional argument used to disable this option for sources
8345 matching <network>
8346 <name> an optional argument to specify a different "X-Original-To"
8347 header name.
8348
8349 Since HAProxy can work in transparent mode, every request from a client can
8350 be redirected to the proxy and HAProxy itself can proxy every request to a
8351 complex SQUID environment and the destination host from SO_ORIGINAL_DST will
8352 be lost. This is annoying when you want access rules based on destination ip
8353 addresses. To solve this problem, a new HTTP header "X-Original-To" may be
8354 added by HAProxy to all requests sent to the server. This header contains a
8355 value representing the original destination IP address. Since this must be
8356 configured to always use the last occurrence of this header only. Note that
8357 only the last occurrence of the header must be used, since it is really
8358 possible that the client has already brought one.
8359
8360 The keyword "header" may be used to supply a different header name to replace
8361 the default "X-Original-To". This can be useful where you might already
8362 have a "X-Original-To" header from a different application, and you need
8363 preserve it. Also if your backend server doesn't use the "X-Original-To"
8364 header and requires different one.
8365
8366 Sometimes, a same HAProxy instance may be shared between a direct client
8367 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
8368 used to decrypt HTTPS traffic). It is possible to disable the addition of the
8369 header for a known source address or network by adding the "except" keyword
8370 followed by the network address. In this case, any source IP matching the
8371 network will not cause an addition of this header. Most common uses are with
8372 private networks or 127.0.0.1.
8373
8374 This option may be specified either in the frontend or in the backend. If at
8375 least one of them uses it, the header will be added. Note that the backend's
8376 setting of the header subargument takes precedence over the frontend's if
8377 both are defined.
8378
Willy Tarreau55165fe2009-05-10 12:02:55 +02008379 Examples :
8380 # Original Destination address
8381 frontend www
8382 mode http
8383 option originalto except 127.0.0.1
8384
8385 # Those servers want the IP Address in X-Client-Dst
8386 backend www
8387 mode http
8388 option originalto header X-Client-Dst
8389
Christopher Faulet315b39c2018-09-21 16:26:19 +02008390 See also : "option httpclose", "option http-server-close".
Willy Tarreau55165fe2009-05-10 12:02:55 +02008391
8392
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008393option persist
8394no option persist
8395 Enable or disable forced persistence on down servers
8396 May be used in sections: defaults | frontend | listen | backend
8397 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008398 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008399
8400 When an HTTP request reaches a backend with a cookie which references a dead
8401 server, by default it is redispatched to another server. It is possible to
8402 force the request to be sent to the dead server first using "option persist"
8403 if absolutely needed. A common use case is when servers are under extreme
8404 load and spend their time flapping. In this case, the users would still be
8405 directed to the server they opened the session on, in the hope they would be
8406 correctly served. It is recommended to use "option redispatch" in conjunction
8407 with this option so that in the event it would not be possible to connect to
8408 the server at all (server definitely dead), the client would finally be
8409 redirected to another valid server.
8410
8411 If this option has been enabled in a "defaults" section, it can be disabled
8412 in a specific instance by prepending the "no" keyword before it.
8413
Willy Tarreau4de91492010-01-22 19:10:05 +01008414 See also : "option redispatch", "retries", "force-persist"
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008415
8416
Willy Tarreau0c122822013-12-15 18:49:01 +01008417option pgsql-check [ user <username> ]
8418 Use PostgreSQL health checks for server testing
8419 May be used in sections : defaults | frontend | listen | backend
8420 yes | no | yes | yes
8421 Arguments :
8422 <username> This is the username which will be used when connecting to
8423 PostgreSQL server.
8424
8425 The check sends a PostgreSQL StartupMessage and waits for either
8426 Authentication request or ErrorResponse message. It is a basic but useful
8427 test which does not produce error nor aborted connect on the server.
8428 This check is identical with the "mysql-check".
8429
8430 See also: "option httpchk"
8431
8432
Willy Tarreau9420b122013-12-15 18:58:25 +01008433option prefer-last-server
8434no option prefer-last-server
8435 Allow multiple load balanced requests to remain on the same server
8436 May be used in sections: defaults | frontend | listen | backend
8437 yes | no | yes | yes
8438 Arguments : none
8439
8440 When the load balancing algorithm in use is not deterministic, and a previous
8441 request was sent to a server to which haproxy still holds a connection, it is
8442 sometimes desirable that subsequent requests on a same session go to the same
8443 server as much as possible. Note that this is different from persistence, as
8444 we only indicate a preference which haproxy tries to apply without any form
8445 of warranty. The real use is for keep-alive connections sent to servers. When
8446 this option is used, haproxy will try to reuse the same connection that is
8447 attached to the server instead of rebalancing to another server, causing a
8448 close of the connection. This can make sense for static file servers. It does
Willy Tarreau068621e2013-12-23 15:11:25 +01008449 not make much sense to use this in combination with hashing algorithms. Note,
8450 haproxy already automatically tries to stick to a server which sends a 401 or
Lukas Tribus80512b12018-10-27 20:07:40 +02008451 to a proxy which sends a 407 (authentication required), when the load
8452 balancing algorithm is not deterministic. This is mandatory for use with the
8453 broken NTLM authentication challenge, and significantly helps in
Willy Tarreau068621e2013-12-23 15:11:25 +01008454 troubleshooting some faulty applications. Option prefer-last-server might be
8455 desirable in these environments as well, to avoid redistributing the traffic
8456 after every other response.
Willy Tarreau9420b122013-12-15 18:58:25 +01008457
8458 If this option has been enabled in a "defaults" section, it can be disabled
8459 in a specific instance by prepending the "no" keyword before it.
8460
8461 See also: "option http-keep-alive"
8462
8463
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008464option redispatch
Joseph Lynch726ab712015-05-11 23:25:34 -07008465option redispatch <interval>
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008466no option redispatch
8467 Enable or disable session redistribution in case of connection failure
8468 May be used in sections: defaults | frontend | listen | backend
8469 yes | no | yes | yes
Joseph Lynch726ab712015-05-11 23:25:34 -07008470 Arguments :
8471 <interval> The optional integer value that controls how often redispatches
8472 occur when retrying connections. Positive value P indicates a
8473 redispatch is desired on every Pth retry, and negative value
Davor Ocelice9ed2812017-12-25 17:49:28 +01008474 N indicate a redispatch is desired on the Nth retry prior to the
Joseph Lynch726ab712015-05-11 23:25:34 -07008475 last retry. For example, the default of -1 preserves the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008476 historical behavior of redispatching on the last retry, a
Joseph Lynch726ab712015-05-11 23:25:34 -07008477 positive value of 1 would indicate a redispatch on every retry,
8478 and a positive value of 3 would indicate a redispatch on every
8479 third retry. You can disable redispatches with a value of 0.
8480
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008481
8482 In HTTP mode, if a server designated by a cookie is down, clients may
8483 definitely stick to it because they cannot flush the cookie, so they will not
8484 be able to access the service anymore.
8485
Willy Tarreau59884a62019-01-02 14:48:31 +01008486 Specifying "option redispatch" will allow the proxy to break cookie or
8487 consistent hash based persistence and redistribute them to a working server.
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008488
Olivier Carrère6e6f59b2020-04-15 11:30:18 +02008489 Active servers are selected from a subset of the list of available
8490 servers. Active servers that are not down or in maintenance (i.e., whose
8491 health is not checked or that have been checked as "up"), are selected in the
8492 following order:
8493
8494 1. Any active, non-backup server, if any, or,
8495
8496 2. If the "allbackups" option is not set, the first backup server in the
8497 list, or
8498
8499 3. If the "allbackups" option is set, any backup server.
8500
8501 When a retry occurs, HAProxy tries to select another server than the last
8502 one. The new server is selected from the current list of servers.
8503
8504 Sometimes, if the list is updated between retries (e.g., if numerous retries
8505 occur and last longer than the time needed to check that a server is down,
8506 remove it from the list and fall back on the list of backup servers),
8507 connections may be redirected to a backup server, though.
8508
Joseph Lynch726ab712015-05-11 23:25:34 -07008509 It also allows to retry connections to another server in case of multiple
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008510 connection failures. Of course, it requires having "retries" set to a nonzero
8511 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01008512
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008513 If this option has been enabled in a "defaults" section, it can be disabled
8514 in a specific instance by prepending the "no" keyword before it.
8515
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02008516 See also : "retries", "force-persist"
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008517
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008518
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02008519option redis-check
8520 Use redis health checks for server testing
8521 May be used in sections : defaults | frontend | listen | backend
8522 yes | no | yes | yes
8523 Arguments : none
8524
8525 It is possible to test that the server correctly talks REDIS protocol instead
8526 of just testing that it accepts the TCP connection. When this option is set,
8527 a PING redis command is sent to the server, and the response is analyzed to
8528 find the "+PONG" response message.
8529
8530 Example :
8531 option redis-check
8532
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03008533 See also : "option httpchk", "option tcp-check", "tcp-check expect"
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02008534
8535
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008536option smtpchk
8537option smtpchk <hello> <domain>
8538 Use SMTP health checks for server testing
8539 May be used in sections : defaults | frontend | listen | backend
8540 yes | no | yes | yes
Willy Tarreaud72758d2010-01-12 10:42:19 +01008541 Arguments :
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008542 <hello> is an optional argument. It is the "hello" command to use. It can
Lukas Tribus27935782018-10-01 02:00:16 +02008543 be either "HELO" (for SMTP) or "EHLO" (for ESMTP). All other
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008544 values will be turned into the default command ("HELO").
8545
8546 <domain> is the domain name to present to the server. It may only be
8547 specified (and is mandatory) if the hello command has been
8548 specified. By default, "localhost" is used.
8549
8550 When "option smtpchk" is set, the health checks will consist in TCP
8551 connections followed by an SMTP command. By default, this command is
8552 "HELO localhost". The server's return code is analyzed and only return codes
8553 starting with a "2" will be considered as valid. All other responses,
8554 including a lack of response will constitute an error and will indicate a
8555 dead server.
8556
8557 This test is meant to be used with SMTP servers or relays. Depending on the
8558 request, it is possible that some servers do not log each connection attempt,
Davor Ocelice9ed2812017-12-25 17:49:28 +01008559 so you may want to experiment to improve the behavior. Using telnet on port
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008560 25 is often easier than adjusting the configuration.
8561
8562 Most often, an incoming SMTP server needs to see the client's IP address for
8563 various purposes, including spam filtering, anti-spoofing and logging. When
8564 possible, it is often wise to masquerade the client's IP address when
8565 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02008566 which requires the transparent proxy feature to be compiled in.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008567
8568 Example :
8569 option smtpchk HELO mydomain.org
8570
8571 See also : "option httpchk", "source"
8572
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008573
Krzysztof Piotr Oledzkiaeebf9b2009-10-04 15:43:17 +02008574option socket-stats
8575no option socket-stats
8576
8577 Enable or disable collecting & providing separate statistics for each socket.
8578 May be used in sections : defaults | frontend | listen | backend
8579 yes | yes | yes | no
8580
8581 Arguments : none
8582
8583
Willy Tarreauff4f82d2009-02-06 11:28:13 +01008584option splice-auto
8585no option splice-auto
8586 Enable or disable automatic kernel acceleration on sockets in both directions
8587 May be used in sections : defaults | frontend | listen | backend
8588 yes | yes | yes | yes
8589 Arguments : none
8590
8591 When this option is enabled either on a frontend or on a backend, haproxy
8592 will automatically evaluate the opportunity to use kernel tcp splicing to
Davor Ocelice9ed2812017-12-25 17:49:28 +01008593 forward data between the client and the server, in either direction. HAProxy
Willy Tarreauff4f82d2009-02-06 11:28:13 +01008594 uses heuristics to estimate if kernel splicing might improve performance or
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01008595 not. Both directions are handled independently. Note that the heuristics used
Willy Tarreauff4f82d2009-02-06 11:28:13 +01008596 are not much aggressive in order to limit excessive use of splicing. This
8597 option requires splicing to be enabled at compile time, and may be globally
8598 disabled with the global option "nosplice". Since splice uses pipes, using it
8599 requires that there are enough spare pipes.
8600
8601 Important note: kernel-based TCP splicing is a Linux-specific feature which
8602 first appeared in kernel 2.6.25. It offers kernel-based acceleration to
8603 transfer data between sockets without copying these data to user-space, thus
8604 providing noticeable performance gains and CPU cycles savings. Since many
8605 early implementations are buggy, corrupt data and/or are inefficient, this
8606 feature is not enabled by default, and it should be used with extreme care.
8607 While it is not possible to detect the correctness of an implementation,
8608 2.6.29 is the first version offering a properly working implementation. In
8609 case of doubt, splicing may be globally disabled using the global "nosplice"
8610 keyword.
8611
8612 Example :
8613 option splice-auto
8614
8615 If this option has been enabled in a "defaults" section, it can be disabled
8616 in a specific instance by prepending the "no" keyword before it.
8617
8618 See also : "option splice-request", "option splice-response", and global
8619 options "nosplice" and "maxpipes"
8620
8621
8622option splice-request
8623no option splice-request
8624 Enable or disable automatic kernel acceleration on sockets for requests
8625 May be used in sections : defaults | frontend | listen | backend
8626 yes | yes | yes | yes
8627 Arguments : none
8628
8629 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04008630 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01008631 the client to the server. It might still use the recv/send scheme if there
8632 are no spare pipes left. This option requires splicing to be enabled at
8633 compile time, and may be globally disabled with the global option "nosplice".
8634 Since splice uses pipes, using it requires that there are enough spare pipes.
8635
8636 Important note: see "option splice-auto" for usage limitations.
8637
8638 Example :
8639 option splice-request
8640
8641 If this option has been enabled in a "defaults" section, it can be disabled
8642 in a specific instance by prepending the "no" keyword before it.
8643
8644 See also : "option splice-auto", "option splice-response", and global options
8645 "nosplice" and "maxpipes"
8646
8647
8648option splice-response
8649no option splice-response
8650 Enable or disable automatic kernel acceleration on sockets for responses
8651 May be used in sections : defaults | frontend | listen | backend
8652 yes | yes | yes | yes
8653 Arguments : none
8654
8655 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04008656 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01008657 the server to the client. It might still use the recv/send scheme if there
8658 are no spare pipes left. This option requires splicing to be enabled at
8659 compile time, and may be globally disabled with the global option "nosplice".
8660 Since splice uses pipes, using it requires that there are enough spare pipes.
8661
8662 Important note: see "option splice-auto" for usage limitations.
8663
8664 Example :
8665 option splice-response
8666
8667 If this option has been enabled in a "defaults" section, it can be disabled
8668 in a specific instance by prepending the "no" keyword before it.
8669
8670 See also : "option splice-auto", "option splice-request", and global options
8671 "nosplice" and "maxpipes"
8672
8673
Christopher Fauletba7bc162016-11-07 21:07:38 +01008674option spop-check
8675 Use SPOP health checks for server testing
8676 May be used in sections : defaults | frontend | listen | backend
8677 no | no | no | yes
8678 Arguments : none
8679
8680 It is possible to test that the server correctly talks SPOP protocol instead
8681 of just testing that it accepts the TCP connection. When this option is set,
8682 a HELLO handshake is performed between HAProxy and the server, and the
8683 response is analyzed to check no error is reported.
8684
8685 Example :
8686 option spop-check
8687
8688 See also : "option httpchk"
8689
8690
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008691option srvtcpka
8692no option srvtcpka
8693 Enable or disable the sending of TCP keepalive packets on the server side
8694 May be used in sections : defaults | frontend | listen | backend
8695 yes | no | yes | yes
8696 Arguments : none
8697
8698 When there is a firewall or any session-aware component between a client and
8699 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01008700 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008701 components decides to expire a session which has remained idle for too long.
8702
8703 Enabling socket-level TCP keep-alives makes the system regularly send packets
8704 to the other end of the connection, leaving it active. The delay between
8705 keep-alive probes is controlled by the system only and depends both on the
8706 operating system and its tuning parameters.
8707
8708 It is important to understand that keep-alive packets are neither emitted nor
8709 received at the application level. It is only the network stacks which sees
8710 them. For this reason, even if one side of the proxy already uses keep-alives
8711 to maintain its connection alive, those keep-alive packets will not be
8712 forwarded to the other side of the proxy.
8713
8714 Please note that this has nothing to do with HTTP keep-alive.
8715
8716 Using option "srvtcpka" enables the emission of TCP keep-alive probes on the
8717 server side of a connection, which should help when session expirations are
8718 noticed between HAProxy and a server.
8719
8720 If this option has been enabled in a "defaults" section, it can be disabled
8721 in a specific instance by prepending the "no" keyword before it.
8722
8723 See also : "option clitcpka", "option tcpka"
8724
8725
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008726option ssl-hello-chk
8727 Use SSLv3 client hello health checks for server testing
8728 May be used in sections : defaults | frontend | listen | backend
8729 yes | no | yes | yes
8730 Arguments : none
8731
8732 When some SSL-based protocols are relayed in TCP mode through HAProxy, it is
8733 possible to test that the server correctly talks SSL instead of just testing
8734 that it accepts the TCP connection. When "option ssl-hello-chk" is set, pure
8735 SSLv3 client hello messages are sent once the connection is established to
8736 the server, and the response is analyzed to find an SSL server hello message.
8737 The server is considered valid only when the response contains this server
8738 hello message.
8739
8740 All servers tested till there correctly reply to SSLv3 client hello messages,
8741 and most servers tested do not even log the requests containing only hello
8742 messages, which is appreciable.
8743
Willy Tarreau763a95b2012-10-04 23:15:39 +02008744 Note that this check works even when SSL support was not built into haproxy
8745 because it forges the SSL message. When SSL support is available, it is best
8746 to use native SSL health checks instead of this one.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008747
Willy Tarreau763a95b2012-10-04 23:15:39 +02008748 See also: "option httpchk", "check-ssl"
8749
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008750
Willy Tarreaued179852013-12-16 01:07:00 +01008751option tcp-check
8752 Perform health checks using tcp-check send/expect sequences
8753 May be used in sections: defaults | frontend | listen | backend
8754 yes | no | yes | yes
8755
8756 This health check method is intended to be combined with "tcp-check" command
8757 lists in order to support send/expect types of health check sequences.
8758
8759 TCP checks currently support 4 modes of operations :
8760 - no "tcp-check" directive : the health check only consists in a connection
8761 attempt, which remains the default mode.
8762
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008763 - "tcp-check send" or "tcp-check send-binary" only is mentioned : this is
Willy Tarreaued179852013-12-16 01:07:00 +01008764 used to send a string along with a connection opening. With some
8765 protocols, it helps sending a "QUIT" message for example that prevents
8766 the server from logging a connection error for each health check. The
8767 check result will still be based on the ability to open the connection
8768 only.
8769
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008770 - "tcp-check expect" only is mentioned : this is used to test a banner.
Willy Tarreaued179852013-12-16 01:07:00 +01008771 The connection is opened and haproxy waits for the server to present some
8772 contents which must validate some rules. The check result will be based
8773 on the matching between the contents and the rules. This is suited for
8774 POP, IMAP, SMTP, FTP, SSH, TELNET.
8775
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008776 - both "tcp-check send" and "tcp-check expect" are mentioned : this is
Davor Ocelice9ed2812017-12-25 17:49:28 +01008777 used to test a hello-type protocol. HAProxy sends a message, the server
8778 responds and its response is analyzed. the check result will be based on
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008779 the matching between the response contents and the rules. This is often
Willy Tarreaued179852013-12-16 01:07:00 +01008780 suited for protocols which require a binding or a request/response model.
8781 LDAP, MySQL, Redis and SSL are example of such protocols, though they
8782 already all have their dedicated checks with a deeper understanding of
8783 the respective protocols.
8784 In this mode, many questions may be sent and many answers may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01008785 analyzed.
Willy Tarreaued179852013-12-16 01:07:00 +01008786
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02008787 A fifth mode can be used to insert comments in different steps of the script.
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02008788
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02008789 For each tcp-check rule you create, you can add a "comment" directive,
8790 followed by a string. This string will be reported in the log and stderr in
8791 debug mode. It is useful to make user-friendly error reporting. The
8792 "comment" is of course optional.
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02008793
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02008794 During the execution of a health check, a variable scope is made available to
8795 store data samples, using the "tcp-check set-var" operation. Freeing those
8796 variable is possible using "tcp-check unset-var".
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +01008797
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02008798
Willy Tarreaued179852013-12-16 01:07:00 +01008799 Examples :
Davor Ocelice9ed2812017-12-25 17:49:28 +01008800 # perform a POP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01008801 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02008802 tcp-check expect string +OK\ POP3\ ready comment POP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01008803
Davor Ocelice9ed2812017-12-25 17:49:28 +01008804 # perform an IMAP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01008805 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02008806 tcp-check expect string *\ OK\ IMAP4\ ready comment IMAP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01008807
8808 # look for the redis master server after ensuring it speaks well
8809 # redis protocol, then it exits properly.
Davor Ocelice9ed2812017-12-25 17:49:28 +01008810 # (send a command then analyze the response 3 times)
Willy Tarreaued179852013-12-16 01:07:00 +01008811 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02008812 tcp-check comment PING\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01008813 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02008814 tcp-check expect string +PONG
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02008815 tcp-check comment role\ check
Willy Tarreaued179852013-12-16 01:07:00 +01008816 tcp-check send info\ replication\r\n
8817 tcp-check expect string role:master
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02008818 tcp-check comment QUIT\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01008819 tcp-check send QUIT\r\n
8820 tcp-check expect string +OK
8821
Davor Ocelice9ed2812017-12-25 17:49:28 +01008822 forge a HTTP request, then analyze the response
Willy Tarreaued179852013-12-16 01:07:00 +01008823 (send many headers before analyzing)
8824 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02008825 tcp-check comment forge\ and\ send\ HTTP\ request
Willy Tarreaued179852013-12-16 01:07:00 +01008826 tcp-check send HEAD\ /\ HTTP/1.1\r\n
8827 tcp-check send Host:\ www.mydomain.com\r\n
8828 tcp-check send User-Agent:\ HAProxy\ tcpcheck\r\n
8829 tcp-check send \r\n
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02008830 tcp-check expect rstring HTTP/1\..\ (2..|3..) comment check\ HTTP\ response
Willy Tarreaued179852013-12-16 01:07:00 +01008831
8832
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02008833 See also : "tcp-check connect", "tcp-check expect" and "tcp-check send".
Willy Tarreaued179852013-12-16 01:07:00 +01008834
8835
Willy Tarreau9ea05a72009-06-14 12:07:01 +02008836option tcp-smart-accept
8837no option tcp-smart-accept
8838 Enable or disable the saving of one ACK packet during the accept sequence
8839 May be used in sections : defaults | frontend | listen | backend
8840 yes | yes | yes | no
8841 Arguments : none
8842
8843 When an HTTP connection request comes in, the system acknowledges it on
8844 behalf of HAProxy, then the client immediately sends its request, and the
8845 system acknowledges it too while it is notifying HAProxy about the new
8846 connection. HAProxy then reads the request and responds. This means that we
8847 have one TCP ACK sent by the system for nothing, because the request could
8848 very well be acknowledged by HAProxy when it sends its response.
8849
8850 For this reason, in HTTP mode, HAProxy automatically asks the system to avoid
8851 sending this useless ACK on platforms which support it (currently at least
8852 Linux). It must not cause any problem, because the system will send it anyway
8853 after 40 ms if the response takes more time than expected to come.
8854
8855 During complex network debugging sessions, it may be desirable to disable
8856 this optimization because delayed ACKs can make troubleshooting more complex
8857 when trying to identify where packets are delayed. It is then possible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01008858 fall back to normal behavior by specifying "no option tcp-smart-accept".
Willy Tarreau9ea05a72009-06-14 12:07:01 +02008859
8860 It is also possible to force it for non-HTTP proxies by simply specifying
8861 "option tcp-smart-accept". For instance, it can make sense with some services
8862 such as SMTP where the server speaks first.
8863
8864 It is recommended to avoid forcing this option in a defaults section. In case
8865 of doubt, consider setting it back to automatic values by prepending the
8866 "default" keyword before it, or disabling it using the "no" keyword.
8867
Willy Tarreaud88edf22009-06-14 15:48:17 +02008868 See also : "option tcp-smart-connect"
8869
8870
8871option tcp-smart-connect
8872no option tcp-smart-connect
8873 Enable or disable the saving of one ACK packet during the connect sequence
8874 May be used in sections : defaults | frontend | listen | backend
8875 yes | no | yes | yes
8876 Arguments : none
8877
8878 On certain systems (at least Linux), HAProxy can ask the kernel not to
8879 immediately send an empty ACK upon a connection request, but to directly
8880 send the buffer request instead. This saves one packet on the network and
8881 thus boosts performance. It can also be useful for some servers, because they
8882 immediately get the request along with the incoming connection.
8883
8884 This feature is enabled when "option tcp-smart-connect" is set in a backend.
8885 It is not enabled by default because it makes network troubleshooting more
8886 complex.
8887
8888 It only makes sense to enable it with protocols where the client speaks first
8889 such as HTTP. In other situations, if there is no data to send in place of
8890 the ACK, a normal ACK is sent.
8891
8892 If this option has been enabled in a "defaults" section, it can be disabled
8893 in a specific instance by prepending the "no" keyword before it.
8894
8895 See also : "option tcp-smart-accept"
8896
Willy Tarreau9ea05a72009-06-14 12:07:01 +02008897
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008898option tcpka
8899 Enable or disable the sending of TCP keepalive packets on both sides
8900 May be used in sections : defaults | frontend | listen | backend
8901 yes | yes | yes | yes
8902 Arguments : none
8903
8904 When there is a firewall or any session-aware component between a client and
8905 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01008906 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008907 components decides to expire a session which has remained idle for too long.
8908
8909 Enabling socket-level TCP keep-alives makes the system regularly send packets
8910 to the other end of the connection, leaving it active. The delay between
8911 keep-alive probes is controlled by the system only and depends both on the
8912 operating system and its tuning parameters.
8913
8914 It is important to understand that keep-alive packets are neither emitted nor
8915 received at the application level. It is only the network stacks which sees
8916 them. For this reason, even if one side of the proxy already uses keep-alives
8917 to maintain its connection alive, those keep-alive packets will not be
8918 forwarded to the other side of the proxy.
8919
8920 Please note that this has nothing to do with HTTP keep-alive.
8921
8922 Using option "tcpka" enables the emission of TCP keep-alive probes on both
8923 the client and server sides of a connection. Note that this is meaningful
8924 only in "defaults" or "listen" sections. If this option is used in a
8925 frontend, only the client side will get keep-alives, and if this option is
8926 used in a backend, only the server side will get keep-alives. For this
8927 reason, it is strongly recommended to explicitly use "option clitcpka" and
8928 "option srvtcpka" when the configuration is split between frontends and
8929 backends.
8930
8931 See also : "option clitcpka", "option srvtcpka"
8932
Willy Tarreau844e3c52008-01-11 16:28:18 +01008933
8934option tcplog
8935 Enable advanced logging of TCP connections with session state and timers
8936 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01008937 yes | yes | yes | no
Willy Tarreau844e3c52008-01-11 16:28:18 +01008938 Arguments : none
8939
8940 By default, the log output format is very poor, as it only contains the
8941 source and destination addresses, and the instance name. By specifying
8942 "option tcplog", each log line turns into a much richer format including, but
8943 not limited to, the connection timers, the session status, the connections
8944 numbers, the frontend, backend and server name, and of course the source
8945 address and ports. This option is useful for pure TCP proxies in order to
8946 find which of the client or server disconnects or times out. For normal HTTP
8947 proxies, it's better to use "option httplog" which is even more complete.
8948
Guillaume de Lafond29f45602017-03-31 19:52:15 +02008949 "option tcplog" overrides any previous "log-format" directive.
8950
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008951 See also : "option httplog", and section 8 about logging.
Willy Tarreau844e3c52008-01-11 16:28:18 +01008952
8953
Willy Tarreau844e3c52008-01-11 16:28:18 +01008954option transparent
8955no option transparent
8956 Enable client-side transparent proxying
8957 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +01008958 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +01008959 Arguments : none
8960
8961 This option was introduced in order to provide layer 7 persistence to layer 3
8962 load balancers. The idea is to use the OS's ability to redirect an incoming
8963 connection for a remote address to a local process (here HAProxy), and let
8964 this process know what address was initially requested. When this option is
8965 used, sessions without cookies will be forwarded to the original destination
8966 IP address of the incoming request (which should match that of another
8967 equipment), while requests with cookies will still be forwarded to the
8968 appropriate server.
8969
8970 Note that contrary to a common belief, this option does NOT make HAProxy
8971 present the client's IP to the server when establishing the connection.
8972
Willy Tarreaua1146052011-03-01 09:51:54 +01008973 See also: the "usesrc" argument of the "source" keyword, and the
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008974 "transparent" option of the "bind" keyword.
Willy Tarreau844e3c52008-01-11 16:28:18 +01008975
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008976
Simon Horman98637e52014-06-20 12:30:16 +09008977external-check command <command>
8978 Executable to run when performing an external-check
8979 May be used in sections : defaults | frontend | listen | backend
8980 yes | no | yes | yes
8981
8982 Arguments :
8983 <command> is the external command to run
8984
Simon Horman98637e52014-06-20 12:30:16 +09008985 The arguments passed to the to the command are:
8986
Cyril Bonté777be862014-12-02 21:21:35 +01008987 <proxy_address> <proxy_port> <server_address> <server_port>
Simon Horman98637e52014-06-20 12:30:16 +09008988
Cyril Bonté777be862014-12-02 21:21:35 +01008989 The <proxy_address> and <proxy_port> are derived from the first listener
8990 that is either IPv4, IPv6 or a UNIX socket. In the case of a UNIX socket
8991 listener the proxy_address will be the path of the socket and the
8992 <proxy_port> will be the string "NOT_USED". In a backend section, it's not
8993 possible to determine a listener, and both <proxy_address> and <proxy_port>
8994 will have the string value "NOT_USED".
Simon Horman98637e52014-06-20 12:30:16 +09008995
Cyril Bonté72cda2a2014-12-27 22:28:39 +01008996 Some values are also provided through environment variables.
8997
8998 Environment variables :
8999 HAPROXY_PROXY_ADDR The first bind address if available (or empty if not
9000 applicable, for example in a "backend" section).
9001
9002 HAPROXY_PROXY_ID The backend id.
9003
9004 HAPROXY_PROXY_NAME The backend name.
9005
9006 HAPROXY_PROXY_PORT The first bind port if available (or empty if not
9007 applicable, for example in a "backend" section or
9008 for a UNIX socket).
9009
9010 HAPROXY_SERVER_ADDR The server address.
9011
9012 HAPROXY_SERVER_CURCONN The current number of connections on the server.
9013
9014 HAPROXY_SERVER_ID The server id.
9015
9016 HAPROXY_SERVER_MAXCONN The server max connections.
9017
9018 HAPROXY_SERVER_NAME The server name.
9019
9020 HAPROXY_SERVER_PORT The server port if available (or empty for a UNIX
9021 socket).
9022
9023 PATH The PATH environment variable used when executing
9024 the command may be set using "external-check path".
9025
William Lallemand4d03e432019-06-14 15:35:37 +02009026 See also "2.3. Environment variables" for other variables.
9027
Simon Horman98637e52014-06-20 12:30:16 +09009028 If the command executed and exits with a zero status then the check is
9029 considered to have passed, otherwise the check is considered to have
9030 failed.
9031
9032 Example :
9033 external-check command /bin/true
9034
9035 See also : "external-check", "option external-check", "external-check path"
9036
9037
9038external-check path <path>
9039 The value of the PATH environment variable used when running an external-check
9040 May be used in sections : defaults | frontend | listen | backend
9041 yes | no | yes | yes
9042
9043 Arguments :
9044 <path> is the path used when executing external command to run
9045
9046 The default path is "".
9047
9048 Example :
9049 external-check path "/usr/bin:/bin"
9050
9051 See also : "external-check", "option external-check",
9052 "external-check command"
9053
9054
Emeric Brun647caf12009-06-30 17:57:00 +02009055persist rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02009056persist rdp-cookie(<name>)
Emeric Brun647caf12009-06-30 17:57:00 +02009057 Enable RDP cookie-based persistence
9058 May be used in sections : defaults | frontend | listen | backend
9059 yes | no | yes | yes
9060 Arguments :
9061 <name> is the optional name of the RDP cookie to check. If omitted, the
Willy Tarreau61e28f22010-05-16 22:31:05 +02009062 default cookie name "msts" will be used. There currently is no
9063 valid reason to change this name.
Emeric Brun647caf12009-06-30 17:57:00 +02009064
9065 This statement enables persistence based on an RDP cookie. The RDP cookie
9066 contains all information required to find the server in the list of known
Davor Ocelice9ed2812017-12-25 17:49:28 +01009067 servers. So when this option is set in the backend, the request is analyzed
Emeric Brun647caf12009-06-30 17:57:00 +02009068 and if an RDP cookie is found, it is decoded. If it matches a known server
9069 which is still UP (or if "option persist" is set), then the connection is
9070 forwarded to this server.
9071
9072 Note that this only makes sense in a TCP backend, but for this to work, the
9073 frontend must have waited long enough to ensure that an RDP cookie is present
9074 in the request buffer. This is the same requirement as with the "rdp-cookie"
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009075 load-balancing method. Thus it is highly recommended to put all statements in
Emeric Brun647caf12009-06-30 17:57:00 +02009076 a single "listen" section.
9077
Willy Tarreau61e28f22010-05-16 22:31:05 +02009078 Also, it is important to understand that the terminal server will emit this
9079 RDP cookie only if it is configured for "token redirection mode", which means
9080 that the "IP address redirection" option is disabled.
9081
Emeric Brun647caf12009-06-30 17:57:00 +02009082 Example :
9083 listen tse-farm
9084 bind :3389
9085 # wait up to 5s for an RDP cookie in the request
9086 tcp-request inspect-delay 5s
9087 tcp-request content accept if RDP_COOKIE
9088 # apply RDP cookie persistence
9089 persist rdp-cookie
9090 # if server is unknown, let's balance on the same cookie.
Cyril Bontédc4d9032012-04-08 21:57:39 +02009091 # alternatively, "balance leastconn" may be useful too.
Emeric Brun647caf12009-06-30 17:57:00 +02009092 balance rdp-cookie
9093 server srv1 1.1.1.1:3389
9094 server srv2 1.1.1.2:3389
9095
Simon Hormanab814e02011-06-24 14:50:20 +09009096 See also : "balance rdp-cookie", "tcp-request", the "req_rdp_cookie" ACL and
9097 the rdp_cookie pattern fetch function.
Emeric Brun647caf12009-06-30 17:57:00 +02009098
9099
Willy Tarreau3a7d2072009-03-05 23:48:25 +01009100rate-limit sessions <rate>
9101 Set a limit on the number of new sessions accepted per second on a frontend
9102 May be used in sections : defaults | frontend | listen | backend
9103 yes | yes | yes | no
9104 Arguments :
9105 <rate> The <rate> parameter is an integer designating the maximum number
9106 of new sessions per second to accept on the frontend.
9107
9108 When the frontend reaches the specified number of new sessions per second, it
9109 stops accepting new connections until the rate drops below the limit again.
9110 During this time, the pending sessions will be kept in the socket's backlog
9111 (in system buffers) and haproxy will not even be aware that sessions are
9112 pending. When applying very low limit on a highly loaded service, it may make
9113 sense to increase the socket's backlog using the "backlog" keyword.
9114
9115 This feature is particularly efficient at blocking connection-based attacks
9116 or service abuse on fragile servers. Since the session rate is measured every
9117 millisecond, it is extremely accurate. Also, the limit applies immediately,
9118 no delay is needed at all to detect the threshold.
9119
9120 Example : limit the connection rate on SMTP to 10 per second max
9121 listen smtp
9122 mode tcp
9123 bind :25
9124 rate-limit sessions 10
Panagiotis Panagiotopoulos7282d8e2016-02-11 16:37:15 +02009125 server smtp1 127.0.0.1:1025
Willy Tarreau3a7d2072009-03-05 23:48:25 +01009126
Willy Tarreaua17c2d92011-07-25 08:16:20 +02009127 Note : when the maximum rate is reached, the frontend's status is not changed
9128 but its sockets appear as "WAITING" in the statistics if the
9129 "socket-stats" option is enabled.
Willy Tarreau3a7d2072009-03-05 23:48:25 +01009130
9131 See also : the "backlog" keyword and the "fe_sess_rate" ACL criterion.
9132
9133
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009134redirect location <loc> [code <code>] <option> [{if | unless} <condition>]
9135redirect prefix <pfx> [code <code>] <option> [{if | unless} <condition>]
9136redirect scheme <sch> [code <code>] <option> [{if | unless} <condition>]
Willy Tarreaub463dfb2008-06-07 23:08:56 +02009137 Return an HTTP redirection if/unless a condition is matched
9138 May be used in sections : defaults | frontend | listen | backend
9139 no | yes | yes | yes
9140
9141 If/unless the condition is matched, the HTTP request will lead to a redirect
Willy Tarreauf285f542010-01-03 20:03:03 +01009142 response. If no condition is specified, the redirect applies unconditionally.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02009143
Willy Tarreau0140f252008-11-19 21:07:09 +01009144 Arguments :
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009145 <loc> With "redirect location", the exact value in <loc> is placed into
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01009146 the HTTP "Location" header. When used in an "http-request" rule,
9147 <loc> value follows the log-format rules and can include some
9148 dynamic values (see Custom Log Format in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009149
9150 <pfx> With "redirect prefix", the "Location" header is built from the
9151 concatenation of <pfx> and the complete URI path, including the
9152 query string, unless the "drop-query" option is specified (see
9153 below). As a special case, if <pfx> equals exactly "/", then
9154 nothing is inserted before the original URI. It allows one to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01009155 redirect to the same URL (for instance, to insert a cookie). When
9156 used in an "http-request" rule, <pfx> value follows the log-format
9157 rules and can include some dynamic values (see Custom Log Format
9158 in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009159
9160 <sch> With "redirect scheme", then the "Location" header is built by
9161 concatenating <sch> with "://" then the first occurrence of the
9162 "Host" header, and then the URI path, including the query string
9163 unless the "drop-query" option is specified (see below). If no
9164 path is found or if the path is "*", then "/" is used instead. If
9165 no "Host" header is found, then an empty host component will be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009166 returned, which most recent browsers interpret as redirecting to
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009167 the same host. This directive is mostly used to redirect HTTP to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01009168 HTTPS. When used in an "http-request" rule, <sch> value follows
9169 the log-format rules and can include some dynamic values (see
9170 Custom Log Format in section 8.2.4).
Willy Tarreau0140f252008-11-19 21:07:09 +01009171
9172 <code> The code is optional. It indicates which type of HTTP redirection
Willy Tarreaub67fdc42013-03-29 19:28:11 +01009173 is desired. Only codes 301, 302, 303, 307 and 308 are supported,
9174 with 302 used by default if no code is specified. 301 means
9175 "Moved permanently", and a browser may cache the Location. 302
Baptiste Assmannea849c02015-08-03 11:42:50 +02009176 means "Moved temporarily" and means that the browser should not
Willy Tarreaub67fdc42013-03-29 19:28:11 +01009177 cache the redirection. 303 is equivalent to 302 except that the
9178 browser will fetch the location with a GET method. 307 is just
9179 like 302 but makes it clear that the same method must be reused.
9180 Likewise, 308 replaces 301 if the same method must be used.
Willy Tarreau0140f252008-11-19 21:07:09 +01009181
9182 <option> There are several options which can be specified to adjust the
Davor Ocelice9ed2812017-12-25 17:49:28 +01009183 expected behavior of a redirection :
Willy Tarreau0140f252008-11-19 21:07:09 +01009184
9185 - "drop-query"
9186 When this keyword is used in a prefix-based redirection, then the
9187 location will be set without any possible query-string, which is useful
9188 for directing users to a non-secure page for instance. It has no effect
9189 with a location-type redirect.
9190
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01009191 - "append-slash"
9192 This keyword may be used in conjunction with "drop-query" to redirect
9193 users who use a URL not ending with a '/' to the same one with the '/'.
9194 It can be useful to ensure that search engines will only see one URL.
9195 For this, a return code 301 is preferred.
9196
Willy Tarreau0140f252008-11-19 21:07:09 +01009197 - "set-cookie NAME[=value]"
9198 A "Set-Cookie" header will be added with NAME (and optionally "=value")
9199 to the response. This is sometimes used to indicate that a user has
9200 been seen, for instance to protect against some types of DoS. No other
9201 cookie option is added, so the cookie will be a session cookie. Note
9202 that for a browser, a sole cookie name without an equal sign is
9203 different from a cookie with an equal sign.
9204
9205 - "clear-cookie NAME[=]"
9206 A "Set-Cookie" header will be added with NAME (and optionally "="), but
9207 with the "Max-Age" attribute set to zero. This will tell the browser to
9208 delete this cookie. It is useful for instance on logout pages. It is
9209 important to note that clearing the cookie "NAME" will not remove a
9210 cookie set with "NAME=value". You have to clear the cookie "NAME=" for
9211 that, because the browser makes the difference.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02009212
9213 Example: move the login URL only to HTTPS.
9214 acl clear dst_port 80
9215 acl secure dst_port 8080
9216 acl login_page url_beg /login
Willy Tarreau0140f252008-11-19 21:07:09 +01009217 acl logout url_beg /logout
Willy Tarreau79da4692008-11-19 20:03:04 +01009218 acl uid_given url_reg /login?userid=[^&]+
Willy Tarreau0140f252008-11-19 21:07:09 +01009219 acl cookie_set hdr_sub(cookie) SEEN=1
9220
9221 redirect prefix https://mysite.com set-cookie SEEN=1 if !cookie_set
Willy Tarreau79da4692008-11-19 20:03:04 +01009222 redirect prefix https://mysite.com if login_page !secure
9223 redirect prefix http://mysite.com drop-query if login_page !uid_given
9224 redirect location http://mysite.com/ if !login_page secure
Willy Tarreau0140f252008-11-19 21:07:09 +01009225 redirect location / clear-cookie USERID= if logout
Willy Tarreaub463dfb2008-06-07 23:08:56 +02009226
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01009227 Example: send redirects for request for articles without a '/'.
9228 acl missing_slash path_reg ^/article/[^/]*$
9229 redirect code 301 prefix / drop-query append-slash if missing_slash
9230
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009231 Example: redirect all HTTP traffic to HTTPS when SSL is handled by haproxy.
David BERARDe7153042012-11-03 00:11:31 +01009232 redirect scheme https if !{ ssl_fc }
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009233
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01009234 Example: append 'www.' prefix in front of all hosts not having it
Coen Rosdorff596659b2016-04-11 11:33:49 +02009235 http-request redirect code 301 location \
9236 http://www.%[hdr(host)]%[capture.req.uri] \
9237 unless { hdr_beg(host) -i www }
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01009238
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009239 See section 7 about ACL usage.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02009240
Willy Tarreau303c0352008-01-17 19:01:39 +01009241
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02009242retries <value>
9243 Set the number of retries to perform on a server after a connection failure
9244 May be used in sections: defaults | frontend | listen | backend
9245 yes | no | yes | yes
9246 Arguments :
9247 <value> is the number of times a connection attempt should be retried on
9248 a server when a connection either is refused or times out. The
9249 default value is 3.
9250
9251 It is important to understand that this value applies to the number of
9252 connection attempts, not full requests. When a connection has effectively
9253 been established to a server, there will be no more retry.
9254
9255 In order to avoid immediate reconnections to a server which is restarting,
Joseph Lynch726ab712015-05-11 23:25:34 -07009256 a turn-around timer of min("timeout connect", one second) is applied before
9257 a retry occurs.
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02009258
9259 When "option redispatch" is set, the last retry may be performed on another
9260 server even if a cookie references a different server.
9261
9262 See also : "option redispatch"
9263
9264
Olivier Houcharda254a372019-04-05 15:30:12 +02009265retry-on [list of keywords]
Jerome Magnin5ce3c142020-05-13 20:09:57 +02009266 Specify when to attempt to automatically retry a failed request.
9267 This setting is only valid when "mode" is set to http and is silently ignored
9268 otherwise.
Olivier Houcharda254a372019-04-05 15:30:12 +02009269 May be used in sections: defaults | frontend | listen | backend
9270 yes | no | yes | yes
9271 Arguments :
9272 <keywords> is a list of keywords or HTTP status codes, each representing a
9273 type of failure event on which an attempt to retry the request
9274 is desired. Please read the notes at the bottom before changing
9275 this setting. The following keywords are supported :
9276
9277 none never retry
9278
9279 conn-failure retry when the connection or the SSL handshake failed
9280 and the request could not be sent. This is the default.
9281
9282 empty-response retry when the server connection was closed after part
9283 of the request was sent, and nothing was received from
9284 the server. This type of failure may be caused by the
9285 request timeout on the server side, poor network
9286 condition, or a server crash or restart while
9287 processing the request.
9288
Olivier Houcharde3249a92019-05-03 23:01:47 +02009289 junk-response retry when the server returned something not looking
9290 like a complete HTTP response. This includes partial
9291 responses headers as well as non-HTTP contents. It
9292 usually is a bad idea to retry on such events, which
9293 may be caused a configuration issue (wrong server port)
9294 or by the request being harmful to the server (buffer
9295 overflow attack for example).
9296
Olivier Houcharda254a372019-04-05 15:30:12 +02009297 response-timeout the server timeout stroke while waiting for the server
9298 to respond to the request. This may be caused by poor
9299 network condition, the reuse of an idle connection
9300 which has expired on the path, or by the request being
9301 extremely expensive to process. It generally is a bad
9302 idea to retry on such events on servers dealing with
9303 heavy database processing (full scans, etc) as it may
9304 amplify denial of service attacks.
9305
Olivier Houchard865d8392019-05-03 22:46:27 +02009306 0rtt-rejected retry requests which were sent over early data and were
9307 rejected by the server. These requests are generally
9308 considered to be safe to retry.
9309
Olivier Houcharda254a372019-04-05 15:30:12 +02009310 <status> any HTTP status code among "404" (Not Found), "408"
9311 (Request Timeout), "425" (Too Early), "500" (Server
9312 Error), "501" (Not Implemented), "502" (Bad Gateway),
9313 "503" (Service Unavailable), "504" (Gateway Timeout).
9314
Olivier Houchardddf0e032019-05-10 18:05:40 +02009315 all-retryable-errors
9316 retry request for any error that are considered
9317 retryable. This currently activates "conn-failure",
9318 "empty-response", "junk-response", "response-timeout",
9319 "0rtt-rejected", "500", "502", "503", and "504".
9320
Olivier Houcharda254a372019-04-05 15:30:12 +02009321 Using this directive replaces any previous settings with the new ones; it is
9322 not cumulative.
9323
9324 Please note that using anything other than "none" and "conn-failure" requires
9325 to allocate a buffer and copy the whole request into it, so it has memory and
9326 performance impacts. Requests not fitting in a single buffer will never be
9327 retried (see the global tune.bufsize setting).
9328
9329 You have to make sure the application has a replay protection mechanism built
9330 in such as a unique transaction IDs passed in requests, or that replaying the
9331 same request has no consequence, or it is very dangerous to use any retry-on
9332 value beside "conn-failure" and "none". Static file servers and caches are
9333 generally considered safe against any type of retry. Using a status code can
9334 be useful to quickly leave a server showing an abnormal behavior (out of
9335 memory, file system issues, etc), but in this case it may be a good idea to
9336 immediately redispatch the connection to another server (please see "option
9337 redispatch" for this). Last, it is important to understand that most causes
9338 of failures are the requests themselves and that retrying a request causing a
9339 server to misbehave will often make the situation even worse for this server,
9340 or for the whole service in case of redispatch.
9341
9342 Unless you know exactly how the application deals with replayed requests, you
9343 should not use this directive.
9344
9345 The default is "conn-failure".
9346
9347 See also: "retries", "option redispatch", "tune.bufsize"
9348
David du Colombier486df472011-03-17 10:40:26 +01009349server <name> <address>[:[port]] [param*]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009350 Declare a server in a backend
9351 May be used in sections : defaults | frontend | listen | backend
9352 no | no | yes | yes
9353 Arguments :
9354 <name> is the internal name assigned to this server. This name will
Davor Ocelice9ed2812017-12-25 17:49:28 +01009355 appear in logs and alerts. If "http-send-name-header" is
Mark Lamourinec2247f02012-01-04 13:02:01 -05009356 set, it will be added to the request header sent to the server.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009357
David du Colombier486df472011-03-17 10:40:26 +01009358 <address> is the IPv4 or IPv6 address of the server. Alternatively, a
9359 resolvable hostname is supported, but this name will be resolved
9360 during start-up. Address "0.0.0.0" or "*" has a special meaning.
9361 It indicates that the connection will be forwarded to the same IP
Willy Tarreaud669a4f2010-07-13 14:49:50 +02009362 address as the one from the client connection. This is useful in
9363 transparent proxy architectures where the client's connection is
9364 intercepted and haproxy must forward to the original destination
9365 address. This is more or less what the "transparent" keyword does
9366 except that with a server it's possible to limit concurrency and
Willy Tarreau24709282013-03-10 21:32:12 +01009367 to report statistics. Optionally, an address family prefix may be
9368 used before the address to force the family regardless of the
9369 address format, which can be useful to specify a path to a unix
9370 socket with no slash ('/'). Currently supported prefixes are :
9371 - 'ipv4@' -> address is always IPv4
9372 - 'ipv6@' -> address is always IPv6
9373 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02009374 - 'abns@' -> address is in abstract namespace (Linux only)
William Lallemand2fe7dd02018-09-11 16:51:29 +02009375 - 'sockpair@' -> address is the FD of a connected unix
9376 socket or of a socketpair. During a connection, the
9377 backend creates a pair of connected sockets, and passes
9378 one of them over the FD. The bind part will use the
9379 received socket as the client FD. Should be used
9380 carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02009381 You may want to reference some environment variables in the
9382 address parameter, see section 2.3 about environment
Willy Tarreau6a031d12016-11-07 19:42:35 +01009383 variables. The "init-addr" setting can be used to modify the way
9384 IP addresses should be resolved upon startup.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009385
Willy Tarreaub6205fd2012-09-24 12:27:33 +02009386 <port> is an optional port specification. If set, all connections will
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009387 be sent to this port. If unset, the same port the client
9388 connected to will be used. The port may also be prefixed by a "+"
9389 or a "-". In this case, the server's port will be determined by
9390 adding this value to the client's port.
9391
9392 <param*> is a list of parameters for this server. The "server" keywords
9393 accepts an important number of options and has a complete section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009394 dedicated to it. Please refer to section 5 for more details.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009395
9396 Examples :
9397 server first 10.1.1.1:1080 cookie first check inter 1000
9398 server second 10.1.1.2:1080 cookie second check inter 1000
Willy Tarreau24709282013-03-10 21:32:12 +01009399 server transp ipv4@
William Lallemandb2f07452015-05-12 14:27:13 +02009400 server backup "${SRV_BACKUP}:1080" backup
9401 server www1_dc1 "${LAN_DC1}.101:80"
9402 server www1_dc2 "${LAN_DC2}.101:80"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009403
Willy Tarreau55dcaf62015-09-27 15:03:15 +02009404 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
9405 sun_path length is used for the address length. Some other programs
9406 such as socat use the string length only by default. Pass the option
9407 ",unix-tightsocklen=0" to any abstract socket definition in socat to
9408 make it compatible with HAProxy's.
9409
Mark Lamourinec2247f02012-01-04 13:02:01 -05009410 See also: "default-server", "http-send-name-header" and section 5 about
9411 server options
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009412
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02009413server-state-file-name [<file>]
9414 Set the server state file to read, load and apply to servers available in
9415 this backend. It only applies when the directive "load-server-state-from-file"
9416 is set to "local". When <file> is not provided or if this directive is not
9417 set, then backend name is used. If <file> starts with a slash '/', then it is
9418 considered as an absolute path. Otherwise, <file> is concatenated to the
9419 global directive "server-state-file-base".
9420
9421 Example: the minimal configuration below would make HAProxy look for the
9422 state server file '/etc/haproxy/states/bk':
9423
9424 global
9425 server-state-file-base /etc/haproxy/states
9426
Willy Tarreauc9c6cdb2020-03-05 16:03:58 +01009427 backend bk
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02009428 load-server-state-from-file
9429
9430 See also: "server-state-file-base", "load-server-state-from-file", and
9431 "show servers state"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009432
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02009433server-template <prefix> <num | range> <fqdn>[:<port>] [params*]
9434 Set a template to initialize servers with shared parameters.
9435 The names of these servers are built from <prefix> and <num | range> parameters.
9436 May be used in sections : defaults | frontend | listen | backend
9437 no | no | yes | yes
9438
9439 Arguments:
9440 <prefix> A prefix for the server names to be built.
9441
9442 <num | range>
9443 If <num> is provided, this template initializes <num> servers
9444 with 1 up to <num> as server name suffixes. A range of numbers
9445 <num_low>-<num_high> may also be used to use <num_low> up to
9446 <num_high> as server name suffixes.
9447
9448 <fqdn> A FQDN for all the servers this template initializes.
9449
9450 <port> Same meaning as "server" <port> argument (see "server" keyword).
9451
9452 <params*>
9453 Remaining server parameters among all those supported by "server"
9454 keyword.
9455
9456 Examples:
9457 # Initializes 3 servers with srv1, srv2 and srv3 as names,
9458 # google.com as FQDN, and health-check enabled.
9459 server-template srv 1-3 google.com:80 check
9460
9461 # or
9462 server-template srv 3 google.com:80 check
9463
9464 # would be equivalent to:
9465 server srv1 google.com:80 check
9466 server srv2 google.com:80 check
9467 server srv3 google.com:80 check
9468
9469
9470
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009471source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +02009472source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreaud53f96b2009-02-04 18:46:54 +01009473source <addr>[:<port>] [interface <name>]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009474 Set the source address for outgoing connections
9475 May be used in sections : defaults | frontend | listen | backend
9476 yes | no | yes | yes
9477 Arguments :
9478 <addr> is the IPv4 address HAProxy will bind to before connecting to a
9479 server. This address is also used as a source for health checks.
Willy Tarreau24709282013-03-10 21:32:12 +01009480
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009481 The default value of 0.0.0.0 means that the system will select
Willy Tarreau24709282013-03-10 21:32:12 +01009482 the most appropriate address to reach its destination. Optionally
9483 an address family prefix may be used before the address to force
9484 the family regardless of the address format, which can be useful
9485 to specify a path to a unix socket with no slash ('/'). Currently
9486 supported prefixes are :
9487 - 'ipv4@' -> address is always IPv4
9488 - 'ipv6@' -> address is always IPv6
9489 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02009490 - 'abns@' -> address is in abstract namespace (Linux only)
Cyril Bonté307ee1e2015-09-28 23:16:06 +02009491 You may want to reference some environment variables in the
9492 address parameter, see section 2.3 about environment variables.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009493
9494 <port> is an optional port. It is normally not needed but may be useful
9495 in some very specific contexts. The default value of zero means
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02009496 the system will select a free port. Note that port ranges are not
9497 supported in the backend. If you want to force port ranges, you
9498 have to specify them on each "server" line.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009499
9500 <addr2> is the IP address to present to the server when connections are
9501 forwarded in full transparent proxy mode. This is currently only
9502 supported on some patched Linux kernels. When this address is
9503 specified, clients connecting to the server will be presented
9504 with this address, while health checks will still use the address
9505 <addr>.
9506
9507 <port2> is the optional port to present to the server when connections
9508 are forwarded in full transparent proxy mode (see <addr2> above).
9509 The default value of zero means the system will select a free
9510 port.
9511
Willy Tarreaubce70882009-09-07 11:51:47 +02009512 <hdr> is the name of a HTTP header in which to fetch the IP to bind to.
9513 This is the name of a comma-separated header list which can
9514 contain multiple IP addresses. By default, the last occurrence is
9515 used. This is designed to work with the X-Forwarded-For header
Baptiste Assmannea3e73b2013-02-02 23:47:49 +01009516 and to automatically bind to the client's IP address as seen
Willy Tarreaubce70882009-09-07 11:51:47 +02009517 by previous proxy, typically Stunnel. In order to use another
9518 occurrence from the last one, please see the <occ> parameter
9519 below. When the header (or occurrence) is not found, no binding
9520 is performed so that the proxy's default IP address is used. Also
9521 keep in mind that the header name is case insensitive, as for any
9522 HTTP header.
9523
9524 <occ> is the occurrence number of a value to be used in a multi-value
9525 header. This is to be used in conjunction with "hdr_ip(<hdr>)",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04009526 in order to specify which occurrence to use for the source IP
Willy Tarreaubce70882009-09-07 11:51:47 +02009527 address. Positive values indicate a position from the first
9528 occurrence, 1 being the first one. Negative values indicate
9529 positions relative to the last one, -1 being the last one. This
9530 is helpful for situations where an X-Forwarded-For header is set
9531 at the entry point of an infrastructure and must be used several
9532 proxy layers away. When this value is not specified, -1 is
9533 assumed. Passing a zero here disables the feature.
9534
Willy Tarreaud53f96b2009-02-04 18:46:54 +01009535 <name> is an optional interface name to which to bind to for outgoing
9536 traffic. On systems supporting this features (currently, only
9537 Linux), this allows one to bind all traffic to the server to
9538 this interface even if it is not the one the system would select
9539 based on routing tables. This should be used with extreme care.
9540 Note that using this option requires root privileges.
9541
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009542 The "source" keyword is useful in complex environments where a specific
9543 address only is allowed to connect to the servers. It may be needed when a
9544 private address must be used through a public gateway for instance, and it is
9545 known that the system cannot determine the adequate source address by itself.
9546
9547 An extension which is available on certain patched Linux kernels may be used
9548 through the "usesrc" optional keyword. It makes it possible to connect to the
9549 servers with an IP address which does not belong to the system itself. This
9550 is called "full transparent proxy mode". For this to work, the destination
9551 servers have to route their traffic back to this address through the machine
9552 running HAProxy, and IP forwarding must generally be enabled on this machine.
9553
9554 In this "full transparent proxy" mode, it is possible to force a specific IP
9555 address to be presented to the servers. This is not much used in fact. A more
9556 common use is to tell HAProxy to present the client's IP address. For this,
9557 there are two methods :
9558
9559 - present the client's IP and port addresses. This is the most transparent
9560 mode, but it can cause problems when IP connection tracking is enabled on
9561 the machine, because a same connection may be seen twice with different
9562 states. However, this solution presents the huge advantage of not
9563 limiting the system to the 64k outgoing address+port couples, because all
9564 of the client ranges may be used.
9565
9566 - present only the client's IP address and select a spare port. This
9567 solution is still quite elegant but slightly less transparent (downstream
9568 firewalls logs will not match upstream's). It also presents the downside
9569 of limiting the number of concurrent connections to the usual 64k ports.
9570 However, since the upstream and downstream ports are different, local IP
9571 connection tracking on the machine will not be upset by the reuse of the
9572 same session.
9573
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009574 This option sets the default source for all servers in the backend. It may
9575 also be specified in a "defaults" section. Finer source address specification
9576 is possible at the server level using the "source" server option. Refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009577 section 5 for more information.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009578
Baptiste Assmann91bd3372015-07-17 21:59:42 +02009579 In order to work, "usesrc" requires root privileges.
9580
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009581 Examples :
9582 backend private
9583 # Connect to the servers using our 192.168.1.200 source address
9584 source 192.168.1.200
9585
9586 backend transparent_ssl1
9587 # Connect to the SSL farm from the client's source address
9588 source 192.168.1.200 usesrc clientip
9589
9590 backend transparent_ssl2
9591 # Connect to the SSL farm from the client's source address and port
9592 # not recommended if IP conntrack is present on the local machine.
9593 source 192.168.1.200 usesrc client
9594
9595 backend transparent_ssl3
9596 # Connect to the SSL farm from the client's source address. It
9597 # is more conntrack-friendly.
9598 source 192.168.1.200 usesrc clientip
9599
9600 backend transparent_smtp
9601 # Connect to the SMTP farm from the client's source address/port
9602 # with Tproxy version 4.
9603 source 0.0.0.0 usesrc clientip
9604
Willy Tarreaubce70882009-09-07 11:51:47 +02009605 backend transparent_http
9606 # Connect to the servers using the client's IP as seen by previous
9607 # proxy.
9608 source 0.0.0.0 usesrc hdr_ip(x-forwarded-for,-1)
9609
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009610 See also : the "source" server option in section 5, the Tproxy patches for
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009611 the Linux kernel on www.balabit.com, the "bind" keyword.
9612
Willy Tarreau844e3c52008-01-11 16:28:18 +01009613
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09009614srvtcpka-cnt <count>
9615 Sets the maximum number of keepalive probes TCP should send before dropping
9616 the connection on the server side.
9617 May be used in sections : defaults | frontend | listen | backend
9618 yes | no | yes | yes
9619 Arguments :
9620 <count> is the maximum number of keepalive probes.
9621
9622 This keyword corresponds to the socket option TCP_KEEPCNT. If this keyword
9623 is not specified, system-wide TCP parameter (tcp_keepalive_probes) is used.
Willy Tarreau52543212020-07-09 05:58:51 +02009624 The availability of this setting depends on the operating system. It is
9625 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09009626
9627 See also : "option srvtcpka", "srvtcpka-idle", "srvtcpka-intvl".
9628
9629
9630srvtcpka-idle <timeout>
9631 Sets the time the connection needs to remain idle before TCP starts sending
9632 keepalive probes, if enabled the sending of TCP keepalive packets on the
9633 server side.
9634 May be used in sections : defaults | frontend | listen | backend
9635 yes | no | yes | yes
9636 Arguments :
9637 <timeout> is the time the connection needs to remain idle before TCP starts
9638 sending keepalive probes. It is specified in seconds by default,
9639 but can be in any other unit if the number is suffixed by the
9640 unit, as explained at the top of this document.
9641
9642 This keyword corresponds to the socket option TCP_KEEPIDLE. If this keyword
9643 is not specified, system-wide TCP parameter (tcp_keepalive_time) is used.
Willy Tarreau52543212020-07-09 05:58:51 +02009644 The availability of this setting depends on the operating system. It is
9645 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09009646
9647 See also : "option srvtcpka", "srvtcpka-cnt", "srvtcpka-intvl".
9648
9649
9650srvtcpka-intvl <timeout>
9651 Sets the time between individual keepalive probes on the server side.
9652 May be used in sections : defaults | frontend | listen | backend
9653 yes | no | yes | yes
9654 Arguments :
9655 <timeout> is the time between individual keepalive probes. It is specified
9656 in seconds by default, but can be in any other unit if the number
9657 is suffixed by the unit, as explained at the top of this
9658 document.
9659
9660 This keyword corresponds to the socket option TCP_KEEPINTVL. If this keyword
9661 is not specified, system-wide TCP parameter (tcp_keepalive_intvl) is used.
Willy Tarreau52543212020-07-09 05:58:51 +02009662 The availability of this setting depends on the operating system. It is
9663 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09009664
9665 See also : "option srvtcpka", "srvtcpka-cnt", "srvtcpka-idle".
9666
9667
Cyril Bonté66c327d2010-10-12 00:14:37 +02009668stats admin { if | unless } <cond>
9669 Enable statistics admin level if/unless a condition is matched
9670 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02009671 no | yes | yes | yes
Cyril Bonté66c327d2010-10-12 00:14:37 +02009672
9673 This statement enables the statistics admin level if/unless a condition is
9674 matched.
9675
9676 The admin level allows to enable/disable servers from the web interface. By
9677 default, statistics page is read-only for security reasons.
9678
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009679 Note : Consider not using this feature in multi-process mode (nbproc > 1)
9680 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01009681 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009682
Cyril Bonté23b39d92011-02-10 22:54:44 +01009683 Currently, the POST request is limited to the buffer size minus the reserved
9684 buffer space, which means that if the list of servers is too long, the
9685 request won't be processed. It is recommended to alter few servers at a
9686 time.
Cyril Bonté66c327d2010-10-12 00:14:37 +02009687
9688 Example :
9689 # statistics admin level only for localhost
9690 backend stats_localhost
9691 stats enable
9692 stats admin if LOCALHOST
9693
9694 Example :
9695 # statistics admin level always enabled because of the authentication
9696 backend stats_auth
9697 stats enable
9698 stats auth admin:AdMiN123
9699 stats admin if TRUE
9700
9701 Example :
9702 # statistics admin level depends on the authenticated user
9703 userlist stats-auth
9704 group admin users admin
9705 user admin insecure-password AdMiN123
9706 group readonly users haproxy
9707 user haproxy insecure-password haproxy
9708
9709 backend stats_auth
9710 stats enable
9711 acl AUTH http_auth(stats-auth)
9712 acl AUTH_ADMIN http_auth_group(stats-auth) admin
9713 stats http-request auth unless AUTH
9714 stats admin if AUTH_ADMIN
9715
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009716 See also : "stats enable", "stats auth", "stats http-request", "nbproc",
9717 "bind-process", section 3.4 about userlists and section 7 about
9718 ACL usage.
Cyril Bonté66c327d2010-10-12 00:14:37 +02009719
9720
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009721stats auth <user>:<passwd>
9722 Enable statistics with authentication and grant access to an account
9723 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02009724 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009725 Arguments :
9726 <user> is a user name to grant access to
9727
9728 <passwd> is the cleartext password associated to this user
9729
9730 This statement enables statistics with default settings, and restricts access
9731 to declared users only. It may be repeated as many times as necessary to
9732 allow as many users as desired. When a user tries to access the statistics
9733 without a valid account, a "401 Forbidden" response will be returned so that
9734 the browser asks the user to provide a valid user and password. The real
9735 which will be returned to the browser is configurable using "stats realm".
9736
9737 Since the authentication method is HTTP Basic Authentication, the passwords
9738 circulate in cleartext on the network. Thus, it was decided that the
9739 configuration file would also use cleartext passwords to remind the users
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02009740 that those ones should not be sensitive and not shared with any other account.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009741
9742 It is also possible to reduce the scope of the proxies which appear in the
9743 report using "stats scope".
9744
9745 Though this statement alone is enough to enable statistics reporting, it is
9746 recommended to set all other settings in order to avoid relying on default
9747 unobvious parameters.
9748
9749 Example :
9750 # public access (limited to this backend only)
9751 backend public_www
9752 server srv1 192.168.0.1:80
9753 stats enable
9754 stats hide-version
9755 stats scope .
9756 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01009757 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009758 stats auth admin1:AdMiN123
9759 stats auth admin2:AdMiN321
9760
9761 # internal monitoring access (unlimited)
9762 backend private_monitoring
9763 stats enable
9764 stats uri /admin?stats
9765 stats refresh 5s
9766
9767 See also : "stats enable", "stats realm", "stats scope", "stats uri"
9768
9769
9770stats enable
9771 Enable statistics reporting with default settings
9772 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02009773 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009774 Arguments : none
9775
9776 This statement enables statistics reporting with default settings defined
9777 at build time. Unless stated otherwise, these settings are used :
9778 - stats uri : /haproxy?stats
9779 - stats realm : "HAProxy Statistics"
9780 - stats auth : no authentication
9781 - stats scope : no restriction
9782
9783 Though this statement alone is enough to enable statistics reporting, it is
9784 recommended to set all other settings in order to avoid relying on default
9785 unobvious parameters.
9786
9787 Example :
9788 # public access (limited to this backend only)
9789 backend public_www
9790 server srv1 192.168.0.1:80
9791 stats enable
9792 stats hide-version
9793 stats scope .
9794 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01009795 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009796 stats auth admin1:AdMiN123
9797 stats auth admin2:AdMiN321
9798
9799 # internal monitoring access (unlimited)
9800 backend private_monitoring
9801 stats enable
9802 stats uri /admin?stats
9803 stats refresh 5s
9804
9805 See also : "stats auth", "stats realm", "stats uri"
9806
9807
Willy Tarreaud63335a2010-02-26 12:56:52 +01009808stats hide-version
9809 Enable statistics and hide HAProxy version reporting
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02009810 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02009811 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01009812 Arguments : none
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02009813
Willy Tarreaud63335a2010-02-26 12:56:52 +01009814 By default, the stats page reports some useful status information along with
9815 the statistics. Among them is HAProxy's version. However, it is generally
9816 considered dangerous to report precise version to anyone, as it can help them
9817 target known weaknesses with specific attacks. The "stats hide-version"
9818 statement removes the version from the statistics report. This is recommended
9819 for public sites or any site with a weak login/password.
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02009820
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02009821 Though this statement alone is enough to enable statistics reporting, it is
9822 recommended to set all other settings in order to avoid relying on default
9823 unobvious parameters.
9824
Willy Tarreaud63335a2010-02-26 12:56:52 +01009825 Example :
9826 # public access (limited to this backend only)
9827 backend public_www
9828 server srv1 192.168.0.1:80
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02009829 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01009830 stats hide-version
9831 stats scope .
9832 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01009833 stats realm HAProxy\ Statistics
Willy Tarreaud63335a2010-02-26 12:56:52 +01009834 stats auth admin1:AdMiN123
9835 stats auth admin2:AdMiN321
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02009836
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02009837 # internal monitoring access (unlimited)
9838 backend private_monitoring
9839 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01009840 stats uri /admin?stats
9841 stats refresh 5s
Krzysztof Piotr Oledzki15514c22010-01-04 16:03:09 +01009842
Willy Tarreaud63335a2010-02-26 12:56:52 +01009843 See also : "stats auth", "stats enable", "stats realm", "stats uri"
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02009844
Willy Tarreau983e01e2010-01-11 18:42:06 +01009845
Cyril Bonté2be1b3f2010-09-30 23:46:30 +02009846stats http-request { allow | deny | auth [realm <realm>] }
9847 [ { if | unless } <condition> ]
9848 Access control for statistics
9849
9850 May be used in sections: defaults | frontend | listen | backend
9851 no | no | yes | yes
9852
9853 As "http-request", these set of options allow to fine control access to
9854 statistics. Each option may be followed by if/unless and acl.
9855 First option with matched condition (or option without condition) is final.
9856 For "deny" a 403 error will be returned, for "allow" normal processing is
9857 performed, for "auth" a 401/407 error code is returned so the client
9858 should be asked to enter a username and password.
9859
9860 There is no fixed limit to the number of http-request statements per
9861 instance.
9862
9863 See also : "http-request", section 3.4 about userlists and section 7
9864 about ACL usage.
9865
9866
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009867stats realm <realm>
9868 Enable statistics and set authentication realm
9869 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02009870 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009871 Arguments :
9872 <realm> is the name of the HTTP Basic Authentication realm reported to
9873 the browser. The browser uses it to display it in the pop-up
9874 inviting the user to enter a valid username and password.
9875
9876 The realm is read as a single word, so any spaces in it should be escaped
9877 using a backslash ('\').
9878
9879 This statement is useful only in conjunction with "stats auth" since it is
9880 only related to authentication.
9881
9882 Though this statement alone is enough to enable statistics reporting, it is
9883 recommended to set all other settings in order to avoid relying on default
9884 unobvious parameters.
9885
9886 Example :
9887 # public access (limited to this backend only)
9888 backend public_www
9889 server srv1 192.168.0.1:80
9890 stats enable
9891 stats hide-version
9892 stats scope .
9893 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01009894 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009895 stats auth admin1:AdMiN123
9896 stats auth admin2:AdMiN321
9897
9898 # internal monitoring access (unlimited)
9899 backend private_monitoring
9900 stats enable
9901 stats uri /admin?stats
9902 stats refresh 5s
9903
9904 See also : "stats auth", "stats enable", "stats uri"
9905
9906
9907stats refresh <delay>
9908 Enable statistics with automatic refresh
9909 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02009910 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009911 Arguments :
9912 <delay> is the suggested refresh delay, specified in seconds, which will
9913 be returned to the browser consulting the report page. While the
9914 browser is free to apply any delay, it will generally respect it
9915 and refresh the page this every seconds. The refresh interval may
9916 be specified in any other non-default time unit, by suffixing the
9917 unit after the value, as explained at the top of this document.
9918
9919 This statement is useful on monitoring displays with a permanent page
9920 reporting the load balancer's activity. When set, the HTML report page will
9921 include a link "refresh"/"stop refresh" so that the user can select whether
Jackie Tapia749f74c2020-07-22 18:59:40 -05009922 they want automatic refresh of the page or not.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009923
9924 Though this statement alone is enough to enable statistics reporting, it is
9925 recommended to set all other settings in order to avoid relying on default
9926 unobvious parameters.
9927
9928 Example :
9929 # public access (limited to this backend only)
9930 backend public_www
9931 server srv1 192.168.0.1:80
9932 stats enable
9933 stats hide-version
9934 stats scope .
9935 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01009936 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009937 stats auth admin1:AdMiN123
9938 stats auth admin2:AdMiN321
9939
9940 # internal monitoring access (unlimited)
9941 backend private_monitoring
9942 stats enable
9943 stats uri /admin?stats
9944 stats refresh 5s
9945
9946 See also : "stats auth", "stats enable", "stats realm", "stats uri"
9947
9948
9949stats scope { <name> | "." }
9950 Enable statistics and limit access scope
9951 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02009952 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009953 Arguments :
9954 <name> is the name of a listen, frontend or backend section to be
9955 reported. The special name "." (a single dot) designates the
9956 section in which the statement appears.
9957
9958 When this statement is specified, only the sections enumerated with this
9959 statement will appear in the report. All other ones will be hidden. This
9960 statement may appear as many times as needed if multiple sections need to be
9961 reported. Please note that the name checking is performed as simple string
9962 comparisons, and that it is never checked that a give section name really
9963 exists.
9964
9965 Though this statement alone is enough to enable statistics reporting, it is
9966 recommended to set all other settings in order to avoid relying on default
9967 unobvious parameters.
9968
9969 Example :
9970 # public access (limited to this backend only)
9971 backend public_www
9972 server srv1 192.168.0.1:80
9973 stats enable
9974 stats hide-version
9975 stats scope .
9976 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01009977 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009978 stats auth admin1:AdMiN123
9979 stats auth admin2:AdMiN321
9980
9981 # internal monitoring access (unlimited)
9982 backend private_monitoring
9983 stats enable
9984 stats uri /admin?stats
9985 stats refresh 5s
9986
9987 See also : "stats auth", "stats enable", "stats realm", "stats uri"
9988
Willy Tarreaud63335a2010-02-26 12:56:52 +01009989
Willy Tarreauc9705a12010-07-27 20:05:50 +02009990stats show-desc [ <desc> ]
Willy Tarreaud63335a2010-02-26 12:56:52 +01009991 Enable reporting of a description on the statistics page.
9992 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02009993 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01009994
Willy Tarreauc9705a12010-07-27 20:05:50 +02009995 <desc> is an optional description to be reported. If unspecified, the
Willy Tarreaud63335a2010-02-26 12:56:52 +01009996 description from global section is automatically used instead.
9997
9998 This statement is useful for users that offer shared services to their
9999 customers, where node or description should be different for each customer.
10000
10001 Though this statement alone is enough to enable statistics reporting, it is
10002 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +010010003 unobvious parameters. By default description is not shown.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010004
10005 Example :
10006 # internal monitoring access (unlimited)
10007 backend private_monitoring
10008 stats enable
10009 stats show-desc Master node for Europe, Asia, Africa
10010 stats uri /admin?stats
10011 stats refresh 5s
10012
10013 See also: "show-node", "stats enable", "stats uri" and "description" in
10014 global section.
10015
10016
10017stats show-legends
Willy Tarreaued2119c2014-04-24 22:10:39 +020010018 Enable reporting additional information on the statistics page
10019 May be used in sections : defaults | frontend | listen | backend
10020 yes | yes | yes | yes
10021 Arguments : none
10022
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010023 Enable reporting additional information on the statistics page :
Willy Tarreaud63335a2010-02-26 12:56:52 +010010024 - cap: capabilities (proxy)
10025 - mode: one of tcp, http or health (proxy)
10026 - id: SNMP ID (proxy, socket, server)
10027 - IP (socket, server)
10028 - cookie (backend, server)
10029
10030 Though this statement alone is enough to enable statistics reporting, it is
10031 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +010010032 unobvious parameters. Default behavior is not to show this information.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010033
10034 See also: "stats enable", "stats uri".
10035
10036
Amaury Denoyelle0b70a8a2020-10-05 11:49:45 +020010037stats show-modules
10038 Enable display of extra statistics module on the statistics page
10039 May be used in sections : defaults | frontend | listen | backend
10040 yes | yes | yes | yes
10041 Arguments : none
10042
10043 New columns are added at the end of the line containing the extra statistics
10044 values as a tooltip.
10045
10046 Though this statement alone is enough to enable statistics reporting, it is
10047 recommended to set all other settings in order to avoid relying on default
10048 unobvious parameters. Default behavior is not to show this information.
10049
10050 See also: "stats enable", "stats uri".
10051
10052
Willy Tarreaud63335a2010-02-26 12:56:52 +010010053stats show-node [ <name> ]
10054 Enable reporting of a host name on the statistics page.
10055 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010056 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +010010057 Arguments:
10058 <name> is an optional name to be reported. If unspecified, the
10059 node name from global section is automatically used instead.
10060
10061 This statement is useful for users that offer shared services to their
10062 customers, where node or description might be different on a stats page
Davor Ocelice9ed2812017-12-25 17:49:28 +010010063 provided for each customer. Default behavior is not to show host name.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010064
10065 Though this statement alone is enough to enable statistics reporting, it is
10066 recommended to set all other settings in order to avoid relying on default
10067 unobvious parameters.
10068
10069 Example:
10070 # internal monitoring access (unlimited)
10071 backend private_monitoring
10072 stats enable
10073 stats show-node Europe-1
10074 stats uri /admin?stats
10075 stats refresh 5s
10076
10077 See also: "show-desc", "stats enable", "stats uri", and "node" in global
10078 section.
10079
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010080
10081stats uri <prefix>
10082 Enable statistics and define the URI prefix to access them
10083 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010084 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010085 Arguments :
10086 <prefix> is the prefix of any URI which will be redirected to stats. This
10087 prefix may contain a question mark ('?') to indicate part of a
10088 query string.
10089
10090 The statistics URI is intercepted on the relayed traffic, so it appears as a
10091 page within the normal application. It is strongly advised to ensure that the
10092 selected URI will never appear in the application, otherwise it will never be
10093 possible to reach it in the application.
10094
10095 The default URI compiled in haproxy is "/haproxy?stats", but this may be
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010010096 changed at build time, so it's better to always explicitly specify it here.
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010097 It is generally a good idea to include a question mark in the URI so that
10098 intermediate proxies refrain from caching the results. Also, since any string
10099 beginning with the prefix will be accepted as a stats request, the question
10100 mark helps ensuring that no valid URI will begin with the same words.
10101
10102 It is sometimes very convenient to use "/" as the URI prefix, and put that
10103 statement in a "listen" instance of its own. That makes it easy to dedicate
10104 an address or a port to statistics only.
10105
10106 Though this statement alone is enough to enable statistics reporting, it is
10107 recommended to set all other settings in order to avoid relying on default
10108 unobvious parameters.
10109
10110 Example :
10111 # public access (limited to this backend only)
10112 backend public_www
10113 server srv1 192.168.0.1:80
10114 stats enable
10115 stats hide-version
10116 stats scope .
10117 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010118 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010119 stats auth admin1:AdMiN123
10120 stats auth admin2:AdMiN321
10121
10122 # internal monitoring access (unlimited)
10123 backend private_monitoring
10124 stats enable
10125 stats uri /admin?stats
10126 stats refresh 5s
10127
10128 See also : "stats auth", "stats enable", "stats realm"
10129
10130
Willy Tarreaud63335a2010-02-26 12:56:52 +010010131stick match <pattern> [table <table>] [{if | unless} <cond>]
10132 Define a request pattern matching condition to stick a user to a server
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010133 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaud63335a2010-02-26 12:56:52 +010010134 no | no | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010135
10136 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +020010137 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010138 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +010010139 will be analyzed in the hope to find a matching entry in a
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010140 stickiness table. This rule is mandatory.
10141
10142 <table> is an optional stickiness table name. If unspecified, the same
10143 backend's table is used. A stickiness table is declared using
10144 the "stick-table" statement.
10145
10146 <cond> is an optional matching condition. It makes it possible to match
10147 on a certain criterion only when other conditions are met (or
10148 not met). For instance, it could be used to match on a source IP
10149 address except when a request passes through a known proxy, in
10150 which case we'd match on a header containing that IP address.
10151
10152 Some protocols or applications require complex stickiness rules and cannot
10153 always simply rely on cookies nor hashing. The "stick match" statement
10154 describes a rule to extract the stickiness criterion from an incoming request
10155 or connection. See section 7 for a complete list of possible patterns and
10156 transformation rules.
10157
10158 The table has to be declared using the "stick-table" statement. It must be of
10159 a type compatible with the pattern. By default it is the one which is present
10160 in the same backend. It is possible to share a table with other backends by
10161 referencing it using the "table" keyword. If another table is referenced,
10162 the server's ID inside the backends are used. By default, all server IDs
10163 start at 1 in each backend, so the server ordering is enough. But in case of
10164 doubt, it is highly recommended to force server IDs using their "id" setting.
10165
10166 It is possible to restrict the conditions where a "stick match" statement
10167 will apply, using "if" or "unless" followed by a condition. See section 7 for
10168 ACL based conditions.
10169
10170 There is no limit on the number of "stick match" statements. The first that
10171 applies and matches will cause the request to be directed to the same server
10172 as was used for the request which created the entry. That way, multiple
10173 matches can be used as fallbacks.
10174
10175 The stick rules are checked after the persistence cookies, so they will not
10176 affect stickiness if a cookie has already been used to select a server. That
10177 way, it becomes very easy to insert cookies and match on IP addresses in
10178 order to maintain stickiness between HTTP and HTTPS.
10179
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010180 Note : Consider not using this feature in multi-process mode (nbproc > 1)
10181 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +010010182 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010183
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010184 Example :
10185 # forward SMTP users to the same server they just used for POP in the
10186 # last 30 minutes
10187 backend pop
10188 mode tcp
10189 balance roundrobin
10190 stick store-request src
10191 stick-table type ip size 200k expire 30m
10192 server s1 192.168.1.1:110
10193 server s2 192.168.1.1:110
10194
10195 backend smtp
10196 mode tcp
10197 balance roundrobin
10198 stick match src table pop
10199 server s1 192.168.1.1:25
10200 server s2 192.168.1.1:25
10201
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010202 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +020010203 about ACLs and samples fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010204
10205
10206stick on <pattern> [table <table>] [{if | unless} <condition>]
10207 Define a request pattern to associate a user to a server
10208 May be used in sections : defaults | frontend | listen | backend
10209 no | no | yes | yes
10210
10211 Note : This form is exactly equivalent to "stick match" followed by
10212 "stick store-request", all with the same arguments. Please refer
10213 to both keywords for details. It is only provided as a convenience
10214 for writing more maintainable configurations.
10215
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010216 Note : Consider not using this feature in multi-process mode (nbproc > 1)
10217 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +010010218 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010219
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010220 Examples :
10221 # The following form ...
Willy Tarreauec579d82010-02-26 19:15:04 +010010222 stick on src table pop if !localhost
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010223
10224 # ...is strictly equivalent to this one :
10225 stick match src table pop if !localhost
10226 stick store-request src table pop if !localhost
10227
10228
10229 # Use cookie persistence for HTTP, and stick on source address for HTTPS as
10230 # well as HTTP without cookie. Share the same table between both accesses.
10231 backend http
10232 mode http
10233 balance roundrobin
10234 stick on src table https
10235 cookie SRV insert indirect nocache
10236 server s1 192.168.1.1:80 cookie s1
10237 server s2 192.168.1.1:80 cookie s2
10238
10239 backend https
10240 mode tcp
10241 balance roundrobin
10242 stick-table type ip size 200k expire 30m
10243 stick on src
10244 server s1 192.168.1.1:443
10245 server s2 192.168.1.1:443
10246
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010247 See also : "stick match", "stick store-request", "nbproc" and "bind-process".
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010248
10249
10250stick store-request <pattern> [table <table>] [{if | unless} <condition>]
10251 Define a request pattern used to create an entry in a stickiness table
10252 May be used in sections : defaults | frontend | listen | backend
10253 no | no | yes | yes
10254
10255 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +020010256 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010257 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +010010258 will be analyzed, extracted and stored in the table once a
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010259 server is selected.
10260
10261 <table> is an optional stickiness table name. If unspecified, the same
10262 backend's table is used. A stickiness table is declared using
10263 the "stick-table" statement.
10264
10265 <cond> is an optional storage condition. It makes it possible to store
10266 certain criteria only when some conditions are met (or not met).
10267 For instance, it could be used to store the source IP address
10268 except when the request passes through a known proxy, in which
10269 case we'd store a converted form of a header containing that IP
10270 address.
10271
10272 Some protocols or applications require complex stickiness rules and cannot
10273 always simply rely on cookies nor hashing. The "stick store-request" statement
10274 describes a rule to decide what to extract from the request and when to do
10275 it, in order to store it into a stickiness table for further requests to
10276 match it using the "stick match" statement. Obviously the extracted part must
10277 make sense and have a chance to be matched in a further request. Storing a
10278 client's IP address for instance often makes sense. Storing an ID found in a
10279 URL parameter also makes sense. Storing a source port will almost never make
10280 any sense because it will be randomly matched. See section 7 for a complete
10281 list of possible patterns and transformation rules.
10282
10283 The table has to be declared using the "stick-table" statement. It must be of
10284 a type compatible with the pattern. By default it is the one which is present
10285 in the same backend. It is possible to share a table with other backends by
10286 referencing it using the "table" keyword. If another table is referenced,
10287 the server's ID inside the backends are used. By default, all server IDs
10288 start at 1 in each backend, so the server ordering is enough. But in case of
10289 doubt, it is highly recommended to force server IDs using their "id" setting.
10290
10291 It is possible to restrict the conditions where a "stick store-request"
10292 statement will apply, using "if" or "unless" followed by a condition. This
10293 condition will be evaluated while parsing the request, so any criteria can be
10294 used. See section 7 for ACL based conditions.
10295
10296 There is no limit on the number of "stick store-request" statements, but
10297 there is a limit of 8 simultaneous stores per request or response. This
10298 makes it possible to store up to 8 criteria, all extracted from either the
10299 request or the response, regardless of the number of rules. Only the 8 first
10300 ones which match will be kept. Using this, it is possible to feed multiple
10301 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +010010302 another protocol or access method. Using multiple store-request rules with
10303 the same table is possible and may be used to find the best criterion to rely
10304 on, by arranging the rules by decreasing preference order. Only the first
10305 extracted criterion for a given table will be stored. All subsequent store-
10306 request rules referencing the same table will be skipped and their ACLs will
10307 not be evaluated.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010308
10309 The "store-request" rules are evaluated once the server connection has been
10310 established, so that the table will contain the real server that processed
10311 the request.
10312
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010313 Note : Consider not using this feature in multi-process mode (nbproc > 1)
10314 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +010010315 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010316
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010317 Example :
10318 # forward SMTP users to the same server they just used for POP in the
10319 # last 30 minutes
10320 backend pop
10321 mode tcp
10322 balance roundrobin
10323 stick store-request src
10324 stick-table type ip size 200k expire 30m
10325 server s1 192.168.1.1:110
10326 server s2 192.168.1.1:110
10327
10328 backend smtp
10329 mode tcp
10330 balance roundrobin
10331 stick match src table pop
10332 server s1 192.168.1.1:25
10333 server s2 192.168.1.1:25
10334
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010335 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +020010336 about ACLs and sample fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010337
10338
Emeric Brun7c6b82e2010-09-24 16:34:28 +020010339stick-table type {ip | integer | string [len <length>] | binary [len <length>]}
Emeric Brunf099e792010-09-27 12:05:28 +020010340 size <size> [expire <expire>] [nopurge] [peers <peersect>]
10341 [store <data_type>]*
Godbach64cef792013-12-04 16:08:22 +080010342 Configure the stickiness table for the current section
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010343 May be used in sections : defaults | frontend | listen | backend
Willy Tarreauc00cdc22010-06-06 16:48:26 +020010344 no | yes | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010345
10346 Arguments :
10347 ip a table declared with "type ip" will only store IPv4 addresses.
10348 This form is very compact (about 50 bytes per entry) and allows
10349 very fast entry lookup and stores with almost no overhead. This
10350 is mainly used to store client source IP addresses.
10351
David du Colombier9a6d3c92011-03-17 10:40:24 +010010352 ipv6 a table declared with "type ipv6" will only store IPv6 addresses.
10353 This form is very compact (about 60 bytes per entry) and allows
10354 very fast entry lookup and stores with almost no overhead. This
10355 is mainly used to store client source IP addresses.
10356
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010357 integer a table declared with "type integer" will store 32bit integers
10358 which can represent a client identifier found in a request for
10359 instance.
10360
10361 string a table declared with "type string" will store substrings of up
10362 to <len> characters. If the string provided by the pattern
10363 extractor is larger than <len>, it will be truncated before
10364 being stored. During matching, at most <len> characters will be
10365 compared between the string in the table and the extracted
10366 pattern. When not specified, the string is automatically limited
Emeric Brun7c6b82e2010-09-24 16:34:28 +020010367 to 32 characters.
10368
10369 binary a table declared with "type binary" will store binary blocks
10370 of <len> bytes. If the block provided by the pattern
10371 extractor is larger than <len>, it will be truncated before
Willy Tarreaube722a22014-06-13 16:31:59 +020010372 being stored. If the block provided by the sample expression
Emeric Brun7c6b82e2010-09-24 16:34:28 +020010373 is shorter than <len>, it will be padded by 0. When not
10374 specified, the block is automatically limited to 32 bytes.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010375
10376 <length> is the maximum number of characters that will be stored in a
Emeric Brun7c6b82e2010-09-24 16:34:28 +020010377 "string" type table (See type "string" above). Or the number
10378 of bytes of the block in "binary" type table. Be careful when
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010379 changing this parameter as memory usage will proportionally
10380 increase.
10381
10382 <size> is the maximum number of entries that can fit in the table. This
Cyril Bonté78caf842010-03-10 22:41:43 +010010383 value directly impacts memory usage. Count approximately
10384 50 bytes per entry, plus the size of a string if any. The size
10385 supports suffixes "k", "m", "g" for 2^10, 2^20 and 2^30 factors.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010386
10387 [nopurge] indicates that we refuse to purge older entries when the table
10388 is full. When not specified and the table is full when haproxy
10389 wants to store an entry in it, it will flush a few of the oldest
10390 entries in order to release some space for the new ones. This is
Davor Ocelice9ed2812017-12-25 17:49:28 +010010391 most often the desired behavior. In some specific cases, it
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010392 be desirable to refuse new entries instead of purging the older
10393 ones. That may be the case when the amount of data to store is
10394 far above the hardware limits and we prefer not to offer access
10395 to new clients than to reject the ones already connected. When
10396 using this parameter, be sure to properly set the "expire"
10397 parameter (see below).
10398
Emeric Brunf099e792010-09-27 12:05:28 +020010399 <peersect> is the name of the peers section to use for replication. Entries
10400 which associate keys to server IDs are kept synchronized with
10401 the remote peers declared in this section. All entries are also
10402 automatically learned from the local peer (old process) during a
10403 soft restart.
10404
Willy Tarreau1abc6732015-05-01 19:21:02 +020010405 NOTE : each peers section may be referenced only by tables
10406 belonging to the same unique process.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010407
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010408 <expire> defines the maximum duration of an entry in the table since it
10409 was last created, refreshed or matched. The expiration delay is
10410 defined using the standard time format, similarly as the various
10411 timeouts. The maximum duration is slightly above 24 days. See
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +030010412 section 2.4 for more information. If this delay is not specified,
Cyril Bontédc4d9032012-04-08 21:57:39 +020010413 the session won't automatically expire, but older entries will
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010414 be removed once full. Be sure not to use the "nopurge" parameter
10415 if not expiration delay is specified.
10416
Willy Tarreau08d5f982010-06-06 13:34:54 +020010417 <data_type> is used to store additional information in the stick-table. This
10418 may be used by ACLs in order to control various criteria related
10419 to the activity of the client matching the stick-table. For each
10420 item specified here, the size of each entry will be inflated so
Willy Tarreauc9705a12010-07-27 20:05:50 +020010421 that the additional data can fit. Several data types may be
10422 stored with an entry. Multiple data types may be specified after
10423 the "store" keyword, as a comma-separated list. Alternatively,
10424 it is possible to repeat the "store" keyword followed by one or
10425 several data types. Except for the "server_id" type which is
10426 automatically detected and enabled, all data types must be
10427 explicitly declared to be stored. If an ACL references a data
10428 type which is not stored, the ACL will simply not match. Some
10429 data types require an argument which must be passed just after
10430 the type between parenthesis. See below for the supported data
10431 types and their arguments.
10432
10433 The data types that can be stored with an entry are the following :
10434 - server_id : this is an integer which holds the numeric ID of the server a
10435 request was assigned to. It is used by the "stick match", "stick store",
10436 and "stick on" rules. It is automatically enabled when referenced.
10437
10438 - gpc0 : first General Purpose Counter. It is a positive 32-bit integer
10439 integer which may be used for anything. Most of the time it will be used
10440 to put a special tag on some entries, for instance to note that a
Davor Ocelice9ed2812017-12-25 17:49:28 +010010441 specific behavior was detected and must be known for future matches.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010442
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010443 - gpc0_rate(<period>) : increment rate of the first General Purpose Counter
10444 over a period. It is a positive 32-bit integer integer which may be used
10445 for anything. Just like <gpc0>, it counts events, but instead of keeping
Davor Ocelice9ed2812017-12-25 17:49:28 +010010446 a cumulative number, it maintains the rate at which the counter is
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010447 incremented. Most of the time it will be used to measure the frequency of
Davor Ocelice9ed2812017-12-25 17:49:28 +010010448 occurrence of certain events (e.g. requests to a specific URL).
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010449
Frédéric Lécaille6778b272018-01-29 15:22:53 +010010450 - gpc1 : second General Purpose Counter. It is a positive 32-bit integer
10451 integer which may be used for anything. Most of the time it will be used
10452 to put a special tag on some entries, for instance to note that a
10453 specific behavior was detected and must be known for future matches.
10454
10455 - gpc1_rate(<period>) : increment rate of the second General Purpose Counter
10456 over a period. It is a positive 32-bit integer integer which may be used
10457 for anything. Just like <gpc1>, it counts events, but instead of keeping
10458 a cumulative number, it maintains the rate at which the counter is
10459 incremented. Most of the time it will be used to measure the frequency of
10460 occurrence of certain events (e.g. requests to a specific URL).
10461
Willy Tarreauc9705a12010-07-27 20:05:50 +020010462 - conn_cnt : Connection Count. It is a positive 32-bit integer which counts
10463 the absolute number of connections received from clients which matched
10464 this entry. It does not mean the connections were accepted, just that
10465 they were received.
10466
10467 - conn_cur : Current Connections. It is a positive 32-bit integer which
10468 stores the concurrent connection counts for the entry. It is incremented
10469 once an incoming connection matches the entry, and decremented once the
10470 connection leaves. That way it is possible to know at any time the exact
10471 number of concurrent connections for an entry.
10472
10473 - conn_rate(<period>) : frequency counter (takes 12 bytes). It takes an
10474 integer parameter <period> which indicates in milliseconds the length
10475 of the period over which the average is measured. It reports the average
10476 incoming connection rate over that period, in connections per period. The
10477 result is an integer which can be matched using ACLs.
10478
10479 - sess_cnt : Session Count. It is a positive 32-bit integer which counts
10480 the absolute number of sessions received from clients which matched this
10481 entry. A session is a connection that was accepted by the layer 4 rules.
10482
10483 - sess_rate(<period>) : frequency counter (takes 12 bytes). It takes an
10484 integer parameter <period> which indicates in milliseconds the length
10485 of the period over which the average is measured. It reports the average
10486 incoming session rate over that period, in sessions per period. The
10487 result is an integer which can be matched using ACLs.
10488
10489 - http_req_cnt : HTTP request Count. It is a positive 32-bit integer which
10490 counts the absolute number of HTTP requests received from clients which
10491 matched this entry. It does not matter whether they are valid requests or
10492 not. Note that this is different from sessions when keep-alive is used on
10493 the client side.
10494
10495 - http_req_rate(<period>) : frequency counter (takes 12 bytes). It takes an
10496 integer parameter <period> which indicates in milliseconds the length
10497 of the period over which the average is measured. It reports the average
10498 HTTP request rate over that period, in requests per period. The result is
10499 an integer which can be matched using ACLs. It does not matter whether
10500 they are valid requests or not. Note that this is different from sessions
10501 when keep-alive is used on the client side.
10502
10503 - http_err_cnt : HTTP Error Count. It is a positive 32-bit integer which
10504 counts the absolute number of HTTP requests errors induced by clients
10505 which matched this entry. Errors are counted on invalid and truncated
10506 requests, as well as on denied or tarpitted requests, and on failed
10507 authentications. If the server responds with 4xx, then the request is
10508 also counted as an error since it's an error triggered by the client
Davor Ocelice9ed2812017-12-25 17:49:28 +010010509 (e.g. vulnerability scan).
Willy Tarreauc9705a12010-07-27 20:05:50 +020010510
10511 - http_err_rate(<period>) : frequency counter (takes 12 bytes). It takes an
10512 integer parameter <period> which indicates in milliseconds the length
10513 of the period over which the average is measured. It reports the average
10514 HTTP request error rate over that period, in requests per period (see
10515 http_err_cnt above for what is accounted as an error). The result is an
10516 integer which can be matched using ACLs.
10517
10518 - bytes_in_cnt : client to server byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +010010519 integer which counts the cumulative number of bytes received from clients
Willy Tarreauc9705a12010-07-27 20:05:50 +020010520 which matched this entry. Headers are included in the count. This may be
10521 used to limit abuse of upload features on photo or video servers.
10522
10523 - bytes_in_rate(<period>) : frequency counter (takes 12 bytes). It takes an
10524 integer parameter <period> which indicates in milliseconds the length
10525 of the period over which the average is measured. It reports the average
10526 incoming bytes rate over that period, in bytes per period. It may be used
10527 to detect users which upload too much and too fast. Warning: with large
10528 uploads, it is possible that the amount of uploaded data will be counted
10529 once upon termination, thus causing spikes in the average transfer speed
10530 instead of having a smooth one. This may partially be smoothed with
10531 "option contstats" though this is not perfect yet. Use of byte_in_cnt is
10532 recommended for better fairness.
10533
10534 - bytes_out_cnt : server to client byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +010010535 integer which counts the cumulative number of bytes sent to clients which
Willy Tarreauc9705a12010-07-27 20:05:50 +020010536 matched this entry. Headers are included in the count. This may be used
10537 to limit abuse of bots sucking the whole site.
10538
10539 - bytes_out_rate(<period>) : frequency counter (takes 12 bytes). It takes
10540 an integer parameter <period> which indicates in milliseconds the length
10541 of the period over which the average is measured. It reports the average
10542 outgoing bytes rate over that period, in bytes per period. It may be used
10543 to detect users which download too much and too fast. Warning: with large
10544 transfers, it is possible that the amount of transferred data will be
10545 counted once upon termination, thus causing spikes in the average
10546 transfer speed instead of having a smooth one. This may partially be
10547 smoothed with "option contstats" though this is not perfect yet. Use of
10548 byte_out_cnt is recommended for better fairness.
Willy Tarreau08d5f982010-06-06 13:34:54 +020010549
Willy Tarreauc00cdc22010-06-06 16:48:26 +020010550 There is only one stick-table per proxy. At the moment of writing this doc,
10551 it does not seem useful to have multiple tables per proxy. If this happens
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010552 to be required, simply create a dummy backend with a stick-table in it and
10553 reference it.
10554
10555 It is important to understand that stickiness based on learning information
10556 has some limitations, including the fact that all learned associations are
Baptiste Assmann123ff042016-03-06 23:29:28 +010010557 lost upon restart unless peers are properly configured to transfer such
10558 information upon restart (recommended). In general it can be good as a
10559 complement but not always as an exclusive stickiness.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010560
Willy Tarreauc9705a12010-07-27 20:05:50 +020010561 Last, memory requirements may be important when storing many data types.
10562 Indeed, storing all indicators above at once in each entry requires 116 bytes
10563 per entry, or 116 MB for a 1-million entries table. This is definitely not
10564 something that can be ignored.
10565
10566 Example:
10567 # Keep track of counters of up to 1 million IP addresses over 5 minutes
10568 # and store a general purpose counter and the average connection rate
10569 # computed over a sliding window of 30 seconds.
10570 stick-table type ip size 1m expire 5m store gpc0,conn_rate(30s)
10571
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +030010572 See also : "stick match", "stick on", "stick store-request", section 2.4
David du Colombiera13d1b92011-03-17 10:40:22 +010010573 about time format and section 7 about ACLs.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010574
10575
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010576stick store-response <pattern> [table <table>] [{if | unless} <condition>]
Baptiste Assmann2f2d2ec2016-03-06 23:27:24 +010010577 Define a response pattern used to create an entry in a stickiness table
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010578 May be used in sections : defaults | frontend | listen | backend
10579 no | no | yes | yes
10580
10581 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +020010582 <pattern> is a sample expression rule as described in section 7.3. It
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010583 describes what elements of the response or connection will
Davor Ocelice9ed2812017-12-25 17:49:28 +010010584 be analyzed, extracted and stored in the table once a
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010585 server is selected.
10586
10587 <table> is an optional stickiness table name. If unspecified, the same
10588 backend's table is used. A stickiness table is declared using
10589 the "stick-table" statement.
10590
10591 <cond> is an optional storage condition. It makes it possible to store
10592 certain criteria only when some conditions are met (or not met).
10593 For instance, it could be used to store the SSL session ID only
10594 when the response is a SSL server hello.
10595
10596 Some protocols or applications require complex stickiness rules and cannot
10597 always simply rely on cookies nor hashing. The "stick store-response"
10598 statement describes a rule to decide what to extract from the response and
10599 when to do it, in order to store it into a stickiness table for further
10600 requests to match it using the "stick match" statement. Obviously the
10601 extracted part must make sense and have a chance to be matched in a further
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010602 request. Storing an ID found in a header of a response makes sense.
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010603 See section 7 for a complete list of possible patterns and transformation
10604 rules.
10605
10606 The table has to be declared using the "stick-table" statement. It must be of
10607 a type compatible with the pattern. By default it is the one which is present
10608 in the same backend. It is possible to share a table with other backends by
10609 referencing it using the "table" keyword. If another table is referenced,
10610 the server's ID inside the backends are used. By default, all server IDs
10611 start at 1 in each backend, so the server ordering is enough. But in case of
10612 doubt, it is highly recommended to force server IDs using their "id" setting.
10613
10614 It is possible to restrict the conditions where a "stick store-response"
10615 statement will apply, using "if" or "unless" followed by a condition. This
10616 condition will be evaluated while parsing the response, so any criteria can
10617 be used. See section 7 for ACL based conditions.
10618
10619 There is no limit on the number of "stick store-response" statements, but
10620 there is a limit of 8 simultaneous stores per request or response. This
10621 makes it possible to store up to 8 criteria, all extracted from either the
10622 request or the response, regardless of the number of rules. Only the 8 first
10623 ones which match will be kept. Using this, it is possible to feed multiple
10624 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +010010625 another protocol or access method. Using multiple store-response rules with
10626 the same table is possible and may be used to find the best criterion to rely
10627 on, by arranging the rules by decreasing preference order. Only the first
10628 extracted criterion for a given table will be stored. All subsequent store-
10629 response rules referencing the same table will be skipped and their ACLs will
10630 not be evaluated. However, even if a store-request rule references a table, a
10631 store-response rule may also use the same table. This means that each table
10632 may learn exactly one element from the request and one element from the
10633 response at once.
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010634
10635 The table will contain the real server that processed the request.
10636
10637 Example :
10638 # Learn SSL session ID from both request and response and create affinity.
10639 backend https
10640 mode tcp
10641 balance roundrobin
Cyril Bontédc4d9032012-04-08 21:57:39 +020010642 # maximum SSL session ID length is 32 bytes.
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010643 stick-table type binary len 32 size 30k expire 30m
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010644
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010645 acl clienthello req_ssl_hello_type 1
10646 acl serverhello rep_ssl_hello_type 2
10647
10648 # use tcp content accepts to detects ssl client and server hello.
10649 tcp-request inspect-delay 5s
10650 tcp-request content accept if clienthello
10651
10652 # no timeout on response inspect delay by default.
10653 tcp-response content accept if serverhello
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010654
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010655 # SSL session ID (SSLID) may be present on a client or server hello.
10656 # Its length is coded on 1 byte at offset 43 and its value starts
10657 # at offset 44.
10658
10659 # Match and learn on request if client hello.
10660 stick on payload_lv(43,1) if clienthello
10661
10662 # Learn on response if server hello.
10663 stick store-response payload_lv(43,1) if serverhello
Cyril Bontédc4d9032012-04-08 21:57:39 +020010664
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010665 server s1 192.168.1.1:443
10666 server s2 192.168.1.1:443
10667
10668 See also : "stick-table", "stick on", and section 7 about ACLs and pattern
10669 extraction.
10670
10671
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020010672tcp-check comment <string>
10673 Defines a comment for the following the tcp-check rule, reported in logs if
10674 it fails.
10675 May be used in sections : defaults | frontend | listen | backend
10676 yes | no | yes | yes
10677
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020010678 Arguments :
10679 <string> is the comment message to add in logs if the following tcp-check
10680 rule fails.
10681
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020010682 It only works for connect, send and expect rules. It is useful to make
10683 user-friendly error reporting.
10684
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020010685 See also : "option tcp-check", "tcp-check connect", "tcp-check send" and
10686 "tcp-check expect".
10687
10688
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020010689tcp-check connect [default] [port <expr>] [addr <ip>] [send-proxy] [via-socks4]
10690 [ssl] [sni <sni>] [alpn <alpn>] [linger]
Christopher Fauletedc6ed92020-04-23 16:27:59 +020010691 [proto <name>] [comment <msg>]
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010692 Opens a new connection
10693 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020010694 yes | no | yes | yes
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010695
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020010696 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020010697 comment <msg> defines a message to report if the rule evaluation fails.
10698
Christopher Faulet4dce5922020-03-30 13:54:42 +020010699 default Use default options of the server line to do the health
Daniel Corbett67a82712020-07-06 23:01:19 -040010700 checks. The server options are used only if not redefined.
Christopher Faulet4dce5922020-03-30 13:54:42 +020010701
Christopher Fauletb7d30092020-03-30 15:19:03 +020010702 port <expr> if not set, check port or server port is used.
Christopher Faulet5c288742020-03-31 08:15:58 +020010703 It tells HAProxy where to open the connection to.
10704 <port> must be a valid TCP port source integer, from 1 to
Christopher Fauletb7d30092020-03-30 15:19:03 +020010705 65535 or an sample-fetch expression.
Christopher Faulet5c288742020-03-31 08:15:58 +020010706
10707 addr <ip> defines the IP address to do the health check.
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010708
10709 send-proxy send a PROXY protocol string
10710
Christopher Faulet085426a2020-03-30 13:07:02 +020010711 via-socks4 enables outgoing health checks using upstream socks4 proxy.
10712
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010713 ssl opens a ciphered connection
10714
Christopher Faulet79b31d42020-03-30 13:00:05 +020010715 sni <sni> specifies the SNI to use to do health checks over SSL.
10716
Christopher Faulet98572322020-03-30 13:16:44 +020010717 alpn <alpn> defines which protocols to advertise with ALPN. The protocol
10718 list consists in a comma-delimited list of protocol names,
10719 for instance: "http/1.1,http/1.0" (without quotes).
10720 If it is not set, the server ALPN is used.
10721
Christopher Fauletedc6ed92020-04-23 16:27:59 +020010722 proto <name> forces the multiplexer's protocol to use for this connection.
10723 It must be a TCP mux protocol and it must be usable on the
10724 backend side. The list of available protocols is reported in
10725 haproxy -vv.
10726
Christopher Faulet5c288742020-03-31 08:15:58 +020010727 linger cleanly close the connection instead of using a single RST.
Gaetan Rivetf8ba6772020-02-07 15:37:17 +010010728
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020010729 When an application lies on more than a single TCP port or when HAProxy
10730 load-balance many services in a single backend, it makes sense to probe all
10731 the services individually before considering a server as operational.
10732
10733 When there are no TCP port configured on the server line neither server port
10734 directive, then the 'tcp-check connect port <port>' must be the first step
10735 of the sequence.
10736
10737 In a tcp-check ruleset a 'connect' is required, it is also mandatory to start
10738 the ruleset with a 'connect' rule. Purpose is to ensure admin know what they
10739 do.
10740
10741 When a connect must start the ruleset, if may still be preceded by set-var,
10742 unset-var or comment rules.
10743
10744 Examples :
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010745 # check HTTP and HTTPs services on a server.
10746 # first open port 80 thanks to server line port directive, then
10747 # tcp-check opens port 443, ciphered and run a request on it:
10748 option tcp-check
10749 tcp-check connect
10750 tcp-check send GET\ /\ HTTP/1.0\r\n
10751 tcp-check send Host:\ haproxy.1wt.eu\r\n
10752 tcp-check send \r\n
10753 tcp-check expect rstring (2..|3..)
10754 tcp-check connect port 443 ssl
10755 tcp-check send GET\ /\ HTTP/1.0\r\n
10756 tcp-check send Host:\ haproxy.1wt.eu\r\n
10757 tcp-check send \r\n
10758 tcp-check expect rstring (2..|3..)
10759 server www 10.0.0.1 check port 80
10760
10761 # check both POP and IMAP from a single server:
10762 option tcp-check
Gaetan Rivetf8ba6772020-02-07 15:37:17 +010010763 tcp-check connect port 110 linger
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010764 tcp-check expect string +OK\ POP3\ ready
10765 tcp-check connect port 143
10766 tcp-check expect string *\ OK\ IMAP4\ ready
10767 server mail 10.0.0.1 check
10768
10769 See also : "option tcp-check", "tcp-check send", "tcp-check expect"
10770
10771
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020010772tcp-check expect [min-recv <int>] [comment <msg>]
Christopher Fauletec07e382020-04-07 14:56:26 +020010773 [ok-status <st>] [error-status <st>] [tout-status <st>]
Christopher Faulet98cc57c2020-04-01 20:52:31 +020010774 [on-success <fmt>] [on-error <fmt>] [status-code <expr>]
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020010775 [!] <match> <pattern>
Davor Ocelice9ed2812017-12-25 17:49:28 +010010776 Specify data to be collected and analyzed during a generic health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010777 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020010778 yes | no | yes | yes
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010779
10780 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020010781 comment <msg> defines a message to report if the rule evaluation fails.
10782
Gaetan Rivet1afd8262020-02-07 15:37:17 +010010783 min-recv is optional and can define the minimum amount of data required to
10784 evaluate the current expect rule. If the number of received bytes
10785 is under this limit, the check will wait for more data. This
10786 option can be used to resolve some ambiguous matching rules or to
10787 avoid executing costly regex matches on content known to be still
10788 incomplete. If an exact string (string or binary) is used, the
10789 minimum between the string length and this parameter is used.
10790 This parameter is ignored if it is set to -1. If the expect rule
10791 does not match, the check will wait for more data. If set to 0,
10792 the evaluation result is always conclusive.
10793
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010794 <match> is a keyword indicating how to look for a specific pattern in the
Gaetan Rivetefab6c62020-02-07 15:37:17 +010010795 response. The keyword may be one of "string", "rstring", "binary" or
10796 "rbinary".
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010797 The keyword may be preceded by an exclamation mark ("!") to negate
10798 the match. Spaces are allowed between the exclamation mark and the
10799 keyword. See below for more details on the supported keywords.
10800
Christopher Fauletec07e382020-04-07 14:56:26 +020010801 ok-status <st> is optional and can be used to set the check status if
10802 the expect rule is successfully evaluated and if it is
10803 the last rule in the tcp-check ruleset. "L7OK", "L7OKC",
Christopher Fauletd888f0f2020-05-07 07:40:17 +020010804 "L6OK" and "L4OK" are supported :
10805 - L7OK : check passed on layer 7
10806 - L7OKC : check conditionally passed on layer 7, for
10807 example 404 with disable-on-404
10808 - L6OK : check passed on layer 6
10809 - L4OK : check passed on layer 4
Christopher Fauletec07e382020-04-07 14:56:26 +020010810 By default "L7OK" is used.
10811
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020010812 error-status <st> is optional and can be used to set the check status if
10813 an error occurred during the expect rule evaluation.
Christopher Fauletd888f0f2020-05-07 07:40:17 +020010814 "L7RSP", "L7STS", "L6RSP" and "L4CON" are supported :
10815 - L7RSP : layer 7 invalid response - protocol error
10816 - L7STS : layer 7 response error, for example HTTP 5xx
10817 - L6RSP : layer 6 invalid response - protocol error
10818 - L4CON : layer 1-4 connection problem
10819 By default "L7RSP" is used.
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020010820
Christopher Fauletec07e382020-04-07 14:56:26 +020010821 tout-status <st> is optional and can be used to set the check status if
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020010822 a timeout occurred during the expect rule evaluation.
Christopher Fauletd888f0f2020-05-07 07:40:17 +020010823 "L7TOUT", "L6TOUT", and "L4TOUT" are supported :
10824 - L7TOUT : layer 7 (HTTP/SMTP) timeout
10825 - L6TOUT : layer 6 (SSL) timeout
10826 - L4TOUT : layer 1-4 timeout
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020010827 By default "L7TOUT" is used.
10828
Christopher Fauletbe52b4d2020-04-01 16:30:22 +020010829 on-success <fmt> is optional and can be used to customize the
10830 informational message reported in logs if the expect
10831 rule is successfully evaluated and if it is the last rule
10832 in the tcp-check ruleset. <fmt> is a log-format string.
10833
10834 on-error <fmt> is optional and can be used to customize the
10835 informational message reported in logs if an error
10836 occurred during the expect rule evaluation. <fmt> is a
10837 log-format string.
10838
Christopher Faulet98cc57c2020-04-01 20:52:31 +020010839 status-code <expr> is optional and can be used to set the check status code
10840 reported in logs, on success or on error. <expr> is a
10841 standard HAProxy expression formed by a sample-fetch
10842 followed by some converters.
10843
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010844 <pattern> is the pattern to look for. It may be a string or a regular
10845 expression. If the pattern contains spaces, they must be escaped
10846 with the usual backslash ('\').
10847 If the match is set to binary, then the pattern must be passed as
Davor Ocelice9ed2812017-12-25 17:49:28 +010010848 a series of hexadecimal digits in an even number. Each sequence of
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010849 two digits will represent a byte. The hexadecimal digits may be
10850 used upper or lower case.
10851
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010852 The available matches are intentionally similar to their http-check cousins :
10853
10854 string <string> : test the exact string matches in the response buffer.
10855 A health check response will be considered valid if the
10856 response's buffer contains this exact string. If the
10857 "string" keyword is prefixed with "!", then the response
10858 will be considered invalid if the body contains this
10859 string. This can be used to look for a mandatory pattern
10860 in a protocol response, or to detect a failure when a
10861 specific error appears in a protocol banner.
10862
10863 rstring <regex> : test a regular expression on the response buffer.
10864 A health check response will be considered valid if the
10865 response's buffer matches this expression. If the
10866 "rstring" keyword is prefixed with "!", then the response
10867 will be considered invalid if the body matches the
10868 expression.
10869
Christopher Fauletaaab0832020-05-05 15:54:22 +020010870 string-lf <fmt> : test a log-format string match in the response's buffer.
10871 A health check response will be considered valid if the
10872 response's buffer contains the string resulting of the
10873 evaluation of <fmt>, which follows the log-format rules.
10874 If prefixed with "!", then the response will be
10875 considered invalid if the buffer contains the string.
10876
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010877 binary <hexstring> : test the exact string in its hexadecimal form matches
10878 in the response buffer. A health check response will
10879 be considered valid if the response's buffer contains
10880 this exact hexadecimal string.
10881 Purpose is to match data on binary protocols.
10882
Gaetan Rivetefab6c62020-02-07 15:37:17 +010010883 rbinary <regex> : test a regular expression on the response buffer, like
10884 "rstring". However, the response buffer is transformed
10885 into its hexadecimal form, including NUL-bytes. This
10886 allows using all regex engines to match any binary
10887 content. The hexadecimal transformation takes twice the
10888 size of the original response. As such, the expected
10889 pattern should work on at-most half the response buffer
10890 size.
10891
Christopher Fauletaaab0832020-05-05 15:54:22 +020010892 binary-lf <hexfmt> : test a log-format string in its hexadecimal form
10893 match in the response's buffer. A health check response
10894 will be considered valid if the response's buffer
10895 contains the hexadecimal string resulting of the
10896 evaluation of <fmt>, which follows the log-format
10897 rules. If prefixed with "!", then the response will be
10898 considered invalid if the buffer contains the
10899 hexadecimal string. The hexadecimal string is converted
10900 in a binary string before matching the response's
10901 buffer.
10902
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010903 It is important to note that the responses will be limited to a certain size
10904 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
10905 Thus, too large responses may not contain the mandatory pattern when using
10906 "string", "rstring" or binary. If a large response is absolutely required, it
10907 is possible to change the default max size by setting the global variable.
10908 However, it is worth keeping in mind that parsing very large responses can
10909 waste some CPU cycles, especially when regular expressions are used, and that
10910 it is always better to focus the checks on smaller resources. Also, in its
10911 current state, the check will not find any string nor regex past a null
10912 character in the response. Similarly it is not possible to request matching
10913 the null character.
10914
10915 Examples :
10916 # perform a POP check
10917 option tcp-check
10918 tcp-check expect string +OK\ POP3\ ready
10919
10920 # perform an IMAP check
10921 option tcp-check
10922 tcp-check expect string *\ OK\ IMAP4\ ready
10923
10924 # look for the redis master server
10925 option tcp-check
10926 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +020010927 tcp-check expect string +PONG
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010928 tcp-check send info\ replication\r\n
10929 tcp-check expect string role:master
10930 tcp-check send QUIT\r\n
10931 tcp-check expect string +OK
10932
10933
10934 See also : "option tcp-check", "tcp-check connect", "tcp-check send",
10935 "tcp-check send-binary", "http-check expect", tune.chksize
10936
10937
Christopher Fauletb50b3e62020-05-05 18:43:43 +020010938tcp-check send <data> [comment <msg>]
10939tcp-check send-lf <fmt> [comment <msg>]
10940 Specify a string or a log-format string to be sent as a question during a
10941 generic health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010942 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020010943 yes | no | yes | yes
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010944
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020010945 Arguments :
10946 comment <msg> defines a message to report if the rule evaluation fails.
10947
Christopher Fauletb50b3e62020-05-05 18:43:43 +020010948 <data> is the string that will be sent during a generic health
10949 check session.
Christopher Faulet16fff672020-04-30 07:50:54 +020010950
Christopher Fauletb50b3e62020-05-05 18:43:43 +020010951 <fmt> is the log-format string that will be sent, once evaluated,
10952 during a generic health check session.
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010953
10954 Examples :
10955 # look for the redis master server
10956 option tcp-check
10957 tcp-check send info\ replication\r\n
10958 tcp-check expect string role:master
10959
10960 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
10961 "tcp-check send-binary", tune.chksize
10962
10963
Christopher Fauletb50b3e62020-05-05 18:43:43 +020010964tcp-check send-binary <hexstring> [comment <msg>]
10965tcp-check send-binary-lf <hexfmt> [comment <msg>]
10966 Specify an hex digits string or an hex digits log-format string to be sent as
10967 a binary question during a raw tcp health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010968 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020010969 yes | no | yes | yes
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010970
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020010971 Arguments :
10972 comment <msg> defines a message to report if the rule evaluation fails.
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020010973
Christopher Fauletb50b3e62020-05-05 18:43:43 +020010974 <hexstring> is the hexadecimal string that will be send, once converted
10975 to binary, during a generic health check session.
Christopher Faulet16fff672020-04-30 07:50:54 +020010976
Christopher Fauletb50b3e62020-05-05 18:43:43 +020010977 <hexfmt> is the hexadecimal log-format string that will be send, once
10978 evaluated and converted to binary, during a generic health
10979 check session.
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010980
10981 Examples :
10982 # redis check in binary
10983 option tcp-check
10984 tcp-check send-binary 50494e470d0a # PING\r\n
10985 tcp-check expect binary 2b504F4e47 # +PONG
10986
10987
10988 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
10989 "tcp-check send", tune.chksize
10990
10991
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010010992tcp-check set-var(<var-name>) <expr>
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010010993 This operation sets the content of a variable. The variable is declared inline.
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010010994 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020010995 yes | no | yes | yes
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010010996
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020010997 Arguments :
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010010998 <var-name> The name of the variable starts with an indication about its
10999 scope. The scopes allowed for tcp-check are:
11000 "proc" : the variable is shared with the whole process.
11001 "sess" : the variable is shared with the tcp-check session.
11002 "check": the variable is declared for the lifetime of the tcp-check.
11003 This prefix is followed by a name. The separator is a '.'.
11004 The name may only contain characters 'a-z', 'A-Z', '0-9', '.',
11005 and '-'.
11006
11007 <expr> Is a sample-fetch expression potentially followed by converters.
11008
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011009 Examples :
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011010 tcp-check set-var(check.port) int(1234)
11011
11012
11013tcp-check unset-var(<var-name>)
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011014 Free a reference to a variable within its scope.
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011015 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020011016 yes | no | yes | yes
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011017
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011018 Arguments :
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011019 <var-name> The name of the variable starts with an indication about its
11020 scope. The scopes allowed for tcp-check are:
11021 "proc" : the variable is shared with the whole process.
11022 "sess" : the variable is shared with the tcp-check session.
11023 "check": the variable is declared for the lifetime of the tcp-check.
11024 This prefix is followed by a name. The separator is a '.'.
11025 The name may only contain characters 'a-z', 'A-Z', '0-9', '.',
11026 and '-'.
11027
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011028 Examples :
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011029 tcp-check unset-var(check.port)
11030
11031
Willy Tarreaue9656522010-08-17 15:40:09 +020011032tcp-request connection <action> [{if | unless} <condition>]
11033 Perform an action on an incoming connection depending on a layer 4 condition
Willy Tarreau1a687942010-05-23 22:40:30 +020011034 May be used in sections : defaults | frontend | listen | backend
11035 no | yes | yes | no
Willy Tarreaue9656522010-08-17 15:40:09 +020011036 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +020011037 <action> defines the action to perform if the condition applies. See
11038 below.
Willy Tarreau1a687942010-05-23 22:40:30 +020011039
Willy Tarreaue9656522010-08-17 15:40:09 +020011040 <condition> is a standard layer4-only ACL-based condition (see section 7).
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011041
11042 Immediately after acceptance of a new incoming connection, it is possible to
11043 evaluate some conditions to decide whether this connection must be accepted
Willy Tarreaue9656522010-08-17 15:40:09 +020011044 or dropped or have its counters tracked. Those conditions cannot make use of
11045 any data contents because the connection has not been read from yet, and the
11046 buffers are not yet allocated. This is used to selectively and very quickly
11047 accept or drop connections from various sources with a very low overhead. If
11048 some contents need to be inspected in order to take the decision, the
11049 "tcp-request content" statements must be used instead.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011050
Willy Tarreaue9656522010-08-17 15:40:09 +020011051 The "tcp-request connection" rules are evaluated in their exact declaration
11052 order. If no rule matches or if there is no rule, the default action is to
11053 accept the incoming connection. There is no specific limit to the number of
11054 rules which may be inserted.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011055
Willy Tarreaua9083d02015-05-08 15:27:59 +020011056 Four types of actions are supported :
Willy Tarreaue9656522010-08-17 15:40:09 +020011057 - accept :
11058 accepts the connection if the condition is true (when used with "if")
11059 or false (when used with "unless"). The first such rule executed ends
11060 the rules evaluation.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011061
Willy Tarreaue9656522010-08-17 15:40:09 +020011062 - reject :
11063 rejects the connection if the condition is true (when used with "if")
11064 or false (when used with "unless"). The first such rule executed ends
11065 the rules evaluation. Rejected connections do not even become a
11066 session, which is why they are accounted separately for in the stats,
11067 as "denied connections". They are not considered for the session
11068 rate-limit and are not logged either. The reason is that these rules
11069 should only be used to filter extremely high connection rates such as
11070 the ones encountered during a massive DDoS attack. Under these extreme
11071 conditions, the simple action of logging each event would make the
11072 system collapse and would considerably lower the filtering capacity. If
11073 logging is absolutely desired, then "tcp-request content" rules should
Willy Tarreau4f614292016-10-21 17:49:36 +020011074 be used instead, as "tcp-request session" rules will not log either.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011075
Willy Tarreau4f0d9192013-06-11 20:40:55 +020011076 - expect-proxy layer4 :
11077 configures the client-facing connection to receive a PROXY protocol
11078 header before any byte is read from the socket. This is equivalent to
11079 having the "accept-proxy" keyword on the "bind" line, except that using
11080 the TCP rule allows the PROXY protocol to be accepted only for certain
11081 IP address ranges using an ACL. This is convenient when multiple layers
11082 of load balancers are passed through by traffic coming from public
11083 hosts.
11084
Bertrand Jacquin90759682016-06-06 15:35:39 +010011085 - expect-netscaler-cip layer4 :
11086 configures the client-facing connection to receive a NetScaler Client
11087 IP insertion protocol header before any byte is read from the socket.
11088 This is equivalent to having the "accept-netscaler-cip" keyword on the
11089 "bind" line, except that using the TCP rule allows the PROXY protocol
11090 to be accepted only for certain IP address ranges using an ACL. This
11091 is convenient when multiple layers of load balancers are passed
11092 through by traffic coming from public hosts.
11093
Willy Tarreau18bf01e2014-06-13 16:18:52 +020011094 - capture <sample> len <length> :
11095 This only applies to "tcp-request content" rules. It captures sample
11096 expression <sample> from the request buffer, and converts it to a
11097 string of at most <len> characters. The resulting string is stored into
11098 the next request "capture" slot, so it will possibly appear next to
11099 some captured HTTP headers. It will then automatically appear in the
11100 logs, and it will be possible to extract it using sample fetch rules to
11101 feed it into headers or anything. The length should be limited given
11102 that this size will be allocated for each capture during the whole
Willy Tarreaua9083d02015-05-08 15:27:59 +020011103 session life. Please check section 7.3 (Fetching samples) and "capture
11104 request header" for more information.
Willy Tarreau18bf01e2014-06-13 16:18:52 +020011105
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011106 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] :
Willy Tarreaue9656522010-08-17 15:40:09 +020011107 enables tracking of sticky counters from current connection. These
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020011108 rules do not stop evaluation and do not change default action. The
11109 number of counters that may be simultaneously tracked by the same
11110 connection is set in MAX_SESS_STKCTR at build time (reported in
John Roeslerfb2fce12019-07-10 15:45:51 -050011111 haproxy -vv) which defaults to 3, so the track-sc number is between 0
Matteo Contrini1857b8c2020-10-16 17:35:54 +020011112 and (MAX_SESS_STKCTR-1). The first "track-sc0" rule executed enables
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020011113 tracking of the counters of the specified table as the first set. The
11114 first "track-sc1" rule executed enables tracking of the counters of the
11115 specified table as the second set. The first "track-sc2" rule executed
11116 enables tracking of the counters of the specified table as the third
11117 set. It is a recommended practice to use the first set of counters for
11118 the per-frontend counters and the second set for the per-backend ones.
11119 But this is just a guideline, all may be used everywhere.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011120
Willy Tarreaue9656522010-08-17 15:40:09 +020011121 These actions take one or two arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +020011122 <key> is mandatory, and is a sample expression rule as described
Willy Tarreau74ca5042013-06-11 23:12:07 +020011123 in section 7.3. It describes what elements of the incoming
Davor Ocelice9ed2812017-12-25 17:49:28 +010011124 request or connection will be analyzed, extracted, combined,
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011125 and used to select which table entry to update the counters.
11126 Note that "tcp-request connection" cannot use content-based
11127 fetches.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011128
Willy Tarreaue9656522010-08-17 15:40:09 +020011129 <table> is an optional table to be used instead of the default one,
11130 which is the stick-table declared in the current proxy. All
11131 the counters for the matches and updates for the key will
11132 then be performed in that table until the session ends.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011133
Willy Tarreaue9656522010-08-17 15:40:09 +020011134 Once a "track-sc*" rule is executed, the key is looked up in the table
11135 and if it is not found, an entry is allocated for it. Then a pointer to
11136 that entry is kept during all the session's life, and this entry's
11137 counters are updated as often as possible, every time the session's
11138 counters are updated, and also systematically when the session ends.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011139 Counters are only updated for events that happen after the tracking has
11140 been started. For example, connection counters will not be updated when
11141 tracking layer 7 information, since the connection event happens before
11142 layer7 information is extracted.
11143
Willy Tarreaue9656522010-08-17 15:40:09 +020011144 If the entry tracks concurrent connection counters, one connection is
11145 counted for as long as the entry is tracked, and the entry will not
11146 expire during that time. Tracking counters also provides a performance
11147 advantage over just checking the keys, because only one table lookup is
11148 performed for all ACL checks that make use of it.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011149
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +020011150 - sc-inc-gpc0(<sc-id>):
11151 The "sc-inc-gpc0" increments the GPC0 counter according to the sticky
11152 counter designated by <sc-id>. If an error occurs, this action silently
11153 fails and the actions evaluation continues.
11154
Frédéric Lécaille6778b272018-01-29 15:22:53 +010011155 - sc-inc-gpc1(<sc-id>):
11156 The "sc-inc-gpc1" increments the GPC1 counter according to the sticky
11157 counter designated by <sc-id>. If an error occurs, this action silently
11158 fails and the actions evaluation continues.
11159
Cédric Dufour0d7712d2019-11-06 18:38:53 +010011160 - sc-set-gpt0(<sc-id>) { <int> | <expr> }:
11161 This action sets the 32-bit unsigned GPT0 tag according to the sticky
11162 counter designated by <sc-id> and the value of <int>/<expr>. The
11163 expected result is a boolean. If an error occurs, this action silently
11164 fails and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +020011165
William Lallemand2e785f22016-05-25 01:48:42 +020011166 - set-src <expr> :
11167 Is used to set the source IP address to the value of specified
11168 expression. Useful if you want to mask source IP for privacy.
11169 If you want to provide an IP from a HTTP header use "http-request
Cyril Bonté6c81d5f2018-10-17 00:14:51 +020011170 set-src".
William Lallemand2e785f22016-05-25 01:48:42 +020011171
Cyril Bonté6c81d5f2018-10-17 00:14:51 +020011172 Arguments:
11173 <expr> Is a standard HAProxy expression formed by a sample-fetch
11174 followed by some converters.
William Lallemand2e785f22016-05-25 01:48:42 +020011175
11176 Example:
William Lallemand2e785f22016-05-25 01:48:42 +020011177 tcp-request connection set-src src,ipmask(24)
11178
Willy Tarreau0c630532016-10-21 17:52:58 +020011179 When possible, set-src preserves the original source port as long as the
11180 address family allows it, otherwise the source port is set to 0.
William Lallemand2e785f22016-05-25 01:48:42 +020011181
William Lallemand44be6402016-05-25 01:51:35 +020011182 - set-src-port <expr> :
11183 Is used to set the source port address to the value of specified
11184 expression.
11185
Cyril Bonté6c81d5f2018-10-17 00:14:51 +020011186 Arguments:
11187 <expr> Is a standard HAProxy expression formed by a sample-fetch
11188 followed by some converters.
William Lallemand44be6402016-05-25 01:51:35 +020011189
11190 Example:
William Lallemand44be6402016-05-25 01:51:35 +020011191 tcp-request connection set-src-port int(4000)
11192
Willy Tarreau0c630532016-10-21 17:52:58 +020011193 When possible, set-src-port preserves the original source address as long
11194 as the address family supports a port, otherwise it forces the source
11195 address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +020011196
William Lallemand13e9b0c2016-05-25 02:34:07 +020011197 - set-dst <expr> :
11198 Is used to set the destination IP address to the value of specified
11199 expression. Useful if you want to mask IP for privacy in log.
11200 If you want to provide an IP from a HTTP header use "http-request
11201 set-dst". If you want to connect to the new address/port, use
11202 '0.0.0.0:0' as a server address in the backend.
11203
11204 <expr> Is a standard HAProxy expression formed by a sample-fetch
11205 followed by some converters.
11206
11207 Example:
11208
11209 tcp-request connection set-dst dst,ipmask(24)
11210 tcp-request connection set-dst ipv4(10.0.0.1)
11211
Willy Tarreau0c630532016-10-21 17:52:58 +020011212 When possible, set-dst preserves the original destination port as long as
11213 the address family allows it, otherwise the destination port is set to 0.
11214
William Lallemand13e9b0c2016-05-25 02:34:07 +020011215 - set-dst-port <expr> :
11216 Is used to set the destination port address to the value of specified
11217 expression. If you want to connect to the new address/port, use
11218 '0.0.0.0:0' as a server address in the backend.
11219
11220
11221 <expr> Is a standard HAProxy expression formed by a sample-fetch
11222 followed by some converters.
11223
11224 Example:
11225
11226 tcp-request connection set-dst-port int(4000)
11227
Willy Tarreau0c630532016-10-21 17:52:58 +020011228 When possible, set-dst-port preserves the original destination address as
11229 long as the address family supports a port, otherwise it forces the
11230 destination address to IPv4 "0.0.0.0" before rewriting the port.
11231
Willy Tarreau2d392c22015-08-24 01:43:45 +020011232 - "silent-drop" :
11233 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +010011234 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +020011235 to prevent the client from being notified. The effect it then that the
11236 client still sees an established connection while there's none on
11237 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
11238 except that it doesn't use any local resource at all on the machine
11239 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +010011240 slow down stronger attackers. It is important to understand the impact
11241 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +020011242 client and HAProxy (firewalls, proxies, load balancers) will also keep
11243 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +010011244 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +020011245 TCP_REPAIR socket option is used to block the emission of a TCP
11246 reset. On other systems, the socket's TTL is reduced to 1 so that the
11247 TCP reset doesn't pass the first router, though it's still delivered to
11248 local networks. Do not use it unless you fully understand how it works.
11249
Willy Tarreaue9656522010-08-17 15:40:09 +020011250 Note that the "if/unless" condition is optional. If no condition is set on
11251 the action, it is simply performed unconditionally. That can be useful for
11252 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011253
Willy Tarreaue9656522010-08-17 15:40:09 +020011254 Example: accept all connections from white-listed hosts, reject too fast
11255 connection without counting them, and track accepted connections.
11256 This results in connection rate being capped from abusive sources.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011257
Willy Tarreaue9656522010-08-17 15:40:09 +020011258 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011259 tcp-request connection reject if { src_conn_rate gt 10 }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011260 tcp-request connection track-sc0 src
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011261
Willy Tarreaue9656522010-08-17 15:40:09 +020011262 Example: accept all connections from white-listed hosts, count all other
11263 connections and reject too fast ones. This results in abusive ones
11264 being blocked as long as they don't slow down.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011265
Willy Tarreaue9656522010-08-17 15:40:09 +020011266 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011267 tcp-request connection track-sc0 src
11268 tcp-request connection reject if { sc0_conn_rate gt 10 }
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011269
Willy Tarreau4f0d9192013-06-11 20:40:55 +020011270 Example: enable the PROXY protocol for traffic coming from all known proxies.
11271
11272 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
11273
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011274 See section 7 about ACL usage.
11275
Willy Tarreau4f614292016-10-21 17:49:36 +020011276 See also : "tcp-request session", "tcp-request content", "stick-table"
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011277
11278
Willy Tarreaue9656522010-08-17 15:40:09 +020011279tcp-request content <action> [{if | unless} <condition>]
11280 Perform an action on a new session depending on a layer 4-7 condition
Willy Tarreau62644772008-07-16 18:36:06 +020011281 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +020011282 no | yes | yes | yes
Willy Tarreaue9656522010-08-17 15:40:09 +020011283 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +020011284 <action> defines the action to perform if the condition applies. See
11285 below.
Willy Tarreau62644772008-07-16 18:36:06 +020011286
Willy Tarreaue9656522010-08-17 15:40:09 +020011287 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
Willy Tarreau62644772008-07-16 18:36:06 +020011288
Davor Ocelice9ed2812017-12-25 17:49:28 +010011289 A request's contents can be analyzed at an early stage of request processing
Willy Tarreaue9656522010-08-17 15:40:09 +020011290 called "TCP content inspection". During this stage, ACL-based rules are
11291 evaluated every time the request contents are updated, until either an
11292 "accept" or a "reject" rule matches, or the TCP request inspection delay
11293 expires with no matching rule.
Willy Tarreau62644772008-07-16 18:36:06 +020011294
Willy Tarreaue9656522010-08-17 15:40:09 +020011295 The first difference between these rules and "tcp-request connection" rules
11296 is that "tcp-request content" rules can make use of contents to take a
11297 decision. Most often, these decisions will consider a protocol recognition or
11298 validity. The second difference is that content-based rules can be used in
Willy Tarreauf3338342014-01-28 21:40:28 +010011299 both frontends and backends. In case of HTTP keep-alive with the client, all
11300 tcp-request content rules are evaluated again, so haproxy keeps a record of
11301 what sticky counters were assigned by a "tcp-request connection" versus a
11302 "tcp-request content" rule, and flushes all the content-related ones after
11303 processing an HTTP request, so that they may be evaluated again by the rules
11304 being evaluated again for the next request. This is of particular importance
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011305 when the rule tracks some L7 information or when it is conditioned by an
Willy Tarreauf3338342014-01-28 21:40:28 +010011306 L7-based ACL, since tracking may change between requests.
Willy Tarreau62644772008-07-16 18:36:06 +020011307
Willy Tarreaue9656522010-08-17 15:40:09 +020011308 Content-based rules are evaluated in their exact declaration order. If no
11309 rule matches or if there is no rule, the default action is to accept the
11310 contents. There is no specific limit to the number of rules which may be
11311 inserted.
Willy Tarreau62644772008-07-16 18:36:06 +020011312
Thierry FOURNIER236657b2015-08-19 08:25:14 +020011313 Several types of actions are supported :
Willy Tarreau18bf01e2014-06-13 16:18:52 +020011314 - accept : the request is accepted
Baptiste Assmann333939c2019-01-21 08:34:50 +010011315 - do-resolve: perform a DNS resolution
Willy Tarreau18bf01e2014-06-13 16:18:52 +020011316 - reject : the request is rejected and the connection is closed
11317 - capture : the specified sample expression is captured
Patrick Hemmer268a7072018-05-11 12:52:31 -040011318 - set-priority-class <expr> | set-priority-offset <expr>
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011319 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +020011320 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +010011321 - sc-inc-gpc1(<sc-id>)
Cédric Dufour0d7712d2019-11-06 18:38:53 +010011322 - sc-set-gpt0(<sc-id>) { <int> | <expr> }
Baptiste Assmanne1afd4f2019-04-18 16:21:13 +020011323 - set-dst <expr>
11324 - set-dst-port <expr>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011325 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +010011326 - unset-var(<var-name>)
Willy Tarreau2d392c22015-08-24 01:43:45 +020011327 - silent-drop
Davor Ocelice9ed2812017-12-25 17:49:28 +010011328 - send-spoe-group <engine-name> <group-name>
Christopher Faulet579d83b2019-11-22 15:34:17 +010011329 - use-service <service-name>
Willy Tarreau62644772008-07-16 18:36:06 +020011330
Willy Tarreaue9656522010-08-17 15:40:09 +020011331 They have the same meaning as their counter-parts in "tcp-request connection"
11332 so please refer to that section for a complete description.
Baptiste Assmann333939c2019-01-21 08:34:50 +010011333 For "do-resolve" action, please check the "http-request do-resolve"
11334 configuration section.
Willy Tarreau62644772008-07-16 18:36:06 +020011335
Willy Tarreauf3338342014-01-28 21:40:28 +010011336 While there is nothing mandatory about it, it is recommended to use the
11337 track-sc0 in "tcp-request connection" rules, track-sc1 for "tcp-request
11338 content" rules in the frontend, and track-sc2 for "tcp-request content"
11339 rules in the backend, because that makes the configuration more readable
11340 and easier to troubleshoot, but this is just a guideline and all counters
11341 may be used everywhere.
Willy Tarreau62644772008-07-16 18:36:06 +020011342
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011343 Note that the "if/unless" condition is optional. If no condition is set on
Willy Tarreaue9656522010-08-17 15:40:09 +020011344 the action, it is simply performed unconditionally. That can be useful for
11345 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau62644772008-07-16 18:36:06 +020011346
Christopher Faulet2079a4a2020-10-02 11:48:57 +020011347 Note also that it is recommended to use a "tcp-request session" rule to track
11348 information that does *not* depend on Layer 7 contents, especially for HTTP
11349 frontends. Some HTTP processing are performed at the session level and may
11350 lead to an early rejection of the requests. Thus, the tracking at the content
11351 level may be disturbed in such case. A warning is emitted during startup to
11352 prevent, as far as possible, such unreliable usage.
11353
Willy Tarreaue9656522010-08-17 15:40:09 +020011354 It is perfectly possible to match layer 7 contents with "tcp-request content"
Christopher Faulet7ea509e2020-10-02 11:38:46 +020011355 rules from a TCP proxy, since HTTP-specific ACL matches are able to
11356 preliminarily parse the contents of a buffer before extracting the required
11357 data. If the buffered contents do not parse as a valid HTTP message, then the
11358 ACL does not match. The parser which is involved there is exactly the same
11359 as for all other HTTP processing, so there is no risk of parsing something
11360 differently. In an HTTP frontend or an HTTP backend, it is guaranteed that
11361 HTTP contents will always be immediately present when the rule is evaluated
11362 first because the HTTP parsing is performed in the early stages of the
11363 connection processing, at the session level. But for such proxies, using
11364 "http-request" rules is much more natural and recommended.
Willy Tarreau62644772008-07-16 18:36:06 +020011365
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011366 Tracking layer7 information is also possible provided that the information
Willy Tarreau4d54c7c2014-09-16 15:48:15 +020011367 are present when the rule is processed. The rule processing engine is able to
11368 wait until the inspect delay expires when the data to be tracked is not yet
11369 available.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011370
Baptiste Assmanne1afd4f2019-04-18 16:21:13 +020011371 The "set-dst" and "set-dst-port" are used to set respectively the destination
11372 IP and port. More information on how to use it at "http-request set-dst".
11373
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011374 The "set-var" is used to set the content of a variable. The variable is
Willy Tarreau4f614292016-10-21 17:49:36 +020011375 declared inline. For "tcp-request session" rules, only session-level
11376 variables can be used, without any layer7 contents.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011377
Daniel Schneller0b547052016-03-21 20:46:57 +010011378 <var-name> The name of the variable starts with an indication about
11379 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010011380 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010011381 "sess" : the variable is shared with the whole session
11382 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011383 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +010011384 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011385 processing
Daniel Schneller0b547052016-03-21 20:46:57 +010011386 "res" : the variable is shared only during response
11387 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011388 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +010011389 The name may only contain characters 'a-z', 'A-Z', '0-9',
11390 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011391
11392 <expr> Is a standard HAProxy expression formed by a sample-fetch
11393 followed by some converters.
11394
Christopher Faulet85d79c92016-11-09 16:54:56 +010011395 The "unset-var" is used to unset a variable. See above for details about
11396 <var-name>.
11397
Patrick Hemmer268a7072018-05-11 12:52:31 -040011398 The "set-priority-class" is used to set the queue priority class of the
11399 current request. The value must be a sample expression which converts to an
11400 integer in the range -2047..2047. Results outside this range will be
11401 truncated. The priority class determines the order in which queued requests
11402 are processed. Lower values have higher priority.
11403
11404 The "set-priority-offset" is used to set the queue priority timestamp offset
11405 of the current request. The value must be a sample expression which converts
11406 to an integer in the range -524287..524287. Results outside this range will be
11407 truncated. When a request is queued, it is ordered first by the priority
11408 class, then by the current timestamp adjusted by the given offset in
11409 milliseconds. Lower values have higher priority.
11410 Note that the resulting timestamp is is only tracked with enough precision for
11411 524,287ms (8m44s287ms). If the request is queued long enough to where the
11412 adjusted timestamp exceeds this value, it will be misidentified as highest
11413 priority. Thus it is important to set "timeout queue" to a value, where when
11414 combined with the offset, does not exceed this limit.
11415
Christopher Faulet76c09ef2017-09-21 11:03:52 +020011416 The "send-spoe-group" is used to trigger sending of a group of SPOE
11417 messages. To do so, the SPOE engine used to send messages must be defined, as
11418 well as the SPOE group to send. Of course, the SPOE engine must refer to an
11419 existing SPOE filter. If not engine name is provided on the SPOE filter line,
11420 the SPOE agent name must be used.
11421
11422 <engine-name> The SPOE engine name.
11423
11424 <group-name> The SPOE group name as specified in the engine configuration.
11425
Christopher Faulet579d83b2019-11-22 15:34:17 +010011426 The "use-service" is used to executes a TCP service which will reply to the
11427 request and stop the evaluation of the rules. This service may choose to
11428 reply by sending any valid response or it may immediately close the
11429 connection without sending anything. Outside natives services, it is possible
11430 to write your own services in Lua. No further "tcp-request" rules are
11431 evaluated.
11432
11433 Example:
11434 tcp-request content use-service lua.deny { src -f /etc/haproxy/blacklist.lst }
11435
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011436 Example:
11437
11438 tcp-request content set-var(sess.my_var) src
Christopher Faulet85d79c92016-11-09 16:54:56 +010011439 tcp-request content unset-var(sess.my_var2)
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011440
Willy Tarreau62644772008-07-16 18:36:06 +020011441 Example:
Willy Tarreaue9656522010-08-17 15:40:09 +020011442 # Accept HTTP requests containing a Host header saying "example.com"
11443 # and reject everything else.
11444 acl is_host_com hdr(Host) -i example.com
11445 tcp-request inspect-delay 30s
Willy Tarreauc0239e02012-04-16 14:42:55 +020011446 tcp-request content accept if is_host_com
Willy Tarreaue9656522010-08-17 15:40:09 +020011447 tcp-request content reject
11448
11449 Example:
Willy Tarreau62644772008-07-16 18:36:06 +020011450 # reject SMTP connection if client speaks first
11451 tcp-request inspect-delay 30s
11452 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011453 tcp-request content reject if content_present
Willy Tarreau62644772008-07-16 18:36:06 +020011454
11455 # Forward HTTPS connection only if client speaks
11456 tcp-request inspect-delay 30s
11457 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011458 tcp-request content accept if content_present
Willy Tarreaue9656522010-08-17 15:40:09 +020011459 tcp-request content reject
11460
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011461 Example:
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030011462 # Track the last IP(stick-table type string) from X-Forwarded-For
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011463 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +020011464 tcp-request content track-sc0 hdr(x-forwarded-for,-1)
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030011465 # Or track the last IP(stick-table type ip|ipv6) from X-Forwarded-For
11466 tcp-request content track-sc0 req.hdr_ip(x-forwarded-for,-1)
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011467
11468 Example:
11469 # track request counts per "base" (concatenation of Host+URL)
11470 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +020011471 tcp-request content track-sc0 base table req-rate
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011472
Willy Tarreaue9656522010-08-17 15:40:09 +020011473 Example: track per-frontend and per-backend counters, block abusers at the
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030011474 frontend when the backend detects abuse(and marks gpc0).
Willy Tarreaue9656522010-08-17 15:40:09 +020011475
11476 frontend http
Davor Ocelice9ed2812017-12-25 17:49:28 +010011477 # Use General Purpose Counter 0 in SC0 as a global abuse counter
Willy Tarreaue9656522010-08-17 15:40:09 +020011478 # protecting all our sites
11479 stick-table type ip size 1m expire 5m store gpc0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011480 tcp-request connection track-sc0 src
11481 tcp-request connection reject if { sc0_get_gpc0 gt 0 }
Willy Tarreaue9656522010-08-17 15:40:09 +020011482 ...
11483 use_backend http_dynamic if { path_end .php }
11484
11485 backend http_dynamic
11486 # if a source makes too fast requests to this dynamic site (tracked
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011487 # by SC1), block it globally in the frontend.
Willy Tarreaue9656522010-08-17 15:40:09 +020011488 stick-table type ip size 1m expire 5m store http_req_rate(10s)
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011489 acl click_too_fast sc1_http_req_rate gt 10
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030011490 acl mark_as_abuser sc0_inc_gpc0(http) gt 0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011491 tcp-request content track-sc1 src
Willy Tarreaue9656522010-08-17 15:40:09 +020011492 tcp-request content reject if click_too_fast mark_as_abuser
Willy Tarreau62644772008-07-16 18:36:06 +020011493
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011494 See section 7 about ACL usage.
Willy Tarreau62644772008-07-16 18:36:06 +020011495
Jarno Huuskonen95b012b2017-04-06 13:59:14 +030011496 See also : "tcp-request connection", "tcp-request session",
11497 "tcp-request inspect-delay", and "http-request".
Willy Tarreau62644772008-07-16 18:36:06 +020011498
11499
11500tcp-request inspect-delay <timeout>
11501 Set the maximum allowed time to wait for data during content inspection
11502 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +020011503 no | yes | yes | yes
Willy Tarreau62644772008-07-16 18:36:06 +020011504 Arguments :
11505 <timeout> is the timeout value specified in milliseconds by default, but
11506 can be in any other unit if the number is suffixed by the unit,
11507 as explained at the top of this document.
11508
11509 People using haproxy primarily as a TCP relay are often worried about the
11510 risk of passing any type of protocol to a server without any analysis. In
11511 order to be able to analyze the request contents, we must first withhold
11512 the data then analyze them. This statement simply enables withholding of
11513 data for at most the specified amount of time.
11514
Willy Tarreaufb356202010-08-03 14:02:05 +020011515 TCP content inspection applies very early when a connection reaches a
11516 frontend, then very early when the connection is forwarded to a backend. This
11517 means that a connection may experience a first delay in the frontend and a
11518 second delay in the backend if both have tcp-request rules.
11519
Willy Tarreau62644772008-07-16 18:36:06 +020011520 Note that when performing content inspection, haproxy will evaluate the whole
11521 rules for every new chunk which gets in, taking into account the fact that
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011522 those data are partial. If no rule matches before the aforementioned delay,
Willy Tarreau62644772008-07-16 18:36:06 +020011523 a last check is performed upon expiration, this time considering that the
Willy Tarreaud869b242009-03-15 14:43:58 +010011524 contents are definitive. If no delay is set, haproxy will not wait at all
11525 and will immediately apply a verdict based on the available information.
11526 Obviously this is unlikely to be very useful and might even be racy, so such
11527 setups are not recommended.
Willy Tarreau62644772008-07-16 18:36:06 +020011528
11529 As soon as a rule matches, the request is released and continues as usual. If
11530 the timeout is reached and no rule matches, the default policy will be to let
11531 it pass through unaffected.
11532
11533 For most protocols, it is enough to set it to a few seconds, as most clients
11534 send the full request immediately upon connection. Add 3 or more seconds to
11535 cover TCP retransmits but that's all. For some protocols, it may make sense
Willy Tarreaud72758d2010-01-12 10:42:19 +010011536 to use large values, for instance to ensure that the client never talks
Davor Ocelice9ed2812017-12-25 17:49:28 +010011537 before the server (e.g. SMTP), or to wait for a client to talk before passing
11538 data to the server (e.g. SSL). Note that the client timeout must cover at
Willy Tarreaub824b002010-09-29 16:36:16 +020011539 least the inspection delay, otherwise it will expire first. If the client
11540 closes the connection or if the buffer is full, the delay immediately expires
11541 since the contents will not be able to change anymore.
Willy Tarreau62644772008-07-16 18:36:06 +020011542
Willy Tarreau55165fe2009-05-10 12:02:55 +020011543 See also : "tcp-request content accept", "tcp-request content reject",
Willy Tarreau62644772008-07-16 18:36:06 +020011544 "timeout client".
11545
11546
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011547tcp-response content <action> [{if | unless} <condition>]
11548 Perform an action on a session response depending on a layer 4-7 condition
11549 May be used in sections : defaults | frontend | listen | backend
11550 no | no | yes | yes
11551 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +020011552 <action> defines the action to perform if the condition applies. See
11553 below.
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011554
11555 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
11556
Davor Ocelice9ed2812017-12-25 17:49:28 +010011557 Response contents can be analyzed at an early stage of response processing
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011558 called "TCP content inspection". During this stage, ACL-based rules are
11559 evaluated every time the response contents are updated, until either an
Willy Tarreaucc1e04b2013-09-11 23:20:29 +020011560 "accept", "close" or a "reject" rule matches, or a TCP response inspection
11561 delay is set and expires with no matching rule.
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011562
11563 Most often, these decisions will consider a protocol recognition or validity.
11564
11565 Content-based rules are evaluated in their exact declaration order. If no
11566 rule matches or if there is no rule, the default action is to accept the
11567 contents. There is no specific limit to the number of rules which may be
11568 inserted.
11569
Thierry FOURNIER236657b2015-08-19 08:25:14 +020011570 Several types of actions are supported :
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011571 - accept :
11572 accepts the response if the condition is true (when used with "if")
11573 or false (when used with "unless"). The first such rule executed ends
11574 the rules evaluation.
11575
Willy Tarreaucc1e04b2013-09-11 23:20:29 +020011576 - close :
11577 immediately closes the connection with the server if the condition is
11578 true (when used with "if"), or false (when used with "unless"). The
11579 first such rule executed ends the rules evaluation. The main purpose of
11580 this action is to force a connection to be finished between a client
11581 and a server after an exchange when the application protocol expects
11582 some long time outs to elapse first. The goal is to eliminate idle
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011583 connections which take significant resources on servers with certain
Willy Tarreaucc1e04b2013-09-11 23:20:29 +020011584 protocols.
11585
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011586 - reject :
11587 rejects the response if the condition is true (when used with "if")
11588 or false (when used with "unless"). The first such rule executed ends
Jamie Gloudonaaa21002012-08-25 00:18:33 -040011589 the rules evaluation. Rejected session are immediately closed.
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011590
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011591 - set-var(<var-name>) <expr>
11592 Sets a variable.
11593
Christopher Faulet85d79c92016-11-09 16:54:56 +010011594 - unset-var(<var-name>)
11595 Unsets a variable.
11596
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +020011597 - sc-inc-gpc0(<sc-id>):
11598 This action increments the GPC0 counter according to the sticky
11599 counter designated by <sc-id>. If an error occurs, this action fails
11600 silently and the actions evaluation continues.
11601
Frédéric Lécaille6778b272018-01-29 15:22:53 +010011602 - sc-inc-gpc1(<sc-id>):
11603 This action increments the GPC1 counter according to the sticky
11604 counter designated by <sc-id>. If an error occurs, this action fails
11605 silently and the actions evaluation continues.
11606
Cédric Dufour0d7712d2019-11-06 18:38:53 +010011607 - sc-set-gpt0(<sc-id>) { <int> | <expr> }
11608 This action sets the 32-bit unsigned GPT0 tag according to the sticky
11609 counter designated by <sc-id> and the value of <int>/<expr>. The
11610 expected result is a boolean. If an error occurs, this action silently
11611 fails and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +020011612
Willy Tarreau2d392c22015-08-24 01:43:45 +020011613 - "silent-drop" :
11614 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +010011615 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +020011616 to prevent the client from being notified. The effect it then that the
11617 client still sees an established connection while there's none on
11618 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
11619 except that it doesn't use any local resource at all on the machine
11620 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +010011621 slow down stronger attackers. It is important to understand the impact
11622 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +020011623 client and HAProxy (firewalls, proxies, load balancers) will also keep
11624 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +010011625 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +020011626 TCP_REPAIR socket option is used to block the emission of a TCP
11627 reset. On other systems, the socket's TTL is reduced to 1 so that the
11628 TCP reset doesn't pass the first router, though it's still delivered to
11629 local networks. Do not use it unless you fully understand how it works.
11630
Christopher Faulet76c09ef2017-09-21 11:03:52 +020011631 - send-spoe-group <engine-name> <group-name>
11632 Send a group of SPOE messages.
11633
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011634 Note that the "if/unless" condition is optional. If no condition is set on
11635 the action, it is simply performed unconditionally. That can be useful for
11636 for changing the default action to a reject.
11637
Jamie Gloudonaaa21002012-08-25 00:18:33 -040011638 It is perfectly possible to match layer 7 contents with "tcp-response
11639 content" rules, but then it is important to ensure that a full response has
11640 been buffered, otherwise no contents will match. In order to achieve this,
11641 the best solution involves detecting the HTTP protocol during the inspection
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011642 period.
11643
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011644 The "set-var" is used to set the content of a variable. The variable is
11645 declared inline.
11646
Daniel Schneller0b547052016-03-21 20:46:57 +010011647 <var-name> The name of the variable starts with an indication about
11648 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010011649 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010011650 "sess" : the variable is shared with the whole session
11651 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011652 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +010011653 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011654 processing
Daniel Schneller0b547052016-03-21 20:46:57 +010011655 "res" : the variable is shared only during response
11656 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011657 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +010011658 The name may only contain characters 'a-z', 'A-Z', '0-9',
11659 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011660
11661 <expr> Is a standard HAProxy expression formed by a sample-fetch
11662 followed by some converters.
11663
11664 Example:
11665
11666 tcp-request content set-var(sess.my_var) src
11667
Christopher Faulet85d79c92016-11-09 16:54:56 +010011668 The "unset-var" is used to unset a variable. See above for details about
11669 <var-name>.
11670
11671 Example:
11672
11673 tcp-request content unset-var(sess.my_var)
11674
Christopher Faulet76c09ef2017-09-21 11:03:52 +020011675 The "send-spoe-group" is used to trigger sending of a group of SPOE
11676 messages. To do so, the SPOE engine used to send messages must be defined, as
11677 well as the SPOE group to send. Of course, the SPOE engine must refer to an
11678 existing SPOE filter. If not engine name is provided on the SPOE filter line,
11679 the SPOE agent name must be used.
11680
11681 <engine-name> The SPOE engine name.
11682
11683 <group-name> The SPOE group name as specified in the engine configuration.
11684
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011685 See section 7 about ACL usage.
11686
11687 See also : "tcp-request content", "tcp-response inspect-delay"
11688
11689
Willy Tarreau4f614292016-10-21 17:49:36 +020011690tcp-request session <action> [{if | unless} <condition>]
11691 Perform an action on a validated session depending on a layer 5 condition
11692 May be used in sections : defaults | frontend | listen | backend
11693 no | yes | yes | no
11694 Arguments :
11695 <action> defines the action to perform if the condition applies. See
11696 below.
11697
11698 <condition> is a standard layer5-only ACL-based condition (see section 7).
11699
Davor Ocelice9ed2812017-12-25 17:49:28 +010011700 Once a session is validated, (i.e. after all handshakes have been completed),
Willy Tarreau4f614292016-10-21 17:49:36 +020011701 it is possible to evaluate some conditions to decide whether this session
11702 must be accepted or dropped or have its counters tracked. Those conditions
11703 cannot make use of any data contents because no buffers are allocated yet and
11704 the processing cannot wait at this stage. The main use case it to copy some
11705 early information into variables (since variables are accessible in the
11706 session), or to keep track of some information collected after the handshake,
11707 such as SSL-level elements (SNI, ciphers, client cert's CN) or information
Davor Ocelice9ed2812017-12-25 17:49:28 +010011708 from the PROXY protocol header (e.g. track a source forwarded this way). The
Willy Tarreau4f614292016-10-21 17:49:36 +020011709 extracted information can thus be copied to a variable or tracked using
11710 "track-sc" rules. Of course it is also possible to decide to accept/reject as
11711 with other rulesets. Most operations performed here could also be performed
11712 in "tcp-request content" rules, except that in HTTP these rules are evaluated
11713 for each new request, and that might not always be acceptable. For example a
11714 rule might increment a counter on each evaluation. It would also be possible
11715 that a country is resolved by geolocation from the source IP address,
11716 assigned to a session-wide variable, then the source address rewritten from
11717 an HTTP header for all requests. If some contents need to be inspected in
11718 order to take the decision, the "tcp-request content" statements must be used
11719 instead.
11720
11721 The "tcp-request session" rules are evaluated in their exact declaration
11722 order. If no rule matches or if there is no rule, the default action is to
11723 accept the incoming session. There is no specific limit to the number of
11724 rules which may be inserted.
11725
11726 Several types of actions are supported :
11727 - accept : the request is accepted
11728 - reject : the request is rejected and the connection is closed
11729 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
11730 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +010011731 - sc-inc-gpc1(<sc-id>)
Cédric Dufour0d7712d2019-11-06 18:38:53 +010011732 - sc-set-gpt0(<sc-id>) { <int> | <expr> }
Willy Tarreau4f614292016-10-21 17:49:36 +020011733 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +010011734 - unset-var(<var-name>)
Willy Tarreau4f614292016-10-21 17:49:36 +020011735 - silent-drop
11736
11737 These actions have the same meaning as their respective counter-parts in
11738 "tcp-request connection" and "tcp-request content", so please refer to these
11739 sections for a complete description.
11740
11741 Note that the "if/unless" condition is optional. If no condition is set on
11742 the action, it is simply performed unconditionally. That can be useful for
11743 "track-sc*" actions as well as for changing the default action to a reject.
11744
11745 Example: track the original source address by default, or the one advertised
11746 in the PROXY protocol header for connection coming from the local
11747 proxies. The first connection-level rule enables receipt of the
11748 PROXY protocol for these ones, the second rule tracks whatever
11749 address we decide to keep after optional decoding.
11750
11751 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
11752 tcp-request session track-sc0 src
11753
11754 Example: accept all sessions from white-listed hosts, reject too fast
11755 sessions without counting them, and track accepted sessions.
11756 This results in session rate being capped from abusive sources.
11757
11758 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
11759 tcp-request session reject if { src_sess_rate gt 10 }
11760 tcp-request session track-sc0 src
11761
11762 Example: accept all sessions from white-listed hosts, count all other
11763 sessions and reject too fast ones. This results in abusive ones
11764 being blocked as long as they don't slow down.
11765
11766 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
11767 tcp-request session track-sc0 src
11768 tcp-request session reject if { sc0_sess_rate gt 10 }
11769
11770 See section 7 about ACL usage.
11771
11772 See also : "tcp-request connection", "tcp-request content", "stick-table"
11773
11774
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011775tcp-response inspect-delay <timeout>
11776 Set the maximum allowed time to wait for a response during content inspection
11777 May be used in sections : defaults | frontend | listen | backend
11778 no | no | yes | yes
11779 Arguments :
11780 <timeout> is the timeout value specified in milliseconds by default, but
11781 can be in any other unit if the number is suffixed by the unit,
11782 as explained at the top of this document.
11783
11784 See also : "tcp-response content", "tcp-request inspect-delay".
11785
11786
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010011787timeout check <timeout>
11788 Set additional check timeout, but only after a connection has been already
11789 established.
11790
11791 May be used in sections: defaults | frontend | listen | backend
11792 yes | no | yes | yes
11793 Arguments:
11794 <timeout> is the timeout value specified in milliseconds by default, but
11795 can be in any other unit if the number is suffixed by the unit,
11796 as explained at the top of this document.
11797
11798 If set, haproxy uses min("timeout connect", "inter") as a connect timeout
11799 for check and "timeout check" as an additional read timeout. The "min" is
Davor Ocelice9ed2812017-12-25 17:49:28 +010011800 used so that people running with *very* long "timeout connect" (e.g. those
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010011801 who needed this due to the queue or tarpit) do not slow down their checks.
Willy Tarreaud7550a22010-02-10 05:10:19 +010011802 (Please also note that there is no valid reason to have such long connect
11803 timeouts, because "timeout queue" and "timeout tarpit" can always be used to
11804 avoid that).
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010011805
11806 If "timeout check" is not set haproxy uses "inter" for complete check
11807 timeout (connect + read) exactly like all <1.3.15 version.
11808
11809 In most cases check request is much simpler and faster to handle than normal
11810 requests and people may want to kick out laggy servers so this timeout should
Willy Tarreau41a340d2008-01-22 12:25:31 +010011811 be smaller than "timeout server".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010011812
11813 This parameter is specific to backends, but can be specified once for all in
11814 "defaults" sections. This is in fact one of the easiest solutions not to
11815 forget about it.
11816
Willy Tarreau41a340d2008-01-22 12:25:31 +010011817 See also: "timeout connect", "timeout queue", "timeout server",
11818 "timeout tarpit".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010011819
11820
Willy Tarreau0ba27502007-12-24 16:55:16 +010011821timeout client <timeout>
Willy Tarreau0ba27502007-12-24 16:55:16 +010011822 Set the maximum inactivity time on the client side.
11823 May be used in sections : defaults | frontend | listen | backend
11824 yes | yes | yes | no
11825 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010011826 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010011827 can be in any other unit if the number is suffixed by the unit,
11828 as explained at the top of this document.
11829
11830 The inactivity timeout applies when the client is expected to acknowledge or
11831 send data. In HTTP mode, this timeout is particularly important to consider
11832 during the first phase, when the client sends the request, and during the
Baptiste Assmann2e1941e2016-03-06 23:24:12 +010011833 response while it is reading data sent by the server. That said, for the
11834 first phase, it is preferable to set the "timeout http-request" to better
11835 protect HAProxy from Slowloris like attacks. The value is specified in
11836 milliseconds by default, but can be in any other unit if the number is
Willy Tarreau0ba27502007-12-24 16:55:16 +010011837 suffixed by the unit, as specified at the top of this document. In TCP mode
11838 (and to a lesser extent, in HTTP mode), it is highly recommended that the
11839 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010011840 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +010011841 losses by specifying timeouts that are slightly above multiples of 3 seconds
Davor Ocelice9ed2812017-12-25 17:49:28 +010011842 (e.g. 4 or 5 seconds). If some long-lived sessions are mixed with short-lived
11843 sessions (e.g. WebSocket and HTTP), it's worth considering "timeout tunnel",
Willy Tarreau05cdd962014-05-10 14:30:07 +020011844 which overrides "timeout client" and "timeout server" for tunnels, as well as
11845 "timeout client-fin" for half-closed connections.
Willy Tarreau0ba27502007-12-24 16:55:16 +010011846
11847 This parameter is specific to frontends, but can be specified once for all in
11848 "defaults" sections. This is in fact one of the easiest solutions not to
11849 forget about it. An unspecified timeout results in an infinite timeout, which
11850 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050011851 during startup because it may result in accumulation of expired sessions in
Willy Tarreau0ba27502007-12-24 16:55:16 +010011852 the system if the system's timeouts are not configured either.
11853
Willy Tarreau95c4e142017-11-26 12:18:55 +010011854 This also applies to HTTP/2 connections, which will be closed with GOAWAY.
Lukas Tribus75df9d72017-11-24 19:05:12 +010011855
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020011856 See also : "timeout server", "timeout tunnel", "timeout http-request".
Willy Tarreau0ba27502007-12-24 16:55:16 +010011857
Willy Tarreau0ba27502007-12-24 16:55:16 +010011858
Willy Tarreau05cdd962014-05-10 14:30:07 +020011859timeout client-fin <timeout>
11860 Set the inactivity timeout on the client side for half-closed connections.
11861 May be used in sections : defaults | frontend | listen | backend
11862 yes | yes | yes | no
11863 Arguments :
11864 <timeout> is the timeout value specified in milliseconds by default, but
11865 can be in any other unit if the number is suffixed by the unit,
11866 as explained at the top of this document.
11867
11868 The inactivity timeout applies when the client is expected to acknowledge or
11869 send data while one direction is already shut down. This timeout is different
11870 from "timeout client" in that it only applies to connections which are closed
11871 in one direction. This is particularly useful to avoid keeping connections in
11872 FIN_WAIT state for too long when clients do not disconnect cleanly. This
11873 problem is particularly common long connections such as RDP or WebSocket.
11874 Note that this timeout can override "timeout tunnel" when a connection shuts
Willy Tarreau599391a2017-11-24 10:16:00 +010011875 down in one direction. It is applied to idle HTTP/2 connections once a GOAWAY
11876 frame was sent, often indicating an expectation that the connection quickly
11877 ends.
Willy Tarreau05cdd962014-05-10 14:30:07 +020011878
11879 This parameter is specific to frontends, but can be specified once for all in
11880 "defaults" sections. By default it is not set, so half-closed connections
11881 will use the other timeouts (timeout.client or timeout.tunnel).
11882
11883 See also : "timeout client", "timeout server-fin", and "timeout tunnel".
11884
11885
Willy Tarreau0ba27502007-12-24 16:55:16 +010011886timeout connect <timeout>
Willy Tarreau0ba27502007-12-24 16:55:16 +010011887 Set the maximum time to wait for a connection attempt to a server to succeed.
11888 May be used in sections : defaults | frontend | listen | backend
11889 yes | no | yes | yes
11890 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010011891 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010011892 can be in any other unit if the number is suffixed by the unit,
11893 as explained at the top of this document.
11894
11895 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010011896 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +010011897 cover one or several TCP packet losses by specifying timeouts that are
Davor Ocelice9ed2812017-12-25 17:49:28 +010011898 slightly above multiples of 3 seconds (e.g. 4 or 5 seconds). By default, the
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010011899 connect timeout also presets both queue and tarpit timeouts to the same value
11900 if these have not been specified.
Willy Tarreau0ba27502007-12-24 16:55:16 +010011901
11902 This parameter is specific to backends, but can be specified once for all in
11903 "defaults" sections. This is in fact one of the easiest solutions not to
11904 forget about it. An unspecified timeout results in an infinite timeout, which
11905 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050011906 during startup because it may result in accumulation of failed sessions in
Willy Tarreau0ba27502007-12-24 16:55:16 +010011907 the system if the system's timeouts are not configured either.
11908
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020011909 See also: "timeout check", "timeout queue", "timeout server", "timeout tarpit".
Willy Tarreau0ba27502007-12-24 16:55:16 +010011910
Willy Tarreau0ba27502007-12-24 16:55:16 +010011911
Willy Tarreaub16a5742010-01-10 14:46:16 +010011912timeout http-keep-alive <timeout>
11913 Set the maximum allowed time to wait for a new HTTP request to appear
11914 May be used in sections : defaults | frontend | listen | backend
11915 yes | yes | yes | yes
11916 Arguments :
11917 <timeout> is the timeout value specified in milliseconds by default, but
11918 can be in any other unit if the number is suffixed by the unit,
11919 as explained at the top of this document.
11920
11921 By default, the time to wait for a new request in case of keep-alive is set
11922 by "timeout http-request". However this is not always convenient because some
11923 people want very short keep-alive timeouts in order to release connections
11924 faster, and others prefer to have larger ones but still have short timeouts
11925 once the request has started to present itself.
11926
11927 The "http-keep-alive" timeout covers these needs. It will define how long to
11928 wait for a new HTTP request to start coming after a response was sent. Once
11929 the first byte of request has been seen, the "http-request" timeout is used
11930 to wait for the complete request to come. Note that empty lines prior to a
11931 new request do not refresh the timeout and are not counted as a new request.
11932
11933 There is also another difference between the two timeouts : when a connection
11934 expires during timeout http-keep-alive, no error is returned, the connection
11935 just closes. If the connection expires in "http-request" while waiting for a
11936 connection to complete, a HTTP 408 error is returned.
11937
11938 In general it is optimal to set this value to a few tens to hundreds of
11939 milliseconds, to allow users to fetch all objects of a page at once but
Davor Ocelice9ed2812017-12-25 17:49:28 +010011940 without waiting for further clicks. Also, if set to a very small value (e.g.
Willy Tarreaub16a5742010-01-10 14:46:16 +010011941 1 millisecond) it will probably only accept pipelined requests but not the
11942 non-pipelined ones. It may be a nice trade-off for very large sites running
Patrick Mézard2382ad62010-05-09 10:43:32 +020011943 with tens to hundreds of thousands of clients.
Willy Tarreaub16a5742010-01-10 14:46:16 +010011944
11945 If this parameter is not set, the "http-request" timeout applies, and if both
11946 are not set, "timeout client" still applies at the lower level. It should be
11947 set in the frontend to take effect, unless the frontend is in TCP mode, in
11948 which case the HTTP backend's timeout will be used.
11949
Willy Tarreau95c4e142017-11-26 12:18:55 +010011950 When using HTTP/2 "timeout client" is applied instead. This is so we can keep
11951 using short keep-alive timeouts in HTTP/1.1 while using longer ones in HTTP/2
Lukas Tribus75df9d72017-11-24 19:05:12 +010011952 (where we only have one connection per client and a connection setup).
11953
Willy Tarreaub16a5742010-01-10 14:46:16 +010011954 See also : "timeout http-request", "timeout client".
11955
11956
Willy Tarreau036fae02008-01-06 13:24:40 +010011957timeout http-request <timeout>
11958 Set the maximum allowed time to wait for a complete HTTP request
11959 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaucd7afc02009-07-12 10:03:17 +020011960 yes | yes | yes | yes
Willy Tarreau036fae02008-01-06 13:24:40 +010011961 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010011962 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau036fae02008-01-06 13:24:40 +010011963 can be in any other unit if the number is suffixed by the unit,
11964 as explained at the top of this document.
11965
11966 In order to offer DoS protection, it may be required to lower the maximum
11967 accepted time to receive a complete HTTP request without affecting the client
11968 timeout. This helps protecting against established connections on which
11969 nothing is sent. The client timeout cannot offer a good protection against
11970 this abuse because it is an inactivity timeout, which means that if the
11971 attacker sends one character every now and then, the timeout will not
11972 trigger. With the HTTP request timeout, no matter what speed the client
Willy Tarreau2705a612014-05-23 17:38:34 +020011973 types, the request will be aborted if it does not complete in time. When the
11974 timeout expires, an HTTP 408 response is sent to the client to inform it
11975 about the problem, and the connection is closed. The logs will report
11976 termination codes "cR". Some recent browsers are having problems with this
Davor Ocelice9ed2812017-12-25 17:49:28 +010011977 standard, well-documented behavior, so it might be needed to hide the 408
Willy Tarreau0f228a02015-05-01 15:37:53 +020011978 code using "option http-ignore-probes" or "errorfile 408 /dev/null". See
11979 more details in the explanations of the "cR" termination code in section 8.5.
Willy Tarreau036fae02008-01-06 13:24:40 +010011980
Baptiste Assmanneccdf432015-10-28 13:49:01 +010011981 By default, this timeout only applies to the header part of the request,
11982 and not to any data. As soon as the empty line is received, this timeout is
11983 not used anymore. When combined with "option http-buffer-request", this
11984 timeout also applies to the body of the request..
11985 It is used again on keep-alive connections to wait for a second
Willy Tarreaub16a5742010-01-10 14:46:16 +010011986 request if "timeout http-keep-alive" is not set.
Willy Tarreau036fae02008-01-06 13:24:40 +010011987
11988 Generally it is enough to set it to a few seconds, as most clients send the
11989 full request immediately upon connection. Add 3 or more seconds to cover TCP
Davor Ocelice9ed2812017-12-25 17:49:28 +010011990 retransmits but that's all. Setting it to very low values (e.g. 50 ms) will
Willy Tarreau036fae02008-01-06 13:24:40 +010011991 generally work on local networks as long as there are no packet losses. This
11992 will prevent people from sending bare HTTP requests using telnet.
11993
11994 If this parameter is not set, the client timeout still applies between each
Willy Tarreaucd7afc02009-07-12 10:03:17 +020011995 chunk of the incoming request. It should be set in the frontend to take
11996 effect, unless the frontend is in TCP mode, in which case the HTTP backend's
11997 timeout will be used.
Willy Tarreau036fae02008-01-06 13:24:40 +010011998
Willy Tarreau0f228a02015-05-01 15:37:53 +020011999 See also : "errorfile", "http-ignore-probes", "timeout http-keep-alive", and
Baptiste Assmanneccdf432015-10-28 13:49:01 +010012000 "timeout client", "option http-buffer-request".
Willy Tarreau036fae02008-01-06 13:24:40 +010012001
Willy Tarreau844e3c52008-01-11 16:28:18 +010012002
12003timeout queue <timeout>
12004 Set the maximum time to wait in the queue for a connection slot to be free
12005 May be used in sections : defaults | frontend | listen | backend
12006 yes | no | yes | yes
12007 Arguments :
12008 <timeout> is the timeout value specified in milliseconds by default, but
12009 can be in any other unit if the number is suffixed by the unit,
12010 as explained at the top of this document.
12011
12012 When a server's maxconn is reached, connections are left pending in a queue
12013 which may be server-specific or global to the backend. In order not to wait
12014 indefinitely, a timeout is applied to requests pending in the queue. If the
12015 timeout is reached, it is considered that the request will almost never be
12016 served, so it is dropped and a 503 error is returned to the client.
12017
12018 The "timeout queue" statement allows to fix the maximum time for a request to
12019 be left pending in a queue. If unspecified, the same value as the backend's
12020 connection timeout ("timeout connect") is used, for backwards compatibility
12021 with older versions with no "timeout queue" parameter.
12022
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020012023 See also : "timeout connect".
Willy Tarreau844e3c52008-01-11 16:28:18 +010012024
12025
12026timeout server <timeout>
Willy Tarreau844e3c52008-01-11 16:28:18 +010012027 Set the maximum inactivity time on the server side.
12028 May be used in sections : defaults | frontend | listen | backend
12029 yes | no | yes | yes
12030 Arguments :
12031 <timeout> is the timeout value specified in milliseconds by default, but
12032 can be in any other unit if the number is suffixed by the unit,
12033 as explained at the top of this document.
12034
12035 The inactivity timeout applies when the server is expected to acknowledge or
12036 send data. In HTTP mode, this timeout is particularly important to consider
12037 during the first phase of the server's response, when it has to send the
12038 headers, as it directly represents the server's processing time for the
12039 request. To find out what value to put there, it's often good to start with
12040 what would be considered as unacceptable response times, then check the logs
12041 to observe the response time distribution, and adjust the value accordingly.
12042
12043 The value is specified in milliseconds by default, but can be in any other
12044 unit if the number is suffixed by the unit, as specified at the top of this
12045 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
12046 recommended that the client timeout remains equal to the server timeout in
12047 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010012048 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +010012049 packet losses by specifying timeouts that are slightly above multiples of 3
Davor Ocelice9ed2812017-12-25 17:49:28 +010012050 seconds (e.g. 4 or 5 seconds minimum). If some long-lived sessions are mixed
12051 with short-lived sessions (e.g. WebSocket and HTTP), it's worth considering
Willy Tarreauce887fd2012-05-12 12:50:00 +020012052 "timeout tunnel", which overrides "timeout client" and "timeout server" for
12053 tunnels.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012054
12055 This parameter is specific to backends, but can be specified once for all in
12056 "defaults" sections. This is in fact one of the easiest solutions not to
12057 forget about it. An unspecified timeout results in an infinite timeout, which
12058 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050012059 during startup because it may result in accumulation of expired sessions in
Willy Tarreau844e3c52008-01-11 16:28:18 +010012060 the system if the system's timeouts are not configured either.
12061
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020012062 See also : "timeout client" and "timeout tunnel".
Willy Tarreau844e3c52008-01-11 16:28:18 +010012063
Willy Tarreau05cdd962014-05-10 14:30:07 +020012064
12065timeout server-fin <timeout>
12066 Set the inactivity timeout on the server side for half-closed connections.
12067 May be used in sections : defaults | frontend | listen | backend
12068 yes | no | yes | yes
12069 Arguments :
12070 <timeout> is the timeout value specified in milliseconds by default, but
12071 can be in any other unit if the number is suffixed by the unit,
12072 as explained at the top of this document.
12073
12074 The inactivity timeout applies when the server is expected to acknowledge or
12075 send data while one direction is already shut down. This timeout is different
12076 from "timeout server" in that it only applies to connections which are closed
12077 in one direction. This is particularly useful to avoid keeping connections in
12078 FIN_WAIT state for too long when a remote server does not disconnect cleanly.
12079 This problem is particularly common long connections such as RDP or WebSocket.
12080 Note that this timeout can override "timeout tunnel" when a connection shuts
12081 down in one direction. This setting was provided for completeness, but in most
12082 situations, it should not be needed.
12083
12084 This parameter is specific to backends, but can be specified once for all in
12085 "defaults" sections. By default it is not set, so half-closed connections
12086 will use the other timeouts (timeout.server or timeout.tunnel).
12087
12088 See also : "timeout client-fin", "timeout server", and "timeout tunnel".
12089
Willy Tarreau844e3c52008-01-11 16:28:18 +010012090
12091timeout tarpit <timeout>
Cyril Bonté78caf842010-03-10 22:41:43 +010012092 Set the duration for which tarpitted connections will be maintained
Willy Tarreau844e3c52008-01-11 16:28:18 +010012093 May be used in sections : defaults | frontend | listen | backend
12094 yes | yes | yes | yes
12095 Arguments :
12096 <timeout> is the tarpit duration specified in milliseconds by default, but
12097 can be in any other unit if the number is suffixed by the unit,
12098 as explained at the top of this document.
12099
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020012100 When a connection is tarpitted using "http-request tarpit", it is maintained
12101 open with no activity for a certain amount of time, then closed. "timeout
12102 tarpit" defines how long it will be maintained open.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012103
12104 The value is specified in milliseconds by default, but can be in any other
12105 unit if the number is suffixed by the unit, as specified at the top of this
12106 document. If unspecified, the same value as the backend's connection timeout
12107 ("timeout connect") is used, for backwards compatibility with older versions
Cyril Bonté78caf842010-03-10 22:41:43 +010012108 with no "timeout tarpit" parameter.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012109
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020012110 See also : "timeout connect".
Willy Tarreau844e3c52008-01-11 16:28:18 +010012111
12112
Willy Tarreauce887fd2012-05-12 12:50:00 +020012113timeout tunnel <timeout>
12114 Set the maximum inactivity time on the client and server side for tunnels.
12115 May be used in sections : defaults | frontend | listen | backend
12116 yes | no | yes | yes
12117 Arguments :
12118 <timeout> is the timeout value specified in milliseconds by default, but
12119 can be in any other unit if the number is suffixed by the unit,
12120 as explained at the top of this document.
12121
Jamie Gloudonaaa21002012-08-25 00:18:33 -040012122 The tunnel timeout applies when a bidirectional connection is established
Willy Tarreauce887fd2012-05-12 12:50:00 +020012123 between a client and a server, and the connection remains inactive in both
12124 directions. This timeout supersedes both the client and server timeouts once
12125 the connection becomes a tunnel. In TCP, this timeout is used as soon as no
Davor Ocelice9ed2812017-12-25 17:49:28 +010012126 analyzer remains attached to either connection (e.g. tcp content rules are
12127 accepted). In HTTP, this timeout is used when a connection is upgraded (e.g.
Willy Tarreauce887fd2012-05-12 12:50:00 +020012128 when switching to the WebSocket protocol, or forwarding a CONNECT request
12129 to a proxy), or after the first response when no keepalive/close option is
12130 specified.
12131
Willy Tarreau05cdd962014-05-10 14:30:07 +020012132 Since this timeout is usually used in conjunction with long-lived connections,
12133 it usually is a good idea to also set "timeout client-fin" to handle the
12134 situation where a client suddenly disappears from the net and does not
12135 acknowledge a close, or sends a shutdown and does not acknowledge pending
12136 data anymore. This can happen in lossy networks where firewalls are present,
12137 and is detected by the presence of large amounts of sessions in a FIN_WAIT
12138 state.
12139
Willy Tarreauce887fd2012-05-12 12:50:00 +020012140 The value is specified in milliseconds by default, but can be in any other
12141 unit if the number is suffixed by the unit, as specified at the top of this
12142 document. Whatever the expected normal idle time, it is a good practice to
12143 cover at least one or several TCP packet losses by specifying timeouts that
Davor Ocelice9ed2812017-12-25 17:49:28 +010012144 are slightly above multiples of 3 seconds (e.g. 4 or 5 seconds minimum).
Willy Tarreauce887fd2012-05-12 12:50:00 +020012145
12146 This parameter is specific to backends, but can be specified once for all in
12147 "defaults" sections. This is in fact one of the easiest solutions not to
12148 forget about it.
12149
12150 Example :
12151 defaults http
12152 option http-server-close
12153 timeout connect 5s
12154 timeout client 30s
Willy Tarreau05cdd962014-05-10 14:30:07 +020012155 timeout client-fin 30s
Willy Tarreauce887fd2012-05-12 12:50:00 +020012156 timeout server 30s
12157 timeout tunnel 1h # timeout to use with WebSocket and CONNECT
12158
Willy Tarreau05cdd962014-05-10 14:30:07 +020012159 See also : "timeout client", "timeout client-fin", "timeout server".
Willy Tarreauce887fd2012-05-12 12:50:00 +020012160
12161
Willy Tarreau844e3c52008-01-11 16:28:18 +010012162transparent (deprecated)
12163 Enable client-side transparent proxying
12164 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +010012165 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +010012166 Arguments : none
12167
12168 This keyword was introduced in order to provide layer 7 persistence to layer
12169 3 load balancers. The idea is to use the OS's ability to redirect an incoming
12170 connection for a remote address to a local process (here HAProxy), and let
12171 this process know what address was initially requested. When this option is
12172 used, sessions without cookies will be forwarded to the original destination
12173 IP address of the incoming request (which should match that of another
12174 equipment), while requests with cookies will still be forwarded to the
12175 appropriate server.
12176
12177 The "transparent" keyword is deprecated, use "option transparent" instead.
12178
12179 Note that contrary to a common belief, this option does NOT make HAProxy
12180 present the client's IP to the server when establishing the connection.
12181
Willy Tarreau844e3c52008-01-11 16:28:18 +010012182 See also: "option transparent"
12183
William Lallemanda73203e2012-03-12 12:48:57 +010012184unique-id-format <string>
12185 Generate a unique ID for each request.
12186 May be used in sections : defaults | frontend | listen | backend
12187 yes | yes | yes | no
12188 Arguments :
12189 <string> is a log-format string.
12190
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012191 This keyword creates a ID for each request using the custom log format. A
12192 unique ID is useful to trace a request passing through many components of
12193 a complex infrastructure. The newly created ID may also be logged using the
12194 %ID tag the log-format string.
William Lallemanda73203e2012-03-12 12:48:57 +010012195
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012196 The format should be composed from elements that are guaranteed to be
12197 unique when combined together. For instance, if multiple haproxy instances
12198 are involved, it might be important to include the node name. It is often
12199 needed to log the incoming connection's source and destination addresses
12200 and ports. Note that since multiple requests may be performed over the same
12201 connection, including a request counter may help differentiate them.
12202 Similarly, a timestamp may protect against a rollover of the counter.
12203 Logging the process ID will avoid collisions after a service restart.
William Lallemanda73203e2012-03-12 12:48:57 +010012204
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012205 It is recommended to use hexadecimal notation for many fields since it
12206 makes them more compact and saves space in logs.
William Lallemanda73203e2012-03-12 12:48:57 +010012207
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012208 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010012209
Julien Vehentf21be322014-03-07 08:27:34 -050012210 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010012211
12212 will generate:
12213
12214 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
12215
12216 See also: "unique-id-header"
12217
12218unique-id-header <name>
12219 Add a unique ID header in the HTTP request.
12220 May be used in sections : defaults | frontend | listen | backend
12221 yes | yes | yes | no
12222 Arguments :
12223 <name> is the name of the header.
12224
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012225 Add a unique-id header in the HTTP request sent to the server, using the
12226 unique-id-format. It can't work if the unique-id-format doesn't exist.
William Lallemanda73203e2012-03-12 12:48:57 +010012227
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012228 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010012229
Julien Vehentf21be322014-03-07 08:27:34 -050012230 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010012231 unique-id-header X-Unique-ID
12232
12233 will generate:
12234
12235 X-Unique-ID: 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
12236
12237 See also: "unique-id-format"
Willy Tarreau844e3c52008-01-11 16:28:18 +010012238
Willy Tarreauf51658d2014-04-23 01:21:56 +020012239use_backend <backend> [{if | unless} <condition>]
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020012240 Switch to a specific backend if/unless an ACL-based condition is matched.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012241 May be used in sections : defaults | frontend | listen | backend
12242 no | yes | yes | no
12243 Arguments :
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010012244 <backend> is the name of a valid backend or "listen" section, or a
12245 "log-format" string resolving to a backend name.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012246
Willy Tarreauf51658d2014-04-23 01:21:56 +020012247 <condition> is a condition composed of ACLs, as described in section 7. If
12248 it is omitted, the rule is unconditionally applied.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012249
12250 When doing content-switching, connections arrive on a frontend and are then
12251 dispatched to various backends depending on a number of conditions. The
12252 relation between the conditions and the backends is described with the
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020012253 "use_backend" keyword. While it is normally used with HTTP processing, it can
Davor Ocelice9ed2812017-12-25 17:49:28 +010012254 also be used in pure TCP, either without content using stateless ACLs (e.g.
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020012255 source address validation) or combined with a "tcp-request" rule to wait for
12256 some payload.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012257
12258 There may be as many "use_backend" rules as desired. All of these rules are
12259 evaluated in their declaration order, and the first one which matches will
12260 assign the backend.
12261
12262 In the first form, the backend will be used if the condition is met. In the
12263 second form, the backend will be used if the condition is not met. If no
12264 condition is valid, the backend defined with "default_backend" will be used.
12265 If no default backend is defined, either the servers in the same section are
12266 used (in case of a "listen" section) or, in case of a frontend, no server is
12267 used and a 503 service unavailable response is returned.
12268
Willy Tarreau51aecc72009-07-12 09:47:04 +020012269 Note that it is possible to switch from a TCP frontend to an HTTP backend. In
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012270 this case, either the frontend has already checked that the protocol is HTTP,
Willy Tarreau51aecc72009-07-12 09:47:04 +020012271 and backend processing will immediately follow, or the backend will wait for
12272 a complete HTTP request to get in. This feature is useful when a frontend
12273 must decode several protocols on a unique port, one of them being HTTP.
12274
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010012275 When <backend> is a simple name, it is resolved at configuration time, and an
12276 error is reported if the specified backend does not exist. If <backend> is
12277 a log-format string instead, no check may be done at configuration time, so
12278 the backend name is resolved dynamically at run time. If the resulting
12279 backend name does not correspond to any valid backend, no other rule is
12280 evaluated, and the default_backend directive is applied instead. Note that
12281 when using dynamic backend names, it is highly recommended to use a prefix
12282 that no other backend uses in order to ensure that an unauthorized backend
12283 cannot be forced from the request.
12284
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012285 It is worth mentioning that "use_backend" rules with an explicit name are
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010012286 used to detect the association between frontends and backends to compute the
12287 backend's "fullconn" setting. This cannot be done for dynamic names.
12288
12289 See also: "default_backend", "tcp-request", "fullconn", "log-format", and
12290 section 7 about ACLs.
Willy Tarreaud72758d2010-01-12 10:42:19 +010012291
Christopher Fauletb30b3102019-09-12 23:03:09 +020012292use-fcgi-app <name>
12293 Defines the FastCGI application to use for the backend.
12294 May be used in sections : defaults | frontend | listen | backend
12295 no | no | yes | yes
12296 Arguments :
12297 <name> is the name of the FastCGI application to use.
12298
12299 See section 10.1 about FastCGI application setup for details.
Willy Tarreau036fae02008-01-06 13:24:40 +010012300
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012301use-server <server> if <condition>
12302use-server <server> unless <condition>
12303 Only use a specific server if/unless an ACL-based condition is matched.
12304 May be used in sections : defaults | frontend | listen | backend
12305 no | no | yes | yes
12306 Arguments :
Jerome Magnin824186b2020-03-29 09:37:12 +020012307 <server> is the name of a valid server in the same backend section
12308 or a "log-format" string resolving to a server name.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012309
12310 <condition> is a condition composed of ACLs, as described in section 7.
12311
12312 By default, connections which arrive to a backend are load-balanced across
12313 the available servers according to the configured algorithm, unless a
12314 persistence mechanism such as a cookie is used and found in the request.
12315
12316 Sometimes it is desirable to forward a particular request to a specific
12317 server without having to declare a dedicated backend for this server. This
12318 can be achieved using the "use-server" rules. These rules are evaluated after
12319 the "redirect" rules and before evaluating cookies, and they have precedence
12320 on them. There may be as many "use-server" rules as desired. All of these
12321 rules are evaluated in their declaration order, and the first one which
12322 matches will assign the server.
12323
12324 If a rule designates a server which is down, and "option persist" is not used
12325 and no force-persist rule was validated, it is ignored and evaluation goes on
12326 with the next rules until one matches.
12327
12328 In the first form, the server will be used if the condition is met. In the
12329 second form, the server will be used if the condition is not met. If no
12330 condition is valid, the processing continues and the server will be assigned
12331 according to other persistence mechanisms.
12332
12333 Note that even if a rule is matched, cookie processing is still performed but
12334 does not assign the server. This allows prefixed cookies to have their prefix
12335 stripped.
12336
12337 The "use-server" statement works both in HTTP and TCP mode. This makes it
12338 suitable for use with content-based inspection. For instance, a server could
Lukas Tribusa267b5d2020-07-19 00:25:06 +020012339 be selected in a farm according to the TLS SNI field when using protocols with
12340 implicit TLS (also see "req_ssl_sni"). And if these servers have their weight
12341 set to zero, they will not be used for other traffic.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012342
12343 Example :
12344 # intercept incoming TLS requests based on the SNI field
12345 use-server www if { req_ssl_sni -i www.example.com }
12346 server www 192.168.0.1:443 weight 0
12347 use-server mail if { req_ssl_sni -i mail.example.com }
Lukas Tribusa267b5d2020-07-19 00:25:06 +020012348 server mail 192.168.0.1:465 weight 0
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012349 use-server imap if { req_ssl_sni -i imap.example.com }
Lukas Tribus98a3e3f2017-03-26 12:55:35 +000012350 server imap 192.168.0.1:993 weight 0
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012351 # all the rest is forwarded to this server
12352 server default 192.168.0.2:443 check
12353
Jerome Magnin824186b2020-03-29 09:37:12 +020012354 When <server> is a simple name, it is checked against existing servers in the
12355 configuration and an error is reported if the specified server does not exist.
12356 If it is a log-format, no check is performed when parsing the configuration,
12357 and if we can't resolve a valid server name at runtime but the use-server rule
Ilya Shipitsin11057a32020-06-21 21:18:27 +050012358 was conditioned by an ACL returning true, no other use-server rule is applied
Jerome Magnin824186b2020-03-29 09:37:12 +020012359 and we fall back to load balancing.
12360
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012361 See also: "use_backend", section 5 about server and section 7 about ACLs.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012362
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012363
Davor Ocelice9ed2812017-12-25 17:49:28 +0100123645. Bind and server options
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012365--------------------------
12366
12367The "bind", "server" and "default-server" keywords support a number of settings
12368depending on some build options and on the system HAProxy was built on. These
12369settings generally each consist in one word sometimes followed by a value,
12370written on the same line as the "bind" or "server" line. All these options are
12371described in this section.
12372
12373
123745.1. Bind options
12375-----------------
12376
12377The "bind" keyword supports a certain number of settings which are all passed
12378as arguments on the same line. The order in which those arguments appear makes
12379no importance, provided that they appear after the bind address. All of these
12380parameters are optional. Some of them consist in a single words (booleans),
12381while other ones expect a value after them. In this case, the value must be
12382provided immediately after the setting name.
12383
12384The currently supported settings are the following ones.
12385
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010012386accept-netscaler-cip <magic number>
12387 Enforces the use of the NetScaler Client IP insertion protocol over any
12388 connection accepted by any of the TCP sockets declared on the same line. The
12389 NetScaler Client IP insertion protocol dictates the layer 3/4 addresses of
12390 the incoming connection to be used everywhere an address is used, with the
12391 only exception of "tcp-request connection" rules which will only see the
12392 real connection address. Logs will reflect the addresses indicated in the
12393 protocol, unless it is violated, in which case the real address will still
12394 be used. This keyword combined with support from external components can be
12395 used as an efficient and reliable alternative to the X-Forwarded-For
Bertrand Jacquin90759682016-06-06 15:35:39 +010012396 mechanism which is not always reliable and not even always usable. See also
12397 "tcp-request connection expect-netscaler-cip" for a finer-grained setting of
12398 which client is allowed to use the protocol.
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010012399
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012400accept-proxy
12401 Enforces the use of the PROXY protocol over any connection accepted by any of
Willy Tarreau77992672014-06-14 11:06:17 +020012402 the sockets declared on the same line. Versions 1 and 2 of the PROXY protocol
12403 are supported and correctly detected. The PROXY protocol dictates the layer
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012404 3/4 addresses of the incoming connection to be used everywhere an address is
12405 used, with the only exception of "tcp-request connection" rules which will
12406 only see the real connection address. Logs will reflect the addresses
12407 indicated in the protocol, unless it is violated, in which case the real
Davor Ocelice9ed2812017-12-25 17:49:28 +010012408 address will still be used. This keyword combined with support from external
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012409 components can be used as an efficient and reliable alternative to the
12410 X-Forwarded-For mechanism which is not always reliable and not even always
Willy Tarreau4f0d9192013-06-11 20:40:55 +020012411 usable. See also "tcp-request connection expect-proxy" for a finer-grained
12412 setting of which client is allowed to use the protocol.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012413
Olivier Houchardc2aae742017-09-22 18:26:28 +020012414allow-0rtt
Bertrand Jacquina25282b2018-08-14 00:56:13 +010012415 Allow receiving early data when using TLSv1.3. This is disabled by default,
Olivier Houchard69752962019-01-08 15:35:32 +010012416 due to security considerations. Because it is vulnerable to replay attacks,
John Roeslerfb2fce12019-07-10 15:45:51 -050012417 you should only allow if for requests that are safe to replay, i.e. requests
Olivier Houchard69752962019-01-08 15:35:32 +010012418 that are idempotent. You can use the "wait-for-handshake" action for any
12419 request that wouldn't be safe with early data.
Olivier Houchardc2aae742017-09-22 18:26:28 +020012420
Willy Tarreauab861d32013-04-02 02:30:41 +020012421alpn <protocols>
12422 This enables the TLS ALPN extension and advertises the specified protocol
12423 list as supported on top of ALPN. The protocol list consists in a comma-
12424 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
John Roeslerfb2fce12019-07-10 15:45:51 -050012425 quotes). This requires that the SSL library is built with support for TLS
Willy Tarreauab861d32013-04-02 02:30:41 +020012426 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
Willy Tarreau95c4e142017-11-26 12:18:55 +010012427 initial NPN extension. ALPN is required to enable HTTP/2 on an HTTP frontend.
12428 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
12429 now obsolete NPN extension. At the time of writing this, most browsers still
12430 support both ALPN and NPN for HTTP/2 so a fallback to NPN may still work for
12431 a while. But ALPN must be used whenever possible. If both HTTP/2 and HTTP/1.1
12432 are expected to be supported, both versions can be advertised, in order of
12433 preference, like below :
12434
12435 bind :443 ssl crt pub.pem alpn h2,http/1.1
Willy Tarreauab861d32013-04-02 02:30:41 +020012436
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012437backlog <backlog>
Willy Tarreaue2711c72019-02-27 15:39:41 +010012438 Sets the socket's backlog to this value. If unspecified or 0, the frontend's
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012439 backlog is used instead, which generally defaults to the maxconn value.
12440
Emmanuel Hocdete7f2b732017-01-09 16:15:54 +010012441curves <curves>
12442 This setting is only available when support for OpenSSL was built in. It sets
12443 the string describing the list of elliptic curves algorithms ("curve suite")
12444 that are negotiated during the SSL/TLS handshake with ECDHE. The format of the
12445 string is a colon-delimited list of curve name.
12446 Example: "X25519:P-256" (without quote)
12447 When "curves" is set, "ecdhe" parameter is ignored.
12448
Emeric Brun7fb34422012-09-28 15:26:15 +020012449ecdhe <named curve>
12450 This setting is only available when support for OpenSSL was built in. It sets
Emeric Brun6924ef82013-03-06 14:08:53 +010012451 the named curve (RFC 4492) used to generate ECDH ephemeral keys. By default,
12452 used named curve is prime256v1.
Emeric Brun7fb34422012-09-28 15:26:15 +020012453
Emeric Brunfd33a262012-10-11 16:28:27 +020012454ca-file <cafile>
Emeric Brun1a073b42012-09-28 17:07:34 +020012455 This setting is only available when support for OpenSSL was built in. It
12456 designates a PEM file from which to load CA certificates used to verify
12457 client's certificate.
12458
Emeric Brunb6dc9342012-09-28 17:55:37 +020012459ca-ignore-err [all|<errorID>,...]
12460 This setting is only available when support for OpenSSL was built in.
12461 Sets a comma separated list of errorIDs to ignore during verify at depth > 0.
12462 If set to 'all', all errors are ignored. SSL handshake is not aborted if an
12463 error is ignored.
12464
Christopher Faulet31af49d2015-06-09 17:29:50 +020012465ca-sign-file <cafile>
12466 This setting is only available when support for OpenSSL was built in. It
12467 designates a PEM file containing both the CA certificate and the CA private
12468 key used to create and sign server's certificates. This is a mandatory
12469 setting when the dynamic generation of certificates is enabled. See
12470 'generate-certificates' for details.
12471
Bertrand Jacquind4d0a232016-11-13 16:37:12 +000012472ca-sign-pass <passphrase>
Christopher Faulet31af49d2015-06-09 17:29:50 +020012473 This setting is only available when support for OpenSSL was built in. It is
12474 the CA private key passphrase. This setting is optional and used only when
12475 the dynamic generation of certificates is enabled. See
12476 'generate-certificates' for details.
12477
Emmanuel Hocdet842e94e2019-12-16 16:39:17 +010012478ca-verify-file <cafile>
12479 This setting designates a PEM file from which to load CA certificates used to
12480 verify client's certificate. It designates CA certificates which must not be
12481 included in CA names sent in server hello message. Typically, "ca-file" must
12482 be defined with intermediate certificates, and "ca-verify-file" with
12483 certificates to ending the chain, like root CA.
12484
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012485ciphers <ciphers>
12486 This setting is only available when support for OpenSSL was built in. It sets
12487 the string describing the list of cipher algorithms ("cipher suite") that are
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +000012488 negotiated during the SSL/TLS handshake up to TLSv1.2. The format of the
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000012489 string is defined in "man 1 ciphers" from OpenSSL man pages. For background
Dirkjan Bussink415150f2018-09-14 11:14:21 +020012490 information and recommendations see e.g.
12491 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
12492 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
12493 cipher configuration, please check the "ciphersuites" keyword.
12494
12495ciphersuites <ciphersuites>
12496 This setting is only available when support for OpenSSL was built in and
12497 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the string describing
12498 the list of cipher algorithms ("cipher suite") that are negotiated during the
12499 TLSv1.3 handshake. The format of the string is defined in "man 1 ciphers" from
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000012500 OpenSSL man pages under the "ciphersuites" section. For cipher configuration
12501 for TLSv1.2 and earlier, please check the "ciphers" keyword.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012502
Emeric Brunfd33a262012-10-11 16:28:27 +020012503crl-file <crlfile>
Emeric Brun1a073b42012-09-28 17:07:34 +020012504 This setting is only available when support for OpenSSL was built in. It
12505 designates a PEM file from which to load certificate revocation list used
12506 to verify client's certificate.
12507
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012508crt <cert>
Alex Davies0fbf0162013-03-02 16:04:50 +000012509 This setting is only available when support for OpenSSL was built in. It
12510 designates a PEM file containing both the required certificates and any
12511 associated private keys. This file can be built by concatenating multiple
12512 PEM files into one (e.g. cat cert.pem key.pem > combined.pem). If your CA
12513 requires an intermediate certificate, this can also be concatenated into this
Emmanuel Hocdet70df7bf2019-01-04 11:08:20 +010012514 file. Intermediate certificate can also be shared in a directory via
12515 "issuers-chain-path" directive.
Alex Davies0fbf0162013-03-02 16:04:50 +000012516
William Lallemand4c5adbf2020-02-24 14:23:22 +010012517 If the file does not contain a private key, HAProxy will try to load
12518 the key at the same path suffixed by a ".key".
12519
Alex Davies0fbf0162013-03-02 16:04:50 +000012520 If the OpenSSL used supports Diffie-Hellman, parameters present in this file
12521 are loaded.
12522
12523 If a directory name is used instead of a PEM file, then all files found in
William Lallemand3f25ae32020-02-24 16:30:12 +010012524 that directory will be loaded in alphabetic order unless their name ends
12525 with '.key', '.issuer', '.ocsp' or '.sctl' (reserved extensions). This
12526 directive may be specified multiple times in order to load certificates from
12527 multiple files or directories. The certificates will be presented to clients
12528 who provide a valid TLS Server Name Indication field matching one of their
12529 CN or alt subjects. Wildcards are supported, where a wildcard character '*'
12530 is used instead of the first hostname component (e.g. *.example.org matches
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010012531 www.example.org but not www.sub.example.org).
Alex Davies0fbf0162013-03-02 16:04:50 +000012532
12533 If no SNI is provided by the client or if the SSL library does not support
12534 TLS extensions, or if the client provides an SNI hostname which does not
12535 match any certificate, then the first loaded certificate will be presented.
12536 This means that when loading certificates from a directory, it is highly
Cyril Bonté3180f7b2015-01-25 00:16:08 +010012537 recommended to load the default one first as a file or to ensure that it will
12538 always be the first one in the directory.
Alex Davies0fbf0162013-03-02 16:04:50 +000012539
Emeric Brune032bfa2012-09-28 13:01:45 +020012540 Note that the same cert may be loaded multiple times without side effects.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012541
Davor Ocelice9ed2812017-12-25 17:49:28 +010012542 Some CAs (such as GoDaddy) offer a drop down list of server types that do not
Alex Davies0fbf0162013-03-02 16:04:50 +000012543 include HAProxy when obtaining a certificate. If this happens be sure to
Davor Ocelice9ed2812017-12-25 17:49:28 +010012544 choose a web server that the CA believes requires an intermediate CA (for
12545 GoDaddy, selection Apache Tomcat will get the correct bundle, but many
Alex Davies0fbf0162013-03-02 16:04:50 +000012546 others, e.g. nginx, result in a wrong bundle that will not work for some
12547 clients).
12548
Emeric Brun4147b2e2014-06-16 18:36:30 +020012549 For each PEM file, haproxy checks for the presence of file at the same path
12550 suffixed by ".ocsp". If such file is found, support for the TLS Certificate
12551 Status Request extension (also known as "OCSP stapling") is automatically
12552 enabled. The content of this file is optional. If not empty, it must contain
12553 a valid OCSP Response in DER format. In order to be valid an OCSP Response
12554 must comply with the following rules: it has to indicate a good status,
12555 it has to be a single response for the certificate of the PEM file, and it
12556 has to be valid at the moment of addition. If these rules are not respected
12557 the OCSP Response is ignored and a warning is emitted. In order to identify
12558 which certificate an OCSP Response applies to, the issuer's certificate is
12559 necessary. If the issuer's certificate is not found in the PEM file, it will
12560 be loaded from a file at the same path as the PEM file suffixed by ".issuer"
12561 if it exists otherwise it will fail with an error.
12562
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010012563 For each PEM file, haproxy also checks for the presence of file at the same
12564 path suffixed by ".sctl". If such file is found, support for Certificate
12565 Transparency (RFC6962) TLS extension is enabled. The file must contain a
12566 valid Signed Certificate Timestamp List, as described in RFC. File is parsed
12567 to check basic syntax, but no signatures are verified.
12568
yanbzhu6c25e9e2016-01-05 12:52:02 -050012569 There are cases where it is desirable to support multiple key types, e.g. RSA
12570 and ECDSA in the cipher suites offered to the clients. This allows clients
12571 that support EC certificates to be able to use EC ciphers, while
12572 simultaneously supporting older, RSA only clients.
yanbzhud19630c2015-12-14 15:10:25 -050012573
William Lallemandf9ff3ec2020-10-02 17:57:44 +020012574 To achieve this, OpenSSL 1.1.1 is required, you can configure this behavior
12575 by providing one crt entry per certificate type, or by configuring a "cert
12576 bundle" like it was required before HAProxy 1.8. See "ssl-load-extra-files".
yanbzhud19630c2015-12-14 15:10:25 -050012577
Emeric Brunb6dc9342012-09-28 17:55:37 +020012578crt-ignore-err <errors>
Alex Davies0fbf0162013-03-02 16:04:50 +000012579 This setting is only available when support for OpenSSL was built in. Sets a
Davor Ocelice9ed2812017-12-25 17:49:28 +010012580 comma separated list of errorIDs to ignore during verify at depth == 0. If
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012581 set to 'all', all errors are ignored. SSL handshake is not aborted if an error
Alex Davies0fbf0162013-03-02 16:04:50 +000012582 is ignored.
Emeric Brunb6dc9342012-09-28 17:55:37 +020012583
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010012584crt-list <file>
12585 This setting is only available when support for OpenSSL was built in. It
Emmanuel Hocdet98263292016-12-29 18:26:15 +010012586 designates a list of PEM file with an optional ssl configuration and a SNI
12587 filter per certificate, with the following format for each line :
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010012588
Emmanuel Hocdet98263292016-12-29 18:26:15 +010012589 <crtfile> [\[<sslbindconf> ...\]] [[!]<snifilter> ...]
12590
William Lallemand5d036392020-06-30 16:11:36 +020012591 sslbindconf supports "allow-0rtt", "alpn", "ca-file", "ca-verify-file",
12592 "ciphers", "ciphersuites", "crl-file", "curves", "ecdhe", "no-ca-names",
12593 "npn", "verify" configuration. With BoringSSL and Openssl >= 1.1.1
12594 "ssl-min-ver" and "ssl-max-ver" are also supported. It overrides the
12595 configuration set in bind line for the certificate.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010012596
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +020012597 Wildcards are supported in the SNI filter. Negative filter are also supported,
12598 only useful in combination with a wildcard filter to exclude a particular SNI.
12599 The certificates will be presented to clients who provide a valid TLS Server
12600 Name Indication field matching one of the SNI filters. If no SNI filter is
12601 specified, the CN and alt subjects are used. This directive may be specified
12602 multiple times. See the "crt" option for more information. The default
12603 certificate is still needed to meet OpenSSL expectations. If it is not used,
12604 the 'strict-sni' option may be used.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010012605
William Lallemandf9ff3ec2020-10-02 17:57:44 +020012606 Multi-cert bundling (see "ssl-load-extra-files") is supported with crt-list,
12607 as long as only the base name is given in the crt-list. SNI filter will do
12608 the same work on all bundled certificates.
yanbzhud19630c2015-12-14 15:10:25 -050012609
William Lallemand7c26ed72020-06-03 17:34:48 +020012610 Empty lines as well as lines beginning with a hash ('#') will be ignored.
12611
Emmanuel Hocdet98263292016-12-29 18:26:15 +010012612 crt-list file example:
12613 cert1.pem
William Lallemand7c26ed72020-06-03 17:34:48 +020012614 # comment
Emmanuel Hocdet05942112017-02-20 16:11:50 +010012615 cert2.pem [alpn h2,http/1.1]
Emmanuel Hocdet98263292016-12-29 18:26:15 +010012616 certW.pem *.domain.tld !secure.domain.tld
Emmanuel Hocdet05942112017-02-20 16:11:50 +010012617 certS.pem [curves X25519:P-256 ciphers ECDHE-ECDSA-AES256-GCM-SHA384] secure.domain.tld
Emmanuel Hocdet98263292016-12-29 18:26:15 +010012618
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012619defer-accept
12620 Is an optional keyword which is supported only on certain Linux kernels. It
12621 states that a connection will only be accepted once some data arrive on it,
12622 or at worst after the first retransmit. This should be used only on protocols
Davor Ocelice9ed2812017-12-25 17:49:28 +010012623 for which the client talks first (e.g. HTTP). It can slightly improve
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012624 performance by ensuring that most of the request is already available when
12625 the connection is accepted. On the other hand, it will not be able to detect
12626 connections which don't talk. It is important to note that this option is
12627 broken in all kernels up to 2.6.31, as the connection is never accepted until
12628 the client talks. This can cause issues with front firewalls which would see
12629 an established connection while the proxy will only see it in SYN_RECV. This
12630 option is only supported on TCPv4/TCPv6 sockets and ignored by other ones.
12631
William Lallemandf6975e92017-05-26 17:42:10 +020012632expose-fd listeners
12633 This option is only usable with the stats socket. It gives your stats socket
12634 the capability to pass listeners FD to another HAProxy process.
William Lallemande202b1e2017-06-01 17:38:56 +020012635 During a reload with the master-worker mode, the process is automatically
12636 reexecuted adding -x and one of the stats socket with this option.
Davor Ocelice9ed2812017-12-25 17:49:28 +010012637 See also "-x" in the management guide.
William Lallemandf6975e92017-05-26 17:42:10 +020012638
Emeric Brun2cb7ae52012-10-05 14:14:21 +020012639force-sslv3
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012640 This option enforces use of SSLv3 only on SSL connections instantiated from
Emeric Brun2cb7ae52012-10-05 14:14:21 +020012641 this listener. SSLv3 is generally less expensive than the TLS counterparts
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012642 for high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012643 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020012644
12645force-tlsv10
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012646 This option enforces use of TLSv1.0 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012647 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012648 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020012649
12650force-tlsv11
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012651 This option enforces use of TLSv1.1 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012652 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012653 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020012654
12655force-tlsv12
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012656 This option enforces use of TLSv1.2 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012657 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012658 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020012659
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020012660force-tlsv13
12661 This option enforces use of TLSv1.3 only on SSL connections instantiated from
12662 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012663 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020012664
Christopher Faulet31af49d2015-06-09 17:29:50 +020012665generate-certificates
12666 This setting is only available when support for OpenSSL was built in. It
12667 enables the dynamic SSL certificates generation. A CA certificate and its
12668 private key are necessary (see 'ca-sign-file'). When HAProxy is configured as
12669 a transparent forward proxy, SSL requests generate errors because of a common
12670 name mismatch on the certificate presented to the client. With this option
12671 enabled, HAProxy will try to forge a certificate using the SNI hostname
12672 indicated by the client. This is done only if no certificate matches the SNI
12673 hostname (see 'crt-list'). If an error occurs, the default certificate is
12674 used, else the 'strict-sni' option is set.
12675 It can also be used when HAProxy is configured as a reverse proxy to ease the
12676 deployment of an architecture with many backends.
12677
12678 Creating a SSL certificate is an expensive operation, so a LRU cache is used
12679 to store forged certificates (see 'tune.ssl.ssl-ctx-cache-size'). It
Davor Ocelice9ed2812017-12-25 17:49:28 +010012680 increases the HAProxy's memory footprint to reduce latency when the same
Christopher Faulet31af49d2015-06-09 17:29:50 +020012681 certificate is used many times.
12682
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012683gid <gid>
12684 Sets the group of the UNIX sockets to the designated system gid. It can also
12685 be set by default in the global section's "unix-bind" statement. Note that
12686 some platforms simply ignore this. This setting is equivalent to the "group"
12687 setting except that the group ID is used instead of its name. This setting is
12688 ignored by non UNIX sockets.
12689
12690group <group>
12691 Sets the group of the UNIX sockets to the designated system group. It can
12692 also be set by default in the global section's "unix-bind" statement. Note
12693 that some platforms simply ignore this. This setting is equivalent to the
12694 "gid" setting except that the group name is used instead of its gid. This
12695 setting is ignored by non UNIX sockets.
12696
12697id <id>
12698 Fixes the socket ID. By default, socket IDs are automatically assigned, but
12699 sometimes it is more convenient to fix them to ease monitoring. This value
12700 must be strictly positive and unique within the listener/frontend. This
12701 option can only be used when defining only a single socket.
12702
12703interface <interface>
Lukas Tribusfce2e962013-02-12 22:13:19 +010012704 Restricts the socket to a specific interface. When specified, only packets
12705 received from that particular interface are processed by the socket. This is
12706 currently only supported on Linux. The interface must be a primary system
12707 interface, not an aliased interface. It is also possible to bind multiple
12708 frontends to the same address if they are bound to different interfaces. Note
12709 that binding to a network interface requires root privileges. This parameter
Jérôme Magnin61275192018-02-07 11:39:58 +010012710 is only compatible with TCPv4/TCPv6 sockets. When specified, return traffic
12711 uses the same interface as inbound traffic, and its associated routing table,
12712 even if there are explicit routes through different interfaces configured.
12713 This can prove useful to address asymmetric routing issues when the same
12714 client IP addresses need to be able to reach frontends hosted on different
12715 interfaces.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012716
Willy Tarreauabb175f2012-09-24 12:43:26 +020012717level <level>
12718 This setting is used with the stats sockets only to restrict the nature of
12719 the commands that can be issued on the socket. It is ignored by other
12720 sockets. <level> can be one of :
Davor Ocelice9ed2812017-12-25 17:49:28 +010012721 - "user" is the least privileged level; only non-sensitive stats can be
Willy Tarreauabb175f2012-09-24 12:43:26 +020012722 read, and no change is allowed. It would make sense on systems where it
12723 is not easy to restrict access to the socket.
12724 - "operator" is the default level and fits most common uses. All data can
Davor Ocelice9ed2812017-12-25 17:49:28 +010012725 be read, and only non-sensitive changes are permitted (e.g. clear max
Willy Tarreauabb175f2012-09-24 12:43:26 +020012726 counters).
Davor Ocelice9ed2812017-12-25 17:49:28 +010012727 - "admin" should be used with care, as everything is permitted (e.g. clear
Willy Tarreauabb175f2012-09-24 12:43:26 +020012728 all counters).
12729
Andjelko Iharosc4df59e2017-07-20 11:59:48 +020012730severity-output <format>
12731 This setting is used with the stats sockets only to configure severity
12732 level output prepended to informational feedback messages. Severity
12733 level of messages can range between 0 and 7, conforming to syslog
12734 rfc5424. Valid and successful socket commands requesting data
12735 (i.e. "show map", "get acl foo" etc.) will never have a severity level
12736 prepended. It is ignored by other sockets. <format> can be one of :
12737 - "none" (default) no severity level is prepended to feedback messages.
12738 - "number" severity level is prepended as a number.
12739 - "string" severity level is prepended as a string following the
12740 rfc5424 convention.
12741
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012742maxconn <maxconn>
12743 Limits the sockets to this number of concurrent connections. Extraneous
12744 connections will remain in the system's backlog until a connection is
12745 released. If unspecified, the limit will be the same as the frontend's
12746 maxconn. Note that in case of port ranges or multiple addresses, the same
12747 value will be applied to each socket. This setting enables different
12748 limitations on expensive sockets, for instance SSL entries which may easily
12749 eat all memory.
12750
12751mode <mode>
12752 Sets the octal mode used to define access permissions on the UNIX socket. It
12753 can also be set by default in the global section's "unix-bind" statement.
12754 Note that some platforms simply ignore this. This setting is ignored by non
12755 UNIX sockets.
12756
12757mss <maxseg>
12758 Sets the TCP Maximum Segment Size (MSS) value to be advertised on incoming
12759 connections. This can be used to force a lower MSS for certain specific
12760 ports, for instance for connections passing through a VPN. Note that this
12761 relies on a kernel feature which is theoretically supported under Linux but
12762 was buggy in all versions prior to 2.6.28. It may or may not work on other
12763 operating systems. It may also not change the advertised value but change the
12764 effective size of outgoing segments. The commonly advertised value for TCPv4
12765 over Ethernet networks is 1460 = 1500(MTU) - 40(IP+TCP). If this value is
12766 positive, it will be used as the advertised MSS. If it is negative, it will
12767 indicate by how much to reduce the incoming connection's advertised MSS for
12768 outgoing segments. This parameter is only compatible with TCP v4/v6 sockets.
12769
12770name <name>
12771 Sets an optional name for these sockets, which will be reported on the stats
12772 page.
12773
Willy Tarreaud72f0f32015-10-13 14:50:22 +020012774namespace <name>
12775 On Linux, it is possible to specify which network namespace a socket will
12776 belong to. This directive makes it possible to explicitly bind a listener to
12777 a namespace different from the default one. Please refer to your operating
12778 system's documentation to find more details about network namespaces.
12779
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012780nice <nice>
12781 Sets the 'niceness' of connections initiated from the socket. Value must be
12782 in the range -1024..1024 inclusive, and defaults to zero. Positive values
12783 means that such connections are more friendly to others and easily offer
12784 their place in the scheduler. On the opposite, negative values mean that
12785 connections want to run with a higher priority than others. The difference
12786 only happens under high loads when the system is close to saturation.
12787 Negative values are appropriate for low-latency or administration services,
12788 and high values are generally recommended for CPU intensive tasks such as SSL
12789 processing or bulk transfers which are less sensible to latency. For example,
12790 it may make sense to use a positive value for an SMTP socket and a negative
12791 one for an RDP socket.
12792
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020012793no-ca-names
12794 This setting is only available when support for OpenSSL was built in. It
12795 prevents from send CA names in server hello message when ca-file is used.
Emmanuel Hocdet842e94e2019-12-16 16:39:17 +010012796 Use "ca-verify-file" instead of "ca-file" with "no-ca-names".
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020012797
Emeric Brun9b3009b2012-10-05 11:55:06 +020012798no-sslv3
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012799 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012800 disables support for SSLv3 on any sockets instantiated from the listener when
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012801 SSL is supported. Note that SSLv2 is forced disabled in the code and cannot
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012802 be enabled using any configuration option. This option is also available on
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012803 global statement "ssl-default-bind-options". Use "ssl-min-ver" and
12804 "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012805
Emeric Brun90ad8722012-10-02 14:00:59 +020012806no-tls-tickets
12807 This setting is only available when support for OpenSSL was built in. It
12808 disables the stateless session resumption (RFC 5077 TLS Ticket
12809 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012810 session resumption is more expensive in CPU usage. This option is also
12811 available on global statement "ssl-default-bind-options".
Lukas Tribusbdb386d2020-03-10 00:56:09 +010012812 The TLS ticket mechanism is only used up to TLS 1.2.
12813 Forward Secrecy is compromised with TLS tickets, unless ticket keys
12814 are periodically rotated (via reload or by using "tls-ticket-keys").
Emeric Brun90ad8722012-10-02 14:00:59 +020012815
Emeric Brun9b3009b2012-10-05 11:55:06 +020012816no-tlsv10
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012817 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012818 disables support for TLSv1.0 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020012819 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012820 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012821 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
12822 and "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012823
Emeric Brun9b3009b2012-10-05 11:55:06 +020012824no-tlsv11
Emeric Brunf5da4932012-09-28 19:42:54 +020012825 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012826 disables support for TLSv1.1 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020012827 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012828 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012829 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
12830 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020012831
Emeric Brun9b3009b2012-10-05 11:55:06 +020012832no-tlsv12
Emeric Brunf5da4932012-09-28 19:42:54 +020012833 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012834 disables support for TLSv1.2 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020012835 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012836 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012837 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
12838 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020012839
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020012840no-tlsv13
12841 This setting is only available when support for OpenSSL was built in. It
12842 disables support for TLSv1.3 on any sockets instantiated from the listener
12843 when SSL is supported. Note that SSLv2 is forced disabled in the code and
12844 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012845 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
12846 and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020012847
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020012848npn <protocols>
12849 This enables the NPN TLS extension and advertises the specified protocol list
12850 as supported on top of NPN. The protocol list consists in a comma-delimited
12851 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
John Roeslerfb2fce12019-07-10 15:45:51 -050012852 This requires that the SSL library is built with support for TLS extensions
Willy Tarreauab861d32013-04-02 02:30:41 +020012853 enabled (check with haproxy -vv). Note that the NPN extension has been
Willy Tarreau95c4e142017-11-26 12:18:55 +010012854 replaced with the ALPN extension (see the "alpn" keyword), though this one is
12855 only available starting with OpenSSL 1.0.2. If HTTP/2 is desired on an older
12856 version of OpenSSL, NPN might still be used as most clients still support it
12857 at the time of writing this. It is possible to enable both NPN and ALPN
12858 though it probably doesn't make any sense out of testing.
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020012859
Lukas Tribus53ae85c2017-05-04 15:45:40 +000012860prefer-client-ciphers
12861 Use the client's preference when selecting the cipher suite, by default
12862 the server's preference is enforced. This option is also available on
12863 global statement "ssl-default-bind-options".
Lukas Tribus926594f2018-05-18 17:55:57 +020012864 Note that with OpenSSL >= 1.1.1 ChaCha20-Poly1305 is reprioritized anyway
12865 (without setting this option), if a ChaCha20-Poly1305 cipher is at the top of
12866 the client cipher list.
Lukas Tribus53ae85c2017-05-04 15:45:40 +000012867
Christopher Fauletc644fa92017-11-23 22:44:11 +010012868process <process-set>[/<thread-set>]
Willy Tarreaua36b3242019-02-02 13:14:34 +010012869 This restricts the list of processes or threads on which this listener is
Christopher Fauletc644fa92017-11-23 22:44:11 +010012870 allowed to run. It does not enforce any process but eliminates those which do
Davor Ocelice9ed2812017-12-25 17:49:28 +010012871 not match. If the frontend uses a "bind-process" setting, the intersection
Christopher Fauletc644fa92017-11-23 22:44:11 +010012872 between the two is applied. If in the end the listener is not allowed to run
12873 on any remaining process, a warning is emitted, and the listener will either
12874 run on the first process of the listener if a single process was specified,
12875 or on all of its processes if multiple processes were specified. If a thread
Davor Ocelice9ed2812017-12-25 17:49:28 +010012876 set is specified, it limits the threads allowed to process incoming
Willy Tarreaua36b3242019-02-02 13:14:34 +010012877 connections for this listener, for the the process set. If multiple processes
12878 and threads are configured, a warning is emitted, as it either results from a
12879 configuration error or a misunderstanding of these models. For the unlikely
12880 case where several ranges are needed, this directive may be repeated.
12881 <process-set> and <thread-set> must use the format
Christopher Fauletc644fa92017-11-23 22:44:11 +010012882
12883 all | odd | even | number[-[number]]
12884
12885 Ranges can be partially defined. The higher bound can be omitted. In such
12886 case, it is replaced by the corresponding maximum value. The main purpose of
12887 this directive is to be used with the stats sockets and have one different
12888 socket per process. The second purpose is to have multiple bind lines sharing
12889 the same IP:port but not the same process in a listener, so that the system
12890 can distribute the incoming connections into multiple queues and allow a
12891 smoother inter-process load balancing. Currently Linux 3.9 and above is known
12892 for supporting this. See also "bind-process" and "nbproc".
Willy Tarreau6ae1ba62014-05-07 19:01:58 +020012893
Christopher Fauleta717b992018-04-10 14:43:00 +020012894proto <name>
12895 Forces the multiplexer's protocol to use for the incoming connections. It
12896 must be compatible with the mode of the frontend (TCP or HTTP). It must also
12897 be usable on the frontend side. The list of available protocols is reported
12898 in haproxy -vv.
Daniel Corbett67a82712020-07-06 23:01:19 -040012899 Idea behind this option is to bypass the selection of the best multiplexer's
Christopher Fauleta717b992018-04-10 14:43:00 +020012900 protocol for all connections instantiated from this listening socket. For
Joseph Herlant71b4b152018-11-13 16:55:16 -080012901 instance, it is possible to force the http/2 on clear TCP by specifying "proto
Christopher Fauleta717b992018-04-10 14:43:00 +020012902 h2" on the bind line.
12903
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012904ssl
12905 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012906 enables SSL deciphering on connections instantiated from this listener. A
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012907 certificate is necessary (see "crt" above). All contents in the buffers will
12908 appear in clear text, so that ACLs and HTTP processing will only have access
Emmanuel Hocdetbd695fe2017-05-15 15:53:41 +020012909 to deciphered contents. SSLv3 is disabled per default, use "ssl-min-ver SSLv3"
12910 to enable it.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012911
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012912ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
12913 This option enforces use of <version> or lower on SSL connections instantiated
William Lallemand50df1cb2020-06-02 10:52:24 +020012914 from this listener. Using this setting without "ssl-min-ver" can be
12915 ambiguous because the default ssl-min-ver value could change in future HAProxy
12916 versions. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012917 "ssl-default-bind-options". See also "ssl-min-ver".
12918
12919ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
William Lallemand50df1cb2020-06-02 10:52:24 +020012920 This option enforces use of <version> or upper on SSL connections
12921 instantiated from this listener. The default value is "TLSv1.2". This option
12922 is also available on global statement "ssl-default-bind-options".
12923 See also "ssl-max-ver".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012924
Emmanuel Hocdet65623372013-01-24 17:17:15 +010012925strict-sni
12926 This setting is only available when support for OpenSSL was built in. The
12927 SSL/TLS negotiation is allow only if the client provided an SNI which match
12928 a certificate. The default certificate is not used.
12929 See the "crt" option for more information.
12930
Willy Tarreau2af207a2015-02-04 00:45:58 +010012931tcp-ut <delay>
Tim Düsterhus4896c442016-11-29 02:15:19 +010012932 Sets the TCP User Timeout for all incoming connections instantiated from this
Willy Tarreau2af207a2015-02-04 00:45:58 +010012933 listening socket. This option is available on Linux since version 2.6.37. It
12934 allows haproxy to configure a timeout for sockets which contain data not
Davor Ocelice9ed2812017-12-25 17:49:28 +010012935 receiving an acknowledgment for the configured delay. This is especially
Willy Tarreau2af207a2015-02-04 00:45:58 +010012936 useful on long-lived connections experiencing long idle periods such as
12937 remote terminals or database connection pools, where the client and server
12938 timeouts must remain high to allow a long period of idle, but where it is
12939 important to detect that the client has disappeared in order to release all
12940 resources associated with its connection (and the server's session). The
12941 argument is a delay expressed in milliseconds by default. This only works
12942 for regular TCP connections, and is ignored for other protocols.
12943
Willy Tarreau1c862c52012-10-05 16:21:00 +020012944tfo
Lukas Tribus0defb902013-02-13 23:35:39 +010012945 Is an optional keyword which is supported only on Linux kernels >= 3.7. It
Willy Tarreau1c862c52012-10-05 16:21:00 +020012946 enables TCP Fast Open on the listening socket, which means that clients which
12947 support this feature will be able to send a request and receive a response
12948 during the 3-way handshake starting from second connection, thus saving one
12949 round-trip after the first connection. This only makes sense with protocols
12950 that use high connection rates and where each round trip matters. This can
12951 possibly cause issues with many firewalls which do not accept data on SYN
12952 packets, so this option should only be enabled once well tested. This option
Lukas Tribus0999f762013-04-02 16:43:24 +020012953 is only supported on TCPv4/TCPv6 sockets and ignored by other ones. You may
12954 need to build HAProxy with USE_TFO=1 if your libc doesn't define
12955 TCP_FASTOPEN.
Willy Tarreau1c862c52012-10-05 16:21:00 +020012956
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010012957tls-ticket-keys <keyfile>
12958 Sets the TLS ticket keys file to load the keys from. The keys need to be 48
Emeric Brun9e754772019-01-10 17:51:55 +010012959 or 80 bytes long, depending if aes128 or aes256 is used, encoded with base64
12960 with one line per key (ex. openssl rand 80 | openssl base64 -A | xargs echo).
12961 The first key determines the key length used for next keys: you can't mix
12962 aes128 and aes256 keys. Number of keys is specified by the TLS_TICKETS_NO
12963 build option (default 3) and at least as many keys need to be present in
12964 the file. Last TLS_TICKETS_NO keys will be used for decryption and the
12965 penultimate one for encryption. This enables easy key rotation by just
12966 appending new key to the file and reloading the process. Keys must be
12967 periodically rotated (ex. every 12h) or Perfect Forward Secrecy is
12968 compromised. It is also a good idea to keep the keys off any permanent
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010012969 storage such as hard drives (hint: use tmpfs and don't swap those files).
12970 Lifetime hint can be changed using tune.ssl.timeout.
12971
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012972transparent
12973 Is an optional keyword which is supported only on certain Linux kernels. It
12974 indicates that the addresses will be bound even if they do not belong to the
12975 local machine, and that packets targeting any of these addresses will be
12976 intercepted just as if the addresses were locally configured. This normally
12977 requires that IP forwarding is enabled. Caution! do not use this with the
12978 default address '*', as it would redirect any traffic for the specified port.
12979 This keyword is available only when HAProxy is built with USE_LINUX_TPROXY=1.
12980 This parameter is only compatible with TCPv4 and TCPv6 sockets, depending on
12981 kernel version. Some distribution kernels include backports of the feature,
12982 so check for support with your vendor.
12983
Willy Tarreau77e3af92012-11-24 15:07:23 +010012984v4v6
12985 Is an optional keyword which is supported only on most recent systems
12986 including Linux kernels >= 2.4.21. It is used to bind a socket to both IPv4
12987 and IPv6 when it uses the default address. Doing so is sometimes necessary
12988 on systems which bind to IPv6 only by default. It has no effect on non-IPv6
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012989 sockets, and is overridden by the "v6only" option.
Willy Tarreau77e3af92012-11-24 15:07:23 +010012990
Willy Tarreau9b6700f2012-11-24 11:55:28 +010012991v6only
12992 Is an optional keyword which is supported only on most recent systems
12993 including Linux kernels >= 2.4.21. It is used to bind a socket to IPv6 only
12994 when it uses the default address. Doing so is sometimes preferred to doing it
Willy Tarreau77e3af92012-11-24 15:07:23 +010012995 system-wide as it is per-listener. It has no effect on non-IPv6 sockets and
12996 has precedence over the "v4v6" option.
Willy Tarreau9b6700f2012-11-24 11:55:28 +010012997
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012998uid <uid>
12999 Sets the owner of the UNIX sockets to the designated system uid. It can also
13000 be set by default in the global section's "unix-bind" statement. Note that
13001 some platforms simply ignore this. This setting is equivalent to the "user"
13002 setting except that the user numeric ID is used instead of its name. This
13003 setting is ignored by non UNIX sockets.
13004
13005user <user>
13006 Sets the owner of the UNIX sockets to the designated system user. It can also
13007 be set by default in the global section's "unix-bind" statement. Note that
13008 some platforms simply ignore this. This setting is equivalent to the "uid"
13009 setting except that the user name is used instead of its uid. This setting is
13010 ignored by non UNIX sockets.
13011
Emeric Brun1a073b42012-09-28 17:07:34 +020013012verify [none|optional|required]
13013 This setting is only available when support for OpenSSL was built in. If set
13014 to 'none', client certificate is not requested. This is the default. In other
13015 cases, a client certificate is requested. If the client does not provide a
13016 certificate after the request and if 'verify' is set to 'required', then the
13017 handshake is aborted, while it would have succeeded if set to 'optional'. The
Emeric Brunfd33a262012-10-11 16:28:27 +020013018 certificate provided by the client is always verified using CAs from
13019 'ca-file' and optional CRLs from 'crl-file'. On verify failure the handshake
13020 is aborted, regardless of the 'verify' option, unless the error code exactly
13021 matches one of those listed with 'ca-ignore-err' or 'crt-ignore-err'.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020013022
Willy Tarreaub6205fd2012-09-24 12:27:33 +0200130235.2. Server and default-server options
Cyril Bontéf0c60612010-02-06 14:44:47 +010013024------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020013025
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010013026The "server" and "default-server" keywords support a certain number of settings
13027which are all passed as arguments on the server line. The order in which those
13028arguments appear does not count, and they are all optional. Some of those
13029settings are single words (booleans) while others expect one or several values
13030after them. In this case, the values must immediately follow the setting name.
13031Except default-server, all those settings must be specified after the server's
13032address if they are used:
Willy Tarreau6a06a402007-07-15 20:15:28 +020013033
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013034 server <name> <address>[:port] [settings ...]
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010013035 default-server [settings ...]
Willy Tarreau6a06a402007-07-15 20:15:28 +020013036
Frédéric Lécailled2376272017-03-21 18:52:12 +010013037Note that all these settings are supported both by "server" and "default-server"
13038keywords, except "id" which is only supported by "server".
13039
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013040The currently supported settings are the following ones.
Willy Tarreau0ba27502007-12-24 16:55:16 +010013041
Willy Tarreauceb4ac92012-04-28 00:41:46 +020013042addr <ipv4|ipv6>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013043 Using the "addr" parameter, it becomes possible to use a different IP address
Baptiste Assmann13f83532016-03-06 23:14:36 +010013044 to send health-checks or to probe the agent-check. On some servers, it may be
13045 desirable to dedicate an IP address to specific component able to perform
13046 complex tests which are more suitable to health-checks than the application.
13047 This parameter is ignored if the "check" parameter is not set. See also the
13048 "port" parameter.
Willy Tarreau6a06a402007-07-15 20:15:28 +020013049
Simon Hormand60d6912013-11-25 10:46:36 +090013050agent-check
13051 Enable an auxiliary agent check which is run independently of a regular
Willy Tarreau81f5d942013-12-09 20:51:51 +010013052 health check. An agent health check is performed by making a TCP connection
Willy Tarreau7a0139e2018-12-16 08:42:56 +010013053 to the port set by the "agent-port" parameter and reading an ASCII string
13054 terminated by the first '\r' or '\n' met. The string is made of a series of
13055 words delimited by spaces, tabs or commas in any order, each consisting of :
Simon Hormand60d6912013-11-25 10:46:36 +090013056
Willy Tarreau81f5d942013-12-09 20:51:51 +010013057 - An ASCII representation of a positive integer percentage, e.g. "75%".
Simon Hormand60d6912013-11-25 10:46:36 +090013058 Values in this format will set the weight proportional to the initial
Willy Tarreauc5af3a62014-10-07 15:27:33 +020013059 weight of a server as configured when haproxy starts. Note that a zero
13060 weight is reported on the stats page as "DRAIN" since it has the same
13061 effect on the server (it's removed from the LB farm).
Simon Hormand60d6912013-11-25 10:46:36 +090013062
Davor Ocelice9ed2812017-12-25 17:49:28 +010013063 - The string "maxconn:" followed by an integer (no space between). Values
13064 in this format will set the maxconn of a server. The maximum number of
13065 connections advertised needs to be multiplied by the number of load
13066 balancers and different backends that use this health check to get the
13067 total number of connections the server might receive. Example: maxconn:30
Nenad Merdanovic174dd372016-04-24 23:10:06 +020013068
Willy Tarreau81f5d942013-12-09 20:51:51 +010013069 - The word "ready". This will turn the server's administrative state to the
Davor Ocelice9ed2812017-12-25 17:49:28 +010013070 READY mode, thus canceling any DRAIN or MAINT state
Simon Hormand60d6912013-11-25 10:46:36 +090013071
Willy Tarreau81f5d942013-12-09 20:51:51 +010013072 - The word "drain". This will turn the server's administrative state to the
13073 DRAIN mode, thus it will not accept any new connections other than those
13074 that are accepted via persistence.
Simon Hormand60d6912013-11-25 10:46:36 +090013075
Willy Tarreau81f5d942013-12-09 20:51:51 +010013076 - The word "maint". This will turn the server's administrative state to the
13077 MAINT mode, thus it will not accept any new connections at all, and health
13078 checks will be stopped.
Simon Hormand60d6912013-11-25 10:46:36 +090013079
William Dauchyf8e795c2020-09-26 13:35:51 +020013080 - The words "down", "fail", or "stopped", optionally followed by a
Willy Tarreau81f5d942013-12-09 20:51:51 +010013081 description string after a sharp ('#'). All of these mark the server's
13082 operating state as DOWN, but since the word itself is reported on the stats
13083 page, the difference allows an administrator to know if the situation was
13084 expected or not : the service may intentionally be stopped, may appear up
Davor Ocelice9ed2812017-12-25 17:49:28 +010013085 but fail some validity tests, or may be seen as down (e.g. missing process,
Willy Tarreau81f5d942013-12-09 20:51:51 +010013086 or port not responding).
Simon Hormand60d6912013-11-25 10:46:36 +090013087
Willy Tarreau81f5d942013-12-09 20:51:51 +010013088 - The word "up" sets back the server's operating state as UP if health checks
13089 also report that the service is accessible.
Simon Hormand60d6912013-11-25 10:46:36 +090013090
Willy Tarreau81f5d942013-12-09 20:51:51 +010013091 Parameters which are not advertised by the agent are not changed. For
13092 example, an agent might be designed to monitor CPU usage and only report a
13093 relative weight and never interact with the operating status. Similarly, an
13094 agent could be designed as an end-user interface with 3 radio buttons
13095 allowing an administrator to change only the administrative state. However,
13096 it is important to consider that only the agent may revert its own actions,
13097 so if a server is set to DRAIN mode or to DOWN state using the agent, the
13098 agent must implement the other equivalent actions to bring the service into
13099 operations again.
Simon Hormand60d6912013-11-25 10:46:36 +090013100
Simon Horman2f1f9552013-11-25 10:46:37 +090013101 Failure to connect to the agent is not considered an error as connectivity
13102 is tested by the regular health check which is enabled by the "check"
Willy Tarreau81f5d942013-12-09 20:51:51 +010013103 parameter. Warning though, it is not a good idea to stop an agent after it
13104 reports "down", since only an agent reporting "up" will be able to turn the
13105 server up again. Note that the CLI on the Unix stats socket is also able to
Willy Tarreau989222a2016-01-15 10:26:26 +010013106 force an agent's result in order to work around a bogus agent if needed.
Simon Horman2f1f9552013-11-25 10:46:37 +090013107
Willy Tarreau81f5d942013-12-09 20:51:51 +010013108 Requires the "agent-port" parameter to be set. See also the "agent-inter"
Frédéric Lécailled2376272017-03-21 18:52:12 +010013109 and "no-agent-check" parameters.
Simon Hormand60d6912013-11-25 10:46:36 +090013110
James Brown55f9ff12015-10-21 18:19:05 -070013111agent-send <string>
13112 If this option is specified, haproxy will send the given string (verbatim)
13113 to the agent server upon connection. You could, for example, encode
13114 the backend name into this string, which would enable your agent to send
13115 different responses based on the backend. Make sure to include a '\n' if
13116 you want to terminate your request with a newline.
13117
Simon Hormand60d6912013-11-25 10:46:36 +090013118agent-inter <delay>
13119 The "agent-inter" parameter sets the interval between two agent checks
13120 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
13121
13122 Just as with every other time-based parameter, it may be entered in any
13123 other explicit unit among { us, ms, s, m, h, d }. The "agent-inter"
13124 parameter also serves as a timeout for agent checks "timeout check" is
13125 not set. In order to reduce "resonance" effects when multiple servers are
13126 hosted on the same hardware, the agent and health checks of all servers
13127 are started with a small time offset between them. It is also possible to
13128 add some random noise in the agent and health checks interval using the
13129 global "spread-checks" keyword. This makes sense for instance when a lot
13130 of backends use the same servers.
13131
13132 See also the "agent-check" and "agent-port" parameters.
13133
Misiek768d8602017-01-09 09:52:43 +010013134agent-addr <addr>
13135 The "agent-addr" parameter sets address for agent check.
13136
13137 You can offload agent-check to another target, so you can make single place
13138 managing status and weights of servers defined in haproxy in case you can't
13139 make self-aware and self-managing services. You can specify both IP or
13140 hostname, it will be resolved.
13141
Simon Hormand60d6912013-11-25 10:46:36 +090013142agent-port <port>
13143 The "agent-port" parameter sets the TCP port used for agent checks.
13144
13145 See also the "agent-check" and "agent-inter" parameters.
13146
Olivier Houchard8cb2d2e2019-05-06 18:58:48 +020013147allow-0rtt
13148 Allow sending early data to the server when using TLS 1.3.
Olivier Houchard22c9b442019-05-06 19:01:04 +020013149 Note that early data will be sent only if the client used early data, or
13150 if the backend uses "retry-on" with the "0rtt-rejected" keyword.
Olivier Houchard8cb2d2e2019-05-06 18:58:48 +020013151
Olivier Houchardc7566002018-11-20 23:33:50 +010013152alpn <protocols>
13153 This enables the TLS ALPN extension and advertises the specified protocol
13154 list as supported on top of ALPN. The protocol list consists in a comma-
13155 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
John Roeslerfb2fce12019-07-10 15:45:51 -050013156 quotes). This requires that the SSL library is built with support for TLS
Olivier Houchardc7566002018-11-20 23:33:50 +010013157 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
13158 initial NPN extension. ALPN is required to connect to HTTP/2 servers.
13159 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
13160 now obsolete NPN extension.
13161 If both HTTP/2 and HTTP/1.1 are expected to be supported, both versions can
13162 be advertised, in order of preference, like below :
13163
13164 server 127.0.0.1:443 ssl crt pub.pem alpn h2,http/1.1
13165
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013166backup
13167 When "backup" is present on a server line, the server is only used in load
13168 balancing when all other non-backup servers are unavailable. Requests coming
13169 with a persistence cookie referencing the server will always be served
13170 though. By default, only the first operational backup server is used, unless
Frédéric Lécailled2376272017-03-21 18:52:12 +010013171 the "allbackups" option is set in the backend. See also the "no-backup" and
13172 "allbackups" options.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013173
Emeric Brunef42d922012-10-11 16:11:36 +020013174ca-file <cafile>
13175 This setting is only available when support for OpenSSL was built in. It
13176 designates a PEM file from which to load CA certificates used to verify
13177 server's certificate.
13178
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013179check
Jerome Magnin90702bc2020-04-26 14:23:04 +020013180 This option enables health checks on a server:
13181 - when not set, no health checking is performed, and the server is always
13182 considered available.
13183 - when set and no other check method is configured, the server is considered
13184 available when a connection can be established at the highest configured
13185 transport layer. This means TCP by default, or SSL/TLS when "ssl" or
13186 "check-ssl" are set, both possibly combined with connection prefixes such
13187 as a PROXY protocol header when "send-proxy" or "check-send-proxy" are
13188 set.
13189 - when set and an application-level health check is defined, the
13190 application-level exchanges are performed on top of the configured
13191 transport layer and the server is considered available if all of the
13192 exchanges succeed.
13193
13194 By default, health checks are performed on the same address and port as
13195 configured on the server, using the same encapsulation parameters (SSL/TLS,
13196 proxy-protocol header, etc... ). It is possible to change the destination
13197 address using "addr" and the port using "port". When done, it is assumed the
13198 server isn't checked on the service port, and configured encapsulation
Ilya Shipitsin4329a9a2020-05-05 21:17:10 +050013199 parameters are not reused. One must explicitly set "check-send-proxy" to send
Jerome Magnin90702bc2020-04-26 14:23:04 +020013200 connection headers, "check-ssl" to use SSL/TLS.
13201
13202 When "sni" or "alpn" are set on the server line, their value is not used for
13203 health checks and one must use "check-sni" or "check-alpn".
13204
13205 The default source address for health check traffic is the same as the one
13206 defined in the backend. It can be changed with the "source" keyword.
13207
13208 The interval between checks can be set using the "inter" keyword, and the
13209 "rise" and "fall" keywords can be used to define how many successful or
13210 failed health checks are required to flag a server available or not
13211 available.
13212
13213 Optional application-level health checks can be configured with "option
13214 httpchk", "option mysql-check" "option smtpchk", "option pgsql-check",
13215 "option ldap-check", or "option redis-check".
13216
13217 Example:
13218 # simple tcp check
13219 backend foo
13220 server s1 192.168.0.1:80 check
13221 # this does a tcp connect + tls handshake
13222 backend foo
13223 server s1 192.168.0.1:443 ssl check
13224 # simple tcp check is enough for check success
13225 backend foo
13226 option tcp-check
13227 tcp-check connect
13228 server s1 192.168.0.1:443 ssl check
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013229
Willy Tarreau6c16adc2012-10-05 00:04:16 +020013230check-send-proxy
13231 This option forces emission of a PROXY protocol line with outgoing health
13232 checks, regardless of whether the server uses send-proxy or not for the
13233 normal traffic. By default, the PROXY protocol is enabled for health checks
13234 if it is already enabled for normal traffic and if no "port" nor "addr"
13235 directive is present. However, if such a directive is present, the
13236 "check-send-proxy" option needs to be used to force the use of the
13237 protocol. See also the "send-proxy" option for more information.
13238
Olivier Houchard92150142018-12-21 19:47:01 +010013239check-alpn <protocols>
13240 Defines which protocols to advertise with ALPN. The protocol list consists in
13241 a comma-delimited list of protocol names, for instance: "http/1.1,http/1.0"
13242 (without quotes). If it is not set, the server ALPN is used.
13243
Christopher Fauletedc6ed92020-04-23 16:27:59 +020013244check-proto <name>
13245 Forces the multiplexer's protocol to use for the server's health-check
13246 connections. It must be compatible with the health-check type (TCP or
13247 HTTP). It must also be usable on the backend side. The list of available
13248 protocols is reported in haproxy -vv.
Daniel Corbett67a82712020-07-06 23:01:19 -040013249 Idea behind this option is to bypass the selection of the best multiplexer's
Christopher Fauletedc6ed92020-04-23 16:27:59 +020013250 protocol for health-check connections established to this server.
13251 If not defined, the server one will be used, if set.
13252
Jérôme Magninae9bb762018-12-09 16:08:26 +010013253check-sni <sni>
Olivier Houchard9130a962017-10-17 17:33:43 +020013254 This option allows you to specify the SNI to be used when doing health checks
Jérôme Magninae9bb762018-12-09 16:08:26 +010013255 over SSL. It is only possible to use a string to set <sni>. If you want to
13256 set a SNI for proxied traffic, see "sni".
Olivier Houchard9130a962017-10-17 17:33:43 +020013257
Willy Tarreau763a95b2012-10-04 23:15:39 +020013258check-ssl
13259 This option forces encryption of all health checks over SSL, regardless of
13260 whether the server uses SSL or not for the normal traffic. This is generally
13261 used when an explicit "port" or "addr" directive is specified and SSL health
13262 checks are not inherited. It is important to understand that this option
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013263 inserts an SSL transport layer below the checks, so that a simple TCP connect
Willy Tarreau763a95b2012-10-04 23:15:39 +020013264 check becomes an SSL connect, which replaces the old ssl-hello-chk. The most
13265 common use is to send HTTPS checks by combining "httpchk" with SSL checks.
Davor Ocelice9ed2812017-12-25 17:49:28 +010013266 All SSL settings are common to health checks and traffic (e.g. ciphers).
Frédéric Lécailled2376272017-03-21 18:52:12 +010013267 See the "ssl" option for more information and "no-check-ssl" to disable
13268 this option.
Willy Tarreau763a95b2012-10-04 23:15:39 +020013269
Alexander Liu2a54bb72019-05-22 19:44:48 +080013270check-via-socks4
John Roeslerfb2fce12019-07-10 15:45:51 -050013271 This option enables outgoing health checks using upstream socks4 proxy. By
Alexander Liu2a54bb72019-05-22 19:44:48 +080013272 default, the health checks won't go through socks tunnel even it was enabled
13273 for normal traffic.
13274
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020013275ciphers <ciphers>
Dirkjan Bussink415150f2018-09-14 11:14:21 +020013276 This setting is only available when support for OpenSSL was built in. This
13277 option sets the string describing the list of cipher algorithms that is
13278 negotiated during the SSL/TLS handshake with the server. The format of the
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000013279 string is defined in "man 1 ciphers" from OpenSSL man pages. For background
13280 information and recommendations see e.g.
13281 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
13282 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
13283 cipher configuration, please check the "ciphersuites" keyword.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020013284
Dirkjan Bussink415150f2018-09-14 11:14:21 +020013285ciphersuites <ciphersuites>
13286 This setting is only available when support for OpenSSL was built in and
13287 OpenSSL 1.1.1 or later was used to build HAProxy. This option sets the string
13288 describing the list of cipher algorithms that is negotiated during the TLS
13289 1.3 handshake with the server. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000013290 "man 1 ciphers" from OpenSSL man pages under the "ciphersuites" section.
13291 For cipher configuration for TLSv1.2 and earlier, please check the "ciphers"
13292 keyword.
Dirkjan Bussink415150f2018-09-14 11:14:21 +020013293
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013294cookie <value>
13295 The "cookie" parameter sets the cookie value assigned to the server to
13296 <value>. This value will be checked in incoming requests, and the first
13297 operational server possessing the same value will be selected. In return, in
13298 cookie insertion or rewrite modes, this value will be assigned to the cookie
13299 sent to the client. There is nothing wrong in having several servers sharing
13300 the same cookie value, and it is in fact somewhat common between normal and
13301 backup servers. See also the "cookie" keyword in backend section.
13302
Emeric Brunef42d922012-10-11 16:11:36 +020013303crl-file <crlfile>
13304 This setting is only available when support for OpenSSL was built in. It
13305 designates a PEM file from which to load certificate revocation list used
13306 to verify server's certificate.
13307
Emeric Bruna7aa3092012-10-26 12:58:00 +020013308crt <cert>
13309 This setting is only available when support for OpenSSL was built in.
13310 It designates a PEM file from which to load both a certificate and the
13311 associated private key. This file can be built by concatenating both PEM
13312 files into one. This certificate will be sent if the server send a client
13313 certificate request.
13314
Willy Tarreau96839092010-03-29 10:02:24 +020013315disabled
13316 The "disabled" keyword starts the server in the "disabled" state. That means
13317 that it is marked down in maintenance mode, and no connection other than the
13318 ones allowed by persist mode will reach it. It is very well suited to setup
13319 new servers, because normal traffic will never reach them, while it is still
13320 possible to test the service by making use of the force-persist mechanism.
Frédéric Lécailled2376272017-03-21 18:52:12 +010013321 See also "enabled" setting.
Willy Tarreau96839092010-03-29 10:02:24 +020013322
Frédéric Lécailled2376272017-03-21 18:52:12 +010013323enabled
13324 This option may be used as 'server' setting to reset any 'disabled'
13325 setting which would have been inherited from 'default-server' directive as
13326 default value.
13327 It may also be used as 'default-server' setting to reset any previous
13328 'default-server' 'disabled' setting.
Willy Tarreau96839092010-03-29 10:02:24 +020013329
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013330error-limit <count>
Willy Tarreau983e01e2010-01-11 18:42:06 +010013331 If health observing is enabled, the "error-limit" parameter specifies the
13332 number of consecutive errors that triggers event selected by the "on-error"
13333 option. By default it is set to 10 consecutive errors.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010013334
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013335 See also the "check", "error-limit" and "on-error".
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010013336
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013337fall <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013338 The "fall" parameter states that a server will be considered as dead after
13339 <count> consecutive unsuccessful health checks. This value defaults to 3 if
13340 unspecified. See also the "check", "inter" and "rise" parameters.
13341
Emeric Brun8694b9a2012-10-05 14:39:07 +020013342force-sslv3
13343 This option enforces use of SSLv3 only when SSL is used to communicate with
13344 the server. SSLv3 is generally less expensive than the TLS counterparts for
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013345 high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013346 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020013347
13348force-tlsv10
13349 This option enforces use of TLSv1.0 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013350 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013351 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020013352
13353force-tlsv11
13354 This option enforces use of TLSv1.1 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013355 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013356 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020013357
13358force-tlsv12
13359 This option enforces use of TLSv1.2 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013360 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013361 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020013362
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020013363force-tlsv13
13364 This option enforces use of TLSv1.3 only when SSL is used to communicate with
13365 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013366 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020013367
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013368id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +020013369 Set a persistent ID for the server. This ID must be positive and unique for
13370 the proxy. An unused ID will automatically be assigned if unset. The first
13371 assigned value will be 1. This ID is currently only returned in statistics.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013372
Willy Tarreau6a031d12016-11-07 19:42:35 +010013373init-addr {last | libc | none | <ip>},[...]*
13374 Indicate in what order the server's address should be resolved upon startup
13375 if it uses an FQDN. Attempts are made to resolve the address by applying in
Davor Ocelice9ed2812017-12-25 17:49:28 +010013376 turn each of the methods mentioned in the comma-delimited list. The first
Willy Tarreau6a031d12016-11-07 19:42:35 +010013377 method which succeeds is used. If the end of the list is reached without
13378 finding a working method, an error is thrown. Method "last" suggests to pick
13379 the address which appears in the state file (see "server-state-file"). Method
13380 "libc" uses the libc's internal resolver (gethostbyname() or getaddrinfo()
13381 depending on the operating system and build options). Method "none"
13382 specifically indicates that the server should start without any valid IP
13383 address in a down state. It can be useful to ignore some DNS issues upon
13384 startup, waiting for the situation to get fixed later. Finally, an IP address
13385 (IPv4 or IPv6) may be provided. It can be the currently known address of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010013386 server (e.g. filled by a configuration generator), or the address of a dummy
Willy Tarreau6a031d12016-11-07 19:42:35 +010013387 server used to catch old sessions and present them with a decent error
13388 message for example. When the "first" load balancing algorithm is used, this
13389 IP address could point to a fake server used to trigger the creation of new
13390 instances on the fly. This option defaults to "last,libc" indicating that the
13391 previous address found in the state file (if any) is used first, otherwise
13392 the libc's resolver is used. This ensures continued compatibility with the
Davor Ocelice9ed2812017-12-25 17:49:28 +010013393 historic behavior.
Willy Tarreau6a031d12016-11-07 19:42:35 +010013394
13395 Example:
13396 defaults
13397 # never fail on address resolution
13398 default-server init-addr last,libc,none
13399
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013400inter <delay>
13401fastinter <delay>
13402downinter <delay>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013403 The "inter" parameter sets the interval between two consecutive health checks
13404 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
13405 It is also possible to use "fastinter" and "downinter" to optimize delays
13406 between checks depending on the server state :
13407
Pieter Baauw44fc9df2015-09-17 21:30:46 +020013408 Server state | Interval used
13409 ----------------------------------------+----------------------------------
13410 UP 100% (non-transitional) | "inter"
13411 ----------------------------------------+----------------------------------
13412 Transitionally UP (going down "fall"), | "fastinter" if set,
13413 Transitionally DOWN (going up "rise"), | "inter" otherwise.
13414 or yet unchecked. |
13415 ----------------------------------------+----------------------------------
13416 DOWN 100% (non-transitional) | "downinter" if set,
13417 | "inter" otherwise.
13418 ----------------------------------------+----------------------------------
Willy Tarreaud72758d2010-01-12 10:42:19 +010013419
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013420 Just as with every other time-based parameter, they can be entered in any
13421 other explicit unit among { us, ms, s, m, h, d }. The "inter" parameter also
13422 serves as a timeout for health checks sent to servers if "timeout check" is
13423 not set. In order to reduce "resonance" effects when multiple servers are
Simon Hormand60d6912013-11-25 10:46:36 +090013424 hosted on the same hardware, the agent and health checks of all servers
13425 are started with a small time offset between them. It is also possible to
13426 add some random noise in the agent and health checks interval using the
13427 global "spread-checks" keyword. This makes sense for instance when a lot
13428 of backends use the same servers.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013429
Emeric Brun97556472020-05-30 01:42:45 +020013430log-proto <logproto>
13431 The "log-proto" specifies the protocol used to forward event messages to
13432 a server configured in a ring section. Possible values are "legacy"
13433 and "octet-count" corresponding respectively to "Non-transparent-framing"
13434 and "Octet counting" in rfc6587. "legacy" is the default.
13435
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013436maxconn <maxconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013437 The "maxconn" parameter specifies the maximal number of concurrent
13438 connections that will be sent to this server. If the number of incoming
Tim Duesterhuscefbbd92019-11-27 22:35:27 +010013439 concurrent connections goes higher than this value, they will be queued,
13440 waiting for a slot to be released. This parameter is very important as it can
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013441 save fragile servers from going down under extreme loads. If a "minconn"
13442 parameter is specified, the limit becomes dynamic. The default value is "0"
13443 which means unlimited. See also the "minconn" and "maxqueue" parameters, and
13444 the backend's "fullconn" keyword.
13445
Tim Duesterhuscefbbd92019-11-27 22:35:27 +010013446 In HTTP mode this parameter limits the number of concurrent requests instead
13447 of the number of connections. Multiple requests might be multiplexed over a
13448 single TCP connection to the server. As an example if you specify a maxconn
13449 of 50 you might see between 1 and 50 actual server connections, but no more
13450 than 50 concurrent requests.
13451
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013452maxqueue <maxqueue>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013453 The "maxqueue" parameter specifies the maximal number of connections which
13454 will wait in the queue for this server. If this limit is reached, next
13455 requests will be redispatched to other servers instead of indefinitely
13456 waiting to be served. This will break persistence but may allow people to
Willy Tarreau8ae8c482020-10-22 17:19:07 +020013457 quickly re-log in when the server they try to connect to is dying. Some load
13458 balancing algorithms such as leastconn take this into account and accept to
13459 add requests into a server's queue up to this value if it is explicitly set
13460 to a value greater than zero, which often allows to better smooth the load
13461 when dealing with single-digit maxconn values. The default value is "0" which
13462 means the queue is unlimited. See also the "maxconn" and "minconn" parameters
13463 and "balance leastconn".
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013464
Willy Tarreau9c538e02019-01-23 10:21:49 +010013465max-reuse <count>
13466 The "max-reuse" argument indicates the HTTP connection processors that they
13467 should not reuse a server connection more than this number of times to send
13468 new requests. Permitted values are -1 (the default), which disables this
13469 limit, or any positive value. Value zero will effectively disable keep-alive.
13470 This is only used to work around certain server bugs which cause them to leak
13471 resources over time. The argument is not necessarily respected by the lower
13472 layers as there might be technical limitations making it impossible to
13473 enforce. At least HTTP/2 connections to servers will respect it.
13474
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013475minconn <minconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013476 When the "minconn" parameter is set, the maxconn limit becomes a dynamic
13477 limit following the backend's load. The server will always accept at least
13478 <minconn> connections, never more than <maxconn>, and the limit will be on
13479 the ramp between both values when the backend has less than <fullconn>
13480 concurrent connections. This makes it possible to limit the load on the
13481 server during normal loads, but push it further for important loads without
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010013482 overloading the server during exceptional loads. See also the "maxconn"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013483 and "maxqueue" parameters, as well as the "fullconn" backend keyword.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010013484
Willy Tarreaud72f0f32015-10-13 14:50:22 +020013485namespace <name>
13486 On Linux, it is possible to specify which network namespace a socket will
13487 belong to. This directive makes it possible to explicitly bind a server to
13488 a namespace different from the default one. Please refer to your operating
13489 system's documentation to find more details about network namespaces.
13490
Frédéric Lécailled2376272017-03-21 18:52:12 +010013491no-agent-check
13492 This option may be used as "server" setting to reset any "agent-check"
13493 setting which would have been inherited from "default-server" directive as
13494 default value.
13495 It may also be used as "default-server" setting to reset any previous
13496 "default-server" "agent-check" setting.
13497
13498no-backup
13499 This option may be used as "server" setting to reset any "backup"
13500 setting which would have been inherited from "default-server" directive as
13501 default value.
13502 It may also be used as "default-server" setting to reset any previous
13503 "default-server" "backup" setting.
13504
13505no-check
13506 This option may be used as "server" setting to reset any "check"
13507 setting which would have been inherited from "default-server" directive as
13508 default value.
13509 It may also be used as "default-server" setting to reset any previous
13510 "default-server" "check" setting.
13511
13512no-check-ssl
13513 This option may be used as "server" setting to reset any "check-ssl"
13514 setting which would have been inherited from "default-server" directive as
13515 default value.
13516 It may also be used as "default-server" setting to reset any previous
13517 "default-server" "check-ssl" setting.
13518
Frédéric Lécailled2376272017-03-21 18:52:12 +010013519no-send-proxy
13520 This option may be used as "server" setting to reset any "send-proxy"
13521 setting which would have been inherited from "default-server" directive as
13522 default value.
13523 It may also be used as "default-server" setting to reset any previous
13524 "default-server" "send-proxy" setting.
13525
13526no-send-proxy-v2
13527 This option may be used as "server" setting to reset any "send-proxy-v2"
13528 setting which would have been inherited from "default-server" directive as
13529 default value.
13530 It may also be used as "default-server" setting to reset any previous
13531 "default-server" "send-proxy-v2" setting.
13532
13533no-send-proxy-v2-ssl
13534 This option may be used as "server" setting to reset any "send-proxy-v2-ssl"
13535 setting which would have been inherited from "default-server" directive as
13536 default value.
13537 It may also be used as "default-server" setting to reset any previous
13538 "default-server" "send-proxy-v2-ssl" setting.
13539
13540no-send-proxy-v2-ssl-cn
13541 This option may be used as "server" setting to reset any "send-proxy-v2-ssl-cn"
13542 setting which would have been inherited from "default-server" directive as
13543 default value.
13544 It may also be used as "default-server" setting to reset any previous
13545 "default-server" "send-proxy-v2-ssl-cn" setting.
13546
13547no-ssl
13548 This option may be used as "server" setting to reset any "ssl"
13549 setting which would have been inherited from "default-server" directive as
13550 default value.
13551 It may also be used as "default-server" setting to reset any previous
13552 "default-server" "ssl" setting.
13553
Willy Tarreau2a3fb1c2015-02-05 16:47:07 +010013554no-ssl-reuse
13555 This option disables SSL session reuse when SSL is used to communicate with
13556 the server. It will force the server to perform a full handshake for every
13557 new connection. It's probably only useful for benchmarking, troubleshooting,
13558 and for paranoid users.
13559
Emeric Brun9b3009b2012-10-05 11:55:06 +020013560no-sslv3
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020013561 This option disables support for SSLv3 when SSL is used to communicate with
13562 the server. Note that SSLv2 is disabled in the code and cannot be enabled
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013563 using any configuration option. Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020013564
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020013565 Supported in default-server: No
13566
Emeric Brunf9c5c472012-10-11 15:28:34 +020013567no-tls-tickets
13568 This setting is only available when support for OpenSSL was built in. It
13569 disables the stateless session resumption (RFC 5077 TLS Ticket
13570 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013571 session resumption is more expensive in CPU usage for servers. This option
13572 is also available on global statement "ssl-default-server-options".
Lukas Tribusbdb386d2020-03-10 00:56:09 +010013573 The TLS ticket mechanism is only used up to TLS 1.2.
13574 Forward Secrecy is compromised with TLS tickets, unless ticket keys
13575 are periodically rotated (via reload or by using "tls-ticket-keys").
Frédéric Lécailled2376272017-03-21 18:52:12 +010013576 See also "tls-tickets".
Emeric Brunf9c5c472012-10-11 15:28:34 +020013577
Emeric Brun9b3009b2012-10-05 11:55:06 +020013578no-tlsv10
Emeric Brun8694b9a2012-10-05 14:39:07 +020013579 This option disables support for TLSv1.0 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020013580 the server. Note that SSLv2 is disabled in the code and cannot be enabled
13581 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013582 often makes sense to disable it when communicating with local servers. This
13583 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013584 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020013585
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020013586 Supported in default-server: No
13587
Emeric Brun9b3009b2012-10-05 11:55:06 +020013588no-tlsv11
Emeric Brun8694b9a2012-10-05 14:39:07 +020013589 This option disables support for TLSv1.1 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020013590 the server. Note that SSLv2 is disabled in the code and cannot be enabled
13591 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013592 often makes sense to disable it when communicating with local servers. This
13593 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013594 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020013595
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020013596 Supported in default-server: No
13597
Emeric Brun9b3009b2012-10-05 11:55:06 +020013598no-tlsv12
Emeric Brun8694b9a2012-10-05 14:39:07 +020013599 This option disables support for TLSv1.2 when SSL is used to communicate with
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020013600 the server. Note that SSLv2 is disabled in the code and cannot be enabled
13601 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013602 often makes sense to disable it when communicating with local servers. This
13603 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013604 Use "ssl-min-ver" and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020013605
13606 Supported in default-server: No
13607
13608no-tlsv13
13609 This option disables support for TLSv1.3 when SSL is used to communicate with
13610 the server. Note that SSLv2 is disabled in the code and cannot be enabled
13611 using any configuration option. TLSv1 is more expensive than SSLv3 so it
13612 often makes sense to disable it when communicating with local servers. This
13613 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013614 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020013615
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020013616 Supported in default-server: No
13617
Frédéric Lécailled2376272017-03-21 18:52:12 +010013618no-verifyhost
13619 This option may be used as "server" setting to reset any "verifyhost"
13620 setting which would have been inherited from "default-server" directive as
13621 default value.
13622 It may also be used as "default-server" setting to reset any previous
13623 "default-server" "verifyhost" setting.
Willy Tarreau763a95b2012-10-04 23:15:39 +020013624
Frédéric Lécaille1b9423d2019-07-04 14:19:06 +020013625no-tfo
13626 This option may be used as "server" setting to reset any "tfo"
13627 setting which would have been inherited from "default-server" directive as
13628 default value.
13629 It may also be used as "default-server" setting to reset any previous
13630 "default-server" "tfo" setting.
13631
Simon Hormanfa461682011-06-25 09:39:49 +090013632non-stick
13633 Never add connections allocated to this sever to a stick-table.
13634 This may be used in conjunction with backup to ensure that
13635 stick-table persistence is disabled for backup servers.
13636
Olivier Houchardc7566002018-11-20 23:33:50 +010013637npn <protocols>
13638 This enables the NPN TLS extension and advertises the specified protocol list
13639 as supported on top of NPN. The protocol list consists in a comma-delimited
13640 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
John Roeslerfb2fce12019-07-10 15:45:51 -050013641 This requires that the SSL library is built with support for TLS extensions
Olivier Houchardc7566002018-11-20 23:33:50 +010013642 enabled (check with haproxy -vv). Note that the NPN extension has been
13643 replaced with the ALPN extension (see the "alpn" keyword), though this one is
13644 only available starting with OpenSSL 1.0.2.
13645
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010013646observe <mode>
13647 This option enables health adjusting based on observing communication with
13648 the server. By default this functionality is disabled and enabling it also
13649 requires to enable health checks. There are two supported modes: "layer4" and
13650 "layer7". In layer4 mode, only successful/unsuccessful tcp connections are
13651 significant. In layer7, which is only allowed for http proxies, responses
13652 received from server are verified, like valid/wrong http code, unparsable
Willy Tarreau150d1462012-03-10 08:19:02 +010013653 headers, a timeout, etc. Valid status codes include 100 to 499, 501 and 505.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010013654
13655 See also the "check", "on-error" and "error-limit".
13656
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013657on-error <mode>
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010013658 Select what should happen when enough consecutive errors are detected.
13659 Currently, four modes are available:
13660 - fastinter: force fastinter
13661 - fail-check: simulate a failed check, also forces fastinter (default)
13662 - sudden-death: simulate a pre-fatal failed health check, one more failed
13663 check will mark a server down, forces fastinter
13664 - mark-down: mark the server immediately down and force fastinter
13665
13666 See also the "check", "observe" and "error-limit".
13667
Simon Hormane0d1bfb2011-06-21 14:34:58 +090013668on-marked-down <action>
13669 Modify what occurs when a server is marked down.
13670 Currently one action is available:
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070013671 - shutdown-sessions: Shutdown peer sessions. When this setting is enabled,
13672 all connections to the server are immediately terminated when the server
13673 goes down. It might be used if the health check detects more complex cases
13674 than a simple connection status, and long timeouts would cause the service
13675 to remain unresponsive for too long a time. For instance, a health check
13676 might detect that a database is stuck and that there's no chance to reuse
13677 existing connections anymore. Connections killed this way are logged with
13678 a 'D' termination code (for "Down").
Simon Hormane0d1bfb2011-06-21 14:34:58 +090013679
13680 Actions are disabled by default
13681
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070013682on-marked-up <action>
13683 Modify what occurs when a server is marked up.
13684 Currently one action is available:
13685 - shutdown-backup-sessions: Shutdown sessions on all backup servers. This is
13686 done only if the server is not in backup state and if it is not disabled
13687 (it must have an effective weight > 0). This can be used sometimes to force
13688 an active server to take all the traffic back after recovery when dealing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013689 with long sessions (e.g. LDAP, SQL, ...). Doing this can cause more trouble
13690 than it tries to solve (e.g. incomplete transactions), so use this feature
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070013691 with extreme care. Sessions killed because a server comes up are logged
13692 with an 'U' termination code (for "Up").
13693
13694 Actions are disabled by default
13695
Willy Tarreau2f3f4d32020-07-01 07:43:51 +020013696pool-low-conn <max>
13697 Set a low threshold on the number of idling connections for a server, below
13698 which a thread will not try to steal a connection from another thread. This
13699 can be useful to improve CPU usage patterns in scenarios involving many very
13700 fast servers, in order to ensure all threads will keep a few idle connections
13701 all the time instead of letting them accumulate over one thread and migrating
13702 them from thread to thread. Typical values of twice the number of threads
13703 seem to show very good performance already with sub-millisecond response
13704 times. The default is zero, indicating that any idle connection can be used
13705 at any time. It is the recommended setting for normal use. This only applies
13706 to connections that can be shared according to the same principles as those
13707 applying to "http-reuse".
13708
Olivier Houchard006e3102018-12-10 18:30:32 +010013709pool-max-conn <max>
13710 Set the maximum number of idling connections for a server. -1 means unlimited
13711 connections, 0 means no idle connections. The default is -1. When idle
13712 connections are enabled, orphaned idle connections which do not belong to any
13713 client session anymore are moved to a dedicated pool so that they remain
13714 usable by future clients. This only applies to connections that can be shared
13715 according to the same principles as those applying to "http-reuse".
13716
Olivier Houchardb7b3faa2018-12-14 18:15:36 +010013717pool-purge-delay <delay>
13718 Sets the delay to start purging idle connections. Each <delay> interval, half
Olivier Houcharda56eebf2019-03-19 16:44:02 +010013719 of the idle connections are closed. 0 means we don't keep any idle connection.
Willy Tarreaufb553652019-06-04 14:06:31 +020013720 The default is 5s.
Olivier Houchardb7b3faa2018-12-14 18:15:36 +010013721
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013722port <port>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013723 Using the "port" parameter, it becomes possible to use a different port to
13724 send health-checks. On some servers, it may be desirable to dedicate a port
13725 to a specific component able to perform complex tests which are more suitable
13726 to health-checks than the application. It is common to run a simple script in
13727 inetd for instance. This parameter is ignored if the "check" parameter is not
13728 set. See also the "addr" parameter.
13729
Christopher Faulet8ed0a3e2018-04-10 14:45:45 +020013730proto <name>
Christopher Faulet8ed0a3e2018-04-10 14:45:45 +020013731 Forces the multiplexer's protocol to use for the outgoing connections to this
13732 server. It must be compatible with the mode of the backend (TCP or HTTP). It
13733 must also be usable on the backend side. The list of available protocols is
13734 reported in haproxy -vv.
Daniel Corbett67a82712020-07-06 23:01:19 -040013735 Idea behind this option is to bypass the selection of the best multiplexer's
Christopher Faulet8ed0a3e2018-04-10 14:45:45 +020013736 protocol for all connections established to this server.
13737
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013738redir <prefix>
13739 The "redir" parameter enables the redirection mode for all GET and HEAD
13740 requests addressing this server. This means that instead of having HAProxy
13741 forward the request to the server, it will send an "HTTP 302" response with
13742 the "Location" header composed of this prefix immediately followed by the
13743 requested URI beginning at the leading '/' of the path component. That means
13744 that no trailing slash should be used after <prefix>. All invalid requests
13745 will be rejected, and all non-GET or HEAD requests will be normally served by
13746 the server. Note that since the response is completely forged, no header
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010013747 mangling nor cookie insertion is possible in the response. However, cookies in
Davor Ocelice9ed2812017-12-25 17:49:28 +010013748 requests are still analyzed, making this solution completely usable to direct
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013749 users to a remote location in case of local disaster. Main use consists in
13750 increasing bandwidth for static servers by having the clients directly
13751 connect to them. Note: never use a relative location here, it would cause a
13752 loop between the client and HAProxy!
13753
13754 Example : server srv1 192.168.1.1:80 redir http://image1.mydomain.com check
13755
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013756rise <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013757 The "rise" parameter states that a server will be considered as operational
13758 after <count> consecutive successful health checks. This value defaults to 2
13759 if unspecified. See also the "check", "inter" and "fall" parameters.
13760
Baptiste Assmann8e2d9432018-06-22 15:04:43 +020013761resolve-opts <option>,<option>,...
13762 Comma separated list of options to apply to DNS resolution linked to this
13763 server.
13764
13765 Available options:
13766
13767 * allow-dup-ip
13768 By default, HAProxy prevents IP address duplication in a backend when DNS
13769 resolution at runtime is in operation.
13770 That said, for some cases, it makes sense that two servers (in the same
13771 backend, being resolved by the same FQDN) have the same IP address.
13772 For such case, simply enable this option.
13773 This is the opposite of prevent-dup-ip.
13774
Daniel Corbettf8716912019-11-17 09:48:56 -050013775 * ignore-weight
13776 Ignore any weight that is set within an SRV record. This is useful when
13777 you would like to control the weights using an alternate method, such as
13778 using an "agent-check" or through the runtime api.
13779
Baptiste Assmann8e2d9432018-06-22 15:04:43 +020013780 * prevent-dup-ip
13781 Ensure HAProxy's default behavior is enforced on a server: prevent re-using
13782 an IP address already set to a server in the same backend and sharing the
13783 same fqdn.
13784 This is the opposite of allow-dup-ip.
13785
13786 Example:
13787 backend b_myapp
13788 default-server init-addr none resolvers dns
13789 server s1 myapp.example.com:80 check resolve-opts allow-dup-ip
13790 server s2 myapp.example.com:81 check resolve-opts allow-dup-ip
13791
13792 With the option allow-dup-ip set:
13793 * if the nameserver returns a single IP address, then both servers will use
13794 it
13795 * If the nameserver returns 2 IP addresses, then each server will pick up a
13796 different address
13797
13798 Default value: not set
13799
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013800resolve-prefer <family>
13801 When DNS resolution is enabled for a server and multiple IP addresses from
13802 different families are returned, HAProxy will prefer using an IP address
13803 from the family mentioned in the "resolve-prefer" parameter.
13804 Available families: "ipv4" and "ipv6"
13805
Baptiste Assmannc4aabae2015-08-04 22:43:06 +020013806 Default value: ipv6
13807
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020013808 Example:
13809
13810 server s1 app1.domain.com:80 resolvers mydns resolve-prefer ipv6
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013811
Thierry Fournierac88cfe2016-02-17 22:05:30 +010013812resolve-net <network>[,<network[,...]]
John Roeslerfb2fce12019-07-10 15:45:51 -050013813 This option prioritizes the choice of an ip address matching a network. This is
Thierry Fournierac88cfe2016-02-17 22:05:30 +010013814 useful with clouds to prefer a local ip. In some cases, a cloud high
Tim Düsterhus4896c442016-11-29 02:15:19 +010013815 availability service can be announced with many ip addresses on many
Davor Ocelice9ed2812017-12-25 17:49:28 +010013816 different datacenters. The latency between datacenter is not negligible, so
13817 this patch permits to prefer a local datacenter. If no address matches the
Thierry Fournierac88cfe2016-02-17 22:05:30 +010013818 configured network, another address is selected.
13819
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020013820 Example:
13821
13822 server s1 app1.domain.com:80 resolvers mydns resolve-net 10.0.0.0/8
Thierry Fournierac88cfe2016-02-17 22:05:30 +010013823
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013824resolvers <id>
13825 Points to an existing "resolvers" section to resolve current server's
13826 hostname.
13827
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020013828 Example:
13829
13830 server s1 app1.domain.com:80 check resolvers mydns
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013831
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020013832 See also section 5.3
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013833
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010013834send-proxy
13835 The "send-proxy" parameter enforces use of the PROXY protocol over any
13836 connection established to this server. The PROXY protocol informs the other
13837 end about the layer 3/4 addresses of the incoming connection, so that it can
13838 know the client's address or the public address it accessed to, whatever the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010013839 upper layer protocol. For connections accepted by an "accept-proxy" or
13840 "accept-netscaler-cip" listener, the advertised address will be used. Only
13841 TCPv4 and TCPv6 address families are supported. Other families such as
13842 Unix sockets, will report an UNKNOWN family. Servers using this option can
13843 fully be chained to another instance of haproxy listening with an
13844 "accept-proxy" setting. This setting must not be used if the server isn't
13845 aware of the protocol. When health checks are sent to the server, the PROXY
13846 protocol is automatically used when this option is set, unless there is an
13847 explicit "port" or "addr" directive, in which case an explicit
13848 "check-send-proxy" directive would also be needed to use the PROXY protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010013849 See also the "no-send-proxy" option of this section and "accept-proxy" and
13850 "accept-netscaler-cip" option of the "bind" keyword.
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010013851
David Safb76832014-05-08 23:42:08 -040013852send-proxy-v2
13853 The "send-proxy-v2" parameter enforces use of the PROXY protocol version 2
13854 over any connection established to this server. The PROXY protocol informs
13855 the other end about the layer 3/4 addresses of the incoming connection, so
13856 that it can know the client's address or the public address it accessed to,
Emmanuel Hocdet404d9782017-10-24 10:55:14 +020013857 whatever the upper layer protocol. It also send ALPN information if an alpn
13858 have been negotiated. This setting must not be used if the server isn't aware
13859 of this version of the protocol. See also the "no-send-proxy-v2" option of
13860 this section and send-proxy" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040013861
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010013862proxy-v2-options <option>[,<option>]*
Tim Duesterhuscf6e0c82020-03-13 12:34:24 +010013863 The "proxy-v2-options" parameter add options to send in PROXY protocol
13864 version 2 when "send-proxy-v2" is used. Options available are:
13865
13866 - ssl : See also "send-proxy-v2-ssl".
13867 - cert-cn : See also "send-proxy-v2-ssl-cn".
13868 - ssl-cipher: Name of the used cipher.
13869 - cert-sig : Signature algorithm of the used certificate.
13870 - cert-key : Key algorithm of the used certificate
13871 - authority : Host name value passed by the client (only SNI from a TLS
13872 connection is supported).
13873 - crc32c : Checksum of the PROXYv2 header.
13874 - unique-id : Send a unique ID generated using the frontend's
13875 "unique-id-format" within the PROXYv2 header.
13876 This unique-id is primarily meant for "mode tcp". It can
13877 lead to unexpected results in "mode http", because the
13878 generated unique ID is also used for the first HTTP request
13879 within a Keep-Alive connection.
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010013880
David Safb76832014-05-08 23:42:08 -040013881send-proxy-v2-ssl
13882 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
13883 2 over any connection established to this server. The PROXY protocol informs
13884 the other end about the layer 3/4 addresses of the incoming connection, so
13885 that it can know the client's address or the public address it accessed to,
13886 whatever the upper layer protocol. In addition, the SSL information extension
13887 of the PROXY protocol is added to the PROXY protocol header. This setting
13888 must not be used if the server isn't aware of this version of the protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010013889 See also the "no-send-proxy-v2-ssl" option of this section and the
13890 "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040013891
13892send-proxy-v2-ssl-cn
13893 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
13894 2 over any connection established to this server. The PROXY protocol informs
13895 the other end about the layer 3/4 addresses of the incoming connection, so
13896 that it can know the client's address or the public address it accessed to,
13897 whatever the upper layer protocol. In addition, the SSL information extension
13898 of the PROXY protocol, along along with the Common Name from the subject of
13899 the client certificate (if any), is added to the PROXY protocol header. This
13900 setting must not be used if the server isn't aware of this version of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010013901 protocol. See also the "no-send-proxy-v2-ssl-cn" option of this section and
13902 the "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040013903
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013904slowstart <start_time_in_ms>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013905 The "slowstart" parameter for a server accepts a value in milliseconds which
13906 indicates after how long a server which has just come back up will run at
13907 full speed. Just as with every other time-based parameter, it can be entered
13908 in any other explicit unit among { us, ms, s, m, h, d }. The speed grows
13909 linearly from 0 to 100% during this time. The limitation applies to two
13910 parameters :
13911
13912 - maxconn: the number of connections accepted by the server will grow from 1
13913 to 100% of the usual dynamic limit defined by (minconn,maxconn,fullconn).
13914
13915 - weight: when the backend uses a dynamic weighted algorithm, the weight
13916 grows linearly from 1 to 100%. In this case, the weight is updated at every
13917 health-check. For this reason, it is important that the "inter" parameter
13918 is smaller than the "slowstart", in order to maximize the number of steps.
13919
13920 The slowstart never applies when haproxy starts, otherwise it would cause
13921 trouble to running servers. It only applies when a server has been previously
13922 seen as failed.
13923
Willy Tarreau732eac42015-07-09 11:40:25 +020013924sni <expression>
13925 The "sni" parameter evaluates the sample fetch expression, converts it to a
13926 string and uses the result as the host name sent in the SNI TLS extension to
13927 the server. A typical use case is to send the SNI received from the client in
13928 a bridged HTTPS scenario, using the "ssl_fc_sni" sample fetch for the
Willy Tarreau2ab88672017-07-05 18:23:03 +020013929 expression, though alternatives such as req.hdr(host) can also make sense. If
13930 "verify required" is set (which is the recommended setting), the resulting
Willy Tarreauad92a9a2017-07-28 11:38:41 +020013931 name will also be matched against the server certificate's names. See the
Jérôme Magninb36a6d22018-12-09 16:03:40 +010013932 "verify" directive for more details. If you want to set a SNI for health
13933 checks, see the "check-sni" directive for more details.
Willy Tarreau732eac42015-07-09 11:40:25 +020013934
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020013935source <addr>[:<pl>[-<ph>]] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +020013936source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020013937source <addr>[:<pl>[-<ph>]] [interface <name>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013938 The "source" parameter sets the source address which will be used when
13939 connecting to the server. It follows the exact same parameters and principle
13940 as the backend "source" keyword, except that it only applies to the server
13941 referencing it. Please consult the "source" keyword for details.
13942
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020013943 Additionally, the "source" statement on a server line allows one to specify a
13944 source port range by indicating the lower and higher bounds delimited by a
13945 dash ('-'). Some operating systems might require a valid IP address when a
13946 source port range is specified. It is permitted to have the same IP/range for
13947 several servers. Doing so makes it possible to bypass the maximum of 64k
13948 total concurrent connections. The limit will then reach 64k connections per
13949 server.
13950
Lukas Tribus7d56c6d2016-09-13 09:51:15 +000013951 Since Linux 4.2/libc 2.23 IP_BIND_ADDRESS_NO_PORT is set for connections
13952 specifying the source address without port(s).
13953
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020013954ssl
Willy Tarreau44f65392013-06-25 07:56:20 +020013955 This option enables SSL ciphering on outgoing connections to the server. It
13956 is critical to verify server certificates using "verify" when using SSL to
13957 connect to servers, otherwise the communication is prone to trivial man in
13958 the-middle attacks rendering SSL useless. When this option is used, health
13959 checks are automatically sent in SSL too unless there is a "port" or an
13960 "addr" directive indicating the check should be sent to a different location.
Frédéric Lécailled2376272017-03-21 18:52:12 +010013961 See the "no-ssl" to disable "ssl" option and "check-ssl" option to force
13962 SSL health checks.
Willy Tarreau763a95b2012-10-04 23:15:39 +020013963
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013964ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
13965 This option enforces use of <version> or lower when SSL is used to communicate
13966 with the server. This option is also available on global statement
13967 "ssl-default-server-options". See also "ssl-min-ver".
13968
13969ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
13970 This option enforces use of <version> or upper when SSL is used to communicate
13971 with the server. This option is also available on global statement
13972 "ssl-default-server-options". See also "ssl-max-ver".
13973
Frédéric Lécailled2376272017-03-21 18:52:12 +010013974ssl-reuse
13975 This option may be used as "server" setting to reset any "no-ssl-reuse"
13976 setting which would have been inherited from "default-server" directive as
13977 default value.
13978 It may also be used as "default-server" setting to reset any previous
13979 "default-server" "no-ssl-reuse" setting.
13980
13981stick
13982 This option may be used as "server" setting to reset any "non-stick"
13983 setting which would have been inherited from "default-server" directive as
13984 default value.
13985 It may also be used as "default-server" setting to reset any previous
13986 "default-server" "non-stick" setting.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020013987
Alexander Liu2a54bb72019-05-22 19:44:48 +080013988socks4 <addr>:<port>
John Roeslerfb2fce12019-07-10 15:45:51 -050013989 This option enables upstream socks4 tunnel for outgoing connections to the
Alexander Liu2a54bb72019-05-22 19:44:48 +080013990 server. Using this option won't force the health check to go via socks4 by
13991 default. You will have to use the keyword "check-via-socks4" to enable it.
13992
Willy Tarreau163d4622015-10-13 16:16:41 +020013993tcp-ut <delay>
13994 Sets the TCP User Timeout for all outgoing connections to this server. This
13995 option is available on Linux since version 2.6.37. It allows haproxy to
13996 configure a timeout for sockets which contain data not receiving an
Davor Ocelice9ed2812017-12-25 17:49:28 +010013997 acknowledgment for the configured delay. This is especially useful on
Willy Tarreau163d4622015-10-13 16:16:41 +020013998 long-lived connections experiencing long idle periods such as remote
13999 terminals or database connection pools, where the client and server timeouts
14000 must remain high to allow a long period of idle, but where it is important to
14001 detect that the server has disappeared in order to release all resources
14002 associated with its connection (and the client's session). One typical use
14003 case is also to force dead server connections to die when health checks are
14004 too slow or during a soft reload since health checks are then disabled. The
14005 argument is a delay expressed in milliseconds by default. This only works for
14006 regular TCP connections, and is ignored for other protocols.
14007
Willy Tarreau034c88c2017-01-23 23:36:45 +010014008tfo
14009 This option enables using TCP fast open when connecting to servers, on
14010 systems that support it (currently only the Linux kernel >= 4.11).
14011 See the "tfo" bind option for more information about TCP fast open.
14012 Please note that when using tfo, you should also use the "conn-failure",
14013 "empty-response" and "response-timeout" keywords for "retry-on", or haproxy
Frédéric Lécaille1b9423d2019-07-04 14:19:06 +020014014 won't be able to retry the connection on failure. See also "no-tfo".
Willy Tarreau034c88c2017-01-23 23:36:45 +010014015
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014016track [<proxy>/]<server>
Willy Tarreau32091232014-05-16 13:52:00 +020014017 This option enables ability to set the current state of the server by tracking
14018 another one. It is possible to track a server which itself tracks another
14019 server, provided that at the end of the chain, a server has health checks
14020 enabled. If <proxy> is omitted the current one is used. If disable-on-404 is
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014021 used, it has to be enabled on both proxies.
14022
Frédéric Lécailled2376272017-03-21 18:52:12 +010014023tls-tickets
14024 This option may be used as "server" setting to reset any "no-tls-tickets"
14025 setting which would have been inherited from "default-server" directive as
14026 default value.
Lukas Tribusbdb386d2020-03-10 00:56:09 +010014027 The TLS ticket mechanism is only used up to TLS 1.2.
14028 Forward Secrecy is compromised with TLS tickets, unless ticket keys
14029 are periodically rotated (via reload or by using "tls-ticket-keys").
Frédéric Lécailled2376272017-03-21 18:52:12 +010014030 It may also be used as "default-server" setting to reset any previous
Bjoern Jacke5ab7eb62020-02-13 14:16:16 +010014031 "default-server" "no-tls-tickets" setting.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014032
Emeric Brunef42d922012-10-11 16:11:36 +020014033verify [none|required]
14034 This setting is only available when support for OpenSSL was built in. If set
Emeric Brun850efd52014-01-29 12:24:34 +010014035 to 'none', server certificate is not verified. In the other case, The
Willy Tarreauad92a9a2017-07-28 11:38:41 +020014036 certificate provided by the server is verified using CAs from 'ca-file' and
14037 optional CRLs from 'crl-file' after having checked that the names provided in
Davor Ocelice9ed2812017-12-25 17:49:28 +010014038 the certificate's subject and subjectAlternateNames attributes match either
Willy Tarreauad92a9a2017-07-28 11:38:41 +020014039 the name passed using the "sni" directive, or if not provided, the static
14040 host name passed using the "verifyhost" directive. When no name is found, the
14041 certificate's names are ignored. For this reason, without SNI it's important
14042 to use "verifyhost". On verification failure the handshake is aborted. It is
14043 critically important to verify server certificates when using SSL to connect
14044 to servers, otherwise the communication is prone to trivial man-in-the-middle
14045 attacks rendering SSL totally useless. Unless "ssl_server_verify" appears in
14046 the global section, "verify" is set to "required" by default.
Emeric Brunef42d922012-10-11 16:11:36 +020014047
Evan Broderbe554312013-06-27 00:05:25 -070014048verifyhost <hostname>
14049 This setting is only available when support for OpenSSL was built in, and
Willy Tarreauad92a9a2017-07-28 11:38:41 +020014050 only takes effect if 'verify required' is also specified. This directive sets
14051 a default static hostname to check the server's certificate against when no
14052 SNI was used to connect to the server. If SNI is not used, this is the only
14053 way to enable hostname verification. This static hostname, when set, will
14054 also be used for health checks (which cannot provide an SNI value). If none
14055 of the hostnames in the certificate match the specified hostname, the
14056 handshake is aborted. The hostnames in the server-provided certificate may
14057 include wildcards. See also "verify", "sni" and "no-verifyhost" options.
Evan Broderbe554312013-06-27 00:05:25 -070014058
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014059weight <weight>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014060 The "weight" parameter is used to adjust the server's weight relative to
14061 other servers. All servers will receive a load proportional to their weight
14062 relative to the sum of all weights, so the higher the weight, the higher the
Willy Tarreau6704d672009-06-15 10:56:05 +020014063 load. The default weight is 1, and the maximal value is 256. A value of 0
14064 means the server will not participate in load-balancing but will still accept
14065 persistent connections. If this parameter is used to distribute the load
14066 according to server's capacity, it is recommended to start with values which
14067 can both grow and shrink, for instance between 10 and 100 to leave enough
14068 room above and below for later adjustments.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014069
14070
Cyril Bonté46175dd2015-07-02 22:45:32 +0200140715.3. Server IP address resolution using DNS
14072-------------------------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014073
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014074HAProxy allows using a host name on the server line to retrieve its IP address
14075using name servers. By default, HAProxy resolves the name when parsing the
14076configuration file, at startup and cache the result for the process' life.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014077This is not sufficient in some cases, such as in Amazon where a server's IP
14078can change after a reboot or an ELB Virtual IP can change based on current
14079workload.
14080This chapter describes how HAProxy can be configured to process server's name
14081resolution at run time.
14082Whether run time server name resolution has been enable or not, HAProxy will
14083carry on doing the first resolution when parsing the configuration.
14084
14085
Cyril Bonté46175dd2015-07-02 22:45:32 +0200140865.3.1. Global overview
14087----------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014088
14089As we've seen in introduction, name resolution in HAProxy occurs at two
14090different steps of the process life:
14091
14092 1. when starting up, HAProxy parses the server line definition and matches a
14093 host name. It uses libc functions to get the host name resolved. This
14094 resolution relies on /etc/resolv.conf file.
14095
Christopher Faulet67957bd2017-09-27 11:00:59 +020014096 2. at run time, HAProxy performs periodically name resolutions for servers
14097 requiring DNS resolutions.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014098
14099A few other events can trigger a name resolution at run time:
14100 - when a server's health check ends up in a connection timeout: this may be
14101 because the server has a new IP address. So we need to trigger a name
14102 resolution to know this new IP.
14103
Christopher Faulet67957bd2017-09-27 11:00:59 +020014104When using resolvers, the server name can either be a hostname, or a SRV label.
Davor Ocelice9ed2812017-12-25 17:49:28 +010014105HAProxy considers anything that starts with an underscore as a SRV label. If a
Christopher Faulet67957bd2017-09-27 11:00:59 +020014106SRV label is specified, then the corresponding SRV records will be retrieved
14107from the DNS server, and the provided hostnames will be used. The SRV label
14108will be checked periodically, and if any server are added or removed, haproxy
14109will automatically do the same.
Olivier Houchardecfa18d2017-08-07 17:30:03 +020014110
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014111A few things important to notice:
John Roeslerfb2fce12019-07-10 15:45:51 -050014112 - all the name servers are queried in the meantime. HAProxy will process the
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014113 first valid response.
14114
14115 - a resolution is considered as invalid (NX, timeout, refused), when all the
14116 servers return an error.
14117
14118
Cyril Bonté46175dd2015-07-02 22:45:32 +0200141195.3.2. The resolvers section
14120----------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014121
14122This section is dedicated to host information related to name resolution in
Christopher Faulet67957bd2017-09-27 11:00:59 +020014123HAProxy. There can be as many as resolvers section as needed. Each section can
14124contain many name servers.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014125
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014126When multiple name servers are configured in a resolvers section, then HAProxy
14127uses the first valid response. In case of invalid responses, only the last one
14128is treated. Purpose is to give the chance to a slow server to deliver a valid
14129answer after a fast faulty or outdated server.
14130
14131When each server returns a different error type, then only the last error is
Christopher Faulet67957bd2017-09-27 11:00:59 +020014132used by HAProxy. The following processing is applied on this error:
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014133
Christopher Faulet67957bd2017-09-27 11:00:59 +020014134 1. HAProxy retries the same DNS query with a new query type. The A queries are
14135 switch to AAAA or the opposite. SRV queries are not concerned here. Timeout
14136 errors are also excluded.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014137
Christopher Faulet67957bd2017-09-27 11:00:59 +020014138 2. When the fallback on the query type was done (or not applicable), HAProxy
14139 retries the original DNS query, with the preferred query type.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014140
Christopher Faulet67957bd2017-09-27 11:00:59 +020014141 3. HAProxy retries previous steps <resolve_retires> times. If no valid
14142 response is received after that, it stops the DNS resolution and reports
14143 the error.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014144
Christopher Faulet67957bd2017-09-27 11:00:59 +020014145For example, with 2 name servers configured in a resolvers section, the
14146following scenarios are possible:
14147
14148 - First response is valid and is applied directly, second response is
14149 ignored
14150
14151 - First response is invalid and second one is valid, then second response is
14152 applied
14153
14154 - First response is a NX domain and second one a truncated response, then
14155 HAProxy retries the query with a new type
14156
14157 - First response is a NX domain and second one is a timeout, then HAProxy
14158 retries the query with a new type
14159
14160 - Query timed out for both name servers, then HAProxy retries it with the
14161 same query type
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014162
Olivier Houcharda8c6db82017-07-06 18:46:47 +020014163As a DNS server may not answer all the IPs in one DNS request, haproxy keeps
14164a cache of previous answers, an answer will be considered obsolete after
Christopher Faulet67957bd2017-09-27 11:00:59 +020014165<hold obsolete> seconds without the IP returned.
Olivier Houcharda8c6db82017-07-06 18:46:47 +020014166
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014167
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014168resolvers <resolvers id>
Davor Ocelice9ed2812017-12-25 17:49:28 +010014169 Creates a new name server list labeled <resolvers id>
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014170
14171A resolvers section accept the following parameters:
14172
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020014173accepted_payload_size <nb>
Davor Ocelice9ed2812017-12-25 17:49:28 +010014174 Defines the maximum payload size accepted by HAProxy and announced to all the
Christopher Faulet67957bd2017-09-27 11:00:59 +020014175 name servers configured in this resolvers section.
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020014176 <nb> is in bytes. If not set, HAProxy announces 512. (minimal value defined
14177 by RFC 6891)
14178
Baptiste Assmann9d8dbbc2017-08-18 23:35:08 +020014179 Note: the maximum allowed value is 8192.
14180
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014181nameserver <id> <ip>:<port>
14182 DNS server description:
14183 <id> : label of the server, should be unique
14184 <ip> : IP address of the server
14185 <port> : port where the DNS service actually runs
14186
Ben Draut44e609b2018-05-29 15:40:08 -060014187parse-resolv-conf
14188 Adds all nameservers found in /etc/resolv.conf to this resolvers nameservers
14189 list. Ordered as if each nameserver in /etc/resolv.conf was individually
14190 placed in the resolvers section in place of this directive.
14191
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014192hold <status> <period>
14193 Defines <period> during which the last name resolution should be kept based
14194 on last resolution <status>
Baptiste Assmann987e16d2016-11-02 22:23:31 +010014195 <status> : last name resolution status. Acceptable values are "nx",
Olivier Houcharda8c6db82017-07-06 18:46:47 +020014196 "other", "refused", "timeout", "valid", "obsolete".
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014197 <period> : interval between two successive name resolution when the last
14198 answer was in <status>. It follows the HAProxy time format.
14199 <period> is in milliseconds by default.
14200
Baptiste Assmann686408b2017-08-18 10:15:42 +020014201 Default value is 10s for "valid", 0s for "obsolete" and 30s for others.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014202
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014203resolve_retries <nb>
14204 Defines the number <nb> of queries to send to resolve a server name before
14205 giving up.
14206 Default value: 3
14207
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014208 A retry occurs on name server timeout or when the full sequence of DNS query
14209 type failover is over and we need to start up from the default ANY query
14210 type.
14211
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014212timeout <event> <time>
14213 Defines timeouts related to name resolution
14214 <event> : the event on which the <time> timeout period applies to.
14215 events available are:
Frédéric Lécaille93d33162019-03-06 09:35:59 +010014216 - resolve : default time to trigger name resolutions when no
14217 other time applied.
Christopher Faulet67957bd2017-09-27 11:00:59 +020014218 Default value: 1s
14219 - retry : time between two DNS queries, when no valid response
Frédéric Lécaille93d33162019-03-06 09:35:59 +010014220 have been received.
Christopher Faulet67957bd2017-09-27 11:00:59 +020014221 Default value: 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014222 <time> : time related to the event. It follows the HAProxy time format.
14223 <time> is expressed in milliseconds.
14224
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020014225 Example:
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014226
14227 resolvers mydns
14228 nameserver dns1 10.0.0.1:53
14229 nameserver dns2 10.0.0.2:53
Ben Draut44e609b2018-05-29 15:40:08 -060014230 parse-resolv-conf
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014231 resolve_retries 3
Christopher Faulet67957bd2017-09-27 11:00:59 +020014232 timeout resolve 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014233 timeout retry 1s
Baptiste Assmann987e16d2016-11-02 22:23:31 +010014234 hold other 30s
14235 hold refused 30s
14236 hold nx 30s
14237 hold timeout 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014238 hold valid 10s
Olivier Houcharda8c6db82017-07-06 18:46:47 +020014239 hold obsolete 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014240
14241
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200142426. Cache
14243---------
14244
14245HAProxy provides a cache, which was designed to perform cache on small objects
14246(favicon, css...). This is a minimalist low-maintenance cache which runs in
14247RAM.
14248
14249The cache is based on a memory which is shared between processes and threads,
14250this memory is split in blocks of 1k.
14251
14252If an object is not used anymore, it can be deleted to store a new object
14253independently of its expiration date. The oldest objects are deleted first
14254when we try to allocate a new one.
14255
14256The cache uses a hash of the host header and the URI as the key.
14257
14258It's possible to view the status of a cache using the Unix socket command
14259"show cache" consult section 9.3 "Unix Socket commands" of Management Guide
14260for more details.
14261
14262When an object is delivered from the cache, the server name in the log is
14263replaced by "<CACHE>".
14264
14265
142666.1. Limitation
14267----------------
14268
14269The cache won't store and won't deliver objects in these cases:
14270
14271- If the response is not a 200
14272- If the response contains a Vary header
14273- If the Content-Length + the headers size is greater than "max-object-size"
14274- If the response is not cacheable
14275
14276- If the request is not a GET
14277- If the HTTP version of the request is smaller than 1.1
14278- If the request contains an Authorization header
14279
14280
142816.2. Setup
14282-----------
14283
14284To setup a cache, you must define a cache section and use it in a proxy with
14285the corresponding http-request and response actions.
14286
14287
142886.2.1. Cache section
14289---------------------
14290
14291cache <name>
14292 Declare a cache section, allocate a shared cache memory named <name>, the
14293 size of cache is mandatory.
14294
14295total-max-size <megabytes>
14296 Define the size in RAM of the cache in megabytes. This size is split in
14297 blocks of 1kB which are used by the cache entries. Its maximum value is 4095.
14298
14299max-object-size <bytes>
14300 Define the maximum size of the objects to be cached. Must not be greater than
14301 an half of "total-max-size". If not set, it equals to a 256th of the cache size.
14302 All objects with sizes larger than "max-object-size" will not be cached.
14303
14304max-age <seconds>
14305 Define the maximum expiration duration. The expiration is set has the lowest
14306 value between the s-maxage or max-age (in this order) directive in the
14307 Cache-Control response header and this value. The default value is 60
14308 seconds, which means that you can't cache an object more than 60 seconds by
14309 default.
14310
14311
143126.2.2. Proxy section
14313---------------------
14314
14315http-request cache-use <name> [ { if | unless } <condition> ]
14316 Try to deliver a cached object from the cache <name>. This directive is also
14317 mandatory to store the cache as it calculates the cache hash. If you want to
14318 use a condition for both storage and delivering that's a good idea to put it
14319 after this one.
14320
14321http-response cache-store <name> [ { if | unless } <condition> ]
14322 Store an http-response within the cache. The storage of the response headers
14323 is done at this step, which means you can use others http-response actions
14324 to modify headers before or after the storage of the response. This action
14325 is responsible for the setup of the cache storage filter.
14326
14327
14328Example:
14329
14330 backend bck1
14331 mode http
14332
14333 http-request cache-use foobar
14334 http-response cache-store foobar
14335 server srv1 127.0.0.1:80
14336
14337 cache foobar
14338 total-max-size 4
14339 max-age 240
14340
14341
Willy Tarreau74ca5042013-06-11 23:12:07 +0200143427. Using ACLs and fetching samples
14343----------------------------------
14344
Davor Ocelice9ed2812017-12-25 17:49:28 +010014345HAProxy is capable of extracting data from request or response streams, from
Willy Tarreau74ca5042013-06-11 23:12:07 +020014346client or server information, from tables, environmental information etc...
14347The action of extracting such data is called fetching a sample. Once retrieved,
14348these samples may be used for various purposes such as a key to a stick-table,
14349but most common usages consist in matching them against predefined constant
14350data called patterns.
14351
14352
143537.1. ACL basics
14354---------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014355
14356The use of Access Control Lists (ACL) provides a flexible solution to perform
14357content switching and generally to take decisions based on content extracted
14358from the request, the response or any environmental status. The principle is
14359simple :
14360
Willy Tarreau74ca5042013-06-11 23:12:07 +020014361 - extract a data sample from a stream, table or the environment
Willy Tarreaue6b11e42013-11-26 19:02:32 +010014362 - optionally apply some format conversion to the extracted sample
Willy Tarreau74ca5042013-06-11 23:12:07 +020014363 - apply one or multiple pattern matching methods on this sample
14364 - perform actions only when a pattern matches the sample
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014365
Willy Tarreau74ca5042013-06-11 23:12:07 +020014366The actions generally consist in blocking a request, selecting a backend, or
14367adding a header.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014368
14369In order to define a test, the "acl" keyword is used. The syntax is :
14370
Willy Tarreau74ca5042013-06-11 23:12:07 +020014371 acl <aclname> <criterion> [flags] [operator] [<value>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014372
14373This creates a new ACL <aclname> or completes an existing one with new tests.
14374Those tests apply to the portion of request/response specified in <criterion>
14375and may be adjusted with optional flags [flags]. Some criteria also support
Willy Tarreaue6b11e42013-11-26 19:02:32 +010014376an operator which may be specified before the set of values. Optionally some
14377conversion operators may be applied to the sample, and they will be specified
14378as a comma-delimited list of keywords just after the first keyword. The values
14379are of the type supported by the criterion, and are separated by spaces.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014380
14381ACL names must be formed from upper and lower case letters, digits, '-' (dash),
14382'_' (underscore) , '.' (dot) and ':' (colon). ACL names are case-sensitive,
14383which means that "my_acl" and "My_Acl" are two different ACLs.
14384
14385There is no enforced limit to the number of ACLs. The unused ones do not affect
14386performance, they just consume a small amount of memory.
14387
Willy Tarreau74ca5042013-06-11 23:12:07 +020014388The criterion generally is the name of a sample fetch method, or one of its ACL
14389specific declinations. The default test method is implied by the output type of
14390this sample fetch method. The ACL declinations can describe alternate matching
Willy Tarreaue6b11e42013-11-26 19:02:32 +010014391methods of a same sample fetch method. The sample fetch methods are the only
14392ones supporting a conversion.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014393
14394Sample fetch methods return data which can be of the following types :
14395 - boolean
14396 - integer (signed or unsigned)
14397 - IPv4 or IPv6 address
14398 - string
14399 - data block
14400
Willy Tarreaue6b11e42013-11-26 19:02:32 +010014401Converters transform any of these data into any of these. For example, some
14402converters might convert a string to a lower-case string while other ones
14403would turn a string to an IPv4 address, or apply a netmask to an IP address.
14404The resulting sample is of the type of the last converter applied to the list,
14405which defaults to the type of the sample fetch method.
14406
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020014407Each sample or converter returns data of a specific type, specified with its
14408keyword in this documentation. When an ACL is declared using a standard sample
14409fetch method, certain types automatically involved a default matching method
14410which are summarized in the table below :
14411
14412 +---------------------+-----------------+
14413 | Sample or converter | Default |
14414 | output type | matching method |
14415 +---------------------+-----------------+
14416 | boolean | bool |
14417 +---------------------+-----------------+
14418 | integer | int |
14419 +---------------------+-----------------+
14420 | ip | ip |
14421 +---------------------+-----------------+
14422 | string | str |
14423 +---------------------+-----------------+
14424 | binary | none, use "-m" |
14425 +---------------------+-----------------+
14426
14427Note that in order to match a binary samples, it is mandatory to specify a
14428matching method, see below.
14429
Willy Tarreau74ca5042013-06-11 23:12:07 +020014430The ACL engine can match these types against patterns of the following types :
14431 - boolean
14432 - integer or integer range
14433 - IP address / network
14434 - string (exact, substring, suffix, prefix, subdir, domain)
14435 - regular expression
14436 - hex block
14437
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014438The following ACL flags are currently supported :
14439
Willy Tarreau2b5285d2010-05-09 23:45:24 +020014440 -i : ignore case during matching of all subsequent patterns.
14441 -f : load patterns from a file.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014442 -m : use a specific pattern matching method
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010014443 -n : forbid the DNS resolutions
Thierry FOURNIER9860c412014-01-29 14:23:29 +010014444 -M : load the file pointed by -f like a map file.
Thierry FOURNIER3534d882014-01-20 17:01:44 +010014445 -u : force the unique id of the ACL
Willy Tarreau6a06a402007-07-15 20:15:28 +020014446 -- : force end of flags. Useful when a string looks like one of the flags.
14447
Willy Tarreau74ca5042013-06-11 23:12:07 +020014448The "-f" flag is followed by the name of a file from which all lines will be
14449read as individual values. It is even possible to pass multiple "-f" arguments
14450if the patterns are to be loaded from multiple files. Empty lines as well as
14451lines beginning with a sharp ('#') will be ignored. All leading spaces and tabs
14452will be stripped. If it is absolutely necessary to insert a valid pattern
14453beginning with a sharp, just prefix it with a space so that it is not taken for
14454a comment. Depending on the data type and match method, haproxy may load the
14455lines into a binary tree, allowing very fast lookups. This is true for IPv4 and
14456exact string matching. In this case, duplicates will automatically be removed.
14457
Thierry FOURNIER9860c412014-01-29 14:23:29 +010014458The "-M" flag allows an ACL to use a map file. If this flag is set, the file is
14459parsed as two column file. The first column contains the patterns used by the
14460ACL, and the second column contain the samples. The sample can be used later by
14461a map. This can be useful in some rare cases where an ACL would just be used to
14462check for the existence of a pattern in a map before a mapping is applied.
14463
Thierry FOURNIER3534d882014-01-20 17:01:44 +010014464The "-u" flag forces the unique id of the ACL. This unique id is used with the
14465socket interface to identify ACL and dynamically change its values. Note that a
14466file is always identified by its name even if an id is set.
14467
Willy Tarreau74ca5042013-06-11 23:12:07 +020014468Also, note that the "-i" flag applies to subsequent entries and not to entries
14469loaded from files preceding it. For instance :
14470
14471 acl valid-ua hdr(user-agent) -f exact-ua.lst -i -f generic-ua.lst test
14472
14473In this example, each line of "exact-ua.lst" will be exactly matched against
14474the "user-agent" header of the request. Then each line of "generic-ua" will be
14475case-insensitively matched. Then the word "test" will be insensitively matched
14476as well.
14477
14478The "-m" flag is used to select a specific pattern matching method on the input
14479sample. All ACL-specific criteria imply a pattern matching method and generally
14480do not need this flag. However, this flag is useful with generic sample fetch
14481methods to describe how they're going to be matched against the patterns. This
14482is required for sample fetches which return data type for which there is no
Davor Ocelice9ed2812017-12-25 17:49:28 +010014483obvious matching method (e.g. string or binary). When "-m" is specified and
Willy Tarreau74ca5042013-06-11 23:12:07 +020014484followed by a pattern matching method name, this method is used instead of the
14485default one for the criterion. This makes it possible to match contents in ways
14486that were not initially planned, or with sample fetch methods which return a
14487string. The matching method also affects the way the patterns are parsed.
14488
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010014489The "-n" flag forbids the dns resolutions. It is used with the load of ip files.
14490By default, if the parser cannot parse ip address it considers that the parsed
14491string is maybe a domain name and try dns resolution. The flag "-n" disable this
14492resolution. It is useful for detecting malformed ip lists. Note that if the DNS
14493server is not reachable, the haproxy configuration parsing may last many minutes
John Roeslerfb2fce12019-07-10 15:45:51 -050014494waiting for the timeout. During this time no error messages are displayed. The
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010014495flag "-n" disable this behavior. Note also that during the runtime, this
14496function is disabled for the dynamic acl modifications.
14497
Willy Tarreau74ca5042013-06-11 23:12:07 +020014498There are some restrictions however. Not all methods can be used with all
14499sample fetch methods. Also, if "-m" is used in conjunction with "-f", it must
14500be placed first. The pattern matching method must be one of the following :
Willy Tarreau5adeda12013-03-31 22:13:34 +020014501
14502 - "found" : only check if the requested sample could be found in the stream,
14503 but do not compare it against any pattern. It is recommended not
Willy Tarreau74ca5042013-06-11 23:12:07 +020014504 to pass any pattern to avoid confusion. This matching method is
14505 particularly useful to detect presence of certain contents such
14506 as headers, cookies, etc... even if they are empty and without
14507 comparing them to anything nor counting them.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014508
14509 - "bool" : check the value as a boolean. It can only be applied to fetches
14510 which return a boolean or integer value, and takes no pattern.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014511 Value zero or false does not match, all other values do match.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014512
14513 - "int" : match the value as an integer. It can be used with integer and
Willy Tarreau74ca5042013-06-11 23:12:07 +020014514 boolean samples. Boolean false is integer 0, true is integer 1.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014515
14516 - "ip" : match the value as an IPv4 or IPv6 address. It is compatible
Willy Tarreau74ca5042013-06-11 23:12:07 +020014517 with IP address samples only, so it is implied and never needed.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014518
Davor Ocelice9ed2812017-12-25 17:49:28 +010014519 - "bin" : match the contents against a hexadecimal string representing a
Willy Tarreau5adeda12013-03-31 22:13:34 +020014520 binary sequence. This may be used with binary or string samples.
14521
14522 - "len" : match the sample's length as an integer. This may be used with
14523 binary or string samples.
14524
Willy Tarreau74ca5042013-06-11 23:12:07 +020014525 - "str" : exact match : match the contents against a string. This may be
14526 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014527
Willy Tarreau74ca5042013-06-11 23:12:07 +020014528 - "sub" : substring match : check that the contents contain at least one of
14529 the provided string patterns. This may be used with binary or
14530 string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014531
Willy Tarreau74ca5042013-06-11 23:12:07 +020014532 - "reg" : regex match : match the contents against a list of regular
14533 expressions. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014534
Willy Tarreau74ca5042013-06-11 23:12:07 +020014535 - "beg" : prefix match : check that the contents begin like the provided
14536 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014537
Willy Tarreau74ca5042013-06-11 23:12:07 +020014538 - "end" : suffix match : check that the contents end like the provided
14539 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014540
Willy Tarreau74ca5042013-06-11 23:12:07 +020014541 - "dir" : subdir match : check that a slash-delimited portion of the
14542 contents exactly matches one of the provided string patterns.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014543 This may be used with binary or string samples.
14544
Willy Tarreau74ca5042013-06-11 23:12:07 +020014545 - "dom" : domain match : check that a dot-delimited portion of the contents
14546 exactly match one of the provided string patterns. This may be
14547 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014548
14549For example, to quickly detect the presence of cookie "JSESSIONID" in an HTTP
14550request, it is possible to do :
14551
14552 acl jsess_present cook(JSESSIONID) -m found
14553
14554In order to apply a regular expression on the 500 first bytes of data in the
14555buffer, one would use the following acl :
14556
14557 acl script_tag payload(0,500) -m reg -i <script>
14558
Willy Tarreaue6b11e42013-11-26 19:02:32 +010014559On systems where the regex library is much slower when using "-i", it is
14560possible to convert the sample to lowercase before matching, like this :
14561
14562 acl script_tag payload(0,500),lower -m reg <script>
14563
Willy Tarreau74ca5042013-06-11 23:12:07 +020014564All ACL-specific criteria imply a default matching method. Most often, these
14565criteria are composed by concatenating the name of the original sample fetch
14566method and the matching method. For example, "hdr_beg" applies the "beg" match
14567to samples retrieved using the "hdr" fetch method. Since all ACL-specific
14568criteria rely on a sample fetch method, it is always possible instead to use
14569the original sample fetch method and the explicit matching method using "-m".
Willy Tarreau2b5285d2010-05-09 23:45:24 +020014570
Willy Tarreau74ca5042013-06-11 23:12:07 +020014571If an alternate match is specified using "-m" on an ACL-specific criterion,
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014572the matching method is simply applied to the underlying sample fetch method.
14573For example, all ACLs below are exact equivalent :
Willy Tarreau2b5285d2010-05-09 23:45:24 +020014574
Willy Tarreau74ca5042013-06-11 23:12:07 +020014575 acl short_form hdr_beg(host) www.
14576 acl alternate1 hdr_beg(host) -m beg www.
14577 acl alternate2 hdr_dom(host) -m beg www.
14578 acl alternate3 hdr(host) -m beg www.
Willy Tarreau2b5285d2010-05-09 23:45:24 +020014579
Willy Tarreau2b5285d2010-05-09 23:45:24 +020014580
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020014581The table below summarizes the compatibility matrix between sample or converter
14582types and the pattern types to fetch against. It indicates for each compatible
14583combination the name of the matching method to be used, surrounded with angle
14584brackets ">" and "<" when the method is the default one and will work by
14585default without "-m".
Willy Tarreau0ba27502007-12-24 16:55:16 +010014586
Willy Tarreau74ca5042013-06-11 23:12:07 +020014587 +-------------------------------------------------+
14588 | Input sample type |
14589 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020014590 | pattern type | boolean | integer | ip | string | binary |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014591 +----------------------+---------+---------+---------+---------+---------+
14592 | none (presence only) | found | found | found | found | found |
14593 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020014594 | none (boolean value) |> bool <| bool | | bool | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014595 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020014596 | integer (value) | int |> int <| int | int | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014597 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010014598 | integer (length) | len | len | len | len | len |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014599 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020014600 | IP address | | |> ip <| ip | ip |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014601 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020014602 | exact string | str | str | str |> str <| str |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014603 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010014604 | prefix | beg | beg | beg | beg | beg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014605 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010014606 | suffix | end | end | end | end | end |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014607 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010014608 | substring | sub | sub | sub | sub | sub |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014609 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010014610 | subdir | dir | dir | dir | dir | dir |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014611 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010014612 | domain | dom | dom | dom | dom | dom |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014613 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010014614 | regex | reg | reg | reg | reg | reg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014615 +----------------------+---------+---------+---------+---------+---------+
14616 | hex block | | | | bin | bin |
14617 +----------------------+---------+---------+---------+---------+---------+
Willy Tarreau6a06a402007-07-15 20:15:28 +020014618
14619
Willy Tarreau74ca5042013-06-11 23:12:07 +0200146207.1.1. Matching booleans
14621------------------------
14622
14623In order to match a boolean, no value is needed and all values are ignored.
14624Boolean matching is used by default for all fetch methods of type "boolean".
14625When boolean matching is used, the fetched value is returned as-is, which means
14626that a boolean "true" will always match and a boolean "false" will never match.
14627
14628Boolean matching may also be enforced using "-m bool" on fetch methods which
14629return an integer value. Then, integer value 0 is converted to the boolean
14630"false" and all other values are converted to "true".
14631
Willy Tarreau6a06a402007-07-15 20:15:28 +020014632
Willy Tarreau74ca5042013-06-11 23:12:07 +0200146337.1.2. Matching integers
14634------------------------
14635
14636Integer matching applies by default to integer fetch methods. It can also be
14637enforced on boolean fetches using "-m int". In this case, "false" is converted
14638to the integer 0, and "true" is converted to the integer 1.
14639
14640Integer matching also supports integer ranges and operators. Note that integer
14641matching only applies to positive values. A range is a value expressed with a
14642lower and an upper bound separated with a colon, both of which may be omitted.
Willy Tarreau6a06a402007-07-15 20:15:28 +020014643
14644For instance, "1024:65535" is a valid range to represent a range of
14645unprivileged ports, and "1024:" would also work. "0:1023" is a valid
14646representation of privileged ports, and ":1023" would also work.
14647
Willy Tarreau62644772008-07-16 18:36:06 +020014648As a special case, some ACL functions support decimal numbers which are in fact
14649two integers separated by a dot. This is used with some version checks for
14650instance. All integer properties apply to those decimal numbers, including
14651ranges and operators.
14652
Willy Tarreau6a06a402007-07-15 20:15:28 +020014653For an easier usage, comparison operators are also supported. Note that using
Willy Tarreau0ba27502007-12-24 16:55:16 +010014654operators with ranges does not make much sense and is strongly discouraged.
14655Similarly, it does not make much sense to perform order comparisons with a set
14656of values.
Willy Tarreau6a06a402007-07-15 20:15:28 +020014657
Willy Tarreau0ba27502007-12-24 16:55:16 +010014658Available operators for integer matching are :
Willy Tarreau6a06a402007-07-15 20:15:28 +020014659
14660 eq : true if the tested value equals at least one value
14661 ge : true if the tested value is greater than or equal to at least one value
14662 gt : true if the tested value is greater than at least one value
14663 le : true if the tested value is less than or equal to at least one value
14664 lt : true if the tested value is less than at least one value
14665
Willy Tarreau0ba27502007-12-24 16:55:16 +010014666For instance, the following ACL matches any negative Content-Length header :
Willy Tarreau6a06a402007-07-15 20:15:28 +020014667
14668 acl negative-length hdr_val(content-length) lt 0
14669
Willy Tarreau62644772008-07-16 18:36:06 +020014670This one matches SSL versions between 3.0 and 3.1 (inclusive) :
14671
14672 acl sslv3 req_ssl_ver 3:3.1
14673
Willy Tarreau6a06a402007-07-15 20:15:28 +020014674
Willy Tarreau74ca5042013-06-11 23:12:07 +0200146757.1.3. Matching strings
14676-----------------------
14677
14678String matching applies to string or binary fetch methods, and exists in 6
14679different forms :
14680
14681 - exact match (-m str) : the extracted string must exactly match the
Davor Ocelice9ed2812017-12-25 17:49:28 +010014682 patterns;
Willy Tarreau74ca5042013-06-11 23:12:07 +020014683
14684 - substring match (-m sub) : the patterns are looked up inside the
Davor Ocelice9ed2812017-12-25 17:49:28 +010014685 extracted string, and the ACL matches if any of them is found inside;
Willy Tarreau74ca5042013-06-11 23:12:07 +020014686
14687 - prefix match (-m beg) : the patterns are compared with the beginning of
14688 the extracted string, and the ACL matches if any of them matches.
14689
14690 - suffix match (-m end) : the patterns are compared with the end of the
14691 extracted string, and the ACL matches if any of them matches.
14692
Baptiste Assmann33db6002016-03-06 23:32:10 +010014693 - subdir match (-m dir) : the patterns are looked up inside the extracted
Willy Tarreau74ca5042013-06-11 23:12:07 +020014694 string, delimited with slashes ("/"), and the ACL matches if any of them
14695 matches.
14696
14697 - domain match (-m dom) : the patterns are looked up inside the extracted
14698 string, delimited with dots ("."), and the ACL matches if any of them
14699 matches.
Willy Tarreau6a06a402007-07-15 20:15:28 +020014700
14701String matching applies to verbatim strings as they are passed, with the
14702exception of the backslash ("\") which makes it possible to escape some
14703characters such as the space. If the "-i" flag is passed before the first
14704string, then the matching will be performed ignoring the case. In order
14705to match the string "-i", either set it second, or pass the "--" flag
Willy Tarreau0ba27502007-12-24 16:55:16 +010014706before the first string. Same applies of course to match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020014707
Mathias Weiersmuellercb250fc2019-12-02 09:43:40 +010014708Do not use string matches for binary fetches which might contain null bytes
14709(0x00), as the comparison stops at the occurrence of the first null byte.
14710Instead, convert the binary fetch to a hex string with the hex converter first.
14711
14712Example:
14713 # matches if the string <tag> is present in the binary sample
14714 acl tag_found req.payload(0,0),hex -m sub 3C7461673E
14715
Willy Tarreau6a06a402007-07-15 20:15:28 +020014716
Willy Tarreau74ca5042013-06-11 23:12:07 +0200147177.1.4. Matching regular expressions (regexes)
14718---------------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020014719
14720Just like with string matching, regex matching applies to verbatim strings as
14721they are passed, with the exception of the backslash ("\") which makes it
14722possible to escape some characters such as the space. If the "-i" flag is
14723passed before the first regex, then the matching will be performed ignoring
14724the case. In order to match the string "-i", either set it second, or pass
Willy Tarreau0ba27502007-12-24 16:55:16 +010014725the "--" flag before the first string. Same principle applies of course to
14726match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020014727
14728
Willy Tarreau74ca5042013-06-11 23:12:07 +0200147297.1.5. Matching arbitrary data blocks
14730-------------------------------------
14731
14732It is possible to match some extracted samples against a binary block which may
14733not safely be represented as a string. For this, the patterns must be passed as
14734a series of hexadecimal digits in an even number, when the match method is set
14735to binary. Each sequence of two digits will represent a byte. The hexadecimal
14736digits may be used upper or lower case.
14737
14738Example :
14739 # match "Hello\n" in the input stream (\x48 \x65 \x6c \x6c \x6f \x0a)
14740 acl hello payload(0,6) -m bin 48656c6c6f0a
14741
14742
147437.1.6. Matching IPv4 and IPv6 addresses
14744---------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020014745
14746IPv4 addresses values can be specified either as plain addresses or with a
14747netmask appended, in which case the IPv4 address matches whenever it is
14748within the network. Plain addresses may also be replaced with a resolvable
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010014749host name, but this practice is generally discouraged as it makes it more
Willy Tarreau0ba27502007-12-24 16:55:16 +010014750difficult to read and debug configurations. If hostnames are used, you should
14751at least ensure that they are present in /etc/hosts so that the configuration
14752does not depend on any random DNS match at the moment the configuration is
14753parsed.
Willy Tarreau6a06a402007-07-15 20:15:28 +020014754
Daniel Schnellereba56342016-04-13 00:26:52 +020014755The dotted IPv4 address notation is supported in both regular as well as the
14756abbreviated form with all-0-octets omitted:
14757
14758 +------------------+------------------+------------------+
14759 | Example 1 | Example 2 | Example 3 |
14760 +------------------+------------------+------------------+
14761 | 192.168.0.1 | 10.0.0.12 | 127.0.0.1 |
14762 | 192.168.1 | 10.12 | 127.1 |
14763 | 192.168.0.1/22 | 10.0.0.12/8 | 127.0.0.1/8 |
14764 | 192.168.1/22 | 10.12/8 | 127.1/8 |
14765 +------------------+------------------+------------------+
14766
14767Notice that this is different from RFC 4632 CIDR address notation in which
14768192.168.42/24 would be equivalent to 192.168.42.0/24.
14769
Willy Tarreauceb4ac92012-04-28 00:41:46 +020014770IPv6 may be entered in their usual form, with or without a netmask appended.
14771Only bit counts are accepted for IPv6 netmasks. In order to avoid any risk of
14772trouble with randomly resolved IP addresses, host names are never allowed in
14773IPv6 patterns.
14774
14775HAProxy is also able to match IPv4 addresses with IPv6 addresses in the
14776following situations :
14777 - tested address is IPv4, pattern address is IPv4, the match applies
14778 in IPv4 using the supplied mask if any.
14779 - tested address is IPv6, pattern address is IPv6, the match applies
14780 in IPv6 using the supplied mask if any.
14781 - tested address is IPv6, pattern address is IPv4, the match applies in IPv4
14782 using the pattern's mask if the IPv6 address matches with 2002:IPV4::,
14783 ::IPV4 or ::ffff:IPV4, otherwise it fails.
14784 - tested address is IPv4, pattern address is IPv6, the IPv4 address is first
14785 converted to IPv6 by prefixing ::ffff: in front of it, then the match is
14786 applied in IPv6 using the supplied IPv6 mask.
14787
Willy Tarreau74ca5042013-06-11 23:12:07 +020014788
147897.2. Using ACLs to form conditions
14790----------------------------------
14791
14792Some actions are only performed upon a valid condition. A condition is a
14793combination of ACLs with operators. 3 operators are supported :
14794
14795 - AND (implicit)
14796 - OR (explicit with the "or" keyword or the "||" operator)
14797 - Negation with the exclamation mark ("!")
Willy Tarreau6a06a402007-07-15 20:15:28 +020014798
Willy Tarreau74ca5042013-06-11 23:12:07 +020014799A condition is formed as a disjunctive form:
Willy Tarreau6a06a402007-07-15 20:15:28 +020014800
Willy Tarreau74ca5042013-06-11 23:12:07 +020014801 [!]acl1 [!]acl2 ... [!]acln { or [!]acl1 [!]acl2 ... [!]acln } ...
Willy Tarreaubef91e72013-03-31 23:14:46 +020014802
Willy Tarreau74ca5042013-06-11 23:12:07 +020014803Such conditions are generally used after an "if" or "unless" statement,
14804indicating when the condition will trigger the action.
Willy Tarreaubef91e72013-03-31 23:14:46 +020014805
Willy Tarreau74ca5042013-06-11 23:12:07 +020014806For instance, to block HTTP requests to the "*" URL with methods other than
14807"OPTIONS", as well as POST requests without content-length, and GET or HEAD
14808requests with a content-length greater than 0, and finally every request which
14809is not either GET/HEAD/POST/OPTIONS !
14810
14811 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030014812 http-request deny if HTTP_URL_STAR !METH_OPTIONS || METH_POST missing_cl
14813 http-request deny if METH_GET HTTP_CONTENT
14814 http-request deny unless METH_GET or METH_POST or METH_OPTIONS
Willy Tarreau74ca5042013-06-11 23:12:07 +020014815
14816To select a different backend for requests to static contents on the "www" site
14817and to every request on the "img", "video", "download" and "ftp" hosts :
14818
14819 acl url_static path_beg /static /images /img /css
14820 acl url_static path_end .gif .png .jpg .css .js
14821 acl host_www hdr_beg(host) -i www
14822 acl host_static hdr_beg(host) -i img. video. download. ftp.
14823
Davor Ocelice9ed2812017-12-25 17:49:28 +010014824 # now use backend "static" for all static-only hosts, and for static URLs
Willy Tarreau74ca5042013-06-11 23:12:07 +020014825 # of host "www". Use backend "www" for the rest.
14826 use_backend static if host_static or host_www url_static
14827 use_backend www if host_www
14828
14829It is also possible to form rules using "anonymous ACLs". Those are unnamed ACL
14830expressions that are built on the fly without needing to be declared. They must
14831be enclosed between braces, with a space before and after each brace (because
14832the braces must be seen as independent words). Example :
14833
14834 The following rule :
14835
14836 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030014837 http-request deny if METH_POST missing_cl
Willy Tarreau74ca5042013-06-11 23:12:07 +020014838
14839 Can also be written that way :
14840
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030014841 http-request deny if METH_POST { hdr_cnt(Content-length) eq 0 }
Willy Tarreau74ca5042013-06-11 23:12:07 +020014842
14843It is generally not recommended to use this construct because it's a lot easier
14844to leave errors in the configuration when written that way. However, for very
14845simple rules matching only one source IP address for instance, it can make more
14846sense to use them than to declare ACLs with random names. Another example of
14847good use is the following :
14848
14849 With named ACLs :
14850
14851 acl site_dead nbsrv(dynamic) lt 2
14852 acl site_dead nbsrv(static) lt 2
14853 monitor fail if site_dead
14854
14855 With anonymous ACLs :
14856
14857 monitor fail if { nbsrv(dynamic) lt 2 } || { nbsrv(static) lt 2 }
14858
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030014859See section 4.2 for detailed help on the "http-request deny" and "use_backend"
14860keywords.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014861
14862
148637.3. Fetching samples
14864---------------------
14865
14866Historically, sample fetch methods were only used to retrieve data to match
14867against patterns using ACLs. With the arrival of stick-tables, a new class of
14868sample fetch methods was created, most often sharing the same syntax as their
14869ACL counterpart. These sample fetch methods are also known as "fetches". As
14870of now, ACLs and fetches have converged. All ACL fetch methods have been made
14871available as fetch methods, and ACLs may use any sample fetch method as well.
14872
14873This section details all available sample fetch methods and their output type.
14874Some sample fetch methods have deprecated aliases that are used to maintain
14875compatibility with existing configurations. They are then explicitly marked as
14876deprecated and should not be used in new setups.
14877
14878The ACL derivatives are also indicated when available, with their respective
14879matching methods. These ones all have a well defined default pattern matching
14880method, so it is never necessary (though allowed) to pass the "-m" option to
14881indicate how the sample will be matched using ACLs.
14882
14883As indicated in the sample type versus matching compatibility matrix above,
14884when using a generic sample fetch method in an ACL, the "-m" option is
14885mandatory unless the sample type is one of boolean, integer, IPv4 or IPv6. When
14886the same keyword exists as an ACL keyword and as a standard fetch method, the
14887ACL engine will automatically pick the ACL-only one by default.
14888
14889Some of these keywords support one or multiple mandatory arguments, and one or
14890multiple optional arguments. These arguments are strongly typed and are checked
14891when the configuration is parsed so that there is no risk of running with an
Davor Ocelice9ed2812017-12-25 17:49:28 +010014892incorrect argument (e.g. an unresolved backend name). Fetch function arguments
14893are passed between parenthesis and are delimited by commas. When an argument
Willy Tarreau74ca5042013-06-11 23:12:07 +020014894is optional, it will be indicated below between square brackets ('[ ]'). When
14895all arguments are optional, the parenthesis may be omitted.
14896
14897Thus, the syntax of a standard sample fetch method is one of the following :
14898 - name
14899 - name(arg1)
14900 - name(arg1,arg2)
14901
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014902
149037.3.1. Converters
14904-----------------
14905
Willy Tarreaue6b11e42013-11-26 19:02:32 +010014906Sample fetch methods may be combined with transformations to be applied on top
14907of the fetched sample (also called "converters"). These combinations form what
14908is called "sample expressions" and the result is a "sample". Initially this
14909was only supported by "stick on" and "stick store-request" directives but this
Davor Ocelice9ed2812017-12-25 17:49:28 +010014910has now be extended to all places where samples may be used (ACLs, log-format,
Willy Tarreaue6b11e42013-11-26 19:02:32 +010014911unique-id-format, add-header, ...).
14912
14913These transformations are enumerated as a series of specific keywords after the
14914sample fetch method. These keywords may equally be appended immediately after
14915the fetch keyword's argument, delimited by a comma. These keywords can also
Davor Ocelice9ed2812017-12-25 17:49:28 +010014916support some arguments (e.g. a netmask) which must be passed in parenthesis.
Willy Tarreau0ba27502007-12-24 16:55:16 +010014917
Willy Tarreau97707872015-01-27 15:12:13 +010014918A certain category of converters are bitwise and arithmetic operators which
14919support performing basic operations on integers. Some bitwise operations are
14920supported (and, or, xor, cpl) and some arithmetic operations are supported
14921(add, sub, mul, div, mod, neg). Some comparators are provided (odd, even, not,
14922bool) which make it possible to report a match without having to write an ACL.
14923
Willy Tarreau74ca5042013-06-11 23:12:07 +020014924The currently available list of transformation keywords include :
Willy Tarreau0ba27502007-12-24 16:55:16 +010014925
Ben Shillitof25e8e52016-12-02 14:25:37 +00001492651d.single(<prop>[,<prop>*])
14927 Returns values for the properties requested as a string, where values are
14928 separated by the delimiter specified with "51degrees-property-separator".
14929 The device is identified using the User-Agent header passed to the
14930 converter. The function can be passed up to five property names, and if a
14931 property name can't be found, the value "NoData" is returned.
14932
14933 Example :
Davor Ocelice9ed2812017-12-25 17:49:28 +010014934 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request,
14935 # containing values for the three properties requested by using the
Ben Shillitof25e8e52016-12-02 14:25:37 +000014936 # User-Agent passed to the converter.
14937 frontend http-in
14938 bind *:8081
14939 default_backend servers
14940 http-request set-header X-51D-DeviceTypeMobileTablet \
14941 %[req.fhdr(User-Agent),51d.single(DeviceType,IsMobile,IsTablet)]
14942
Willy Tarreau97707872015-01-27 15:12:13 +010014943add(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014944 Adds <value> to the input value of type signed integer, and returns the
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014945 result as a signed integer. <value> can be a numeric value or a variable
Daniel Schneller0b547052016-03-21 20:46:57 +010014946 name. The name of the variable starts with an indication about its scope. The
14947 scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010014948 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010014949 "sess" : the variable is shared with the whole session
14950 "txn" : the variable is shared with the transaction (request and response)
14951 "req" : the variable is shared only during request processing
14952 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010014953 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014954 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010014955
Nenad Merdanovicc31499d2019-03-23 11:00:32 +010014956aes_gcm_dec(<bits>,<nonce>,<key>,<aead_tag>)
14957 Decrypts the raw byte input using the AES128-GCM, AES192-GCM or
14958 AES256-GCM algorithm, depending on the <bits> parameter. All other parameters
14959 need to be base64 encoded and the returned result is in raw byte format.
14960 If the <aead_tag> validation fails, the converter doesn't return any data.
14961 The <nonce>, <key> and <aead_tag> can either be strings or variables. This
14962 converter requires at least OpenSSL 1.0.1.
14963
14964 Example:
14965 http-response set-header X-Decrypted-Text %[var(txn.enc),\
14966 aes_gcm_dec(128,txn.nonce,Zm9vb2Zvb29mb29wZm9vbw==,txn.aead_tag)]
14967
Willy Tarreau97707872015-01-27 15:12:13 +010014968and(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014969 Performs a bitwise "AND" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014970 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010014971 numeric value or a variable name. The name of the variable starts with an
14972 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010014973 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010014974 "sess" : the variable is shared with the whole session
14975 "txn" : the variable is shared with the transaction (request and response)
14976 "req" : the variable is shared only during request processing
14977 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010014978 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014979 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010014980
Holger Just1bfc24b2017-05-06 00:56:53 +020014981b64dec
14982 Converts (decodes) a base64 encoded input string to its binary
14983 representation. It performs the inverse operation of base64().
14984
Emeric Brun53d1a982014-04-30 18:21:37 +020014985base64
14986 Converts a binary input sample to a base64 string. It is used to log or
Davor Ocelice9ed2812017-12-25 17:49:28 +010014987 transfer binary content in a way that can be reliably transferred (e.g.
Emeric Brun53d1a982014-04-30 18:21:37 +020014988 an SSL ID can be copied in a header).
14989
Willy Tarreau97707872015-01-27 15:12:13 +010014990bool
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014991 Returns a boolean TRUE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010014992 non-null, otherwise returns FALSE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010014993 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010014994 presence of a flag).
14995
Emeric Brun54c4ac82014-11-03 15:32:43 +010014996bytes(<offset>[,<length>])
14997 Extracts some bytes from an input binary sample. The result is a binary
14998 sample starting at an offset (in bytes) of the original sample and
Tim Düsterhus4896c442016-11-29 02:15:19 +010014999 optionally truncated at the given length.
Emeric Brun54c4ac82014-11-03 15:32:43 +010015000
Willy Tarreau280f42b2018-02-19 15:34:12 +010015001concat([<start>],[<var>],[<end>])
15002 Concatenates up to 3 fields after the current sample which is then turned to
15003 a string. The first one, <start>, is a constant string, that will be appended
15004 immediately after the existing sample. It may be omitted if not used. The
15005 second one, <var>, is a variable name. The variable will be looked up, its
15006 contents converted to a string, and it will be appended immediately after the
15007 <first> part. If the variable is not found, nothing is appended. It may be
15008 omitted as well. The third field, <end> is a constant string that will be
15009 appended after the variable. It may also be omitted. Together, these elements
15010 allow to concatenate variables with delimiters to an existing set of
15011 variables. This can be used to build new variables made of a succession of
Willy Tarreauef21fac2020-02-14 13:37:20 +010015012 other variables, such as colon-delimited values. If commas or closing
Daniel Corbett67a82712020-07-06 23:01:19 -040015013 parenthesis are needed as delimiters, they must be protected by quotes or
Willy Tarreauef21fac2020-02-14 13:37:20 +010015014 backslashes, themselves protected so that they are not stripped by the first
15015 level parser. See examples below.
Willy Tarreau280f42b2018-02-19 15:34:12 +010015016
15017 Example:
15018 tcp-request session set-var(sess.src) src
15019 tcp-request session set-var(sess.dn) ssl_c_s_dn
15020 tcp-request session set-var(txn.sig) str(),concat(<ip=,sess.ip,>),concat(<dn=,sess.dn,>)
Willy Tarreauef21fac2020-02-14 13:37:20 +010015021 tcp-request session set-var(txn.ipport) "str(),concat('addr=(',sess.ip),concat(',',sess.port,')')"
Willy Tarreau280f42b2018-02-19 15:34:12 +010015022 http-request set-header x-hap-sig %[var(txn.sig)]
15023
Willy Tarreau97707872015-01-27 15:12:13 +010015024cpl
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015025 Takes the input value of type signed integer, applies a ones-complement
15026 (flips all bits) and returns the result as an signed integer.
Willy Tarreau97707872015-01-27 15:12:13 +010015027
Willy Tarreau80599772015-01-20 19:35:24 +010015028crc32([<avalanche>])
15029 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32
15030 hash function. Optionally, it is possible to apply a full avalanche hash
15031 function to the output if the optional <avalanche> argument equals 1. This
15032 converter uses the same functions as used by the various hash-based load
15033 balancing algorithms, so it will provide exactly the same results. It is
15034 provided for compatibility with other software which want a CRC32 to be
15035 computed on some input keys, so it follows the most common implementation as
15036 found in Ethernet, Gzip, PNG, etc... It is slower than the other algorithms
15037 but may provide a better or at least less predictable distribution. It must
15038 not be used for security purposes as a 32-bit hash is trivial to break. See
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010015039 also "djb2", "sdbm", "wt6", "crc32c" and the "hash-type" directive.
15040
15041crc32c([<avalanche>])
15042 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32C
15043 hash function. Optionally, it is possible to apply a full avalanche hash
15044 function to the output if the optional <avalanche> argument equals 1. This
15045 converter uses the same functions as described in RFC4960, Appendix B [8].
15046 It is provided for compatibility with other software which want a CRC32C to be
15047 computed on some input keys. It is slower than the other algorithms and it must
15048 not be used for security purposes as a 32-bit hash is trivial to break. See
15049 also "djb2", "sdbm", "wt6", "crc32" and the "hash-type" directive.
Willy Tarreau80599772015-01-20 19:35:24 +010015050
Christopher Fauletea159d62020-04-01 16:21:44 +020015051cut_crlf
15052 Cuts the string representation of the input sample on the first carriage
15053 return ('\r') or newline ('\n') character found. Only the string length is
15054 updated.
15055
David Carlier29b3ca32015-09-25 14:09:21 +010015056da-csv-conv(<prop>[,<prop>*])
David Carlier4542b102015-06-01 13:54:29 +020015057 Asks the DeviceAtlas converter to identify the User Agent string passed on
15058 input, and to emit a string made of the concatenation of the properties
15059 enumerated in argument, delimited by the separator defined by the global
15060 keyword "deviceatlas-property-separator", or by default the pipe character
David Carlier840b0242016-03-16 10:09:55 +000015061 ('|'). There's a limit of 12 different properties imposed by the haproxy
David Carlier4542b102015-06-01 13:54:29 +020015062 configuration language.
15063
15064 Example:
15065 frontend www
Cyril Bonté307ee1e2015-09-28 23:16:06 +020015066 bind *:8881
15067 default_backend servers
David Carlier840b0242016-03-16 10:09:55 +000015068 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 +020015069
Willy Tarreau0851fd52019-12-17 10:07:25 +010015070debug([<prefix][,<destination>])
15071 This converter is used as debug tool. It takes a capture of the input sample
15072 and sends it to event sink <destination>, which may designate a ring buffer
15073 such as "buf0", as well as "stdout", or "stderr". Available sinks may be
15074 checked at run time by issuing "show events" on the CLI. When not specified,
15075 the output will be "buf0", which may be consulted via the CLI's "show events"
15076 command. An optional prefix <prefix> may be passed to help distinguish
15077 outputs from multiple expressions. It will then appear before the colon in
15078 the output message. The input sample is passed as-is on the output, so that
15079 it is safe to insert the debug converter anywhere in a chain, even with non-
15080 printable sample types.
15081
15082 Example:
15083 tcp-request connection track-sc0 src,debug(track-sc)
Thierry FOURNIER9687c772015-05-07 15:46:29 +020015084
Patrick Gansterer8e366512020-04-22 16:47:57 +020015085digest(<algorithm>)
15086 Converts a binary input sample to a message digest. The result is a binary
15087 sample. The <algorithm> must be an OpenSSL message digest name (e.g. sha256).
15088
15089 Please note that this converter is only available when haproxy has been
15090 compiled with USE_OPENSSL.
15091
Willy Tarreau97707872015-01-27 15:12:13 +010015092div(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015093 Divides the input value of type signed integer by <value>, and returns the
15094 result as an signed integer. If <value> is null, the largest unsigned
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015095 integer is returned (typically 2^63-1). <value> can be a numeric value or a
Daniel Schneller0b547052016-03-21 20:46:57 +010015096 variable name. The name of the variable starts with an indication about its
15097 scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015098 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015099 "sess" : the variable is shared with the whole session
15100 "txn" : the variable is shared with the transaction (request and response)
15101 "req" : the variable is shared only during request processing
15102 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010015103 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015104 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010015105
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015106djb2([<avalanche>])
15107 Hashes a binary input sample into an unsigned 32-bit quantity using the DJB2
15108 hash function. Optionally, it is possible to apply a full avalanche hash
15109 function to the output if the optional <avalanche> argument equals 1. This
15110 converter uses the same functions as used by the various hash-based load
15111 balancing algorithms, so it will provide exactly the same results. It is
15112 mostly intended for debugging, but can be used as a stick-table entry to
15113 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010015114 32-bit hash is trivial to break. See also "crc32", "sdbm", "wt6", "crc32c",
15115 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015116
Willy Tarreau97707872015-01-27 15:12:13 +010015117even
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015118 Returns a boolean TRUE if the input value of type signed integer is even
Willy Tarreau97707872015-01-27 15:12:13 +010015119 otherwise returns FALSE. It is functionally equivalent to "not,and(1),bool".
15120
Marcin Deranek9631a282018-04-16 14:30:46 +020015121field(<index>,<delimiters>[,<count>])
15122 Extracts the substring at the given index counting from the beginning
15123 (positive index) or from the end (negative index) considering given delimiters
15124 from an input string. Indexes start at 1 or -1 and delimiters are a string
15125 formatted list of chars. Optionally you can specify <count> of fields to
15126 extract (default: 1). Value of 0 indicates extraction of all remaining
15127 fields.
15128
15129 Example :
15130 str(f1_f2_f3__f5),field(5,_) # f5
15131 str(f1_f2_f3__f5),field(2,_,0) # f2_f3__f5
15132 str(f1_f2_f3__f5),field(2,_,2) # f2_f3
15133 str(f1_f2_f3__f5),field(-2,_,3) # f2_f3_
15134 str(f1_f2_f3__f5),field(-3,_,0) # f1_f2_f3
Emeric Brunf399b0d2014-11-03 17:07:03 +010015135
Baptiste Assmanne138dda2020-10-22 15:39:03 +020015136fix_is_valid
15137 Parses a binary payload and performs sanity checks regarding FIX (Financial
15138 Information eXchange):
15139
15140 - checks that all tag IDs and values are not empty and the tags IDs are well
15141 numeric
15142 - checks the BeginString tag is the first tag with a valide FIX version
15143 - checks the BodyLength tag is the second one with the right body length
15144 - checks the MstType tag is the third tag.
15145 - checks that last tag in the message is the CheckSum tag with a valid
15146 checksum
15147
15148 Due to current HAProxy design, only the first message sent by the client and
15149 the server can be parsed.
15150
15151 This converter returns a boolean, true if the payload contains a valid FIX
15152 message, false if not.
15153
15154 See also the fix_tag_value converter.
15155
15156 Example:
15157 tcp-request inspect-delay 10s
15158 tcp-request content reject unless { req.payload(0,0),fix_is_valid }
15159
15160fix_tag_value(<tag>)
15161 Parses a FIX (Financial Information eXchange) message and extracts the value
15162 from the tag <tag>. <tag> can be a string or an integer pointing to the
15163 desired tag. Any integer value is accepted, but only the following strings
15164 are translated into their integer equivalent: BeginString, BodyLength,
15165 MsgType, SenderComID, TargetComID, CheckSum. More tag names can be easily
15166 added.
15167
15168 Due to current HAProxy design, only the first message sent by the client and
15169 the server can be parsed. No message validation is performed by this
15170 converter. It is highly recommended to validate the message first using
15171 fix_is_valid converter.
15172
15173 See also the fix_is_valid converter.
15174
15175 Example:
15176 tcp-request inspect-delay 10s
15177 tcp-request content reject unless { req.payload(0,0),fix_is_valid }
15178 # MsgType tag ID is 35, so both lines below will return the same content
15179 tcp-request content set-var(txn.foo) req.payload(0,0),fix_tag_value(35)
15180 tcp-request content set-var(txn.bar) req.payload(0,0),fix_tag_value(MsgType)
15181
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015182hex
Davor Ocelice9ed2812017-12-25 17:49:28 +010015183 Converts a binary input sample to a hex string containing two hex digits per
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015184 input byte. It is used to log or transfer hex dumps of some binary input data
Davor Ocelice9ed2812017-12-25 17:49:28 +010015185 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 +020015186 header).
Thierry FOURNIER2f49d6d2014-03-12 15:01:52 +010015187
Dragan Dosen3f957b22017-10-24 09:27:34 +020015188hex2i
15189 Converts a hex string containing two hex digits per input byte to an
John Roeslerfb2fce12019-07-10 15:45:51 -050015190 integer. If the input value cannot be converted, then zero is returned.
Dragan Dosen3f957b22017-10-24 09:27:34 +020015191
Christopher Faulet4ccc12f2020-04-01 09:08:32 +020015192htonl
15193 Converts the input integer value to its 32-bit binary representation in the
15194 network byte order. Because sample fetches own signed 64-bit integer, when
15195 this converter is used, the input integer value is first casted to an
15196 unsigned 32-bit integer.
15197
Patrick Gansterer8e366512020-04-22 16:47:57 +020015198hmac(<algorithm>, <key>)
15199 Converts a binary input sample to a message authentication code with the given
15200 key. The result is a binary sample. The <algorithm> must be one of the
15201 registered OpenSSL message digest names (e.g. sha256). The <key> parameter must
15202 be base64 encoded and can either be a string or a variable.
15203
15204 Please note that this converter is only available when haproxy has been
15205 compiled with USE_OPENSSL.
15206
Cyril Bonté6bcd1822019-11-05 23:13:59 +010015207http_date([<offset],[<unit>])
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015208 Converts an integer supposed to contain a date since epoch to a string
15209 representing this date in a format suitable for use in HTTP header fields. If
Damien Claisseae6f1252019-10-30 15:57:28 +000015210 an offset value is specified, then it is added to the date before the
15211 conversion is operated. This is particularly useful to emit Date header fields,
15212 Expires values in responses when combined with a positive offset, or
15213 Last-Modified values when the offset is negative.
15214 If a unit value is specified, then consider the timestamp as either
15215 "s" for seconds (default behavior), "ms" for milliseconds, or "us" for
15216 microseconds since epoch. Offset is assumed to have the same unit as
15217 input timestamp.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015218
Tim Duesterhus3943e4f2020-09-11 14:25:23 +020015219iif(<true>,<false>)
15220 Returns the <true> string if the input value is true. Returns the <false>
15221 string otherwise.
15222
15223 Example:
Tim Duesterhus870713b2020-09-11 17:13:12 +020015224 http-request set-header x-forwarded-proto %[ssl_fc,iif(https,http)]
Tim Duesterhus3943e4f2020-09-11 14:25:23 +020015225
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015226in_table(<table>)
15227 Uses the string representation of the input sample to perform a look up in
15228 the specified table. If the key is not found in the table, a boolean false
15229 is returned. Otherwise a boolean true is returned. This can be used to verify
Davor Ocelice9ed2812017-12-25 17:49:28 +010015230 the presence of a certain key in a table tracking some elements (e.g. whether
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015231 or not a source IP address or an Authorization header was already seen).
15232
Tim Duesterhus1478aa72018-01-25 16:24:51 +010015233ipmask(<mask4>, [<mask6>])
15234 Apply a mask to an IP address, and use the result for lookups and storage.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020015235 This can be used to make all hosts within a certain mask to share the same
Tim Duesterhus1478aa72018-01-25 16:24:51 +010015236 table entries and as such use the same server. The mask4 can be passed in
15237 dotted form (e.g. 255.255.255.0) or in CIDR form (e.g. 24). The mask6 can
15238 be passed in quadruplet form (e.g. ffff:ffff::) or in CIDR form (e.g. 64).
15239 If no mask6 is given IPv6 addresses will fail to convert for backwards
15240 compatibility reasons.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020015241
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015242json([<input-code>])
Davor Ocelice9ed2812017-12-25 17:49:28 +010015243 Escapes the input string and produces an ASCII output string ready to use as a
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015244 JSON string. The converter tries to decode the input string according to the
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020015245 <input-code> parameter. It can be "ascii", "utf8", "utf8s", "utf8p" or
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015246 "utf8ps". The "ascii" decoder never fails. The "utf8" decoder detects 3 types
15247 of errors:
15248 - bad UTF-8 sequence (lone continuation byte, bad number of continuation
15249 bytes, ...)
15250 - invalid range (the decoded value is within a UTF-8 prohibited range),
15251 - code overlong (the value is encoded with more bytes than necessary).
15252
15253 The UTF-8 JSON encoding can produce a "too long value" error when the UTF-8
15254 character is greater than 0xffff because the JSON string escape specification
15255 only authorizes 4 hex digits for the value encoding. The UTF-8 decoder exists
15256 in 4 variants designated by a combination of two suffix letters : "p" for
15257 "permissive" and "s" for "silently ignore". The behaviors of the decoders
15258 are :
Davor Ocelice9ed2812017-12-25 17:49:28 +010015259 - "ascii" : never fails;
15260 - "utf8" : fails on any detected errors;
15261 - "utf8s" : never fails, but removes characters corresponding to errors;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015262 - "utf8p" : accepts and fixes the overlong errors, but fails on any other
Davor Ocelice9ed2812017-12-25 17:49:28 +010015263 error;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015264 - "utf8ps" : never fails, accepts and fixes the overlong errors, but removes
15265 characters corresponding to the other errors.
15266
15267 This converter is particularly useful for building properly escaped JSON for
Davor Ocelice9ed2812017-12-25 17:49:28 +010015268 logging to servers which consume JSON-formatted traffic logs.
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015269
15270 Example:
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015271 capture request header Host len 15
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020015272 capture request header user-agent len 150
15273 log-format '{"ip":"%[src]","user-agent":"%[capture.req.hdr(1),json(utf8s)]"}'
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015274
15275 Input request from client 127.0.0.1:
15276 GET / HTTP/1.0
15277 User-Agent: Very "Ugly" UA 1/2
15278
15279 Output log:
15280 {"ip":"127.0.0.1","user-agent":"Very \"Ugly\" UA 1\/2"}
15281
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015282language(<value>[,<default>])
15283 Returns the value with the highest q-factor from a list as extracted from the
15284 "accept-language" header using "req.fhdr". Values with no q-factor have a
15285 q-factor of 1. Values with a q-factor of 0 are dropped. Only values which
15286 belong to the list of semi-colon delimited <values> will be considered. The
15287 argument <value> syntax is "lang[;lang[;lang[;...]]]". If no value matches the
15288 given list and a default value is provided, it is returned. Note that language
15289 names may have a variant after a dash ('-'). If this variant is present in the
15290 list, it will be matched, but if it is not, only the base language is checked.
15291 The match is case-sensitive, and the output string is always one of those
Davor Ocelice9ed2812017-12-25 17:49:28 +010015292 provided in arguments. The ordering of arguments is meaningless, only the
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015293 ordering of the values in the request counts, as the first value among
15294 multiple sharing the same q-factor is used.
Thierry FOURNIERad903512014-04-11 17:51:01 +020015295
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015296 Example :
Thierry FOURNIERad903512014-04-11 17:51:01 +020015297
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015298 # this configuration switches to the backend matching a
15299 # given language based on the request :
Thierry FOURNIERad903512014-04-11 17:51:01 +020015300
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015301 acl es req.fhdr(accept-language),language(es;fr;en) -m str es
15302 acl fr req.fhdr(accept-language),language(es;fr;en) -m str fr
15303 acl en req.fhdr(accept-language),language(es;fr;en) -m str en
15304 use_backend spanish if es
15305 use_backend french if fr
15306 use_backend english if en
15307 default_backend choose_your_language
Thierry FOURNIERad903512014-04-11 17:51:01 +020015308
Willy Tarreau60a2ee72017-12-15 07:13:48 +010015309length
Etienne Carriereed0d24e2017-12-13 13:41:34 +010015310 Get the length of the string. This can only be placed after a string
15311 sample fetch function or after a transformation keyword returning a string
15312 type. The result is of type integer.
15313
Willy Tarreauffcb2e42014-07-10 16:29:08 +020015314lower
15315 Convert a string sample to lower case. This can only be placed after a string
15316 sample fetch function or after a transformation keyword returning a string
15317 type. The result is of type string.
15318
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020015319ltime(<format>[,<offset>])
15320 Converts an integer supposed to contain a date since epoch to a string
15321 representing this date in local time using a format defined by the <format>
15322 string using strftime(3). The purpose is to allow any date format to be used
15323 in logs. An optional <offset> in seconds may be applied to the input date
15324 (positive or negative). See the strftime() man page for the format supported
15325 by your operating system. See also the utime converter.
15326
15327 Example :
15328
15329 # Emit two colons, one with the local time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010015330 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020015331 log-format %[date,ltime(%Y%m%d%H%M%S)]\ %ci:%cp
15332
Christopher Faulet51fc9d12020-04-01 17:24:41 +020015333ltrim(<chars>)
15334 Skips any characters from <chars> from the beginning of the string
15335 representation of the input sample.
15336
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015337map(<map_file>[,<default_value>])
15338map_<match_type>(<map_file>[,<default_value>])
15339map_<match_type>_<output_type>(<map_file>[,<default_value>])
15340 Search the input value from <map_file> using the <match_type> matching method,
15341 and return the associated value converted to the type <output_type>. If the
15342 input value cannot be found in the <map_file>, the converter returns the
15343 <default_value>. If the <default_value> is not set, the converter fails and
15344 acts as if no input value could be fetched. If the <match_type> is not set, it
15345 defaults to "str". Likewise, if the <output_type> is not set, it defaults to
15346 "str". For convenience, the "map" keyword is an alias for "map_str" and maps a
15347 string to another string.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010015348
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015349 It is important to avoid overlapping between the keys : IP addresses and
15350 strings are stored in trees, so the first of the finest match will be used.
15351 Other keys are stored in lists, so the first matching occurrence will be used.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010015352
Tim Düsterhus4896c442016-11-29 02:15:19 +010015353 The following array contains the list of all map functions available sorted by
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015354 input type, match type and output type.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010015355
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015356 input type | match method | output type str | output type int | output type ip
15357 -----------+--------------+-----------------+-----------------+---------------
15358 str | str | map_str | map_str_int | map_str_ip
15359 -----------+--------------+-----------------+-----------------+---------------
Willy Tarreau787a4c02014-05-10 07:55:30 +020015360 str | beg | map_beg | map_beg_int | map_end_ip
15361 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015362 str | sub | map_sub | map_sub_int | map_sub_ip
15363 -----------+--------------+-----------------+-----------------+---------------
15364 str | dir | map_dir | map_dir_int | map_dir_ip
15365 -----------+--------------+-----------------+-----------------+---------------
15366 str | dom | map_dom | map_dom_int | map_dom_ip
15367 -----------+--------------+-----------------+-----------------+---------------
15368 str | end | map_end | map_end_int | map_end_ip
15369 -----------+--------------+-----------------+-----------------+---------------
Ruoshan Huang3c5e3742016-12-02 16:25:31 +080015370 str | reg | map_reg | map_reg_int | map_reg_ip
15371 -----------+--------------+-----------------+-----------------+---------------
15372 str | reg | map_regm | map_reg_int | map_reg_ip
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015373 -----------+--------------+-----------------+-----------------+---------------
15374 int | int | map_int | map_int_int | map_int_ip
15375 -----------+--------------+-----------------+-----------------+---------------
15376 ip | ip | map_ip | map_ip_int | map_ip_ip
15377 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010015378
Thierry Fournier8feaa662016-02-10 22:55:20 +010015379 The special map called "map_regm" expect matching zone in the regular
15380 expression and modify the output replacing back reference (like "\1") by
15381 the corresponding match text.
15382
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015383 The file contains one key + value per line. Lines which start with '#' are
15384 ignored, just like empty lines. Leading tabs and spaces are stripped. The key
15385 is then the first "word" (series of non-space/tabs characters), and the value
15386 is what follows this series of space/tab till the end of the line excluding
15387 trailing spaces/tabs.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010015388
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015389 Example :
15390
15391 # this is a comment and is ignored
15392 2.22.246.0/23 United Kingdom \n
15393 <-><-----------><--><------------><---->
15394 | | | | `- trailing spaces ignored
15395 | | | `---------- value
15396 | | `-------------------- middle spaces ignored
15397 | `---------------------------- key
15398 `------------------------------------ leading spaces ignored
15399
Willy Tarreau97707872015-01-27 15:12:13 +010015400mod(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015401 Divides the input value of type signed integer by <value>, and returns the
15402 remainder as an signed integer. If <value> is null, then zero is returned.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015403 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010015404 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015405 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015406 "sess" : the variable is shared with the whole session
15407 "txn" : the variable is shared with the transaction (request and response)
15408 "req" : the variable is shared only during request processing
15409 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010015410 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015411 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010015412
Baptiste Assmanne279ca62020-10-27 18:10:06 +010015413mqtt_field_value(<packettype>,<fieldname or property ID>)
15414 Returns value of <fieldname> found in input MQTT payload of type
15415 <packettype>.
15416 <packettype> can be either a string (case insensitive matching) or a numeric
15417 value corresponding to the type of packet we're supposed to extract data
15418 from.
15419 Supported string and integers can be found here:
15420 https://docs.oasis-open.org/mqtt/mqtt/v3.1.1/os/mqtt-v3.1.1-os.html#_Toc398718021
15421 https://docs.oasis-open.org/mqtt/mqtt/v5.0/os/mqtt-v5.0-os.html#_Toc3901022
15422
15423 <fieldname> depends on <packettype> and can be any of the following below.
15424 (note that <fieldname> matching is case insensitive).
15425 <property id> can only be found in MQTT v5.0 streams. check this table:
15426 https://docs.oasis-open.org/mqtt/mqtt/v5.0/os/mqtt-v5.0-os.html#_Toc3901029
15427
15428 - CONNECT (or 1): flags, protocol_name, protocol_version, client_identifier,
15429 will_topic, will_payload, username, password, keepalive
15430 OR any property ID as a numeric value (for MQTT v5.0
15431 packets only):
15432 17: Session Expiry Interval
15433 33: Receive Maximum
15434 39: Maximum Packet Size
15435 34: Topic Alias Maximum
15436 25: Request Response Information
15437 23: Request Problem Information
15438 21: Authentication Method
15439 22: Authentication Data
15440 18: Will Delay Interval
15441 1: Payload Format Indicator
15442 2: Message Expiry Interval
15443 3: Content Type
15444 8: Response Topic
15445 9: Correlation Data
15446 Not supported yet:
15447 38: User Property
15448
15449 - CONNACK (or 2): flags, protocol_version, reason_code
15450 OR any property ID as a numeric value (for MQTT v5.0
15451 packets only):
15452 17: Session Expiry Interval
15453 33: Receive Maximum
15454 36: Maximum QoS
15455 37: Retain Available
15456 39: Maximum Packet Size
15457 18: Assigned Client Identifier
15458 34: Topic Alias Maximum
15459 31: Reason String
15460 40; Wildcard Subscription Available
15461 41: Subscription Identifiers Available
15462 42: Shared Subscription Available
15463 19: Server Keep Alive
15464 26: Response Information
15465 28: Server Reference
15466 21: Authentication Method
15467 22: Authentication Data
15468 Not supported yet:
15469 38: User Property
15470
15471 Due to current HAProxy design, only the first message sent by the client and
15472 the server can be parsed. Thus this converter can extract data only from
15473 CONNECT and CONNACK packet types. CONNECT is the first message sent by the
15474 client and CONNACK is the first response sent by the server.
15475
15476 Example:
15477
15478 acl data_in_buffer req.len ge 4
15479 tcp-request content set-var(txn.username) \
15480 req.payload(0,0),mqtt_field_value(connect,protocol_name) \
15481 if data_in_buffer
15482 # do the same as above
15483 tcp-request content set-var(txn.username) \
15484 req.payload(0,0),mqtt_field_value(1,protocol_name) \
15485 if data_in_buffer
15486
15487mqtt_is_valid
15488 Checks that the binary input is a valid MQTT packet. It returns a boolean.
15489
15490 Due to current HAProxy design, only the first message sent by the client and
15491 the server can be parsed. Thus this converter can extract data only from
15492 CONNECT and CONNACK packet types. CONNECT is the first message sent by the
15493 client and CONNACK is the first response sent by the server.
15494
15495 Example:
15496
15497 acl data_in_buffer req.len ge 4
15498 tcp-request content reject unless req.payload(0,0),mqtt_is_valid
15499
Willy Tarreau97707872015-01-27 15:12:13 +010015500mul(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015501 Multiplies the input value of type signed integer by <value>, and returns
Thierry FOURNIER00c005c2015-07-08 01:10:21 +020015502 the product as an signed integer. In case of overflow, the largest possible
15503 value for the sign is returned so that the operation doesn't wrap around.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015504 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010015505 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015506 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015507 "sess" : the variable is shared with the whole session
15508 "txn" : the variable is shared with the transaction (request and response)
15509 "req" : the variable is shared only during request processing
15510 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010015511 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015512 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010015513
Nenad Merdanovicb7e7c472017-03-12 21:56:55 +010015514nbsrv
15515 Takes an input value of type string, interprets it as a backend name and
15516 returns the number of usable servers in that backend. Can be used in places
15517 where we want to look up a backend from a dynamic name, like a result of a
15518 map lookup.
15519
Willy Tarreau97707872015-01-27 15:12:13 +010015520neg
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015521 Takes the input value of type signed integer, computes the opposite value,
15522 and returns the remainder as an signed integer. 0 is identity. This operator
15523 is provided for reversed subtracts : in order to subtract the input from a
15524 constant, simply perform a "neg,add(value)".
Willy Tarreau97707872015-01-27 15:12:13 +010015525
15526not
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015527 Returns a boolean FALSE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010015528 non-null, otherwise returns TRUE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010015529 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010015530 absence of a flag).
15531
15532odd
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015533 Returns a boolean TRUE if the input value of type signed integer is odd
Willy Tarreau97707872015-01-27 15:12:13 +010015534 otherwise returns FALSE. It is functionally equivalent to "and(1),bool".
15535
15536or(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015537 Performs a bitwise "OR" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015538 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010015539 numeric value or a variable name. The name of the variable starts with an
15540 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015541 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015542 "sess" : the variable is shared with the whole session
15543 "txn" : the variable is shared with the transaction (request and response)
15544 "req" : the variable is shared only during request processing
15545 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010015546 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015547 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010015548
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010015549protobuf(<field_number>,[<field_type>])
15550 This extracts the protocol buffers message field in raw mode of an input binary
15551 sample representation of a protocol buffer message with <field_number> as field
15552 number (dotted notation) if <field_type> is not present, or as an integer sample
15553 if this field is present (see also "ungrpc" below).
15554 The list of the authorized types is the following one: "int32", "int64", "uint32",
15555 "uint64", "sint32", "sint64", "bool", "enum" for the "varint" wire type 0
15556 "fixed64", "sfixed64", "double" for the 64bit wire type 1, "fixed32", "sfixed32",
15557 "float" for the wire type 5. Note that "string" is considered as a length-delimited
15558 type, so it does not require any <field_type> argument to be extracted.
15559 More information may be found here about the protocol buffers message field types:
15560 https://developers.google.com/protocol-buffers/docs/encoding
15561
Willy Tarreauc4dc3502015-01-23 20:39:28 +010015562regsub(<regex>,<subst>[,<flags>])
Willy Tarreau7eda8492015-01-20 19:47:06 +010015563 Applies a regex-based substitution to the input string. It does the same
15564 operation as the well-known "sed" utility with "s/<regex>/<subst>/". By
15565 default it will replace in the input string the first occurrence of the
15566 largest part matching the regular expression <regex> with the substitution
15567 string <subst>. It is possible to replace all occurrences instead by adding
15568 the flag "g" in the third argument <flags>. It is also possible to make the
15569 regex case insensitive by adding the flag "i" in <flags>. Since <flags> is a
15570 string, it is made up from the concatenation of all desired flags. Thus if
15571 both "i" and "g" are desired, using "gi" or "ig" will have the same effect.
Willy Tarreauef21fac2020-02-14 13:37:20 +010015572 The first use of this converter is to replace certain characters or sequence
15573 of characters with other ones.
15574
15575 It is highly recommended to enclose the regex part using protected quotes to
15576 improve clarity and never have a closing parenthesis from the regex mixed up
15577 with the parenthesis from the function. Just like in Bourne shell, the first
15578 level of quotes is processed when delimiting word groups on the line, a
15579 second level is usable for argument. It is recommended to use single quotes
15580 outside since these ones do not try to resolve backslashes nor dollar signs.
Willy Tarreau7eda8492015-01-20 19:47:06 +010015581
Willy Tarreaucd0d2ed2020-02-14 17:33:06 +010015582 Examples:
Willy Tarreau7eda8492015-01-20 19:47:06 +010015583
15584 # de-duplicate "/" in header "x-path".
15585 # input: x-path: /////a///b/c/xzxyz/
15586 # output: x-path: /a/b/c/xzxyz/
Willy Tarreauef21fac2020-02-14 13:37:20 +010015587 http-request set-header x-path "%[hdr(x-path),regsub('/+','/','g')]"
Willy Tarreau7eda8492015-01-20 19:47:06 +010015588
Willy Tarreaucd0d2ed2020-02-14 17:33:06 +010015589 # copy query string to x-query and drop all leading '?', ';' and '&'
15590 http-request set-header x-query "%[query,regsub([?;&]*,'')]"
15591
Jerome Magnin07e1e3c2020-02-16 19:20:19 +010015592 # capture groups and backreferences
15593 # both lines do the same.
Willy Tarreau465dc7d2020-10-08 18:05:56 +020015594 http-request redirect location %[url,'regsub("(foo|bar)([0-9]+)?","\2\1",i)']
Jerome Magnin07e1e3c2020-02-16 19:20:19 +010015595 http-request redirect location %[url,regsub(\"(foo|bar)([0-9]+)?\",\"\2\1\",i)]
15596
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020015597capture-req(<id>)
15598 Capture the string entry in the request slot <id> and returns the entry as
15599 is. If the slot doesn't exist, the capture fails silently.
15600
15601 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020015602 "http-response capture", "capture.req.hdr" and
15603 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020015604
15605capture-res(<id>)
15606 Capture the string entry in the response slot <id> and returns the entry as
15607 is. If the slot doesn't exist, the capture fails silently.
15608
15609 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020015610 "http-response capture", "capture.req.hdr" and
15611 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020015612
Christopher Faulet568415a2020-04-01 17:24:47 +020015613rtrim(<chars>)
15614 Skips any characters from <chars> from the end of the string representation
15615 of the input sample.
15616
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015617sdbm([<avalanche>])
15618 Hashes a binary input sample into an unsigned 32-bit quantity using the SDBM
15619 hash function. Optionally, it is possible to apply a full avalanche hash
15620 function to the output if the optional <avalanche> argument equals 1. This
15621 converter uses the same functions as used by the various hash-based load
15622 balancing algorithms, so it will provide exactly the same results. It is
15623 mostly intended for debugging, but can be used as a stick-table entry to
15624 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010015625 32-bit hash is trivial to break. See also "crc32", "djb2", "wt6", "crc32c",
15626 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015627
Tim Duesterhusf38175c2020-06-09 11:48:42 +020015628secure_memcmp(<var>)
15629 Compares the contents of <var> with the input value. Both values are treated
15630 as a binary string. Returns a boolean indicating whether both binary strings
15631 match.
15632
15633 If both binary strings have the same length then the comparison will be
15634 performed in constant time.
15635
15636 Please note that this converter is only available when haproxy has been
15637 compiled with USE_OPENSSL.
15638
15639 Example :
15640
15641 http-request set-var(txn.token) hdr(token)
15642 # Check whether the token sent by the client matches the secret token
15643 # value, without leaking the contents using a timing attack.
15644 acl token_given str(my_secret_token),secure_memcmp(txn.token)
15645
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020015646set-var(<var name>)
Davor Ocelice9ed2812017-12-25 17:49:28 +010015647 Sets a variable with the input content and returns the content on the output
15648 as-is. The variable keeps the value and the associated input type. The name of
15649 the variable starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015650 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015651 "sess" : the variable is shared with the whole session
15652 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020015653 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010015654 "req" : the variable is shared only during request processing,
15655 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020015656 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015657 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020015658
Dragan Dosen6e5a9ca2017-10-24 09:18:23 +020015659sha1
Tim Duesterhusd4376302019-06-17 12:41:44 +020015660 Converts a binary input sample to a SHA-1 digest. The result is a binary
Dragan Dosen6e5a9ca2017-10-24 09:18:23 +020015661 sample with length of 20 bytes.
15662
Tim Duesterhusd4376302019-06-17 12:41:44 +020015663sha2([<bits>])
15664 Converts a binary input sample to a digest in the SHA-2 family. The result
15665 is a binary sample with length of <bits>/8 bytes.
15666
15667 Valid values for <bits> are 224, 256, 384, 512, each corresponding to
15668 SHA-<bits>. The default value is 256.
15669
15670 Please note that this converter is only available when haproxy has been
15671 compiled with USE_OPENSSL.
15672
Nenad Merdanovic177adc92019-08-27 01:58:13 +020015673srv_queue
15674 Takes an input value of type string, either a server name or <backend>/<server>
15675 format and returns the number of queued sessions on that server. Can be used
15676 in places where we want to look up queued sessions from a dynamic name, like a
15677 cookie value (e.g. req.cook(SRVID),srv_queue) and then make a decision to break
15678 persistence or direct a request elsewhere.
15679
Tim Duesterhusca097c12018-04-27 21:18:45 +020015680strcmp(<var>)
15681 Compares the contents of <var> with the input value of type string. Returns
15682 the result as a signed integer compatible with strcmp(3): 0 if both strings
15683 are identical. A value less than 0 if the left string is lexicographically
15684 smaller than the right string or if the left string is shorter. A value greater
15685 than 0 otherwise (right string greater than left string or the right string is
15686 shorter).
15687
Tim Duesterhusf38175c2020-06-09 11:48:42 +020015688 See also the secure_memcmp converter if you need to compare two binary
15689 strings in constant time.
15690
Tim Duesterhusca097c12018-04-27 21:18:45 +020015691 Example :
15692
15693 http-request set-var(txn.host) hdr(host)
15694 # Check whether the client is attempting domain fronting.
15695 acl ssl_sni_http_host_match ssl_fc_sni,strcmp(txn.host) eq 0
15696
15697
Willy Tarreau97707872015-01-27 15:12:13 +010015698sub(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015699 Subtracts <value> from the input value of type signed integer, and returns
15700 the result as an signed integer. Note: in order to subtract the input from
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015701 a constant, simply perform a "neg,add(value)". <value> can be a numeric value
Daniel Schneller0b547052016-03-21 20:46:57 +010015702 or a variable name. The name of the variable starts with an indication about
15703 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015704 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015705 "sess" : the variable is shared with the whole session
15706 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015707 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010015708 "req" : the variable is shared only during request processing,
15709 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015710 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015711 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010015712
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015713table_bytes_in_rate(<table>)
15714 Uses the string representation of the input sample to perform a look up in
15715 the specified table. If the key is not found in the table, integer value zero
15716 is returned. Otherwise the converter returns the average client-to-server
15717 bytes rate associated with the input sample in the designated table, measured
15718 in amount of bytes over the period configured in the table. See also the
15719 sc_bytes_in_rate sample fetch keyword.
15720
15721
15722table_bytes_out_rate(<table>)
15723 Uses the string representation of the input sample to perform a look up in
15724 the specified table. If the key is not found in the table, integer value zero
15725 is returned. Otherwise the converter returns the average server-to-client
15726 bytes rate associated with the input sample in the designated table, measured
15727 in amount of bytes over the period configured in the table. See also the
15728 sc_bytes_out_rate sample fetch keyword.
15729
15730table_conn_cnt(<table>)
15731 Uses the string representation of the input sample to perform a look up in
15732 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010015733 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015734 connections associated with the input sample in the designated table. See
15735 also the sc_conn_cnt sample fetch keyword.
15736
15737table_conn_cur(<table>)
15738 Uses the string representation of the input sample to perform a look up in
15739 the specified table. If the key is not found in the table, integer value zero
15740 is returned. Otherwise the converter returns the current amount of concurrent
15741 tracked connections associated with the input sample in the designated table.
15742 See also the sc_conn_cur sample fetch keyword.
15743
15744table_conn_rate(<table>)
15745 Uses the string representation of the input sample to perform a look up in
15746 the specified table. If the key is not found in the table, integer value zero
15747 is returned. Otherwise the converter returns the average incoming connection
15748 rate associated with the input sample in the designated table. See also the
15749 sc_conn_rate sample fetch keyword.
15750
Thierry FOURNIER236657b2015-08-19 08:25:14 +020015751table_gpt0(<table>)
15752 Uses the string representation of the input sample to perform a look up in
15753 the specified table. If the key is not found in the table, boolean value zero
15754 is returned. Otherwise the converter returns the current value of the first
15755 general purpose tag associated with the input sample in the designated table.
15756 See also the sc_get_gpt0 sample fetch keyword.
15757
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015758table_gpc0(<table>)
15759 Uses the string representation of the input sample to perform a look up in
15760 the specified table. If the key is not found in the table, integer value zero
15761 is returned. Otherwise the converter returns the current value of the first
15762 general purpose counter associated with the input sample in the designated
15763 table. See also the sc_get_gpc0 sample fetch keyword.
15764
15765table_gpc0_rate(<table>)
15766 Uses the string representation of the input sample to perform a look up in
15767 the specified table. If the key is not found in the table, integer value zero
15768 is returned. Otherwise the converter returns the frequency which the gpc0
15769 counter was incremented over the configured period in the table, associated
15770 with the input sample in the designated table. See also the sc_get_gpc0_rate
15771 sample fetch keyword.
15772
Frédéric Lécaille6778b272018-01-29 15:22:53 +010015773table_gpc1(<table>)
15774 Uses the string representation of the input sample to perform a look up in
15775 the specified table. If the key is not found in the table, integer value zero
15776 is returned. Otherwise the converter returns the current value of the second
15777 general purpose counter associated with the input sample in the designated
15778 table. See also the sc_get_gpc1 sample fetch keyword.
15779
15780table_gpc1_rate(<table>)
15781 Uses the string representation of the input sample to perform a look up in
15782 the specified table. If the key is not found in the table, integer value zero
15783 is returned. Otherwise the converter returns the frequency which the gpc1
15784 counter was incremented over the configured period in the table, associated
15785 with the input sample in the designated table. See also the sc_get_gpc1_rate
15786 sample fetch keyword.
15787
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015788table_http_err_cnt(<table>)
15789 Uses the string representation of the input sample to perform a look up in
15790 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010015791 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015792 errors associated with the input sample in the designated table. See also the
15793 sc_http_err_cnt sample fetch keyword.
15794
15795table_http_err_rate(<table>)
15796 Uses the string representation of the input sample to perform a look up in
15797 the specified table. If the key is not found in the table, integer value zero
15798 is returned. Otherwise the average rate of HTTP errors associated with the
15799 input sample in the designated table, measured in amount of errors over the
15800 period configured in the table. See also the sc_http_err_rate sample fetch
15801 keyword.
15802
15803table_http_req_cnt(<table>)
15804 Uses the string representation of the input sample to perform a look up in
15805 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010015806 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015807 requests associated with the input sample in the designated table. See also
15808 the sc_http_req_cnt sample fetch keyword.
15809
15810table_http_req_rate(<table>)
15811 Uses the string representation of the input sample to perform a look up in
15812 the specified table. If the key is not found in the table, integer value zero
15813 is returned. Otherwise the average rate of HTTP requests associated with the
15814 input sample in the designated table, measured in amount of requests over the
15815 period configured in the table. See also the sc_http_req_rate sample fetch
15816 keyword.
15817
15818table_kbytes_in(<table>)
15819 Uses the string representation of the input sample to perform a look up in
15820 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010015821 is returned. Otherwise the converter returns the cumulative number of client-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015822 to-server data associated with the input sample in the designated table,
15823 measured in kilobytes. The test is currently performed on 32-bit integers,
15824 which limits values to 4 terabytes. See also the sc_kbytes_in sample fetch
15825 keyword.
15826
15827table_kbytes_out(<table>)
15828 Uses the string representation of the input sample to perform a look up in
15829 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010015830 is returned. Otherwise the converter returns the cumulative number of server-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015831 to-client data associated with the input sample in the designated table,
15832 measured in kilobytes. The test is currently performed on 32-bit integers,
15833 which limits values to 4 terabytes. See also the sc_kbytes_out sample fetch
15834 keyword.
15835
15836table_server_id(<table>)
15837 Uses the string representation of the input sample to perform a look up in
15838 the specified table. If the key is not found in the table, integer value zero
15839 is returned. Otherwise the converter returns the server ID associated with
15840 the input sample in the designated table. A server ID is associated to a
15841 sample by a "stick" rule when a connection to a server succeeds. A server ID
15842 zero means that no server is associated with this key.
15843
15844table_sess_cnt(<table>)
15845 Uses the string representation of the input sample to perform a look up in
15846 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010015847 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015848 sessions associated with the input sample in the designated table. Note that
15849 a session here refers to an incoming connection being accepted by the
15850 "tcp-request connection" rulesets. See also the sc_sess_cnt sample fetch
15851 keyword.
15852
15853table_sess_rate(<table>)
15854 Uses the string representation of the input sample to perform a look up in
15855 the specified table. If the key is not found in the table, integer value zero
15856 is returned. Otherwise the converter returns the average incoming session
15857 rate associated with the input sample in the designated table. Note that a
15858 session here refers to an incoming connection being accepted by the
15859 "tcp-request connection" rulesets. See also the sc_sess_rate sample fetch
15860 keyword.
15861
15862table_trackers(<table>)
15863 Uses the string representation of the input sample to perform a look up in
15864 the specified table. If the key is not found in the table, integer value zero
15865 is returned. Otherwise the converter returns the current amount of concurrent
15866 connections tracking the same key as the input sample in the designated
15867 table. It differs from table_conn_cur in that it does not rely on any stored
15868 information but on the table's reference count (the "use" value which is
15869 returned by "show table" on the CLI). This may sometimes be more suited for
15870 layer7 tracking. It can be used to tell a server how many concurrent
15871 connections there are from a given address for example. See also the
15872 sc_trackers sample fetch keyword.
15873
Willy Tarreauffcb2e42014-07-10 16:29:08 +020015874upper
15875 Convert a string sample to upper case. This can only be placed after a string
15876 sample fetch function or after a transformation keyword returning a string
15877 type. The result is of type string.
15878
Willy Tarreau62ba9ba2020-04-23 17:54:47 +020015879url_dec([<in_form>])
15880 Takes an url-encoded string provided as input and returns the decoded version
15881 as output. The input and the output are of type string. If the <in_form>
15882 argument is set to a non-zero integer value, the input string is assumed to
15883 be part of a form or query string and the '+' character will be turned into a
15884 space (' '). Otherwise this will only happen after a question mark indicating
15885 a query string ('?').
Thierry FOURNIER82ff3c92015-05-07 15:46:20 +020015886
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010015887ungrpc(<field_number>,[<field_type>])
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010015888 This extracts the protocol buffers message field in raw mode of an input binary
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010015889 sample representation of a gRPC message with <field_number> as field number
15890 (dotted notation) if <field_type> is not present, or as an integer sample if this
15891 field is present.
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010015892 The list of the authorized types is the following one: "int32", "int64", "uint32",
15893 "uint64", "sint32", "sint64", "bool", "enum" for the "varint" wire type 0
15894 "fixed64", "sfixed64", "double" for the 64bit wire type 1, "fixed32", "sfixed32",
15895 "float" for the wire type 5. Note that "string" is considered as a length-delimited
Frédéric Lécaille93d33162019-03-06 09:35:59 +010015896 type, so it does not require any <field_type> argument to be extracted.
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010015897 More information may be found here about the protocol buffers message field types:
15898 https://developers.google.com/protocol-buffers/docs/encoding
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010015899
15900 Example:
15901 // with such a protocol buffer .proto file content adapted from
15902 // https://github.com/grpc/grpc/blob/master/examples/protos/route_guide.proto
15903
15904 message Point {
15905 int32 latitude = 1;
15906 int32 longitude = 2;
15907 }
15908
15909 message PPoint {
15910 Point point = 59;
15911 }
15912
15913 message Rectangle {
15914 // One corner of the rectangle.
15915 PPoint lo = 48;
15916 // The other corner of the rectangle.
15917 PPoint hi = 49;
15918 }
15919
Peter Gervaidf4c9d22020-06-11 18:05:11 +020015920 let's say a body request is made of a "Rectangle" object value (two PPoint
15921 protocol buffers messages), the four protocol buffers fields could be
15922 extracted with these "ungrpc" directives:
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010015923
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010015924 req.body,ungrpc(48.59.1,int32) # "latitude" of "lo" first PPoint
15925 req.body,ungrpc(48.59.2,int32) # "longitude" of "lo" first PPoint
John Roeslerfb2fce12019-07-10 15:45:51 -050015926 req.body,ungrpc(49.59.1,int32) # "latitude" of "hi" second PPoint
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010015927 req.body,ungrpc(49.59.2,int32) # "longitude" of "hi" second PPoint
15928
Peter Gervaidf4c9d22020-06-11 18:05:11 +020015929 We could also extract the intermediary 48.59 field as a binary sample as follows:
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010015930
Frédéric Lécaille93d33162019-03-06 09:35:59 +010015931 req.body,ungrpc(48.59)
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010015932
Peter Gervaidf4c9d22020-06-11 18:05:11 +020015933 As a gRPC message is always made of a gRPC header followed by protocol buffers
15934 messages, in the previous example the "latitude" of "lo" first PPoint
15935 could be extracted with these equivalent directives:
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010015936
15937 req.body,ungrpc(48.59),protobuf(1,int32)
15938 req.body,ungrpc(48),protobuf(59.1,int32)
15939 req.body,ungrpc(48),protobuf(59),protobuf(1,int32)
15940
Peter Gervaidf4c9d22020-06-11 18:05:11 +020015941 Note that the first convert must be "ungrpc", the remaining ones must be
15942 "protobuf" and only the last one may have or not a second argument to
15943 interpret the previous binary sample.
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010015944
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010015945
Christopher Faulet85d79c92016-11-09 16:54:56 +010015946unset-var(<var name>)
15947 Unsets a variable if the input content is defined. The name of the variable
15948 starts with an indication about its scope. The scopes allowed are:
15949 "proc" : the variable is shared with the whole process
15950 "sess" : the variable is shared with the whole session
15951 "txn" : the variable is shared with the transaction (request and
15952 response),
15953 "req" : the variable is shared only during request processing,
15954 "res" : the variable is shared only during response processing.
15955 This prefix is followed by a name. The separator is a '.'. The name may only
15956 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
15957
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020015958utime(<format>[,<offset>])
15959 Converts an integer supposed to contain a date since epoch to a string
15960 representing this date in UTC time using a format defined by the <format>
15961 string using strftime(3). The purpose is to allow any date format to be used
15962 in logs. An optional <offset> in seconds may be applied to the input date
15963 (positive or negative). See the strftime() man page for the format supported
15964 by your operating system. See also the ltime converter.
15965
15966 Example :
15967
15968 # Emit two colons, one with the UTC time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010015969 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020015970 log-format %[date,utime(%Y%m%d%H%M%S)]\ %ci:%cp
15971
Marcin Deranek9631a282018-04-16 14:30:46 +020015972word(<index>,<delimiters>[,<count>])
15973 Extracts the nth word counting from the beginning (positive index) or from
15974 the end (negative index) considering given delimiters from an input string.
15975 Indexes start at 1 or -1 and delimiters are a string formatted list of chars.
Jerome Magnin88209322020-01-28 13:33:44 +010015976 Delimiters at the beginning or end of the input string are ignored.
Marcin Deranek9631a282018-04-16 14:30:46 +020015977 Optionally you can specify <count> of words to extract (default: 1).
15978 Value of 0 indicates extraction of all remaining words.
15979
15980 Example :
15981 str(f1_f2_f3__f5),word(4,_) # f5
15982 str(f1_f2_f3__f5),word(2,_,0) # f2_f3__f5
15983 str(f1_f2_f3__f5),word(3,_,2) # f3__f5
15984 str(f1_f2_f3__f5),word(-2,_,3) # f1_f2_f3
15985 str(f1_f2_f3__f5),word(-3,_,0) # f1_f2
Jerome Magnin88209322020-01-28 13:33:44 +010015986 str(/f1/f2/f3/f4),word(1,/) # f1
Emeric Brunc9a0f6d2014-11-25 14:09:01 +010015987
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015988wt6([<avalanche>])
15989 Hashes a binary input sample into an unsigned 32-bit quantity using the WT6
15990 hash function. Optionally, it is possible to apply a full avalanche hash
15991 function to the output if the optional <avalanche> argument equals 1. This
15992 converter uses the same functions as used by the various hash-based load
15993 balancing algorithms, so it will provide exactly the same results. It is
15994 mostly intended for debugging, but can be used as a stick-table entry to
15995 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010015996 32-bit hash is trivial to break. See also "crc32", "djb2", "sdbm", "crc32c",
15997 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015998
Willy Tarreau97707872015-01-27 15:12:13 +010015999xor(<value>)
16000 Performs a bitwise "XOR" (exclusive OR) between <value> and the input value
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016001 of type signed integer, and returns the result as an signed integer.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016002 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010016003 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010016004 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010016005 "sess" : the variable is shared with the whole session
16006 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016007 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010016008 "req" : the variable is shared only during request processing,
16009 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016010 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010016011 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010016012
Thierry FOURNIER01e09742016-12-26 11:46:11 +010016013xxh32([<seed>])
16014 Hashes a binary input sample into an unsigned 32-bit quantity using the 32-bit
16015 variant of the XXHash hash function. This hash supports a seed which defaults
16016 to zero but a different value maybe passed as the <seed> argument. This hash
16017 is known to be very good and very fast so it can be used to hash URLs and/or
16018 URL parameters for use as stick-table keys to collect statistics with a low
16019 collision rate, though care must be taken as the algorithm is not considered
16020 as cryptographically secure.
16021
16022xxh64([<seed>])
16023 Hashes a binary input sample into a signed 64-bit quantity using the 64-bit
16024 variant of the XXHash hash function. This hash supports a seed which defaults
16025 to zero but a different value maybe passed as the <seed> argument. This hash
16026 is known to be very good and very fast so it can be used to hash URLs and/or
16027 URL parameters for use as stick-table keys to collect statistics with a low
16028 collision rate, though care must be taken as the algorithm is not considered
16029 as cryptographically secure.
16030
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010016031
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200160327.3.2. Fetching samples from internal states
Willy Tarreau74ca5042013-06-11 23:12:07 +020016033--------------------------------------------
16034
16035A first set of sample fetch methods applies to internal information which does
16036not even relate to any client information. These ones are sometimes used with
16037"monitor-fail" directives to report an internal status to external watchers.
16038The sample fetch methods described in this section are usable anywhere.
16039
16040always_false : boolean
16041 Always returns the boolean "false" value. It may be used with ACLs as a
16042 temporary replacement for another one when adjusting configurations.
16043
16044always_true : boolean
16045 Always returns the boolean "true" value. It may be used with ACLs as a
16046 temporary replacement for another one when adjusting configurations.
16047
16048avg_queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010016049 Returns the total number of queued connections of the designated backend
Willy Tarreau74ca5042013-06-11 23:12:07 +020016050 divided by the number of active servers. The current backend is used if no
16051 backend is specified. This is very similar to "queue" except that the size of
16052 the farm is considered, in order to give a more accurate measurement of the
16053 time it may take for a new connection to be processed. The main usage is with
16054 ACL to return a sorry page to new users when it becomes certain they will get
16055 a degraded service, or to pass to the backend servers in a header so that
16056 they decide to work in degraded mode or to disable some functions to speed up
16057 the processing a bit. Note that in the event there would not be any active
16058 server anymore, twice the number of queued connections would be considered as
16059 the measured value. This is a fair estimate, as we expect one server to get
16060 back soon anyway, but we still prefer to send new traffic to another backend
16061 if in better shape. See also the "queue", "be_conn", and "be_sess_rate"
16062 sample fetches.
Krzysztof Piotr Oledzki346f76d2010-01-12 21:59:30 +010016063
Willy Tarreau74ca5042013-06-11 23:12:07 +020016064be_conn([<backend>]) : integer
Willy Tarreaua36af912009-10-10 12:02:45 +020016065 Applies to the number of currently established connections on the backend,
16066 possibly including the connection being evaluated. If no backend name is
16067 specified, the current one is used. But it is also possible to check another
16068 backend. It can be used to use a specific farm when the nominal one is full.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040016069 See also the "fe_conn", "queue", "be_conn_free", and "be_sess_rate" criteria.
16070
16071be_conn_free([<backend>]) : integer
16072 Returns an integer value corresponding to the number of available connections
16073 across available servers in the backend. Queue slots are not included. Backup
16074 servers are also not included, unless all other servers are down. If no
16075 backend name is specified, the current one is used. But it is also possible
16076 to check another backend. It can be used to use a specific farm when the
Patrick Hemmer155e93e2018-06-14 18:01:35 -040016077 nominal one is full. See also the "be_conn", "connslots", and "srv_conn_free"
16078 criteria.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040016079
16080 OTHER CAVEATS AND NOTES: if any of the server maxconn, or maxqueue is 0
16081 (meaning unlimited), then this fetch clearly does not make sense, in which
16082 case the value returned will be -1.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016083
Willy Tarreau74ca5042013-06-11 23:12:07 +020016084be_sess_rate([<backend>]) : integer
16085 Returns an integer value corresponding to the sessions creation rate on the
16086 backend, in number of new sessions per second. This is used with ACLs to
16087 switch to an alternate backend when an expensive or fragile one reaches too
Davor Ocelice9ed2812017-12-25 17:49:28 +010016088 high a session rate, or to limit abuse of service (e.g. prevent sucking of an
Willy Tarreau74ca5042013-06-11 23:12:07 +020016089 online dictionary). It can also be useful to add this element to logs using a
16090 log-format directive.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016091
16092 Example :
16093 # Redirect to an error page if the dictionary is requested too often
16094 backend dynamic
16095 mode http
16096 acl being_scanned be_sess_rate gt 100
16097 redirect location /denied.html if being_scanned
Willy Tarreau0ba27502007-12-24 16:55:16 +010016098
Davor Ocelice9ed2812017-12-25 17:49:28 +010016099bin(<hex>) : bin
Thierry FOURNIERcc103292015-06-06 19:30:17 +020016100 Returns a binary chain. The input is the hexadecimal representation
16101 of the string.
16102
16103bool(<bool>) : bool
16104 Returns a boolean value. <bool> can be 'true', 'false', '1' or '0'.
16105 'false' and '0' are the same. 'true' and '1' are the same.
16106
Willy Tarreau74ca5042013-06-11 23:12:07 +020016107connslots([<backend>]) : integer
16108 Returns an integer value corresponding to the number of connection slots
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016109 still available in the backend, by totaling the maximum amount of
Willy Tarreau74ca5042013-06-11 23:12:07 +020016110 connections on all servers and the maximum queue size. This is probably only
16111 used with ACLs.
Tait Clarridge7896d522012-12-05 21:39:31 -050016112
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080016113 The basic idea here is to be able to measure the number of connection "slots"
Willy Tarreau55165fe2009-05-10 12:02:55 +020016114 still available (connection + queue), so that anything beyond that (intended
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080016115 usage; see "use_backend" keyword) can be redirected to a different backend.
16116
Willy Tarreau55165fe2009-05-10 12:02:55 +020016117 'connslots' = number of available server connection slots, + number of
16118 available server queue slots.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080016119
Willy Tarreaua36af912009-10-10 12:02:45 +020016120 Note that while "fe_conn" may be used, "connslots" comes in especially
Willy Tarreau55165fe2009-05-10 12:02:55 +020016121 useful when you have a case of traffic going to one single ip, splitting into
Willy Tarreau74ca5042013-06-11 23:12:07 +020016122 multiple backends (perhaps using ACLs to do name-based load balancing) and
Willy Tarreau55165fe2009-05-10 12:02:55 +020016123 you want to be able to differentiate between different backends, and their
Davor Ocelice9ed2812017-12-25 17:49:28 +010016124 available "connslots". Also, whereas "nbsrv" only measures servers that are
Willy Tarreau74ca5042013-06-11 23:12:07 +020016125 actually *down*, this fetch is more fine-grained and looks into the number of
Willy Tarreaua36af912009-10-10 12:02:45 +020016126 available connection slots as well. See also "queue" and "avg_queue".
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080016127
Willy Tarreau55165fe2009-05-10 12:02:55 +020016128 OTHER CAVEATS AND NOTES: at this point in time, the code does not take care
16129 of dynamic connections. Also, if any of the server maxconn, or maxqueue is 0,
Willy Tarreau74ca5042013-06-11 23:12:07 +020016130 then this fetch clearly does not make sense, in which case the value returned
Willy Tarreau55165fe2009-05-10 12:02:55 +020016131 will be -1.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080016132
Willy Tarreau70fe9442018-11-22 16:07:39 +010016133cpu_calls : integer
16134 Returns the number of calls to the task processing the stream or current
16135 request since it was allocated. This number is reset for each new request on
16136 the same connections in case of HTTP keep-alive. This value should usually be
16137 low and stable (around 2 calls for a typically simple request) but may become
16138 high if some processing (compression, caching or analysis) is performed. This
16139 is purely for performance monitoring purposes.
16140
16141cpu_ns_avg : integer
16142 Returns the average number of nanoseconds spent in each call to the task
16143 processing the stream or current request. This number is reset for each new
16144 request on the same connections in case of HTTP keep-alive. This value
16145 indicates the overall cost of processing the request or the connection for
16146 each call. There is no good nor bad value but the time spent in a call
16147 automatically causes latency for other processing (see lat_ns_avg below),
16148 and may affect other connection's apparent response time. Certain operations
16149 like compression, complex regex matching or heavy Lua operations may directly
16150 affect this value, and having it in the logs will make it easier to spot the
16151 faulty processing that needs to be fixed to recover decent performance.
16152 Note: this value is exactly cpu_ns_tot divided by cpu_calls.
16153
16154cpu_ns_tot : integer
16155 Returns the total number of nanoseconds spent in each call to the task
16156 processing the stream or current request. This number is reset for each new
16157 request on the same connections in case of HTTP keep-alive. This value
16158 indicates the overall cost of processing the request or the connection for
16159 each call. There is no good nor bad value but the time spent in a call
16160 automatically causes latency for other processing (see lat_ns_avg below),
16161 induces CPU costs on the machine, and may affect other connection's apparent
16162 response time. Certain operations like compression, complex regex matching or
16163 heavy Lua operations may directly affect this value, and having it in the
16164 logs will make it easier to spot the faulty processing that needs to be fixed
16165 to recover decent performance. The value may be artificially high due to a
16166 high cpu_calls count, for example when processing many HTTP chunks, and for
16167 this reason it is often preferred to log cpu_ns_avg instead.
16168
Cyril Bonté6bcd1822019-11-05 23:13:59 +010016169date([<offset>],[<unit>]) : integer
Willy Tarreau6236d3a2013-07-25 14:28:25 +020016170 Returns the current date as the epoch (number of seconds since 01/01/1970).
Damien Claisseae6f1252019-10-30 15:57:28 +000016171
16172 If an offset value is specified, then it is added to the current date before
16173 returning the value. This is particularly useful to compute relative dates,
16174 as both positive and negative offsets are allowed.
Willy Tarreau276fae92013-07-25 14:36:01 +020016175 It is useful combined with the http_date converter.
16176
Damien Claisseae6f1252019-10-30 15:57:28 +000016177 <unit> is facultative, and can be set to "s" for seconds (default behavior),
16178 "ms" for milliseconds or "us" for microseconds.
16179 If unit is set, return value is an integer reflecting either seconds,
16180 milliseconds or microseconds since epoch, plus offset.
16181 It is useful when a time resolution of less than a second is needed.
16182
Willy Tarreau276fae92013-07-25 14:36:01 +020016183 Example :
16184
16185 # set an expires header to now+1 hour in every response
16186 http-response set-header Expires %[date(3600),http_date]
Willy Tarreau6236d3a2013-07-25 14:28:25 +020016187
Damien Claisseae6f1252019-10-30 15:57:28 +000016188 # set an expires header to now+1 hour in every response, with
16189 # millisecond granularity
16190 http-response set-header Expires %[date(3600000,ms),http_date(0,ms)]
16191
Etienne Carrierea792a0a2018-01-17 13:43:24 +010016192date_us : integer
16193 Return the microseconds part of the date (the "second" part is returned by
16194 date sample). This sample is coherent with the date sample as it is comes
16195 from the same timeval structure.
16196
Willy Tarreaud716f9b2017-10-13 11:03:15 +020016197distcc_body(<token>[,<occ>]) : binary
16198 Parses a distcc message and returns the body associated to occurrence #<occ>
16199 of the token <token>. Occurrences start at 1, and when unspecified, any may
16200 match though in practice only the first one is checked for now. This can be
16201 used to extract file names or arguments in files built using distcc through
16202 haproxy. Please refer to distcc's protocol documentation for the complete
16203 list of supported tokens.
16204
16205distcc_param(<token>[,<occ>]) : integer
16206 Parses a distcc message and returns the parameter associated to occurrence
16207 #<occ> of the token <token>. Occurrences start at 1, and when unspecified,
16208 any may match though in practice only the first one is checked for now. This
16209 can be used to extract certain information such as the protocol version, the
16210 file size or the argument in files built using distcc through haproxy.
16211 Another use case consists in waiting for the start of the preprocessed file
16212 contents before connecting to the server to avoid keeping idle connections.
16213 Please refer to distcc's protocol documentation for the complete list of
16214 supported tokens.
16215
16216 Example :
16217 # wait up to 20s for the pre-processed file to be uploaded
16218 tcp-request inspect-delay 20s
16219 tcp-request content accept if { distcc_param(DOTI) -m found }
16220 # send large files to the big farm
16221 use_backend big_farm if { distcc_param(DOTI) gt 1000000 }
16222
Willy Tarreau595ec542013-06-12 21:34:28 +020016223env(<name>) : string
16224 Returns a string containing the value of environment variable <name>. As a
16225 reminder, environment variables are per-process and are sampled when the
16226 process starts. This can be useful to pass some information to a next hop
16227 server, or with ACLs to take specific action when the process is started a
16228 certain way.
16229
16230 Examples :
16231 # Pass the Via header to next hop with the local hostname in it
16232 http-request add-header Via 1.1\ %[env(HOSTNAME)]
16233
16234 # reject cookie-less requests when the STOP environment variable is set
16235 http-request deny if !{ cook(SESSIONID) -m found } { env(STOP) -m found }
16236
Willy Tarreau74ca5042013-06-11 23:12:07 +020016237fe_conn([<frontend>]) : integer
16238 Returns the number of currently established connections on the frontend,
Willy Tarreaud63335a2010-02-26 12:56:52 +010016239 possibly including the connection being evaluated. If no frontend name is
16240 specified, the current one is used. But it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020016241 frontend. It can be used to return a sorry page before hard-blocking, or to
16242 use a specific backend to drain new requests when the farm is considered
Davor Ocelice9ed2812017-12-25 17:49:28 +010016243 full. This is mostly used with ACLs but can also be used to pass some
Willy Tarreau74ca5042013-06-11 23:12:07 +020016244 statistics to servers in HTTP headers. See also the "dst_conn", "be_conn",
16245 "fe_sess_rate" fetches.
Willy Tarreaua36af912009-10-10 12:02:45 +020016246
Nenad Merdanovicad9a7e92016-10-03 04:57:37 +020016247fe_req_rate([<frontend>]) : integer
16248 Returns an integer value corresponding to the number of HTTP requests per
16249 second sent to a frontend. This number can differ from "fe_sess_rate" in
16250 situations where client-side keep-alive is enabled.
16251
Willy Tarreau74ca5042013-06-11 23:12:07 +020016252fe_sess_rate([<frontend>]) : integer
16253 Returns an integer value corresponding to the sessions creation rate on the
16254 frontend, in number of new sessions per second. This is used with ACLs to
16255 limit the incoming session rate to an acceptable range in order to prevent
16256 abuse of service at the earliest moment, for example when combined with other
16257 layer 4 ACLs in order to force the clients to wait a bit for the rate to go
16258 down below the limit. It can also be useful to add this element to logs using
16259 a log-format directive. See also the "rate-limit sessions" directive for use
16260 in frontends.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010016261
16262 Example :
16263 # This frontend limits incoming mails to 10/s with a max of 100
16264 # concurrent connections. We accept any connection below 10/s, and
16265 # force excess clients to wait for 100 ms. Since clients are limited to
16266 # 100 max, there cannot be more than 10 incoming mails per second.
16267 frontend mail
16268 bind :25
16269 mode tcp
16270 maxconn 100
16271 acl too_fast fe_sess_rate ge 10
16272 tcp-request inspect-delay 100ms
16273 tcp-request content accept if ! too_fast
16274 tcp-request content accept if WAIT_END
Willy Tarreaud72758d2010-01-12 10:42:19 +010016275
Nenad Merdanovic807a6e72017-03-12 22:00:00 +010016276hostname : string
16277 Returns the system hostname.
16278
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016279int(<integer>) : signed integer
16280 Returns a signed integer.
16281
Thierry FOURNIERcc103292015-06-06 19:30:17 +020016282ipv4(<ipv4>) : ipv4
16283 Returns an ipv4.
16284
16285ipv6(<ipv6>) : ipv6
16286 Returns an ipv6.
16287
Willy Tarreau70fe9442018-11-22 16:07:39 +010016288lat_ns_avg : integer
16289 Returns the average number of nanoseconds spent between the moment the task
16290 handling the stream is woken up and the moment it is effectively called. This
16291 number is reset for each new request on the same connections in case of HTTP
16292 keep-alive. This value indicates the overall latency inflicted to the current
16293 request by all other requests being processed in parallel, and is a direct
16294 indicator of perceived performance due to noisy neighbours. In order to keep
16295 the value low, it is possible to reduce the scheduler's run queue depth using
16296 "tune.runqueue-depth", to reduce the number of concurrent events processed at
16297 once using "tune.maxpollevents", to decrease the stream's nice value using
Willy Tarreaue7723bd2020-06-24 11:11:02 +020016298 the "nice" option on the "bind" lines or in the frontend, to enable low
16299 latency scheduling using "tune.sched.low-latency", or to look for other heavy
16300 requests in logs (those exhibiting large values of "cpu_ns_avg"), whose
16301 processing needs to be adjusted or fixed. Compression of large buffers could
16302 be a culprit, like heavy regex or long lists of regex. Note: this value is
16303 exactly lat_ns_tot divided by cpu_calls.
Willy Tarreau70fe9442018-11-22 16:07:39 +010016304
16305lat_ns_tot : integer
16306 Returns the total number of nanoseconds spent between the moment the task
16307 handling the stream is woken up and the moment it is effectively called. This
16308 number is reset for each new request on the same connections in case of HTTP
16309 keep-alive. This value indicates the overall latency inflicted to the current
16310 request by all other requests being processed in parallel, and is a direct
16311 indicator of perceived performance due to noisy neighbours. In order to keep
16312 the value low, it is possible to reduce the scheduler's run queue depth using
16313 "tune.runqueue-depth", to reduce the number of concurrent events processed at
16314 once using "tune.maxpollevents", to decrease the stream's nice value using
Willy Tarreaue7723bd2020-06-24 11:11:02 +020016315 the "nice" option on the "bind" lines or in the frontend, to enable low
16316 latency scheduling using "tune.sched.low-latency", or to look for other heavy
16317 requests in logs (those exhibiting large values of "cpu_ns_avg"), whose
16318 processing needs to be adjusted or fixed. Compression of large buffers could
16319 be a culprit, like heavy regex or long lists of regex. Note: while it
Willy Tarreau70fe9442018-11-22 16:07:39 +010016320 may intuitively seem that the total latency adds to a transfer time, it is
16321 almost never true because while a task waits for the CPU, network buffers
16322 continue to fill up and the next call will process more at once. The value
16323 may be artificially high due to a high cpu_calls count, for example when
16324 processing many HTTP chunks, and for this reason it is often preferred to log
16325 lat_ns_avg instead, which is a more relevant performance indicator.
16326
Thierry FOURNIERcc103292015-06-06 19:30:17 +020016327meth(<method>) : method
16328 Returns a method.
16329
Willy Tarreau0f30d262014-11-24 16:02:05 +010016330nbproc : integer
16331 Returns an integer value corresponding to the number of processes that were
16332 started (it equals the global "nbproc" setting). This is useful for logging
16333 and debugging purposes.
16334
Willy Tarreau74ca5042013-06-11 23:12:07 +020016335nbsrv([<backend>]) : integer
16336 Returns an integer value corresponding to the number of usable servers of
16337 either the current backend or the named backend. This is mostly used with
16338 ACLs but can also be useful when added to logs. This is normally used to
Willy Tarreaud63335a2010-02-26 12:56:52 +010016339 switch to an alternate backend when the number of servers is too low to
16340 to handle some load. It is useful to report a failure when combined with
16341 "monitor fail".
Willy Tarreau079ff0a2009-03-05 21:34:28 +010016342
Patrick Hemmerfabb24f2018-08-13 14:07:57 -040016343prio_class : integer
16344 Returns the priority class of the current session for http mode or connection
16345 for tcp mode. The value will be that set by the last call to "http-request
16346 set-priority-class" or "tcp-request content set-priority-class".
16347
16348prio_offset : integer
16349 Returns the priority offset of the current session for http mode or
16350 connection for tcp mode. The value will be that set by the last call to
16351 "http-request set-priority-offset" or "tcp-request content
16352 set-priority-offset".
16353
Willy Tarreau0f30d262014-11-24 16:02:05 +010016354proc : integer
16355 Returns an integer value corresponding to the position of the process calling
16356 the function, between 1 and global.nbproc. This is useful for logging and
16357 debugging purposes.
16358
Willy Tarreau74ca5042013-06-11 23:12:07 +020016359queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010016360 Returns the total number of queued connections of the designated backend,
16361 including all the connections in server queues. If no backend name is
16362 specified, the current one is used, but it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020016363 one. This is useful with ACLs or to pass statistics to backend servers. This
16364 can be used to take actions when queuing goes above a known level, generally
16365 indicating a surge of traffic or a massive slowdown on the servers. One
16366 possible action could be to reject new users but still accept old ones. See
16367 also the "avg_queue", "be_conn", and "be_sess_rate" fetches.
16368
Willy Tarreau84310e22014-02-14 11:59:04 +010016369rand([<range>]) : integer
16370 Returns a random integer value within a range of <range> possible values,
16371 starting at zero. If the range is not specified, it defaults to 2^32, which
16372 gives numbers between 0 and 4294967295. It can be useful to pass some values
16373 needed to take some routing decisions for example, or just for debugging
16374 purposes. This random must not be used for security purposes.
16375
Luca Schimweg8a694b82019-09-10 15:42:52 +020016376uuid([<version>]) : string
16377 Returns a UUID following the RFC4122 standard. If the version is not
16378 specified, a UUID version 4 (fully random) is returned.
16379 Currently, only version 4 is supported.
16380
Willy Tarreau74ca5042013-06-11 23:12:07 +020016381srv_conn([<backend>/]<server>) : integer
16382 Returns an integer value corresponding to the number of currently established
16383 connections on the designated server, possibly including the connection being
16384 evaluated. If <backend> is omitted, then the server is looked up in the
16385 current backend. It can be used to use a specific farm when one server is
16386 full, or to inform the server about our view of the number of active
Patrick Hemmer155e93e2018-06-14 18:01:35 -040016387 connections with it. See also the "fe_conn", "be_conn", "queue", and
16388 "srv_conn_free" fetch methods.
16389
16390srv_conn_free([<backend>/]<server>) : integer
16391 Returns an integer value corresponding to the number of available connections
16392 on the designated server, possibly including the connection being evaluated.
16393 The value does not include queue slots. If <backend> is omitted, then the
16394 server is looked up in the current backend. It can be used to use a specific
16395 farm when one server is full, or to inform the server about our view of the
16396 number of active connections with it. See also the "be_conn_free" and
16397 "srv_conn" fetch methods.
16398
16399 OTHER CAVEATS AND NOTES: If the server maxconn is 0, then this fetch clearly
16400 does not make sense, in which case the value returned will be -1.
Willy Tarreau74ca5042013-06-11 23:12:07 +020016401
16402srv_is_up([<backend>/]<server>) : boolean
16403 Returns true when the designated server is UP, and false when it is either
16404 DOWN or in maintenance mode. If <backend> is omitted, then the server is
16405 looked up in the current backend. It is mainly used to take action based on
Davor Ocelice9ed2812017-12-25 17:49:28 +010016406 an external status reported via a health check (e.g. a geographical site's
Willy Tarreau74ca5042013-06-11 23:12:07 +020016407 availability). Another possible use which is more of a hack consists in
16408 using dummy servers as boolean variables that can be enabled or disabled from
16409 the CLI, so that rules depending on those ACLs can be tweaked in realtime.
16410
Willy Tarreauff2b7af2017-10-13 11:46:26 +020016411srv_queue([<backend>/]<server>) : integer
16412 Returns an integer value corresponding to the number of connections currently
16413 pending in the designated server's queue. If <backend> is omitted, then the
16414 server is looked up in the current backend. It can sometimes be used together
16415 with the "use-server" directive to force to use a known faster server when it
16416 is not much loaded. See also the "srv_conn", "avg_queue" and "queue" sample
16417 fetch methods.
16418
Willy Tarreau74ca5042013-06-11 23:12:07 +020016419srv_sess_rate([<backend>/]<server>) : integer
16420 Returns an integer corresponding to the sessions creation rate on the
16421 designated server, in number of new sessions per second. If <backend> is
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016422 omitted, then the server is looked up in the current backend. This is mostly
Willy Tarreau74ca5042013-06-11 23:12:07 +020016423 used with ACLs but can make sense with logs too. This is used to switch to an
16424 alternate backend when an expensive or fragile one reaches too high a session
Davor Ocelice9ed2812017-12-25 17:49:28 +010016425 rate, or to limit abuse of service (e.g. prevent latent requests from
Willy Tarreau74ca5042013-06-11 23:12:07 +020016426 overloading servers).
16427
16428 Example :
16429 # Redirect to a separate back
16430 acl srv1_full srv_sess_rate(be1/srv1) gt 50
16431 acl srv2_full srv_sess_rate(be1/srv2) gt 50
16432 use_backend be2 if srv1_full or srv2_full
16433
Christopher Faulet1bea8652020-07-10 16:03:45 +020016434srv_iweight([<backend>/]<server>): integer
16435 Returns an integer corresponding to the server's initial weight. If <backend>
16436 is omitted, then the server is looked up in the current backend. See also
16437 "srv_weight" and "srv_uweight".
16438
16439srv_uweight([<backend>/]<server>): integer
16440 Returns an integer corresponding to the user visible server's weight. If
16441 <backend> is omitted, then the server is looked up in the current
16442 backend. See also "srv_weight" and "srv_iweight".
16443
16444srv_weight([<backend>/]<server>): integer
16445 Returns an integer corresponding to the current (or effective) server's
16446 weight. If <backend> is omitted, then the server is looked up in the current
16447 backend. See also "srv_iweight" and "srv_uweight".
16448
Willy Tarreau0f30d262014-11-24 16:02:05 +010016449stopping : boolean
16450 Returns TRUE if the process calling the function is currently stopping. This
16451 can be useful for logging, or for relaxing certain checks or helping close
16452 certain connections upon graceful shutdown.
16453
Thierry FOURNIERcc103292015-06-06 19:30:17 +020016454str(<string>) : string
16455 Returns a string.
16456
Willy Tarreau74ca5042013-06-11 23:12:07 +020016457table_avl([<table>]) : integer
16458 Returns the total number of available entries in the current proxy's
16459 stick-table or in the designated stick-table. See also table_cnt.
16460
16461table_cnt([<table>]) : integer
16462 Returns the total number of entries currently in use in the current proxy's
16463 stick-table or in the designated stick-table. See also src_conn_cnt and
16464 table_avl for other entry counting methods.
16465
Christopher Faulet34adb2a2017-11-21 21:45:38 +010016466thread : integer
16467 Returns an integer value corresponding to the position of the thread calling
16468 the function, between 0 and (global.nbthread-1). This is useful for logging
16469 and debugging purposes.
16470
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020016471var(<var-name>) : undefined
16472 Returns a variable with the stored type. If the variable is not set, the
Daniel Schneller0b547052016-03-21 20:46:57 +010016473 sample fetch fails. The name of the variable starts with an indication
16474 about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010016475 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010016476 "sess" : the variable is shared with the whole session
16477 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020016478 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010016479 "req" : the variable is shared only during request processing,
16480 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020016481 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010016482 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020016483
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200164847.3.3. Fetching samples at Layer 4
Willy Tarreau74ca5042013-06-11 23:12:07 +020016485----------------------------------
16486
16487The layer 4 usually describes just the transport layer which in haproxy is
16488closest to the connection, where no content is yet made available. The fetch
16489methods described here are usable as low as the "tcp-request connection" rule
16490sets unless they require some future information. Those generally include
16491TCP/IP addresses and ports, as well as elements from stick-tables related to
Willy Tarreau4d4149c2013-07-23 19:33:46 +020016492the incoming connection. For retrieving a value from a sticky counters, the
16493counter number can be explicitly set as 0, 1, or 2 using the pre-defined
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020016494"sc0_", "sc1_", or "sc2_" prefix. These three pre-defined prefixes can only be
16495used if MAX_SESS_STKCTR value does not exceed 3, otherwise the counter number
16496can be specified as the first integer argument when using the "sc_" prefix.
16497Starting from "sc_0" to "sc_N" where N is (MAX_SESS_STKCTR-1). An optional
16498table may be specified with the "sc*" form, in which case the currently
16499tracked key will be looked up into this alternate table instead of the table
16500currently being tracked.
Willy Tarreau74ca5042013-06-11 23:12:07 +020016501
Jérôme Magnin35e53a62019-01-16 14:38:37 +010016502bc_http_major : integer
Jérôme Magnin86577422018-12-07 09:03:11 +010016503 Returns the backend connection's HTTP major version encoding, which may be 1
16504 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
16505 encoding and not the version present in the request header.
16506
Willy Tarreau74ca5042013-06-11 23:12:07 +020016507be_id : integer
16508 Returns an integer containing the current backend's id. It can be used in
Christopher Fauletd1b44642020-04-30 09:51:15 +020016509 frontends with responses to check which backend processed the request. It can
16510 also be used in a tcp-check or an http-check ruleset.
Willy Tarreau74ca5042013-06-11 23:12:07 +020016511
Marcin Deranekd2471c22016-12-12 14:08:05 +010016512be_name : string
16513 Returns a string containing the current backend's name. It can be used in
Christopher Fauletd1b44642020-04-30 09:51:15 +020016514 frontends with responses to check which backend processed the request. It can
16515 also be used in a tcp-check or an http-check ruleset.
Marcin Deranekd2471c22016-12-12 14:08:05 +010016516
Willy Tarreau74ca5042013-06-11 23:12:07 +020016517dst : ip
16518 This is the destination IPv4 address of the connection on the client side,
16519 which is the address the client connected to. It can be useful when running
16520 in transparent mode. It is of type IP and works on both IPv4 and IPv6 tables.
16521 On IPv6 tables, IPv4 address is mapped to its IPv6 equivalent, according to
Willy Tarreau64ded3d2019-01-23 10:02:15 +010016522 RFC 4291. When the incoming connection passed through address translation or
16523 redirection involving connection tracking, the original destination address
16524 before the redirection will be reported. On Linux systems, the source and
16525 destination may seldom appear reversed if the nf_conntrack_tcp_loose sysctl
16526 is set, because a late response may reopen a timed out connection and switch
16527 what is believed to be the source and the destination.
Willy Tarreau74ca5042013-06-11 23:12:07 +020016528
16529dst_conn : integer
16530 Returns an integer value corresponding to the number of currently established
16531 connections on the same socket including the one being evaluated. It is
16532 normally used with ACLs but can as well be used to pass the information to
16533 servers in an HTTP header or in logs. It can be used to either return a sorry
16534 page before hard-blocking, or to use a specific backend to drain new requests
16535 when the socket is considered saturated. This offers the ability to assign
16536 different limits to different listening ports or addresses. See also the
16537 "fe_conn" and "be_conn" fetches.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016538
Willy Tarreau16e01562016-08-09 16:46:18 +020016539dst_is_local : boolean
16540 Returns true if the destination address of the incoming connection is local
16541 to the system, or false if the address doesn't exist on the system, meaning
16542 that it was intercepted in transparent mode. It can be useful to apply
16543 certain rules by default to forwarded traffic and other rules to the traffic
Davor Ocelice9ed2812017-12-25 17:49:28 +010016544 targeting the real address of the machine. For example the stats page could
Willy Tarreau16e01562016-08-09 16:46:18 +020016545 be delivered only on this address, or SSH access could be locally redirected.
16546 Please note that the check involves a few system calls, so it's better to do
16547 it only once per connection.
16548
Willy Tarreau74ca5042013-06-11 23:12:07 +020016549dst_port : integer
16550 Returns an integer value corresponding to the destination TCP port of the
16551 connection on the client side, which is the port the client connected to.
16552 This might be used when running in transparent mode, when assigning dynamic
16553 ports to some clients for a whole application session, to stick all users to
16554 a same server, or to pass the destination port information to a server using
16555 an HTTP header.
16556
Willy Tarreau60ca10a2017-08-18 15:26:54 +020016557fc_http_major : integer
16558 Reports the front connection's HTTP major version encoding, which may be 1
16559 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
16560 encoding and not on the version present in the request header.
16561
Geoff Simmons7185b782019-08-27 18:31:16 +020016562fc_pp_authority : string
16563 Returns the authority TLV sent by the client in the PROXY protocol header,
16564 if any.
16565
Tim Duesterhusd1b15b62020-03-13 12:34:23 +010016566fc_pp_unique_id : string
16567 Returns the unique ID TLV sent by the client in the PROXY protocol header,
16568 if any.
16569
Emeric Brun4f603012017-01-05 15:11:44 +010016570fc_rcvd_proxy : boolean
16571 Returns true if the client initiated the connection with a PROXY protocol
16572 header.
16573
Thierry Fournier / OZON.IO6310bef2016-07-24 20:16:50 +020016574fc_rtt(<unit>) : integer
16575 Returns the Round Trip Time (RTT) measured by the kernel for the client
16576 connection. <unit> is facultative, by default the unit is milliseconds. <unit>
16577 can be set to "ms" for milliseconds or "us" for microseconds. If the server
16578 connection is not established, if the connection is not TCP or if the
16579 operating system does not support TCP_INFO, for example Linux kernels before
16580 2.4, the sample fetch fails.
16581
16582fc_rttvar(<unit>) : integer
16583 Returns the Round Trip Time (RTT) variance measured by the kernel for the
16584 client connection. <unit> is facultative, by default the unit is milliseconds.
16585 <unit> can be set to "ms" for milliseconds or "us" for microseconds. If the
16586 server connection is not established, if the connection is not TCP or if the
16587 operating system does not support TCP_INFO, for example Linux kernels before
16588 2.4, the sample fetch fails.
16589
Christopher Fauletba0c53e2019-10-17 14:40:48 +020016590fc_unacked : integer
Joe Williams30fcd392016-08-10 07:06:44 -070016591 Returns the unacked counter measured by the kernel for the client connection.
16592 If the server connection is not established, if the connection is not TCP or
16593 if the operating system does not support TCP_INFO, for example Linux kernels
16594 before 2.4, the sample fetch fails.
16595
Christopher Fauletba0c53e2019-10-17 14:40:48 +020016596fc_sacked : integer
Joe Williams30fcd392016-08-10 07:06:44 -070016597 Returns the sacked counter measured by the kernel for the client connection.
16598 If the server connection is not established, if the connection is not TCP or
16599 if the operating system does not support TCP_INFO, for example Linux kernels
16600 before 2.4, the sample fetch fails.
16601
Christopher Fauletba0c53e2019-10-17 14:40:48 +020016602fc_retrans : integer
Joe Williams30fcd392016-08-10 07:06:44 -070016603 Returns the retransmits counter measured by the kernel for the client
16604 connection. If the server connection is not established, if the connection is
16605 not TCP or if the operating system does not support TCP_INFO, for example
16606 Linux kernels before 2.4, the sample fetch fails.
16607
Christopher Fauletba0c53e2019-10-17 14:40:48 +020016608fc_fackets : integer
Joe Williams30fcd392016-08-10 07:06:44 -070016609 Returns the fack counter measured by the kernel for the client
16610 connection. If the server connection is not established, if the connection is
16611 not TCP or if the operating system does not support TCP_INFO, for example
16612 Linux kernels before 2.4, the sample fetch fails.
16613
Christopher Fauletba0c53e2019-10-17 14:40:48 +020016614fc_lost : integer
Joe Williams30fcd392016-08-10 07:06:44 -070016615 Returns the lost counter measured by the kernel for the client
16616 connection. If the server connection is not established, if the connection is
16617 not TCP or if the operating system does not support TCP_INFO, for example
16618 Linux kernels before 2.4, the sample fetch fails.
16619
Christopher Fauletba0c53e2019-10-17 14:40:48 +020016620fc_reordering : integer
Joe Williams30fcd392016-08-10 07:06:44 -070016621 Returns the reordering counter measured by the kernel for the client
16622 connection. If the server connection is not established, if the connection is
16623 not TCP or if the operating system does not support TCP_INFO, for example
16624 Linux kernels before 2.4, the sample fetch fails.
16625
Marcin Deranek9a66dfb2018-04-13 14:37:50 +020016626fe_defbe : string
16627 Returns a string containing the frontend's default backend name. It can be
16628 used in frontends to check which backend will handle requests by default.
16629
Willy Tarreau74ca5042013-06-11 23:12:07 +020016630fe_id : integer
16631 Returns an integer containing the current frontend's id. It can be used in
Marcin Deranek6e413ed2016-12-13 12:40:01 +010016632 backends to check from which frontend it was called, or to stick all users
Willy Tarreau74ca5042013-06-11 23:12:07 +020016633 coming via a same frontend to the same server.
16634
Marcin Deranekd2471c22016-12-12 14:08:05 +010016635fe_name : string
16636 Returns a string containing the current frontend's name. It can be used in
16637 backends to check from which frontend it was called, or to stick all users
16638 coming via a same frontend to the same server.
16639
Cyril Bonté62ba8702014-04-22 23:52:25 +020016640sc_bytes_in_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016641sc0_bytes_in_rate([<table>]) : integer
16642sc1_bytes_in_rate([<table>]) : integer
16643sc2_bytes_in_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020016644 Returns the average client-to-server bytes rate from the currently tracked
16645 counters, measured in amount of bytes over the period configured in the
16646 table. See also src_bytes_in_rate.
16647
Cyril Bonté62ba8702014-04-22 23:52:25 +020016648sc_bytes_out_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016649sc0_bytes_out_rate([<table>]) : integer
16650sc1_bytes_out_rate([<table>]) : integer
16651sc2_bytes_out_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020016652 Returns the average server-to-client bytes rate from the currently tracked
16653 counters, measured in amount of bytes over the period configured in the
16654 table. See also src_bytes_out_rate.
16655
Cyril Bonté62ba8702014-04-22 23:52:25 +020016656sc_clr_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016657sc0_clr_gpc0([<table>]) : integer
16658sc1_clr_gpc0([<table>]) : integer
16659sc2_clr_gpc0([<table>]) : integer
Willy Tarreauf73cd112011-08-13 01:45:16 +020016660 Clears the first General Purpose Counter associated to the currently tracked
16661 counters, and returns its previous value. Before the first invocation, the
Willy Tarreau869948b2013-01-04 14:14:57 +010016662 stored value is zero, so first invocation will always return zero. This is
16663 typically used as a second ACL in an expression in order to mark a connection
16664 when a first ACL was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020016665
Jarno Huuskonen676f6222017-03-30 09:19:45 +030016666 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020016667 # block if 5 consecutive requests continue to come faster than 10 sess
16668 # per second, and reset the counter as soon as the traffic slows down.
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020016669 acl abuse sc0_http_req_rate gt 10
16670 acl kill sc0_inc_gpc0 gt 5
16671 acl save sc0_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020016672 tcp-request connection accept if !abuse save
16673 tcp-request connection reject if abuse kill
16674
Frédéric Lécaille6778b272018-01-29 15:22:53 +010016675sc_clr_gpc1(<ctr>[,<table>]) : integer
16676sc0_clr_gpc1([<table>]) : integer
16677sc1_clr_gpc1([<table>]) : integer
16678sc2_clr_gpc1([<table>]) : integer
16679 Clears the second General Purpose Counter associated to the currently tracked
16680 counters, and returns its previous value. Before the first invocation, the
16681 stored value is zero, so first invocation will always return zero. This is
16682 typically used as a second ACL in an expression in order to mark a connection
16683 when a first ACL was verified.
16684
Cyril Bonté62ba8702014-04-22 23:52:25 +020016685sc_conn_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016686sc0_conn_cnt([<table>]) : integer
16687sc1_conn_cnt([<table>]) : integer
16688sc2_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010016689 Returns the cumulative number of incoming connections from currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020016690 counters. See also src_conn_cnt.
16691
Cyril Bonté62ba8702014-04-22 23:52:25 +020016692sc_conn_cur(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016693sc0_conn_cur([<table>]) : integer
16694sc1_conn_cur([<table>]) : integer
16695sc2_conn_cur([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020016696 Returns the current amount of concurrent connections tracking the same
16697 tracked counters. This number is automatically incremented when tracking
16698 begins and decremented when tracking stops. See also src_conn_cur.
16699
Cyril Bonté62ba8702014-04-22 23:52:25 +020016700sc_conn_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016701sc0_conn_rate([<table>]) : integer
16702sc1_conn_rate([<table>]) : integer
16703sc2_conn_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020016704 Returns the average connection rate from the currently tracked counters,
16705 measured in amount of connections over the period configured in the table.
16706 See also src_conn_rate.
16707
Cyril Bonté62ba8702014-04-22 23:52:25 +020016708sc_get_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016709sc0_get_gpc0([<table>]) : integer
16710sc1_get_gpc0([<table>]) : integer
16711sc2_get_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020016712 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020016713 currently tracked counters. See also src_get_gpc0 and sc/sc0/sc1/sc2_inc_gpc0.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020016714
Frédéric Lécaille6778b272018-01-29 15:22:53 +010016715sc_get_gpc1(<ctr>[,<table>]) : integer
16716sc0_get_gpc1([<table>]) : integer
16717sc1_get_gpc1([<table>]) : integer
16718sc2_get_gpc1([<table>]) : integer
16719 Returns the value of the second General Purpose Counter associated to the
16720 currently tracked counters. See also src_get_gpc1 and sc/sc0/sc1/sc2_inc_gpc1.
16721
Thierry FOURNIER236657b2015-08-19 08:25:14 +020016722sc_get_gpt0(<ctr>[,<table>]) : integer
16723sc0_get_gpt0([<table>]) : integer
16724sc1_get_gpt0([<table>]) : integer
16725sc2_get_gpt0([<table>]) : integer
16726 Returns the value of the first General Purpose Tag associated to the
16727 currently tracked counters. See also src_get_gpt0.
16728
Cyril Bonté62ba8702014-04-22 23:52:25 +020016729sc_gpc0_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016730sc0_gpc0_rate([<table>]) : integer
16731sc1_gpc0_rate([<table>]) : integer
16732sc2_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020016733 Returns the average increment rate of the first General Purpose Counter
16734 associated to the currently tracked counters. It reports the frequency
16735 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020016736 src_gpc0_rate, sc/sc0/sc1/sc2_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
16737 that the "gpc0_rate" counter must be stored in the stick-table for a value to
16738 be returned, as "gpc0" only holds the event count.
Willy Tarreaue9656522010-08-17 15:40:09 +020016739
Frédéric Lécaille6778b272018-01-29 15:22:53 +010016740sc_gpc1_rate(<ctr>[,<table>]) : integer
16741sc0_gpc1_rate([<table>]) : integer
16742sc1_gpc1_rate([<table>]) : integer
16743sc2_gpc1_rate([<table>]) : integer
16744 Returns the average increment rate of the second General Purpose Counter
16745 associated to the currently tracked counters. It reports the frequency
16746 which the gpc1 counter was incremented over the configured period. See also
16747 src_gpcA_rate, sc/sc0/sc1/sc2_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
16748 that the "gpc1_rate" counter must be stored in the stick-table for a value to
16749 be returned, as "gpc1" only holds the event count.
16750
Cyril Bonté62ba8702014-04-22 23:52:25 +020016751sc_http_err_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016752sc0_http_err_cnt([<table>]) : integer
16753sc1_http_err_cnt([<table>]) : integer
16754sc2_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010016755 Returns the cumulative number of HTTP errors from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020016756 counters. This includes the both request errors and 4xx error responses.
16757 See also src_http_err_cnt.
16758
Cyril Bonté62ba8702014-04-22 23:52:25 +020016759sc_http_err_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016760sc0_http_err_rate([<table>]) : integer
16761sc1_http_err_rate([<table>]) : integer
16762sc2_http_err_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020016763 Returns the average rate of HTTP errors from the currently tracked counters,
16764 measured in amount of errors over the period configured in the table. This
16765 includes the both request errors and 4xx error responses. See also
16766 src_http_err_rate.
16767
Cyril Bonté62ba8702014-04-22 23:52:25 +020016768sc_http_req_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016769sc0_http_req_cnt([<table>]) : integer
16770sc1_http_req_cnt([<table>]) : integer
16771sc2_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010016772 Returns the cumulative number of HTTP requests from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020016773 counters. This includes every started request, valid or not. See also
16774 src_http_req_cnt.
16775
Cyril Bonté62ba8702014-04-22 23:52:25 +020016776sc_http_req_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016777sc0_http_req_rate([<table>]) : integer
16778sc1_http_req_rate([<table>]) : integer
16779sc2_http_req_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020016780 Returns the average rate of HTTP requests from the currently tracked
16781 counters, measured in amount of requests over the period configured in
16782 the table. This includes every started request, valid or not. See also
16783 src_http_req_rate.
16784
Cyril Bonté62ba8702014-04-22 23:52:25 +020016785sc_inc_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016786sc0_inc_gpc0([<table>]) : integer
16787sc1_inc_gpc0([<table>]) : integer
16788sc2_inc_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020016789 Increments the first General Purpose Counter associated to the currently
Willy Tarreau869948b2013-01-04 14:14:57 +010016790 tracked counters, and returns its new value. Before the first invocation,
16791 the stored value is zero, so first invocation will increase it to 1 and will
16792 return 1. This is typically used as a second ACL in an expression in order
16793 to mark a connection when a first ACL was verified :
Willy Tarreaue9656522010-08-17 15:40:09 +020016794
Jarno Huuskonen676f6222017-03-30 09:19:45 +030016795 Example:
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020016796 acl abuse sc0_http_req_rate gt 10
16797 acl kill sc0_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020016798 tcp-request connection reject if abuse kill
16799
Frédéric Lécaille6778b272018-01-29 15:22:53 +010016800sc_inc_gpc1(<ctr>[,<table>]) : integer
16801sc0_inc_gpc1([<table>]) : integer
16802sc1_inc_gpc1([<table>]) : integer
16803sc2_inc_gpc1([<table>]) : integer
16804 Increments the second General Purpose Counter associated to the currently
16805 tracked counters, and returns its new value. Before the first invocation,
16806 the stored value is zero, so first invocation will increase it to 1 and will
16807 return 1. This is typically used as a second ACL in an expression in order
16808 to mark a connection when a first ACL was verified.
16809
Cyril Bonté62ba8702014-04-22 23:52:25 +020016810sc_kbytes_in(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016811sc0_kbytes_in([<table>]) : integer
16812sc1_kbytes_in([<table>]) : integer
16813sc2_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020016814 Returns the total amount of client-to-server data from the currently tracked
16815 counters, measured in kilobytes. The test is currently performed on 32-bit
16816 integers, which limits values to 4 terabytes. See also src_kbytes_in.
Willy Tarreaue9656522010-08-17 15:40:09 +020016817
Cyril Bonté62ba8702014-04-22 23:52:25 +020016818sc_kbytes_out(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016819sc0_kbytes_out([<table>]) : integer
16820sc1_kbytes_out([<table>]) : integer
16821sc2_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020016822 Returns the total amount of server-to-client data from the currently tracked
16823 counters, measured in kilobytes. The test is currently performed on 32-bit
16824 integers, which limits values to 4 terabytes. See also src_kbytes_out.
Willy Tarreaue9656522010-08-17 15:40:09 +020016825
Cyril Bonté62ba8702014-04-22 23:52:25 +020016826sc_sess_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016827sc0_sess_cnt([<table>]) : integer
16828sc1_sess_cnt([<table>]) : integer
16829sc2_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010016830 Returns the cumulative number of incoming connections that were transformed
Willy Tarreaue9656522010-08-17 15:40:09 +020016831 into sessions, which means that they were accepted by a "tcp-request
16832 connection" rule, from the currently tracked counters. A backend may count
16833 more sessions than connections because each connection could result in many
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016834 backend sessions if some HTTP keep-alive is performed over the connection
Willy Tarreaue9656522010-08-17 15:40:09 +020016835 with the client. See also src_sess_cnt.
16836
Cyril Bonté62ba8702014-04-22 23:52:25 +020016837sc_sess_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016838sc0_sess_rate([<table>]) : integer
16839sc1_sess_rate([<table>]) : integer
16840sc2_sess_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020016841 Returns the average session rate from the currently tracked counters,
16842 measured in amount of sessions over the period configured in the table. A
16843 session is a connection that got past the early "tcp-request connection"
16844 rules. A backend may count more sessions than connections because each
16845 connection could result in many backend sessions if some HTTP keep-alive is
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016846 performed over the connection with the client. See also src_sess_rate.
Willy Tarreaue9656522010-08-17 15:40:09 +020016847
Cyril Bonté62ba8702014-04-22 23:52:25 +020016848sc_tracked(<ctr>[,<table>]) : boolean
Willy Tarreau0f791d42013-07-23 19:56:43 +020016849sc0_tracked([<table>]) : boolean
16850sc1_tracked([<table>]) : boolean
16851sc2_tracked([<table>]) : boolean
Willy Tarreau6f1615f2013-06-03 15:15:22 +020016852 Returns true if the designated session counter is currently being tracked by
16853 the current session. This can be useful when deciding whether or not we want
16854 to set some values in a header passed to the server.
16855
Cyril Bonté62ba8702014-04-22 23:52:25 +020016856sc_trackers(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016857sc0_trackers([<table>]) : integer
16858sc1_trackers([<table>]) : integer
16859sc2_trackers([<table>]) : integer
Willy Tarreau2406db42012-12-09 12:16:43 +010016860 Returns the current amount of concurrent connections tracking the same
16861 tracked counters. This number is automatically incremented when tracking
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020016862 begins and decremented when tracking stops. It differs from sc0_conn_cur in
Willy Tarreau2406db42012-12-09 12:16:43 +010016863 that it does not rely on any stored information but on the table's reference
16864 count (the "use" value which is returned by "show table" on the CLI). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020016865 may sometimes be more suited for layer7 tracking. It can be used to tell a
16866 server how many concurrent connections there are from a given address for
16867 example.
Willy Tarreau2406db42012-12-09 12:16:43 +010016868
Willy Tarreau74ca5042013-06-11 23:12:07 +020016869so_id : integer
16870 Returns an integer containing the current listening socket's id. It is useful
16871 in frontends involving many "bind" lines, or to stick all users coming via a
16872 same socket to the same server.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016873
Jerome Magnineb421b22020-03-27 22:08:40 +010016874so_name : string
16875 Returns a string containing the current listening socket's name, as defined
16876 with name on a "bind" line. It can serve the same purposes as so_id but with
16877 strings instead of integers.
16878
Willy Tarreau74ca5042013-06-11 23:12:07 +020016879src : ip
Davor Ocelice9ed2812017-12-25 17:49:28 +010016880 This is the source IPv4 address of the client of the session. It is of type
Willy Tarreau74ca5042013-06-11 23:12:07 +020016881 IP and works on both IPv4 and IPv6 tables. On IPv6 tables, IPv4 addresses are
16882 mapped to their IPv6 equivalent, according to RFC 4291. Note that it is the
16883 TCP-level source address which is used, and not the address of a client
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016884 behind a proxy. However if the "accept-proxy" or "accept-netscaler-cip" bind
16885 directive is used, it can be the address of a client behind another
16886 PROXY-protocol compatible component for all rule sets except
Willy Tarreau64ded3d2019-01-23 10:02:15 +010016887 "tcp-request connection" which sees the real address. When the incoming
16888 connection passed through address translation or redirection involving
16889 connection tracking, the original destination address before the redirection
16890 will be reported. On Linux systems, the source and destination may seldom
16891 appear reversed if the nf_conntrack_tcp_loose sysctl is set, because a late
16892 response may reopen a timed out connection and switch what is believed to be
16893 the source and the destination.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016894
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010016895 Example:
16896 # add an HTTP header in requests with the originating address' country
16897 http-request set-header X-Country %[src,map_ip(geoip.lst)]
16898
Willy Tarreau74ca5042013-06-11 23:12:07 +020016899src_bytes_in_rate([<table>]) : integer
16900 Returns the average bytes rate from the incoming connection's source address
16901 in the current proxy's stick-table or in the designated stick-table, measured
16902 in amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020016903 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_in_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020016904
Willy Tarreau74ca5042013-06-11 23:12:07 +020016905src_bytes_out_rate([<table>]) : integer
16906 Returns the average bytes rate to the incoming connection's source address in
16907 the current proxy's stick-table or in the designated stick-table, measured in
Willy Tarreauc9705a12010-07-27 20:05:50 +020016908 amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020016909 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_out_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020016910
Willy Tarreau74ca5042013-06-11 23:12:07 +020016911src_clr_gpc0([<table>]) : integer
16912 Clears the first General Purpose Counter associated to the incoming
16913 connection's source address in the current proxy's stick-table or in the
16914 designated stick-table, and returns its previous value. If the address is not
16915 found, an entry is created and 0 is returned. This is typically used as a
16916 second ACL in an expression in order to mark a connection when a first ACL
16917 was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020016918
Jarno Huuskonen676f6222017-03-30 09:19:45 +030016919 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020016920 # block if 5 consecutive requests continue to come faster than 10 sess
16921 # per second, and reset the counter as soon as the traffic slows down.
16922 acl abuse src_http_req_rate gt 10
16923 acl kill src_inc_gpc0 gt 5
Willy Tarreau869948b2013-01-04 14:14:57 +010016924 acl save src_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020016925 tcp-request connection accept if !abuse save
16926 tcp-request connection reject if abuse kill
16927
Frédéric Lécaille6778b272018-01-29 15:22:53 +010016928src_clr_gpc1([<table>]) : integer
16929 Clears the second General Purpose Counter associated to the incoming
16930 connection's source address in the current proxy's stick-table or in the
16931 designated stick-table, and returns its previous value. If the address is not
16932 found, an entry is created and 0 is returned. This is typically used as a
16933 second ACL in an expression in order to mark a connection when a first ACL
16934 was verified.
16935
Willy Tarreau74ca5042013-06-11 23:12:07 +020016936src_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010016937 Returns the cumulative number of connections initiated from the current
Willy Tarreau74ca5042013-06-11 23:12:07 +020016938 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020016939 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020016940 See also sc/sc0/sc1/sc2_conn_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020016941
Willy Tarreau74ca5042013-06-11 23:12:07 +020016942src_conn_cur([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020016943 Returns the current amount of concurrent connections initiated from the
Willy Tarreau74ca5042013-06-11 23:12:07 +020016944 current incoming connection's source address in the current proxy's
16945 stick-table or in the designated stick-table. If the address is not found,
Willy Tarreau4d4149c2013-07-23 19:33:46 +020016946 zero is returned. See also sc/sc0/sc1/sc2_conn_cur.
Willy Tarreauc9705a12010-07-27 20:05:50 +020016947
Willy Tarreau74ca5042013-06-11 23:12:07 +020016948src_conn_rate([<table>]) : integer
16949 Returns the average connection rate from the incoming connection's source
16950 address in the current proxy's stick-table or in the designated stick-table,
16951 measured in amount of connections over the period configured in the table. If
Willy Tarreau4d4149c2013-07-23 19:33:46 +020016952 the address is not found, zero is returned. See also sc/sc0/sc1/sc2_conn_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020016953
Willy Tarreau74ca5042013-06-11 23:12:07 +020016954src_get_gpc0([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020016955 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau74ca5042013-06-11 23:12:07 +020016956 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020016957 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020016958 See also sc/sc0/sc1/sc2_get_gpc0 and src_inc_gpc0.
Willy Tarreauc9705a12010-07-27 20:05:50 +020016959
Frédéric Lécaille6778b272018-01-29 15:22:53 +010016960src_get_gpc1([<table>]) : integer
16961 Returns the value of the second General Purpose Counter associated to the
16962 incoming connection's source address in the current proxy's stick-table or in
16963 the designated stick-table. If the address is not found, zero is returned.
16964 See also sc/sc0/sc1/sc2_get_gpc1 and src_inc_gpc1.
16965
Thierry FOURNIER236657b2015-08-19 08:25:14 +020016966src_get_gpt0([<table>]) : integer
16967 Returns the value of the first General Purpose Tag associated to the
16968 incoming connection's source address in the current proxy's stick-table or in
16969 the designated stick-table. If the address is not found, zero is returned.
16970 See also sc/sc0/sc1/sc2_get_gpt0.
16971
Willy Tarreau74ca5042013-06-11 23:12:07 +020016972src_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020016973 Returns the average increment rate of the first General Purpose Counter
Willy Tarreau74ca5042013-06-11 23:12:07 +020016974 associated to the incoming connection's source address in the current proxy's
Willy Tarreauba2ffd12013-05-29 15:54:14 +020016975 stick-table or in the designated stick-table. It reports the frequency
16976 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020016977 sc/sc0/sc1/sc2_gpc0_rate, src_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
16978 that the "gpc0_rate" counter must be stored in the stick-table for a value to
16979 be returned, as "gpc0" only holds the event count.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020016980
Frédéric Lécaille6778b272018-01-29 15:22:53 +010016981src_gpc1_rate([<table>]) : integer
16982 Returns the average increment rate of the second General Purpose Counter
16983 associated to the incoming connection's source address in the current proxy's
16984 stick-table or in the designated stick-table. It reports the frequency
16985 which the gpc1 counter was incremented over the configured period. See also
16986 sc/sc0/sc1/sc2_gpc1_rate, src_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
16987 that the "gpc1_rate" counter must be stored in the stick-table for a value to
16988 be returned, as "gpc1" only holds the event count.
16989
Willy Tarreau74ca5042013-06-11 23:12:07 +020016990src_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010016991 Returns the cumulative number of HTTP errors from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020016992 source address in the current proxy's stick-table or in the designated
Willy Tarreauc9705a12010-07-27 20:05:50 +020016993 stick-table. This includes the both request errors and 4xx error responses.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020016994 See also sc/sc0/sc1/sc2_http_err_cnt. If the address is not found, zero is
Willy Tarreau74ca5042013-06-11 23:12:07 +020016995 returned.
Willy Tarreauc9705a12010-07-27 20:05:50 +020016996
Willy Tarreau74ca5042013-06-11 23:12:07 +020016997src_http_err_rate([<table>]) : integer
16998 Returns the average rate of HTTP errors from the incoming connection's source
16999 address in the current proxy's stick-table or in the designated stick-table,
17000 measured in amount of errors over the period configured in the table. This
17001 includes the both request errors and 4xx error responses. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017002 not found, zero is returned. See also sc/sc0/sc1/sc2_http_err_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017003
Willy Tarreau74ca5042013-06-11 23:12:07 +020017004src_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017005 Returns the cumulative number of HTTP requests from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020017006 source address in the current proxy's stick-table or in the designated stick-
17007 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017008 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017009
Willy Tarreau74ca5042013-06-11 23:12:07 +020017010src_http_req_rate([<table>]) : integer
17011 Returns the average rate of HTTP requests from the incoming connection's
17012 source address in the current proxy's stick-table or in the designated stick-
17013 table, measured in amount of requests over the period configured in the
Willy Tarreauc9705a12010-07-27 20:05:50 +020017014 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017015 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017016
Willy Tarreau74ca5042013-06-11 23:12:07 +020017017src_inc_gpc0([<table>]) : integer
17018 Increments the first General Purpose Counter associated to the incoming
17019 connection's source address in the current proxy's stick-table or in the
17020 designated stick-table, and returns its new value. If the address is not
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020017021 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc0.
Willy Tarreau74ca5042013-06-11 23:12:07 +020017022 This is typically used as a second ACL in an expression in order to mark a
17023 connection when a first ACL was verified :
Willy Tarreauc9705a12010-07-27 20:05:50 +020017024
Jarno Huuskonen676f6222017-03-30 09:19:45 +030017025 Example:
Willy Tarreauc9705a12010-07-27 20:05:50 +020017026 acl abuse src_http_req_rate gt 10
Willy Tarreau869948b2013-01-04 14:14:57 +010017027 acl kill src_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020017028 tcp-request connection reject if abuse kill
Willy Tarreauc9705a12010-07-27 20:05:50 +020017029
Frédéric Lécaille6778b272018-01-29 15:22:53 +010017030src_inc_gpc1([<table>]) : integer
17031 Increments the second General Purpose Counter associated to the incoming
17032 connection's source address in the current proxy's stick-table or in the
17033 designated stick-table, and returns its new value. If the address is not
17034 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc1.
17035 This is typically used as a second ACL in an expression in order to mark a
17036 connection when a first ACL was verified.
17037
Willy Tarreau16e01562016-08-09 16:46:18 +020017038src_is_local : boolean
17039 Returns true if the source address of the incoming connection is local to the
17040 system, or false if the address doesn't exist on the system, meaning that it
17041 comes from a remote machine. Note that UNIX addresses are considered local.
17042 It can be useful to apply certain access restrictions based on where the
Davor Ocelice9ed2812017-12-25 17:49:28 +010017043 client comes from (e.g. require auth or https for remote machines). Please
Willy Tarreau16e01562016-08-09 16:46:18 +020017044 note that the check involves a few system calls, so it's better to do it only
17045 once per connection.
17046
Willy Tarreau74ca5042013-06-11 23:12:07 +020017047src_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020017048 Returns the total amount of data received from the incoming connection's
17049 source address in the current proxy's stick-table or in the designated
17050 stick-table, measured in kilobytes. If the address is not found, zero is
17051 returned. The test is currently performed on 32-bit integers, which limits
17052 values to 4 terabytes. See also sc/sc0/sc1/sc2_kbytes_in.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017053
Willy Tarreau74ca5042013-06-11 23:12:07 +020017054src_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020017055 Returns the total amount of data sent to the incoming connection's source
17056 address in the current proxy's stick-table or in the designated stick-table,
17057 measured in kilobytes. If the address is not found, zero is returned. The
17058 test is currently performed on 32-bit integers, which limits values to 4
17059 terabytes. See also sc/sc0/sc1/sc2_kbytes_out.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020017060
Willy Tarreau74ca5042013-06-11 23:12:07 +020017061src_port : integer
17062 Returns an integer value corresponding to the TCP source port of the
17063 connection on the client side, which is the port the client connected from.
17064 Usage of this function is very limited as modern protocols do not care much
17065 about source ports nowadays.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010017066
Willy Tarreau74ca5042013-06-11 23:12:07 +020017067src_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017068 Returns the cumulative number of connections initiated from the incoming
Willy Tarreauc9705a12010-07-27 20:05:50 +020017069 connection's source IPv4 address in the current proxy's stick-table or in the
17070 designated stick-table, that were transformed into sessions, which means that
17071 they were accepted by "tcp-request" rules. If the address is not found, zero
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017072 is returned. See also sc/sc0/sc1/sc2_sess_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017073
Willy Tarreau74ca5042013-06-11 23:12:07 +020017074src_sess_rate([<table>]) : integer
17075 Returns the average session rate from the incoming connection's source
17076 address in the current proxy's stick-table or in the designated stick-table,
17077 measured in amount of sessions over the period configured in the table. A
17078 session is a connection that went past the early "tcp-request" rules. If the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017079 address is not found, zero is returned. See also sc/sc0/sc1/sc2_sess_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017080
Willy Tarreau74ca5042013-06-11 23:12:07 +020017081src_updt_conn_cnt([<table>]) : integer
17082 Creates or updates the entry associated to the incoming connection's source
17083 address in the current proxy's stick-table or in the designated stick-table.
17084 This table must be configured to store the "conn_cnt" data type, otherwise
17085 the match will be ignored. The current count is incremented by one, and the
17086 expiration timer refreshed. The updated count is returned, so this match
17087 can't return zero. This was used to reject service abusers based on their
17088 source address. Note: it is recommended to use the more complete "track-sc*"
17089 actions in "tcp-request" rules instead.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020017090
17091 Example :
17092 # This frontend limits incoming SSH connections to 3 per 10 second for
17093 # each source address, and rejects excess connections until a 10 second
17094 # silence is observed. At most 20 addresses are tracked.
17095 listen ssh
17096 bind :22
17097 mode tcp
17098 maxconn 100
Willy Tarreauc9705a12010-07-27 20:05:50 +020017099 stick-table type ip size 20 expire 10s store conn_cnt
Willy Tarreau74ca5042013-06-11 23:12:07 +020017100 tcp-request content reject if { src_updt_conn_cnt gt 3 }
Willy Tarreaua975b8f2010-06-05 19:13:27 +020017101 server local 127.0.0.1:22
17102
Willy Tarreau74ca5042013-06-11 23:12:07 +020017103srv_id : integer
17104 Returns an integer containing the server's id when processing the response.
17105 While it's almost only used with ACLs, it may be used for logging or
Christopher Fauletd1b44642020-04-30 09:51:15 +020017106 debugging. It can also be used in a tcp-check or an http-check ruleset.
Hervé COMMOWICKdaa824e2011-08-05 12:09:44 +020017107
vkill1dfd1652019-10-30 16:58:14 +080017108srv_name : string
17109 Returns a string containing the server's name when processing the response.
17110 While it's almost only used with ACLs, it may be used for logging or
Christopher Fauletd1b44642020-04-30 09:51:15 +020017111 debugging. It can also be used in a tcp-check or an http-check ruleset.
vkill1dfd1652019-10-30 16:58:14 +080017112
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200171137.3.4. Fetching samples at Layer 5
Willy Tarreau74ca5042013-06-11 23:12:07 +020017114----------------------------------
Willy Tarreau0b1cd942010-05-16 22:18:27 +020017115
Willy Tarreau74ca5042013-06-11 23:12:07 +020017116The layer 5 usually describes just the session layer which in haproxy is
17117closest to the session once all the connection handshakes are finished, but
17118when no content is yet made available. The fetch methods described here are
17119usable as low as the "tcp-request content" rule sets unless they require some
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017120future information. Those generally include the results of SSL negotiations.
Willy Tarreauc735a072011-03-29 00:57:02 +020017121
Ben Shillitof25e8e52016-12-02 14:25:37 +00001712251d.all(<prop>[,<prop>*]) : string
17123 Returns values for the properties requested as a string, where values are
17124 separated by the delimiter specified with "51degrees-property-separator".
17125 The device is identified using all the important HTTP headers from the
17126 request. The function can be passed up to five property names, and if a
17127 property name can't be found, the value "NoData" is returned.
17128
17129 Example :
17130 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request
17131 # containing the three properties requested using all relevant headers from
17132 # the request.
17133 frontend http-in
17134 bind *:8081
17135 default_backend servers
17136 http-request set-header X-51D-DeviceTypeMobileTablet \
17137 %[51d.all(DeviceType,IsMobile,IsTablet)]
17138
Emeric Brun645ae792014-04-30 14:21:06 +020017139ssl_bc : boolean
17140 Returns true when the back connection was made via an SSL/TLS transport
17141 layer and is locally deciphered. This means the outgoing connection was made
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017142 other a server with the "ssl" option. It can be used in a tcp-check or an
17143 http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017144
17145ssl_bc_alg_keysize : integer
17146 Returns the symmetric cipher key size supported in bits when the outgoing
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017147 connection was made over an SSL/TLS transport layer. It can be used in a
17148 tcp-check or an http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017149
Olivier Houchard6b77f492018-11-22 18:18:29 +010017150ssl_bc_alpn : string
17151 This extracts the Application Layer Protocol Negotiation field from an
17152 outgoing connection made via a TLS transport layer.
Michael Prokop4438c602019-05-24 10:25:45 +020017153 The result is a string containing the protocol name negotiated with the
Olivier Houchard6b77f492018-11-22 18:18:29 +010017154 server. The SSL library must have been built with support for TLS
17155 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
17156 not advertised unless the "alpn" keyword on the "server" line specifies a
17157 protocol list. Also, nothing forces the server to pick a protocol from this
17158 list, any other one may be requested. The TLS ALPN extension is meant to
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017159 replace the TLS NPN extension. See also "ssl_bc_npn". It can be used in a
17160 tcp-check or an http-check ruleset.
Olivier Houchard6b77f492018-11-22 18:18:29 +010017161
Emeric Brun645ae792014-04-30 14:21:06 +020017162ssl_bc_cipher : string
17163 Returns the name of the used cipher when the outgoing connection was made
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017164 over an SSL/TLS transport layer. It can be used in a tcp-check or an
17165 http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017166
Patrick Hemmer65674662019-06-04 08:13:03 -040017167ssl_bc_client_random : binary
17168 Returns the client random of the back connection when the incoming connection
17169 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
17170 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017171 It can be used in a tcp-check or an http-check ruleset.
Patrick Hemmer65674662019-06-04 08:13:03 -040017172
Emeric Brun74f7ffa2018-02-19 16:14:12 +010017173ssl_bc_is_resumed : boolean
17174 Returns true when the back connection was made over an SSL/TLS transport
17175 layer and the newly created SSL session was resumed using a cached
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017176 session or a TLS ticket. It can be used in a tcp-check or an http-check
17177 ruleset.
Emeric Brun74f7ffa2018-02-19 16:14:12 +010017178
Olivier Houchard6b77f492018-11-22 18:18:29 +010017179ssl_bc_npn : string
17180 This extracts the Next Protocol Negotiation field from an outgoing connection
17181 made via a TLS transport layer. The result is a string containing the
Michael Prokop4438c602019-05-24 10:25:45 +020017182 protocol name negotiated with the server . The SSL library must have been
Olivier Houchard6b77f492018-11-22 18:18:29 +010017183 built with support for TLS extensions enabled (check haproxy -vv). Note that
17184 the TLS NPN extension is not advertised unless the "npn" keyword on the
17185 "server" line specifies a protocol list. Also, nothing forces the server to
17186 pick a protocol from this list, any other one may be used. Please note that
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017187 the TLS NPN extension was replaced with ALPN. It can be used in a tcp-check
17188 or an http-check ruleset.
Olivier Houchard6b77f492018-11-22 18:18:29 +010017189
Emeric Brun645ae792014-04-30 14:21:06 +020017190ssl_bc_protocol : string
17191 Returns the name of the used protocol when the outgoing connection was made
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017192 over an SSL/TLS transport layer. It can be used in a tcp-check or an
17193 http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017194
Emeric Brunb73a9b02014-04-30 18:49:19 +020017195ssl_bc_unique_id : binary
Emeric Brun645ae792014-04-30 14:21:06 +020017196 When the outgoing connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020017197 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017198 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64". It
17199 can be used in a tcp-check or an http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017200
Patrick Hemmer65674662019-06-04 08:13:03 -040017201ssl_bc_server_random : binary
17202 Returns the server random of the back connection when the incoming connection
17203 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
17204 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017205 It can be used in a tcp-check or an http-check ruleset.
Patrick Hemmer65674662019-06-04 08:13:03 -040017206
Emeric Brun645ae792014-04-30 14:21:06 +020017207ssl_bc_session_id : binary
17208 Returns the SSL ID of the back connection when the outgoing connection was
17209 made over an SSL/TLS transport layer. It is useful to log if we want to know
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017210 if session was reused or not. It can be used in a tcp-check or an http-check
17211 ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017212
Patrick Hemmere0275472018-04-28 19:15:51 -040017213ssl_bc_session_key : binary
17214 Returns the SSL session master key of the back connection when the outgoing
17215 connection was made over an SSL/TLS transport layer. It is useful to decrypt
17216 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017217 BoringSSL. It can be used in a tcp-check or an http-check ruleset.
Patrick Hemmere0275472018-04-28 19:15:51 -040017218
Emeric Brun645ae792014-04-30 14:21:06 +020017219ssl_bc_use_keysize : integer
17220 Returns the symmetric cipher key size used in bits when the outgoing
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017221 connection was made over an SSL/TLS transport layer. It can be used in a
17222 tcp-check or an http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017223
Willy Tarreau74ca5042013-06-11 23:12:07 +020017224ssl_c_ca_err : integer
17225 When the incoming connection was made over an SSL/TLS transport layer,
17226 returns the ID of the first error detected during verification of the client
17227 certificate at depth > 0, or 0 if no error was encountered during this
17228 verification process. Please refer to your SSL library's documentation to
17229 find the exhaustive list of error codes.
Willy Tarreauc735a072011-03-29 00:57:02 +020017230
Willy Tarreau74ca5042013-06-11 23:12:07 +020017231ssl_c_ca_err_depth : integer
17232 When the incoming connection was made over an SSL/TLS transport layer,
17233 returns the depth in the CA chain of the first error detected during the
17234 verification of the client certificate. If no error is encountered, 0 is
17235 returned.
Willy Tarreau0ba27502007-12-24 16:55:16 +010017236
Christopher Faulet70d10d12020-11-06 12:10:33 +010017237ssl_c_chain_der : binary
William Dauchya598b502020-08-06 18:11:38 +020017238 Returns the DER formatted chain certificate presented by the client when the
17239 incoming connection was made over an SSL/TLS transport layer. When used for
17240 an ACL, the value(s) to match against can be passed in hexadecimal form. One
17241 can parse the result with any lib accepting ASN.1 DER data. It currentlly
17242 does not support resumed sessions.
17243
Christopher Faulet70d10d12020-11-06 12:10:33 +010017244ssl_c_der : binary
17245 Returns the DER formatted certificate presented by the client when the
17246 incoming connection was made over an SSL/TLS transport layer. When used for
17247 an ACL, the value(s) to match against can be passed in hexadecimal form.
17248
Willy Tarreau74ca5042013-06-11 23:12:07 +020017249ssl_c_err : integer
17250 When the incoming connection was made over an SSL/TLS transport layer,
17251 returns the ID of the first error detected during verification at depth 0, or
17252 0 if no error was encountered during this verification process. Please refer
17253 to your SSL library's documentation to find the exhaustive list of error
17254 codes.
Willy Tarreau62644772008-07-16 18:36:06 +020017255
Elliot Otchet71f82972020-01-15 08:12:14 -050017256ssl_c_i_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020017257 When the incoming connection was made over an SSL/TLS transport layer,
17258 returns the full distinguished name of the issuer of the certificate
17259 presented by the client when no <entry> is specified, or the value of the
17260 first given entry found from the beginning of the DN. If a positive/negative
17261 occurrence number is specified as the optional second argument, it returns
17262 the value of the nth given entry value from the beginning/end of the DN.
17263 For instance, "ssl_c_i_dn(OU,2)" the second organization unit, and
17264 "ssl_c_i_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050017265 The <format> parameter allows you to receive the DN suitable for
17266 consumption by different protocols. Currently supported is rfc2253 for
17267 LDAP v3.
17268 If you'd like to modify the format only you can specify an empty string
17269 and zero for the first two parameters. Example: ssl_c_i_dn(,0,rfc2253)
Willy Tarreau62644772008-07-16 18:36:06 +020017270
Willy Tarreau74ca5042013-06-11 23:12:07 +020017271ssl_c_key_alg : string
17272 Returns the name of the algorithm used to generate the key of the certificate
17273 presented by the client when the incoming connection was made over an SSL/TLS
17274 transport layer.
Willy Tarreau62644772008-07-16 18:36:06 +020017275
Willy Tarreau74ca5042013-06-11 23:12:07 +020017276ssl_c_notafter : string
17277 Returns the end date presented by the client as a formatted string
17278 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
17279 transport layer.
Emeric Brunbede3d02009-06-30 17:54:00 +020017280
Willy Tarreau74ca5042013-06-11 23:12:07 +020017281ssl_c_notbefore : string
17282 Returns the start date presented by the client as a formatted string
17283 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
17284 transport layer.
Willy Tarreaub6672b52011-12-12 17:23:41 +010017285
Elliot Otchet71f82972020-01-15 08:12:14 -050017286ssl_c_s_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020017287 When the incoming connection was made over an SSL/TLS transport layer,
17288 returns the full distinguished name of the subject of the certificate
17289 presented by the client when no <entry> is specified, or the value of the
17290 first given entry found from the beginning of the DN. If a positive/negative
17291 occurrence number is specified as the optional second argument, it returns
17292 the value of the nth given entry value from the beginning/end of the DN.
17293 For instance, "ssl_c_s_dn(OU,2)" the second organization unit, and
17294 "ssl_c_s_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050017295 The <format> parameter allows you to receive the DN suitable for
17296 consumption by different protocols. Currently supported is rfc2253 for
17297 LDAP v3.
17298 If you'd like to modify the format only you can specify an empty string
17299 and zero for the first two parameters. Example: ssl_c_s_dn(,0,rfc2253)
Willy Tarreaub6672b52011-12-12 17:23:41 +010017300
Willy Tarreau74ca5042013-06-11 23:12:07 +020017301ssl_c_serial : binary
17302 Returns the serial of the certificate presented by the client when the
17303 incoming connection was made over an SSL/TLS transport layer. When used for
17304 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun2525b6b2012-10-18 15:59:43 +020017305
Willy Tarreau74ca5042013-06-11 23:12:07 +020017306ssl_c_sha1 : binary
17307 Returns the SHA-1 fingerprint of the certificate presented by the client when
17308 the incoming connection was made over an SSL/TLS transport layer. This can be
17309 used to stick a client to a server, or to pass this information to a server.
Willy Tarreau2d0caa32014-07-02 19:01:22 +020017310 Note that the output is binary, so if you want to pass that signature to the
17311 server, you need to encode it in hex or base64, such as in the example below:
17312
Jarno Huuskonen676f6222017-03-30 09:19:45 +030017313 Example:
Willy Tarreau2d0caa32014-07-02 19:01:22 +020017314 http-request set-header X-SSL-Client-SHA1 %[ssl_c_sha1,hex]
Emeric Brun2525b6b2012-10-18 15:59:43 +020017315
Willy Tarreau74ca5042013-06-11 23:12:07 +020017316ssl_c_sig_alg : string
17317 Returns the name of the algorithm used to sign the certificate presented by
17318 the client when the incoming connection was made over an SSL/TLS transport
17319 layer.
Emeric Brun87855892012-10-17 17:39:35 +020017320
Willy Tarreau74ca5042013-06-11 23:12:07 +020017321ssl_c_used : boolean
17322 Returns true if current SSL session uses a client certificate even if current
17323 connection uses SSL session resumption. See also "ssl_fc_has_crt".
Emeric Brun7f56e742012-10-19 18:15:40 +020017324
Willy Tarreau74ca5042013-06-11 23:12:07 +020017325ssl_c_verify : integer
17326 Returns the verify result error ID when the incoming connection was made over
17327 an SSL/TLS transport layer, otherwise zero if no error is encountered. Please
17328 refer to your SSL library's documentation for an exhaustive list of error
17329 codes.
Emeric Brunce5ad802012-10-22 14:11:22 +020017330
Willy Tarreau74ca5042013-06-11 23:12:07 +020017331ssl_c_version : integer
17332 Returns the version of the certificate presented by the client when the
17333 incoming connection was made over an SSL/TLS transport layer.
Emeric Brunce5ad802012-10-22 14:11:22 +020017334
Emeric Brun43e79582014-10-29 19:03:26 +010017335ssl_f_der : binary
17336 Returns the DER formatted certificate presented by the frontend when the
17337 incoming connection was made over an SSL/TLS transport layer. When used for
17338 an ACL, the value(s) to match against can be passed in hexadecimal form.
17339
Elliot Otchet71f82972020-01-15 08:12:14 -050017340ssl_f_i_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020017341 When the incoming connection was made over an SSL/TLS transport layer,
17342 returns the full distinguished name of the issuer of the certificate
17343 presented by the frontend when no <entry> is specified, or the value of the
17344 first given entry found from the beginning of the DN. If a positive/negative
Emeric Brun87855892012-10-17 17:39:35 +020017345 occurrence number is specified as the optional second argument, it returns
Willy Tarreau74ca5042013-06-11 23:12:07 +020017346 the value of the nth given entry value from the beginning/end of the DN.
17347 For instance, "ssl_f_i_dn(OU,2)" the second organization unit, and
17348 "ssl_f_i_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050017349 The <format> parameter allows you to receive the DN suitable for
17350 consumption by different protocols. Currently supported is rfc2253 for
17351 LDAP v3.
17352 If you'd like to modify the format only you can specify an empty string
17353 and zero for the first two parameters. Example: ssl_f_i_dn(,0,rfc2253)
Emeric Brun87855892012-10-17 17:39:35 +020017354
Willy Tarreau74ca5042013-06-11 23:12:07 +020017355ssl_f_key_alg : string
17356 Returns the name of the algorithm used to generate the key of the certificate
17357 presented by the frontend when the incoming connection was made over an
17358 SSL/TLS transport layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020017359
Willy Tarreau74ca5042013-06-11 23:12:07 +020017360ssl_f_notafter : string
17361 Returns the end date presented by the frontend as a formatted string
17362 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
17363 transport layer.
Emeric Brun2525b6b2012-10-18 15:59:43 +020017364
Willy Tarreau74ca5042013-06-11 23:12:07 +020017365ssl_f_notbefore : string
17366 Returns the start date presented by the frontend as a formatted string
17367 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
17368 transport layer.
Emeric Brun87855892012-10-17 17:39:35 +020017369
Elliot Otchet71f82972020-01-15 08:12:14 -050017370ssl_f_s_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020017371 When the incoming connection was made over an SSL/TLS transport layer,
17372 returns the full distinguished name of the subject of the certificate
17373 presented by the frontend when no <entry> is specified, or the value of the
17374 first given entry found from the beginning of the DN. If a positive/negative
17375 occurrence number is specified as the optional second argument, it returns
17376 the value of the nth given entry value from the beginning/end of the DN.
17377 For instance, "ssl_f_s_dn(OU,2)" the second organization unit, and
17378 "ssl_f_s_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050017379 The <format> parameter allows you to receive the DN suitable for
17380 consumption by different protocols. Currently supported is rfc2253 for
17381 LDAP v3.
17382 If you'd like to modify the format only you can specify an empty string
17383 and zero for the first two parameters. Example: ssl_f_s_dn(,0,rfc2253)
Emeric Brunce5ad802012-10-22 14:11:22 +020017384
Willy Tarreau74ca5042013-06-11 23:12:07 +020017385ssl_f_serial : binary
17386 Returns the serial of the certificate presented by the frontend when the
17387 incoming connection was made over an SSL/TLS transport layer. When used for
17388 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun87855892012-10-17 17:39:35 +020017389
Emeric Brun55f4fa82014-04-30 17:11:25 +020017390ssl_f_sha1 : binary
17391 Returns the SHA-1 fingerprint of the certificate presented by the frontend
17392 when the incoming connection was made over an SSL/TLS transport layer. This
17393 can be used to know which certificate was chosen using SNI.
17394
Willy Tarreau74ca5042013-06-11 23:12:07 +020017395ssl_f_sig_alg : string
17396 Returns the name of the algorithm used to sign the certificate presented by
17397 the frontend when the incoming connection was made over an SSL/TLS transport
17398 layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020017399
Willy Tarreau74ca5042013-06-11 23:12:07 +020017400ssl_f_version : integer
17401 Returns the version of the certificate presented by the frontend when the
17402 incoming connection was made over an SSL/TLS transport layer.
17403
17404ssl_fc : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020017405 Returns true when the front connection was made via an SSL/TLS transport
17406 layer and is locally deciphered. This means it has matched a socket declared
17407 with a "bind" line having the "ssl" option.
17408
Willy Tarreau74ca5042013-06-11 23:12:07 +020017409 Example :
17410 # This passes "X-Proto: https" to servers when client connects over SSL
17411 listen http-https
17412 bind :80
17413 bind :443 ssl crt /etc/haproxy.pem
17414 http-request add-header X-Proto https if { ssl_fc }
17415
17416ssl_fc_alg_keysize : integer
17417 Returns the symmetric cipher key size supported in bits when the incoming
17418 connection was made over an SSL/TLS transport layer.
17419
17420ssl_fc_alpn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017421 This extracts the Application Layer Protocol Negotiation field from an
Willy Tarreau74ca5042013-06-11 23:12:07 +020017422 incoming connection made via a TLS transport layer and locally deciphered by
17423 haproxy. The result is a string containing the protocol name advertised by
17424 the client. The SSL library must have been built with support for TLS
17425 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
17426 not advertised unless the "alpn" keyword on the "bind" line specifies a
17427 protocol list. Also, nothing forces the client to pick a protocol from this
17428 list, any other one may be requested. The TLS ALPN extension is meant to
17429 replace the TLS NPN extension. See also "ssl_fc_npn".
17430
Willy Tarreau74ca5042013-06-11 23:12:07 +020017431ssl_fc_cipher : string
17432 Returns the name of the used cipher when the incoming connection was made
17433 over an SSL/TLS transport layer.
Willy Tarreauab861d32013-04-02 02:30:41 +020017434
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010017435ssl_fc_cipherlist_bin : binary
17436 Returns the binary form of the client hello cipher list. The maximum returned
17437 value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010017438 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010017439
17440ssl_fc_cipherlist_hex : string
17441 Returns the binary form of the client hello cipher list encoded as
17442 hexadecimal. The maximum returned value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010017443 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010017444
17445ssl_fc_cipherlist_str : string
17446 Returns the decoded text form of the client hello cipher list. The maximum
17447 number of ciphers returned is according with the value of
17448 "tune.ssl.capture-cipherlist-size". Note that this sample-fetch is only
Davor Ocelice9ed2812017-12-25 17:49:28 +010017449 available with OpenSSL >= 1.0.2. If the function is not enabled, this
Emmanuel Hocdetddcde192017-09-01 17:32:08 +020017450 sample-fetch returns the hash like "ssl_fc_cipherlist_xxh".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010017451
17452ssl_fc_cipherlist_xxh : integer
17453 Returns a xxh64 of the cipher list. This hash can be return only is the value
17454 "tune.ssl.capture-cipherlist-size" is set greater than 0, however the hash
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010017455 take in account all the data of the cipher list.
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010017456
Patrick Hemmer65674662019-06-04 08:13:03 -040017457ssl_fc_client_random : binary
17458 Returns the client random of the front connection when the incoming connection
17459 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
17460 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
17461
William Lallemand7d42ef52020-07-06 11:41:30 +020017462ssl_fc_client_early_traffic_secret : string
17463 Return the CLIENT_EARLY_TRAFFIC_SECRET as an hexadecimal string for the
17464 front connection when the incoming connection was made over a TLS 1.3
17465 transport layer.
17466 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
17467 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
17468 activated with "tune.ssl.keylog on" in the global section. See also
17469 "tune.ssl.keylog"
17470
17471ssl_fc_client_handshake_traffic_secret : string
17472 Return the CLIENT_HANDSHAKE_TRAFFIC_SECRET as an hexadecimal string for the
17473 front connection when the incoming connection was made over a TLS 1.3
17474 transport layer.
17475 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
17476 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
17477 activated with "tune.ssl.keylog on" in the global section. See also
17478 "tune.ssl.keylog"
17479
17480ssl_fc_client_traffic_secret_0 : string
17481 Return the CLIENT_TRAFFIC_SECRET_0 as an hexadecimal string for the
17482 front connection when the incoming connection was made over a TLS 1.3
17483 transport layer.
17484 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
17485 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
17486 activated with "tune.ssl.keylog on" in the global section. See also
17487 "tune.ssl.keylog"
17488
17489ssl_fc_exporter_secret : string
17490 Return the EXPORTER_SECRET as an hexadecimal string for the
17491 front connection when the incoming connection was made over a TLS 1.3
17492 transport layer.
17493 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
17494 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
17495 activated with "tune.ssl.keylog on" in the global section. See also
17496 "tune.ssl.keylog"
17497
17498ssl_fc_early_exporter_secret : string
17499 Return the EARLY_EXPORTER_SECRET as an hexadecimal string for the
17500 front connection when the incoming connection was made over an TLS 1.3
17501 transport layer.
17502 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
17503 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
17504 activated with "tune.ssl.keylog on" in the global section. See also
17505 "tune.ssl.keylog"
17506
Willy Tarreau74ca5042013-06-11 23:12:07 +020017507ssl_fc_has_crt : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020017508 Returns true if a client certificate is present in an incoming connection over
17509 SSL/TLS transport layer. Useful if 'verify' statement is set to 'optional'.
Emeric Brun9143d372012-12-20 15:44:16 +010017510 Note: on SSL session resumption with Session ID or TLS ticket, client
17511 certificate is not present in the current connection but may be retrieved
17512 from the cache or the ticket. So prefer "ssl_c_used" if you want to check if
17513 current SSL session uses a client certificate.
Emeric Brun2525b6b2012-10-18 15:59:43 +020017514
Olivier Houchardccaa7de2017-10-02 11:51:03 +020017515ssl_fc_has_early : boolean
17516 Returns true if early data were sent, and the handshake didn't happen yet. As
17517 it has security implications, it is useful to be able to refuse those, or
17518 wait until the handshake happened.
17519
Willy Tarreau74ca5042013-06-11 23:12:07 +020017520ssl_fc_has_sni : boolean
17521 This checks for the presence of a Server Name Indication TLS extension (SNI)
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020017522 in an incoming connection was made over an SSL/TLS transport layer. Returns
17523 true when the incoming connection presents a TLS SNI field. This requires
John Roeslerfb2fce12019-07-10 15:45:51 -050017524 that the SSL library is built with support for TLS extensions enabled (check
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020017525 haproxy -vv).
Willy Tarreau7875d092012-09-10 08:20:03 +020017526
Nenad Merdanovic1516fe32016-05-17 03:31:21 +020017527ssl_fc_is_resumed : boolean
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020017528 Returns true if the SSL/TLS session has been resumed through the use of
Jérôme Magnin4a326cb2018-01-15 14:01:17 +010017529 SSL session cache or TLS tickets on an incoming connection over an SSL/TLS
17530 transport layer.
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020017531
Willy Tarreau74ca5042013-06-11 23:12:07 +020017532ssl_fc_npn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017533 This extracts the Next Protocol Negotiation field from an incoming connection
Willy Tarreau74ca5042013-06-11 23:12:07 +020017534 made via a TLS transport layer and locally deciphered by haproxy. The result
17535 is a string containing the protocol name advertised by the client. The SSL
17536 library must have been built with support for TLS extensions enabled (check
17537 haproxy -vv). Note that the TLS NPN extension is not advertised unless the
17538 "npn" keyword on the "bind" line specifies a protocol list. Also, nothing
17539 forces the client to pick a protocol from this list, any other one may be
17540 requested. Please note that the TLS NPN extension was replaced with ALPN.
Willy Tarreaua33c6542012-10-15 13:19:06 +020017541
Willy Tarreau74ca5042013-06-11 23:12:07 +020017542ssl_fc_protocol : string
17543 Returns the name of the used protocol when the incoming connection was made
17544 over an SSL/TLS transport layer.
Willy Tarreau7875d092012-09-10 08:20:03 +020017545
Emeric Brunb73a9b02014-04-30 18:49:19 +020017546ssl_fc_unique_id : binary
David Sc1ad52e2014-04-08 18:48:47 -040017547 When the incoming connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020017548 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
17549 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
David Sc1ad52e2014-04-08 18:48:47 -040017550
William Lallemand7d42ef52020-07-06 11:41:30 +020017551ssl_fc_server_handshake_traffic_secret : string
17552 Return the SERVER_HANDSHAKE_TRAFFIC_SECRET as an hexadecimal string for the
17553 front connection when the incoming connection was made over a TLS 1.3
17554 transport layer.
17555 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
17556 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
17557 activated with "tune.ssl.keylog on" in the global section. See also
17558 "tune.ssl.keylog"
17559
17560ssl_fc_server_traffic_secret_0 : string
17561 Return the SERVER_TRAFFIC_SECRET_0 as an hexadecimal string for the
17562 front connection when the incoming connection was made over an TLS 1.3
17563 transport layer.
17564 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
17565 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
17566 activated with "tune.ssl.keylog on" in the global section. See also
17567 "tune.ssl.keylog"
17568
Patrick Hemmer65674662019-06-04 08:13:03 -040017569ssl_fc_server_random : binary
17570 Returns the server random of the front connection when the incoming connection
17571 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
17572 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
17573
Willy Tarreau74ca5042013-06-11 23:12:07 +020017574ssl_fc_session_id : binary
17575 Returns the SSL ID of the front connection when the incoming connection was
17576 made over an SSL/TLS transport layer. It is useful to stick a given client to
17577 a server. It is important to note that some browsers refresh their session ID
17578 every few minutes.
Willy Tarreau7875d092012-09-10 08:20:03 +020017579
Patrick Hemmere0275472018-04-28 19:15:51 -040017580ssl_fc_session_key : binary
17581 Returns the SSL session master key of the front connection when the incoming
17582 connection was made over an SSL/TLS transport layer. It is useful to decrypt
17583 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
17584 BoringSSL.
17585
17586
Willy Tarreau74ca5042013-06-11 23:12:07 +020017587ssl_fc_sni : string
17588 This extracts the Server Name Indication TLS extension (SNI) field from an
17589 incoming connection made via an SSL/TLS transport layer and locally
17590 deciphered by haproxy. The result (when present) typically is a string
17591 matching the HTTPS host name (253 chars or less). The SSL library must have
17592 been built with support for TLS extensions enabled (check haproxy -vv).
17593
17594 This fetch is different from "req_ssl_sni" above in that it applies to the
17595 connection being deciphered by haproxy and not to SSL contents being blindly
17596 forwarded. See also "ssl_fc_sni_end" and "ssl_fc_sni_reg" below. This
John Roeslerfb2fce12019-07-10 15:45:51 -050017597 requires that the SSL library is built with support for TLS extensions
Cyril Bonté9c1eb1e2012-10-09 22:45:34 +020017598 enabled (check haproxy -vv).
Willy Tarreau62644772008-07-16 18:36:06 +020017599
Willy Tarreau74ca5042013-06-11 23:12:07 +020017600 ACL derivatives :
Willy Tarreau74ca5042013-06-11 23:12:07 +020017601 ssl_fc_sni_end : suffix match
17602 ssl_fc_sni_reg : regex match
Emeric Brun589fcad2012-10-16 14:13:26 +020017603
Willy Tarreau74ca5042013-06-11 23:12:07 +020017604ssl_fc_use_keysize : integer
17605 Returns the symmetric cipher key size used in bits when the incoming
17606 connection was made over an SSL/TLS transport layer.
Willy Tarreaub6fb4202008-07-20 11:18:28 +020017607
William Lallemandbfa3e812020-06-25 20:07:18 +020017608ssl_s_der : binary
17609 Returns the DER formatted certificate presented by the server when the
17610 outgoing connection was made over an SSL/TLS transport layer. When used for
17611 an ACL, the value(s) to match against can be passed in hexadecimal form.
17612
William Dauchya598b502020-08-06 18:11:38 +020017613ssl_s_chain_der : binary
17614 Returns the DER formatted chain certificate presented by the server when the
17615 outgoing connection was made over an SSL/TLS transport layer. When used for
17616 an ACL, the value(s) to match against can be passed in hexadecimal form. One
17617 can parse the result with any lib accepting ASN.1 DER data. It currentlly
17618 does not support resumed sessions.
17619
William Lallemandbfa3e812020-06-25 20:07:18 +020017620ssl_s_key_alg : string
17621 Returns the name of the algorithm used to generate the key of the certificate
17622 presented by the server when the outgoing connection was made over an
17623 SSL/TLS transport layer.
17624
17625ssl_s_notafter : string
17626 Returns the end date presented by the server as a formatted string
17627 YYMMDDhhmmss[Z] when the outgoing connection was made over an SSL/TLS
17628 transport layer.
17629
17630ssl_s_notbefore : string
17631 Returns the start date presented by the server as a formatted string
17632 YYMMDDhhmmss[Z] when the outgoing connection was made over an SSL/TLS
17633 transport layer.
17634
17635ssl_s_i_dn([<entry>[,<occ>[,<format>]]]) : string
17636 When the outgoing connection was made over an SSL/TLS transport layer,
17637 returns the full distinguished name of the issuer of the certificate
17638 presented by the server when no <entry> is specified, or the value of the
17639 first given entry found from the beginning of the DN. If a positive/negative
17640 occurrence number is specified as the optional second argument, it returns
17641 the value of the nth given entry value from the beginning/end of the DN.
William Lallemand8f600c82020-06-26 09:55:06 +020017642 For instance, "ssl_s_i_dn(OU,2)" the second organization unit, and
17643 "ssl_s_i_dn(CN)" retrieves the common name.
William Lallemandbfa3e812020-06-25 20:07:18 +020017644 The <format> parameter allows you to receive the DN suitable for
17645 consumption by different protocols. Currently supported is rfc2253 for
17646 LDAP v3.
17647 If you'd like to modify the format only you can specify an empty string
17648 and zero for the first two parameters. Example: ssl_s_i_dn(,0,rfc2253)
17649
17650ssl_s_s_dn([<entry>[,<occ>[,<format>]]]) : string
17651 When the outgoing connection was made over an SSL/TLS transport layer,
17652 returns the full distinguished name of the subject of the certificate
17653 presented by the server when no <entry> is specified, or the value of the
17654 first given entry found from the beginning of the DN. If a positive/negative
17655 occurrence number is specified as the optional second argument, it returns
17656 the value of the nth given entry value from the beginning/end of the DN.
William Lallemand8f600c82020-06-26 09:55:06 +020017657 For instance, "ssl_s_s_dn(OU,2)" the second organization unit, and
17658 "ssl_s_s_dn(CN)" retrieves the common name.
William Lallemandbfa3e812020-06-25 20:07:18 +020017659 The <format> parameter allows you to receive the DN suitable for
17660 consumption by different protocols. Currently supported is rfc2253 for
17661 LDAP v3.
17662 If you'd like to modify the format only you can specify an empty string
17663 and zero for the first two parameters. Example: ssl_s_s_dn(,0,rfc2253)
17664
17665ssl_s_serial : binary
17666 Returns the serial of the certificate presented by the server when the
17667 outgoing connection was made over an SSL/TLS transport layer. When used for
17668 an ACL, the value(s) to match against can be passed in hexadecimal form.
17669
17670ssl_s_sha1 : binary
17671 Returns the SHA-1 fingerprint of the certificate presented by the server
17672 when the outgoing connection was made over an SSL/TLS transport layer. This
17673 can be used to know which certificate was chosen using SNI.
17674
17675ssl_s_sig_alg : string
17676 Returns the name of the algorithm used to sign the certificate presented by
17677 the server when the outgoing connection was made over an SSL/TLS transport
17678 layer.
17679
17680ssl_s_version : integer
17681 Returns the version of the certificate presented by the server when the
17682 outgoing connection was made over an SSL/TLS transport layer.
Willy Tarreaub6fb4202008-07-20 11:18:28 +020017683
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200176847.3.5. Fetching samples from buffer contents (Layer 6)
Willy Tarreau74ca5042013-06-11 23:12:07 +020017685------------------------------------------------------
Willy Tarreaub6fb4202008-07-20 11:18:28 +020017686
Willy Tarreau74ca5042013-06-11 23:12:07 +020017687Fetching samples from buffer contents is a bit different from the previous
17688sample fetches above because the sampled data are ephemeral. These data can
17689only be used when they're available and will be lost when they're forwarded.
17690For this reason, samples fetched from buffer contents during a request cannot
17691be used in a response for example. Even while the data are being fetched, they
17692can change. Sometimes it is necessary to set some delays or combine multiple
17693sample fetch methods to ensure that the expected data are complete and usable,
17694for example through TCP request content inspection. Please see the "tcp-request
17695content" keyword for more detailed information on the subject.
Willy Tarreau62644772008-07-16 18:36:06 +020017696
Willy Tarreau74ca5042013-06-11 23:12:07 +020017697payload(<offset>,<length>) : binary (deprecated)
Davor Ocelice9ed2812017-12-25 17:49:28 +010017698 This is an alias for "req.payload" when used in the context of a request (e.g.
Willy Tarreau74ca5042013-06-11 23:12:07 +020017699 "stick on", "stick match"), and for "res.payload" when used in the context of
17700 a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010017701
Willy Tarreau74ca5042013-06-11 23:12:07 +020017702payload_lv(<offset1>,<length>[,<offset2>]) : binary (deprecated)
17703 This is an alias for "req.payload_lv" when used in the context of a request
Davor Ocelice9ed2812017-12-25 17:49:28 +010017704 (e.g. "stick on", "stick match"), and for "res.payload_lv" when used in the
Willy Tarreau74ca5042013-06-11 23:12:07 +020017705 context of a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010017706
Thierry FOURNIERd7d88812017-04-19 15:15:14 +020017707req.hdrs : string
17708 Returns the current request headers as string including the last empty line
17709 separating headers from the request body. The last empty line can be used to
17710 detect a truncated header block. This sample fetch is useful for some SPOE
17711 headers analyzers and for advanced logging.
17712
Thierry FOURNIER5617dce2017-04-09 05:38:19 +020017713req.hdrs_bin : binary
17714 Returns the current request headers contained in preparsed binary form. This
17715 is useful for offloading some processing with SPOE. Each string is described
17716 by a length followed by the number of bytes indicated in the length. The
17717 length is represented using the variable integer encoding detailed in the
17718 SPOE documentation. The end of the list is marked by a couple of empty header
17719 names and values (length of 0 for both).
17720
17721 *(<str:header-name><str:header-value>)<empty string><empty string>
17722
17723 int: refer to the SPOE documentation for the encoding
17724 str: <int:length><bytes>
17725
Willy Tarreau74ca5042013-06-11 23:12:07 +020017726req.len : integer
17727req_len : integer (deprecated)
17728 Returns an integer value corresponding to the number of bytes present in the
17729 request buffer. This is mostly used in ACL. It is important to understand
17730 that this test does not return false as long as the buffer is changing. This
17731 means that a check with equality to zero will almost always immediately match
17732 at the beginning of the session, while a test for more data will wait for
17733 that data to come in and return false only when haproxy is certain that no
17734 more data will come in. This test was designed to be used with TCP request
17735 content inspection.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020017736
Willy Tarreau74ca5042013-06-11 23:12:07 +020017737req.payload(<offset>,<length>) : binary
17738 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020017739 in the request buffer. As a special case, if the <length> argument is zero,
17740 the the whole buffer from <offset> to the end is extracted. This can be used
17741 with ACLs in order to check for the presence of some content in a buffer at
17742 any location.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020017743
Willy Tarreau74ca5042013-06-11 23:12:07 +020017744 ACL alternatives :
17745 payload(<offset>,<length>) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020017746
Willy Tarreau74ca5042013-06-11 23:12:07 +020017747req.payload_lv(<offset1>,<length>[,<offset2>]) : binary
17748 This extracts a binary block whose size is specified at <offset1> for <length>
17749 bytes, and which starts at <offset2> if specified or just after the length in
17750 the request buffer. The <offset2> parameter also supports relative offsets if
17751 prepended with a '+' or '-' sign.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020017752
Willy Tarreau74ca5042013-06-11 23:12:07 +020017753 ACL alternatives :
17754 payload_lv(<offset1>,<length>[,<offset2>]) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020017755
Willy Tarreau74ca5042013-06-11 23:12:07 +020017756 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020017757
Willy Tarreau74ca5042013-06-11 23:12:07 +020017758req.proto_http : boolean
17759req_proto_http : boolean (deprecated)
17760 Returns true when data in the request buffer look like HTTP and correctly
17761 parses as such. It is the same parser as the common HTTP request parser which
17762 is used so there should be no surprises. The test does not match until the
17763 request is complete, failed or timed out. This test may be used to report the
17764 protocol in TCP logs, but the biggest use is to block TCP request analysis
17765 until a complete HTTP request is present in the buffer, for example to track
17766 a header.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020017767
Willy Tarreau74ca5042013-06-11 23:12:07 +020017768 Example:
17769 # track request counts per "base" (concatenation of Host+URL)
17770 tcp-request inspect-delay 10s
17771 tcp-request content reject if !HTTP
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020017772 tcp-request content track-sc0 base table req-rate
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020017773
Willy Tarreau74ca5042013-06-11 23:12:07 +020017774req.rdp_cookie([<name>]) : string
17775rdp_cookie([<name>]) : string (deprecated)
17776 When the request buffer looks like the RDP protocol, extracts the RDP cookie
17777 <name>, or any cookie if unspecified. The parser only checks for the first
17778 cookie, as illustrated in the RDP protocol specification. The cookie name is
17779 case insensitive. Generally the "MSTS" cookie name will be used, as it can
17780 contain the user name of the client connecting to the server if properly
17781 configured on the client. The "MSTSHASH" cookie is often used as well for
17782 session stickiness to servers.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020017783
Willy Tarreau74ca5042013-06-11 23:12:07 +020017784 This differs from "balance rdp-cookie" in that any balancing algorithm may be
17785 used and thus the distribution of clients to backend servers is not linked to
17786 a hash of the RDP cookie. It is envisaged that using a balancing algorithm
17787 such as "balance roundrobin" or "balance leastconn" will lead to a more even
17788 distribution of clients to backend servers than the hash used by "balance
17789 rdp-cookie".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020017790
Willy Tarreau74ca5042013-06-11 23:12:07 +020017791 ACL derivatives :
17792 req_rdp_cookie([<name>]) : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020017793
Willy Tarreau74ca5042013-06-11 23:12:07 +020017794 Example :
17795 listen tse-farm
17796 bind 0.0.0.0:3389
17797 # wait up to 5s for an RDP cookie in the request
17798 tcp-request inspect-delay 5s
17799 tcp-request content accept if RDP_COOKIE
17800 # apply RDP cookie persistence
17801 persist rdp-cookie
17802 # Persist based on the mstshash cookie
17803 # This is only useful makes sense if
17804 # balance rdp-cookie is not used
17805 stick-table type string size 204800
17806 stick on req.rdp_cookie(mstshash)
17807 server srv1 1.1.1.1:3389
17808 server srv1 1.1.1.2:3389
Willy Tarreau04aa6a92012-04-06 18:57:55 +020017809
Willy Tarreau74ca5042013-06-11 23:12:07 +020017810 See also : "balance rdp-cookie", "persist rdp-cookie", "tcp-request" and the
17811 "req_rdp_cookie" ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020017812
Willy Tarreau74ca5042013-06-11 23:12:07 +020017813req.rdp_cookie_cnt([name]) : integer
17814rdp_cookie_cnt([name]) : integer (deprecated)
17815 Tries to parse the request buffer as RDP protocol, then returns an integer
17816 corresponding to the number of RDP cookies found. If an optional cookie name
17817 is passed, only cookies matching this name are considered. This is mostly
17818 used in ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020017819
Willy Tarreau74ca5042013-06-11 23:12:07 +020017820 ACL derivatives :
17821 req_rdp_cookie_cnt([<name>]) : integer match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020017822
Alex Zorin4afdd132018-12-30 13:56:28 +110017823req.ssl_alpn : string
17824 Returns a string containing the values of the Application-Layer Protocol
17825 Negotiation (ALPN) TLS extension (RFC7301), sent by the client within the SSL
17826 ClientHello message. Note that this only applies to raw contents found in the
17827 request buffer and not to the contents deciphered via an SSL data layer, so
17828 this will not work with "bind" lines having the "ssl" option. This is useful
17829 in ACL to make a routing decision based upon the ALPN preferences of a TLS
Jarno Huuskonene504f812019-01-03 07:56:49 +020017830 client, like in the example below. See also "ssl_fc_alpn".
Alex Zorin4afdd132018-12-30 13:56:28 +110017831
17832 Examples :
17833 # Wait for a client hello for at most 5 seconds
17834 tcp-request inspect-delay 5s
17835 tcp-request content accept if { req_ssl_hello_type 1 }
Jarno Huuskonene504f812019-01-03 07:56:49 +020017836 use_backend bk_acme if { req.ssl_alpn acme-tls/1 }
Alex Zorin4afdd132018-12-30 13:56:28 +110017837 default_backend bk_default
17838
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020017839req.ssl_ec_ext : boolean
17840 Returns a boolean identifying if client sent the Supported Elliptic Curves
17841 Extension as defined in RFC4492, section 5.1. within the SSL ClientHello
Cyril Bonté307ee1e2015-09-28 23:16:06 +020017842 message. This can be used to present ECC compatible clients with EC
17843 certificate and to use RSA for all others, on the same IP address. Note that
17844 this only applies to raw contents found in the request buffer and not to
17845 contents deciphered via an SSL data layer, so this will not work with "bind"
17846 lines having the "ssl" option.
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020017847
Willy Tarreau74ca5042013-06-11 23:12:07 +020017848req.ssl_hello_type : integer
17849req_ssl_hello_type : integer (deprecated)
17850 Returns an integer value containing the type of the SSL hello message found
17851 in the request buffer if the buffer contains data that parse as a complete
17852 SSL (v3 or superior) client hello message. Note that this only applies to raw
17853 contents found in the request buffer and not to contents deciphered via an
17854 SSL data layer, so this will not work with "bind" lines having the "ssl"
17855 option. This is mostly used in ACL to detect presence of an SSL hello message
17856 that is supposed to contain an SSL session ID usable for stickiness.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020017857
Willy Tarreau74ca5042013-06-11 23:12:07 +020017858req.ssl_sni : string
17859req_ssl_sni : string (deprecated)
17860 Returns a string containing the value of the Server Name TLS extension sent
17861 by a client in a TLS stream passing through the request buffer if the buffer
17862 contains data that parse as a complete SSL (v3 or superior) client hello
17863 message. Note that this only applies to raw contents found in the request
17864 buffer and not to contents deciphered via an SSL data layer, so this will not
Lukas Tribusa267b5d2020-07-19 00:25:06 +020017865 work with "bind" lines having the "ssl" option. This will only work for actual
17866 implicit TLS based protocols like HTTPS (443), IMAPS (993), SMTPS (465),
17867 however it will not work for explicit TLS based protocols, like SMTP (25/587)
17868 or IMAP (143). SNI normally contains the name of the host the client tries to
17869 connect to (for recent browsers). SNI is useful for allowing or denying access
17870 to certain hosts when SSL/TLS is used by the client. This test was designed to
17871 be used with TCP request content inspection. If content switching is needed,
17872 it is recommended to first wait for a complete client hello (type 1), like in
17873 the example below. See also "ssl_fc_sni".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020017874
Willy Tarreau74ca5042013-06-11 23:12:07 +020017875 ACL derivatives :
17876 req_ssl_sni : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020017877
Willy Tarreau74ca5042013-06-11 23:12:07 +020017878 Examples :
17879 # Wait for a client hello for at most 5 seconds
17880 tcp-request inspect-delay 5s
17881 tcp-request content accept if { req_ssl_hello_type 1 }
17882 use_backend bk_allow if { req_ssl_sni -f allowed_sites }
17883 default_backend bk_sorry_page
Willy Tarreau04aa6a92012-04-06 18:57:55 +020017884
Pradeep Jindalbb2acf52015-09-29 10:12:57 +053017885req.ssl_st_ext : integer
17886 Returns 0 if the client didn't send a SessionTicket TLS Extension (RFC5077)
17887 Returns 1 if the client sent SessionTicket TLS Extension
17888 Returns 2 if the client also sent non-zero length TLS SessionTicket
17889 Note that this only applies to raw contents found in the request buffer and
17890 not to contents deciphered via an SSL data layer, so this will not work with
17891 "bind" lines having the "ssl" option. This can for example be used to detect
17892 whether the client sent a SessionTicket or not and stick it accordingly, if
17893 no SessionTicket then stick on SessionID or don't stick as there's no server
17894 side state is there when SessionTickets are in use.
17895
Willy Tarreau74ca5042013-06-11 23:12:07 +020017896req.ssl_ver : integer
17897req_ssl_ver : integer (deprecated)
17898 Returns an integer value containing the version of the SSL/TLS protocol of a
17899 stream present in the request buffer. Both SSLv2 hello messages and SSLv3
17900 messages are supported. TLSv1 is announced as SSL version 3.1. The value is
17901 composed of the major version multiplied by 65536, added to the minor
17902 version. Note that this only applies to raw contents found in the request
17903 buffer and not to contents deciphered via an SSL data layer, so this will not
17904 work with "bind" lines having the "ssl" option. The ACL version of the test
Davor Ocelice9ed2812017-12-25 17:49:28 +010017905 matches against a decimal notation in the form MAJOR.MINOR (e.g. 3.1). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020017906 fetch is mostly used in ACL.
Willy Tarreaud63335a2010-02-26 12:56:52 +010017907
Willy Tarreau74ca5042013-06-11 23:12:07 +020017908 ACL derivatives :
17909 req_ssl_ver : decimal match
Willy Tarreaud63335a2010-02-26 12:56:52 +010017910
Willy Tarreau47e8eba2013-09-11 23:28:46 +020017911res.len : integer
17912 Returns an integer value corresponding to the number of bytes present in the
17913 response buffer. This is mostly used in ACL. It is important to understand
17914 that this test does not return false as long as the buffer is changing. This
17915 means that a check with equality to zero will almost always immediately match
17916 at the beginning of the session, while a test for more data will wait for
17917 that data to come in and return false only when haproxy is certain that no
17918 more data will come in. This test was designed to be used with TCP response
Christopher Faulete596d182020-05-05 17:46:34 +020017919 content inspection. But it may also be used in tcp-check based expect rules.
Willy Tarreau47e8eba2013-09-11 23:28:46 +020017920
Willy Tarreau74ca5042013-06-11 23:12:07 +020017921res.payload(<offset>,<length>) : binary
17922 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020017923 in the response buffer. As a special case, if the <length> argument is zero,
Christopher Faulete596d182020-05-05 17:46:34 +020017924 the whole buffer from <offset> to the end is extracted. This can be used
Willy Tarreau00f00842013-08-02 11:07:32 +020017925 with ACLs in order to check for the presence of some content in a buffer at
Christopher Faulete596d182020-05-05 17:46:34 +020017926 any location. It may also be used in tcp-check based expect rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +010017927
Willy Tarreau74ca5042013-06-11 23:12:07 +020017928res.payload_lv(<offset1>,<length>[,<offset2>]) : binary
17929 This extracts a binary block whose size is specified at <offset1> for <length>
17930 bytes, and which starts at <offset2> if specified or just after the length in
17931 the response buffer. The <offset2> parameter also supports relative offsets
Christopher Faulete596d182020-05-05 17:46:34 +020017932 if prepended with a '+' or '-' sign. It may also be used in tcp-check based
17933 expect rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +010017934
Willy Tarreau74ca5042013-06-11 23:12:07 +020017935 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaud63335a2010-02-26 12:56:52 +010017936
Willy Tarreau971f7b62015-09-29 14:06:59 +020017937res.ssl_hello_type : integer
17938rep_ssl_hello_type : integer (deprecated)
17939 Returns an integer value containing the type of the SSL hello message found
17940 in the response buffer if the buffer contains data that parses as a complete
17941 SSL (v3 or superior) hello message. Note that this only applies to raw
17942 contents found in the response buffer and not to contents deciphered via an
17943 SSL data layer, so this will not work with "server" lines having the "ssl"
17944 option. This is mostly used in ACL to detect presence of an SSL hello message
17945 that is supposed to contain an SSL session ID usable for stickiness.
17946
Willy Tarreau74ca5042013-06-11 23:12:07 +020017947wait_end : boolean
17948 This fetch either returns true when the inspection period is over, or does
17949 not fetch. It is only used in ACLs, in conjunction with content analysis to
Davor Ocelice9ed2812017-12-25 17:49:28 +010017950 avoid returning a wrong verdict early. It may also be used to delay some
Willy Tarreau74ca5042013-06-11 23:12:07 +020017951 actions, such as a delayed reject for some special addresses. Since it either
17952 stops the rules evaluation or immediately returns true, it is recommended to
Davor Ocelice9ed2812017-12-25 17:49:28 +010017953 use this acl as the last one in a rule. Please note that the default ACL
Willy Tarreau74ca5042013-06-11 23:12:07 +020017954 "WAIT_END" is always usable without prior declaration. This test was designed
17955 to be used with TCP request content inspection.
Willy Tarreaud63335a2010-02-26 12:56:52 +010017956
Willy Tarreau74ca5042013-06-11 23:12:07 +020017957 Examples :
17958 # delay every incoming request by 2 seconds
17959 tcp-request inspect-delay 2s
17960 tcp-request content accept if WAIT_END
Willy Tarreaud63335a2010-02-26 12:56:52 +010017961
Willy Tarreau74ca5042013-06-11 23:12:07 +020017962 # don't immediately tell bad guys they are rejected
17963 tcp-request inspect-delay 10s
17964 acl goodguys src 10.0.0.0/24
17965 acl badguys src 10.0.1.0/24
17966 tcp-request content accept if goodguys
17967 tcp-request content reject if badguys WAIT_END
17968 tcp-request content reject
17969
17970
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200179717.3.6. Fetching HTTP samples (Layer 7)
Willy Tarreau74ca5042013-06-11 23:12:07 +020017972--------------------------------------
17973
17974It is possible to fetch samples from HTTP contents, requests and responses.
17975This application layer is also called layer 7. It is only possible to fetch the
17976data in this section when a full HTTP request or response has been parsed from
17977its respective request or response buffer. This is always the case with all
17978HTTP specific rules and for sections running with "mode http". When using TCP
17979content inspection, it may be necessary to support an inspection delay in order
17980to let the request or response come in first. These fetches may require a bit
17981more CPU resources than the layer 4 ones, but not much since the request and
17982response are indexed.
17983
17984base : string
17985 This returns the concatenation of the first Host header and the path part of
17986 the request, which starts at the first slash and ends before the question
17987 mark. It can be useful in virtual hosted environments to detect URL abuses as
17988 well as to improve shared caches efficiency. Using this with a limited size
17989 stick table also allows one to collect statistics about most commonly
17990 requested objects by host/path. With ACLs it can allow simple content
17991 switching rules involving the host and the path at the same time, such as
17992 "www.example.com/favicon.ico". See also "path" and "uri".
17993
17994 ACL derivatives :
17995 base : exact string match
17996 base_beg : prefix match
17997 base_dir : subdir match
17998 base_dom : domain match
17999 base_end : suffix match
18000 base_len : length match
18001 base_reg : regex match
18002 base_sub : substring match
18003
18004base32 : integer
18005 This returns a 32-bit hash of the value returned by the "base" fetch method
18006 above. This is useful to track per-URL activity on high traffic sites without
18007 having to store all URLs. Instead a shorter hash is stored, saving a lot of
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020018008 memory. The output type is an unsigned integer. The hash function used is
18009 SDBM with full avalanche on the output. Technically, base32 is exactly equal
18010 to "base,sdbm(1)".
Willy Tarreau74ca5042013-06-11 23:12:07 +020018011
18012base32+src : binary
18013 This returns the concatenation of the base32 fetch above and the src fetch
18014 below. The resulting type is of type binary, with a size of 8 or 20 bytes
18015 depending on the source address family. This can be used to track per-IP,
18016 per-URL counters.
18017
William Lallemand65ad6e12014-01-31 15:08:02 +010018018capture.req.hdr(<idx>) : string
18019 This extracts the content of the header captured by the "capture request
18020 header", idx is the position of the capture keyword in the configuration.
18021 The first entry is an index of 0. See also: "capture request header".
18022
18023capture.req.method : string
18024 This extracts the METHOD of an HTTP request. It can be used in both request
18025 and response. Unlike "method", it can be used in both request and response
18026 because it's allocated.
18027
18028capture.req.uri : string
18029 This extracts the request's URI, which starts at the first slash and ends
18030 before the first space in the request (without the host part). Unlike "path"
18031 and "url", it can be used in both request and response because it's
18032 allocated.
18033
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020018034capture.req.ver : string
18035 This extracts the request's HTTP version and returns either "HTTP/1.0" or
18036 "HTTP/1.1". Unlike "req.ver", it can be used in both request, response, and
18037 logs because it relies on a persistent flag.
18038
William Lallemand65ad6e12014-01-31 15:08:02 +010018039capture.res.hdr(<idx>) : string
18040 This extracts the content of the header captured by the "capture response
18041 header", idx is the position of the capture keyword in the configuration.
18042 The first entry is an index of 0.
18043 See also: "capture response header"
18044
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020018045capture.res.ver : string
18046 This extracts the response's HTTP version and returns either "HTTP/1.0" or
18047 "HTTP/1.1". Unlike "res.ver", it can be used in logs because it relies on a
18048 persistent flag.
18049
Willy Tarreaua5910cc2015-05-02 00:46:08 +020018050req.body : binary
Christopher Fauletaf4dc4c2020-05-05 17:33:25 +020018051 This returns the HTTP request's available body as a block of data. It is
18052 recommended to use "option http-buffer-request" to be sure to wait, as much
18053 as possible, for the request's body.
Willy Tarreaua5910cc2015-05-02 00:46:08 +020018054
Thierry FOURNIER9826c772015-05-20 15:50:54 +020018055req.body_param([<name>) : string
18056 This fetch assumes that the body of the POST request is url-encoded. The user
18057 can check if the "content-type" contains the value
18058 "application/x-www-form-urlencoded". This extracts the first occurrence of the
18059 parameter <name> in the body, which ends before '&'. The parameter name is
18060 case-sensitive. If no name is given, any parameter will match, and the first
18061 one will be returned. The result is a string corresponding to the value of the
18062 parameter <name> as presented in the request body (no URL decoding is
18063 performed). Note that the ACL version of this fetch iterates over multiple
18064 parameters and will iteratively report all parameters values if no name is
18065 given.
18066
Willy Tarreaua5910cc2015-05-02 00:46:08 +020018067req.body_len : integer
18068 This returns the length of the HTTP request's available body in bytes. It may
18069 be lower than the advertised length if the body is larger than the buffer. It
Christopher Fauletaf4dc4c2020-05-05 17:33:25 +020018070 is recommended to use "option http-buffer-request" to be sure to wait, as
18071 much as possible, for the request's body.
Willy Tarreaua5910cc2015-05-02 00:46:08 +020018072
18073req.body_size : integer
18074 This returns the advertised length of the HTTP request's body in bytes. It
Christopher Fauletaf4dc4c2020-05-05 17:33:25 +020018075 will represent the advertised Content-Length header, or the size of the
18076 available data in case of chunked encoding.
Willy Tarreaua5910cc2015-05-02 00:46:08 +020018077
Willy Tarreau74ca5042013-06-11 23:12:07 +020018078req.cook([<name>]) : string
18079cook([<name>]) : string (deprecated)
18080 This extracts the last occurrence of the cookie name <name> on a "Cookie"
18081 header line from the request, and returns its value as string. If no name is
18082 specified, the first cookie value is returned. When used with ACLs, all
18083 matching cookies are evaluated. Spaces around the name and the value are
18084 ignored as requested by the Cookie header specification (RFC6265). The cookie
18085 name is case-sensitive. Empty cookies are valid, so an empty cookie may very
18086 well return an empty value if it is present. Use the "found" match to detect
18087 presence. Use the res.cook() variant for response cookies sent by the server.
18088
18089 ACL derivatives :
18090 cook([<name>]) : exact string match
18091 cook_beg([<name>]) : prefix match
18092 cook_dir([<name>]) : subdir match
18093 cook_dom([<name>]) : domain match
18094 cook_end([<name>]) : suffix match
18095 cook_len([<name>]) : length match
18096 cook_reg([<name>]) : regex match
18097 cook_sub([<name>]) : substring match
Willy Tarreaud63335a2010-02-26 12:56:52 +010018098
Willy Tarreau74ca5042013-06-11 23:12:07 +020018099req.cook_cnt([<name>]) : integer
18100cook_cnt([<name>]) : integer (deprecated)
18101 Returns an integer value representing the number of occurrences of the cookie
18102 <name> in the request, or all cookies if <name> is not specified.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018103
Willy Tarreau74ca5042013-06-11 23:12:07 +020018104req.cook_val([<name>]) : integer
18105cook_val([<name>]) : integer (deprecated)
18106 This extracts the last occurrence of the cookie name <name> on a "Cookie"
18107 header line from the request, and converts its value to an integer which is
18108 returned. If no name is specified, the first cookie value is returned. When
18109 used in ACLs, all matching names are iterated over until a value matches.
Willy Tarreau0e698542011-09-16 08:32:32 +020018110
Willy Tarreau74ca5042013-06-11 23:12:07 +020018111cookie([<name>]) : string (deprecated)
18112 This extracts the last occurrence of the cookie name <name> on a "Cookie"
18113 header line from the request, or a "Set-Cookie" header from the response, and
18114 returns its value as a string. A typical use is to get multiple clients
18115 sharing a same profile use the same server. This can be similar to what
Willy Tarreau294d0f02015-08-10 19:40:12 +020018116 "appsession" did with the "request-learn" statement, but with support for
Willy Tarreau74ca5042013-06-11 23:12:07 +020018117 multi-peer synchronization and state keeping across restarts. If no name is
18118 specified, the first cookie value is returned. This fetch should not be used
18119 anymore and should be replaced by req.cook() or res.cook() instead as it
18120 ambiguously uses the direction based on the context where it is used.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018121
Willy Tarreau74ca5042013-06-11 23:12:07 +020018122hdr([<name>[,<occ>]]) : string
18123 This is equivalent to req.hdr() when used on requests, and to res.hdr() when
18124 used on responses. Please refer to these respective fetches for more details.
18125 In case of doubt about the fetch direction, please use the explicit ones.
18126 Note that contrary to the hdr() sample fetch method, the hdr_* ACL keywords
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030018127 unambiguously apply to the request headers.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018128
Willy Tarreau74ca5042013-06-11 23:12:07 +020018129req.fhdr(<name>[,<occ>]) : string
18130 This extracts the last occurrence of header <name> in an HTTP request. When
18131 used from an ACL, all occurrences are iterated over until a match is found.
18132 Optionally, a specific occurrence might be specified as a position number.
18133 Positive values indicate a position from the first occurrence, with 1 being
18134 the first one. Negative values indicate positions relative to the last one,
18135 with -1 being the last one. It differs from req.hdr() in that any commas
18136 present in the value are returned and are not used as delimiters. This is
18137 sometimes useful with headers such as User-Agent.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018138
Willy Tarreau74ca5042013-06-11 23:12:07 +020018139req.fhdr_cnt([<name>]) : integer
18140 Returns an integer value representing the number of occurrences of request
18141 header field name <name>, or the total number of header fields if <name> is
18142 not specified. Contrary to its req.hdr_cnt() cousin, this function returns
18143 the number of full line headers and does not stop on commas.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018144
Willy Tarreau74ca5042013-06-11 23:12:07 +020018145req.hdr([<name>[,<occ>]]) : string
18146 This extracts the last occurrence of header <name> in an HTTP request. When
18147 used from an ACL, all occurrences are iterated over until a match is found.
18148 Optionally, a specific occurrence might be specified as a position number.
18149 Positive values indicate a position from the first occurrence, with 1 being
18150 the first one. Negative values indicate positions relative to the last one,
18151 with -1 being the last one. A typical use is with the X-Forwarded-For header
18152 once converted to IP, associated with an IP stick-table. The function
18153 considers any comma as a delimiter for distinct values. If full-line headers
Lukas Tribus23953682017-04-28 13:24:30 +000018154 are desired instead, use req.fhdr(). Please carefully check RFC7231 to know
Willy Tarreau74ca5042013-06-11 23:12:07 +020018155 how certain headers are supposed to be parsed. Also, some of them are case
Davor Ocelice9ed2812017-12-25 17:49:28 +010018156 insensitive (e.g. Connection).
Willy Tarreaud63335a2010-02-26 12:56:52 +010018157
Willy Tarreau74ca5042013-06-11 23:12:07 +020018158 ACL derivatives :
18159 hdr([<name>[,<occ>]]) : exact string match
18160 hdr_beg([<name>[,<occ>]]) : prefix match
18161 hdr_dir([<name>[,<occ>]]) : subdir match
18162 hdr_dom([<name>[,<occ>]]) : domain match
18163 hdr_end([<name>[,<occ>]]) : suffix match
18164 hdr_len([<name>[,<occ>]]) : length match
18165 hdr_reg([<name>[,<occ>]]) : regex match
18166 hdr_sub([<name>[,<occ>]]) : substring match
18167
18168req.hdr_cnt([<name>]) : integer
18169hdr_cnt([<header>]) : integer (deprecated)
18170 Returns an integer value representing the number of occurrences of request
18171 header field name <name>, or the total number of header field values if
18172 <name> is not specified. It is important to remember that one header line may
18173 count as several headers if it has several values. The function considers any
18174 comma as a delimiter for distinct values. If full-line headers are desired
18175 instead, req.fhdr_cnt() should be used instead. With ACLs, it can be used to
18176 detect presence, absence or abuse of a specific header, as well as to block
18177 request smuggling attacks by rejecting requests which contain more than one
18178 of certain headers. See "req.hdr" for more information on header matching.
18179
18180req.hdr_ip([<name>[,<occ>]]) : ip
18181hdr_ip([<name>[,<occ>]]) : ip (deprecated)
18182 This extracts the last occurrence of header <name> in an HTTP request,
18183 converts it to an IPv4 or IPv6 address and returns this address. When used
18184 with ACLs, all occurrences are checked, and if <name> is omitted, every value
18185 of every header is checked. Optionally, a specific occurrence might be
18186 specified as a position number. Positive values indicate a position from the
Davor Ocelice9ed2812017-12-25 17:49:28 +010018187 first occurrence, with 1 being the first one. Negative values indicate
Willy Tarreau74ca5042013-06-11 23:12:07 +020018188 positions relative to the last one, with -1 being the last one. A typical use
18189 is with the X-Forwarded-For and X-Client-IP headers.
18190
18191req.hdr_val([<name>[,<occ>]]) : integer
18192hdr_val([<name>[,<occ>]]) : integer (deprecated)
18193 This extracts the last occurrence of header <name> in an HTTP request, and
18194 converts it to an integer value. When used with ACLs, all occurrences are
18195 checked, and if <name> is omitted, every value of every header is checked.
18196 Optionally, a specific occurrence might be specified as a position number.
18197 Positive values indicate a position from the first occurrence, with 1 being
18198 the first one. Negative values indicate positions relative to the last one,
18199 with -1 being the last one. A typical use is with the X-Forwarded-For header.
18200
Frédéric Lécailleec891192019-02-26 15:02:35 +010018201
18202
Willy Tarreau74ca5042013-06-11 23:12:07 +020018203http_auth(<userlist>) : boolean
18204 Returns a boolean indicating whether the authentication data received from
18205 the client match a username & password stored in the specified userlist. This
18206 fetch function is not really useful outside of ACLs. Currently only http
18207 basic auth is supported.
18208
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010018209http_auth_group(<userlist>) : string
18210 Returns a string corresponding to the user name found in the authentication
18211 data received from the client if both the user name and password are valid
18212 according to the specified userlist. The main purpose is to use it in ACLs
18213 where it is then checked whether the user belongs to any group within a list.
Willy Tarreau74ca5042013-06-11 23:12:07 +020018214 This fetch function is not really useful outside of ACLs. Currently only http
18215 basic auth is supported.
18216
18217 ACL derivatives :
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010018218 http_auth_group(<userlist>) : group ...
18219 Returns true when the user extracted from the request and whose password is
18220 valid according to the specified userlist belongs to at least one of the
18221 groups.
Willy Tarreau74ca5042013-06-11 23:12:07 +020018222
Christopher Fauleta4063562019-08-02 11:51:37 +020018223http_auth_pass : string
Willy Tarreauc9c6cdb2020-03-05 16:03:58 +010018224 Returns the user's password found in the authentication data received from
18225 the client, as supplied in the Authorization header. Not checks are
18226 performed by this sample fetch. Only Basic authentication is supported.
Christopher Fauleta4063562019-08-02 11:51:37 +020018227
18228http_auth_type : string
Willy Tarreauc9c6cdb2020-03-05 16:03:58 +010018229 Returns the authentication method found in the authentication data received from
18230 the client, as supplied in the Authorization header. Not checks are
18231 performed by this sample fetch. Only Basic authentication is supported.
Christopher Fauleta4063562019-08-02 11:51:37 +020018232
18233http_auth_user : string
Willy Tarreauc9c6cdb2020-03-05 16:03:58 +010018234 Returns the user name found in the authentication data received from the
18235 client, as supplied in the Authorization header. Not checks are performed by
18236 this sample fetch. Only Basic authentication is supported.
Christopher Fauleta4063562019-08-02 11:51:37 +020018237
Willy Tarreau74ca5042013-06-11 23:12:07 +020018238http_first_req : boolean
Willy Tarreau7f18e522010-10-22 20:04:13 +020018239 Returns true when the request being processed is the first one of the
18240 connection. This can be used to add or remove headers that may be missing
Willy Tarreau74ca5042013-06-11 23:12:07 +020018241 from some requests when a request is not the first one, or to help grouping
18242 requests in the logs.
Willy Tarreau7f18e522010-10-22 20:04:13 +020018243
Willy Tarreau74ca5042013-06-11 23:12:07 +020018244method : integer + string
18245 Returns an integer value corresponding to the method in the HTTP request. For
18246 example, "GET" equals 1 (check sources to establish the matching). Value 9
18247 means "other method" and may be converted to a string extracted from the
18248 stream. This should not be used directly as a sample, this is only meant to
18249 be used from ACLs, which transparently convert methods from patterns to these
18250 integer + string values. Some predefined ACL already check for most common
18251 methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020018252
Willy Tarreau74ca5042013-06-11 23:12:07 +020018253 ACL derivatives :
18254 method : case insensitive method match
Willy Tarreau6a06a402007-07-15 20:15:28 +020018255
Willy Tarreau74ca5042013-06-11 23:12:07 +020018256 Example :
18257 # only accept GET and HEAD requests
18258 acl valid_method method GET HEAD
18259 http-request deny if ! valid_method
Willy Tarreau6a06a402007-07-15 20:15:28 +020018260
Willy Tarreau74ca5042013-06-11 23:12:07 +020018261path : string
18262 This extracts the request's URL path, which starts at the first slash and
18263 ends before the question mark (without the host part). A typical use is with
18264 prefetch-capable caches, and with portals which need to aggregate multiple
18265 information from databases and keep them in caches. Note that with outgoing
18266 caches, it would be wiser to use "url" instead. With ACLs, it's typically
Davor Ocelice9ed2812017-12-25 17:49:28 +010018267 used to match exact file names (e.g. "/login.php"), or directory parts using
Willy Tarreau74ca5042013-06-11 23:12:07 +020018268 the derivative forms. See also the "url" and "base" fetch methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020018269
Willy Tarreau74ca5042013-06-11 23:12:07 +020018270 ACL derivatives :
18271 path : exact string match
18272 path_beg : prefix match
18273 path_dir : subdir match
18274 path_dom : domain match
18275 path_end : suffix match
18276 path_len : length match
18277 path_reg : regex match
18278 path_sub : substring match
Willy Tarreau6a06a402007-07-15 20:15:28 +020018279
Christopher Faulete720c322020-09-02 17:25:18 +020018280pathq : string
18281 This extracts the request's URL path with the query-string, which starts at
18282 the first slash. This sample fetch is pretty handy to always retrieve a
18283 relative URI, excluding the scheme and the authority part, if any. Indeed,
18284 while it is the common representation for an HTTP/1.1 request target, in
18285 HTTP/2, an absolute URI is often used. This sample fetch will return the same
18286 result in both cases.
18287
Willy Tarreau49ad95c2015-01-19 15:06:26 +010018288query : string
18289 This extracts the request's query string, which starts after the first
18290 question mark. If no question mark is present, this fetch returns nothing. If
18291 a question mark is present but nothing follows, it returns an empty string.
18292 This means it's possible to easily know whether a query string is present
Tim Düsterhus4896c442016-11-29 02:15:19 +010018293 using the "found" matching method. This fetch is the complement of "path"
Willy Tarreau49ad95c2015-01-19 15:06:26 +010018294 which stops before the question mark.
18295
Willy Tarreaueb27ec72015-02-20 13:55:29 +010018296req.hdr_names([<delim>]) : string
18297 This builds a string made from the concatenation of all header names as they
18298 appear in the request when the rule is evaluated. The default delimiter is
18299 the comma (',') but it may be overridden as an optional argument <delim>. In
18300 this case, only the first character of <delim> is considered.
18301
Willy Tarreau74ca5042013-06-11 23:12:07 +020018302req.ver : string
18303req_ver : string (deprecated)
18304 Returns the version string from the HTTP request, for example "1.1". This can
18305 be useful for logs, but is mostly there for ACL. Some predefined ACL already
18306 check for versions 1.0 and 1.1.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018307
Willy Tarreau74ca5042013-06-11 23:12:07 +020018308 ACL derivatives :
18309 req_ver : exact string match
Willy Tarreau0e698542011-09-16 08:32:32 +020018310
Christopher Faulete596d182020-05-05 17:46:34 +020018311res.body : binary
18312 This returns the HTTP response's available body as a block of data. Unlike
18313 the request side, there is no directive to wait for the response's body. This
18314 sample fetch is really useful (and usable) in the health-check context. It
18315 may be used in tcp-check based expect rules.
18316
18317res.body_len : integer
18318 This returns the length of the HTTP response available body in bytes. Unlike
18319 the request side, there is no directive to wait for the response's body. This
18320 sample fetch is really useful (and usable) in the health-check context. It
18321 may be used in tcp-check based expect rules.
18322
18323res.body_size : integer
18324 This returns the advertised length of the HTTP response body in bytes. It
18325 will represent the advertised Content-Length header, or the size of the
18326 available data in case of chunked encoding. Unlike the request side, there is
18327 no directive to wait for the response body. This sample fetch is really
18328 useful (and usable) in the health-check context. It may be used in tcp-check
18329 based expect rules.
18330
Remi Tricot-Le Bretonbf971212020-10-27 11:55:57 +010018331res.cache_hit : boolean
18332 Returns the boolean "true" value if the response has been built out of an
18333 HTTP cache entry, otherwise returns boolean "false".
18334
18335res.cache_name : string
18336 Returns a string containing the name of the HTTP cache that was used to
18337 build the HTTP response if res.cache_hit is true, otherwise returns an
18338 empty string.
18339
Willy Tarreau74ca5042013-06-11 23:12:07 +020018340res.comp : boolean
18341 Returns the boolean "true" value if the response has been compressed by
18342 HAProxy, otherwise returns boolean "false". This may be used to add
18343 information in the logs.
Willy Tarreau6a06a402007-07-15 20:15:28 +020018344
Willy Tarreau74ca5042013-06-11 23:12:07 +020018345res.comp_algo : string
18346 Returns a string containing the name of the algorithm used if the response
18347 was compressed by HAProxy, for example : "deflate". This may be used to add
18348 some information in the logs.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018349
Willy Tarreau74ca5042013-06-11 23:12:07 +020018350res.cook([<name>]) : string
18351scook([<name>]) : string (deprecated)
18352 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
18353 header line from the response, and returns its value as string. If no name is
Christopher Faulete596d182020-05-05 17:46:34 +020018354 specified, the first cookie value is returned. It may be used in tcp-check
18355 based expect rules.
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020018356
Willy Tarreau74ca5042013-06-11 23:12:07 +020018357 ACL derivatives :
18358 scook([<name>] : exact string match
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020018359
Willy Tarreau74ca5042013-06-11 23:12:07 +020018360res.cook_cnt([<name>]) : integer
18361scook_cnt([<name>]) : integer (deprecated)
18362 Returns an integer value representing the number of occurrences of the cookie
18363 <name> in the response, or all cookies if <name> is not specified. This is
Christopher Faulete596d182020-05-05 17:46:34 +020018364 mostly useful when combined with ACLs to detect suspicious responses. It may
18365 be used in tcp-check based expect rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018366
Willy Tarreau74ca5042013-06-11 23:12:07 +020018367res.cook_val([<name>]) : integer
18368scook_val([<name>]) : integer (deprecated)
18369 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
18370 header line from the response, and converts its value to an integer which is
Christopher Faulete596d182020-05-05 17:46:34 +020018371 returned. If no name is specified, the first cookie value is returned. It may
18372 be used in tcp-check based expect rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018373
Willy Tarreau74ca5042013-06-11 23:12:07 +020018374res.fhdr([<name>[,<occ>]]) : string
18375 This extracts the last occurrence of header <name> in an HTTP response, or of
18376 the last header if no <name> is specified. Optionally, a specific occurrence
18377 might be specified as a position number. Positive values indicate a position
18378 from the first occurrence, with 1 being the first one. Negative values
18379 indicate positions relative to the last one, with -1 being the last one. It
18380 differs from res.hdr() in that any commas present in the value are returned
18381 and are not used as delimiters. If this is not desired, the res.hdr() fetch
18382 should be used instead. This is sometimes useful with headers such as Date or
Christopher Faulete596d182020-05-05 17:46:34 +020018383 Expires. It may be used in tcp-check based expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020018384
Willy Tarreau74ca5042013-06-11 23:12:07 +020018385res.fhdr_cnt([<name>]) : integer
18386 Returns an integer value representing the number of occurrences of response
18387 header field name <name>, or the total number of header fields if <name> is
18388 not specified. Contrary to its res.hdr_cnt() cousin, this function returns
18389 the number of full line headers and does not stop on commas. If this is not
Christopher Faulete596d182020-05-05 17:46:34 +020018390 desired, the res.hdr_cnt() fetch should be used instead. It may be used in
18391 tcp-check based expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020018392
Willy Tarreau74ca5042013-06-11 23:12:07 +020018393res.hdr([<name>[,<occ>]]) : string
18394shdr([<name>[,<occ>]]) : string (deprecated)
18395 This extracts the last occurrence of header <name> in an HTTP response, or of
18396 the last header if no <name> is specified. Optionally, a specific occurrence
18397 might be specified as a position number. Positive values indicate a position
18398 from the first occurrence, with 1 being the first one. Negative values
18399 indicate positions relative to the last one, with -1 being the last one. This
18400 can be useful to learn some data into a stick-table. The function considers
18401 any comma as a delimiter for distinct values. If this is not desired, the
Christopher Faulete596d182020-05-05 17:46:34 +020018402 res.fhdr() fetch should be used instead. It may be used in tcp-check based
18403 expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020018404
Willy Tarreau74ca5042013-06-11 23:12:07 +020018405 ACL derivatives :
18406 shdr([<name>[,<occ>]]) : exact string match
18407 shdr_beg([<name>[,<occ>]]) : prefix match
18408 shdr_dir([<name>[,<occ>]]) : subdir match
18409 shdr_dom([<name>[,<occ>]]) : domain match
18410 shdr_end([<name>[,<occ>]]) : suffix match
18411 shdr_len([<name>[,<occ>]]) : length match
18412 shdr_reg([<name>[,<occ>]]) : regex match
18413 shdr_sub([<name>[,<occ>]]) : substring match
18414
18415res.hdr_cnt([<name>]) : integer
18416shdr_cnt([<name>]) : integer (deprecated)
18417 Returns an integer value representing the number of occurrences of response
18418 header field name <name>, or the total number of header fields if <name> is
18419 not specified. The function considers any comma as a delimiter for distinct
18420 values. If this is not desired, the res.fhdr_cnt() fetch should be used
Christopher Faulete596d182020-05-05 17:46:34 +020018421 instead. It may be used in tcp-check based expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020018422
Willy Tarreau74ca5042013-06-11 23:12:07 +020018423res.hdr_ip([<name>[,<occ>]]) : ip
18424shdr_ip([<name>[,<occ>]]) : ip (deprecated)
18425 This extracts the last occurrence of header <name> in an HTTP response,
18426 convert it to an IPv4 or IPv6 address and returns this address. Optionally, a
18427 specific occurrence might be specified as a position number. Positive values
18428 indicate a position from the first occurrence, with 1 being the first one.
18429 Negative values indicate positions relative to the last one, with -1 being
Christopher Faulete596d182020-05-05 17:46:34 +020018430 the last one. This can be useful to learn some data into a stick table. It
18431 may be used in tcp-check based expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020018432
Willy Tarreaueb27ec72015-02-20 13:55:29 +010018433res.hdr_names([<delim>]) : string
18434 This builds a string made from the concatenation of all header names as they
18435 appear in the response when the rule is evaluated. The default delimiter is
18436 the comma (',') but it may be overridden as an optional argument <delim>. In
Christopher Faulete596d182020-05-05 17:46:34 +020018437 this case, only the first character of <delim> is considered. It may be used
18438 in tcp-check based expect rules.
Willy Tarreaueb27ec72015-02-20 13:55:29 +010018439
Willy Tarreau74ca5042013-06-11 23:12:07 +020018440res.hdr_val([<name>[,<occ>]]) : integer
18441shdr_val([<name>[,<occ>]]) : integer (deprecated)
18442 This extracts the last occurrence of header <name> in an HTTP response, and
18443 converts it to an integer value. Optionally, a specific occurrence might be
18444 specified as a position number. Positive values indicate a position from the
18445 first occurrence, with 1 being the first one. Negative values indicate
18446 positions relative to the last one, with -1 being the last one. This can be
Christopher Faulete596d182020-05-05 17:46:34 +020018447 useful to learn some data into a stick table. It may be used in tcp-check
18448 based expect rules.
18449
18450res.hdrs : string
18451 Returns the current response headers as string including the last empty line
18452 separating headers from the request body. The last empty line can be used to
18453 detect a truncated header block. This sample fetch is useful for some SPOE
18454 headers analyzers and for advanced logging. It may also be used in tcp-check
18455 based expect rules.
18456
18457res.hdrs_bin : binary
18458 Returns the current response headers contained in preparsed binary form. This
18459 is useful for offloading some processing with SPOE. It may be used in
18460 tcp-check based expect rules. Each string is described by a length followed
18461 by the number of bytes indicated in the length. The length is represented
18462 using the variable integer encoding detailed in the SPOE documentation. The
18463 end of the list is marked by a couple of empty header names and values
18464 (length of 0 for both).
18465
18466 *(<str:header-name><str:header-value>)<empty string><empty string>
18467
18468 int: refer to the SPOE documentation for the encoding
18469 str: <int:length><bytes>
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010018470
Willy Tarreau74ca5042013-06-11 23:12:07 +020018471res.ver : string
18472resp_ver : string (deprecated)
18473 Returns the version string from the HTTP response, for example "1.1". This
Christopher Faulete596d182020-05-05 17:46:34 +020018474 can be useful for logs, but is mostly there for ACL. It may be used in
18475 tcp-check based expect rules.
Willy Tarreau0e698542011-09-16 08:32:32 +020018476
Willy Tarreau74ca5042013-06-11 23:12:07 +020018477 ACL derivatives :
18478 resp_ver : exact string match
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010018479
Willy Tarreau74ca5042013-06-11 23:12:07 +020018480set-cookie([<name>]) : string (deprecated)
18481 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
18482 header line from the response and uses the corresponding value to match. This
Willy Tarreau294d0f02015-08-10 19:40:12 +020018483 can be comparable to what "appsession" did with default options, but with
Willy Tarreau74ca5042013-06-11 23:12:07 +020018484 support for multi-peer synchronization and state keeping across restarts.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010018485
Willy Tarreau74ca5042013-06-11 23:12:07 +020018486 This fetch function is deprecated and has been superseded by the "res.cook"
18487 fetch. This keyword will disappear soon.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010018488
Willy Tarreau74ca5042013-06-11 23:12:07 +020018489status : integer
18490 Returns an integer containing the HTTP status code in the HTTP response, for
18491 example, 302. It is mostly used within ACLs and integer ranges, for example,
Christopher Faulete596d182020-05-05 17:46:34 +020018492 to remove any Location header if the response is not a 3xx. It may be used in
18493 tcp-check based expect rules.
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018494
Thierry Fournier0e00dca2016-04-07 15:47:40 +020018495unique-id : string
18496 Returns the unique-id attached to the request. The directive
18497 "unique-id-format" must be set. If it is not set, the unique-id sample fetch
18498 fails. Note that the unique-id is usually used with HTTP requests, however this
18499 sample fetch can be used with other protocols. Obviously, if it is used with
18500 other protocols than HTTP, the unique-id-format directive must not contain
18501 HTTP parts. See: unique-id-format and unique-id-header
18502
Willy Tarreau74ca5042013-06-11 23:12:07 +020018503url : string
18504 This extracts the request's URL as presented in the request. A typical use is
18505 with prefetch-capable caches, and with portals which need to aggregate
18506 multiple information from databases and keep them in caches. With ACLs, using
18507 "path" is preferred over using "url", because clients may send a full URL as
18508 is normally done with proxies. The only real use is to match "*" which does
18509 not match in "path", and for which there is already a predefined ACL. See
18510 also "path" and "base".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018511
Willy Tarreau74ca5042013-06-11 23:12:07 +020018512 ACL derivatives :
18513 url : exact string match
18514 url_beg : prefix match
18515 url_dir : subdir match
18516 url_dom : domain match
18517 url_end : suffix match
18518 url_len : length match
18519 url_reg : regex match
18520 url_sub : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018521
Willy Tarreau74ca5042013-06-11 23:12:07 +020018522url_ip : ip
18523 This extracts the IP address from the request's URL when the host part is
18524 presented as an IP address. Its use is very limited. For instance, a
18525 monitoring system might use this field as an alternative for the source IP in
18526 order to test what path a given source address would follow, or to force an
18527 entry in a table for a given source address. With ACLs it can be used to
18528 restrict access to certain systems through a proxy, for example when combined
18529 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018530
Willy Tarreau74ca5042013-06-11 23:12:07 +020018531url_port : integer
18532 This extracts the port part from the request's URL. Note that if the port is
18533 not specified in the request, port 80 is assumed. With ACLs it can be used to
18534 restrict access to certain systems through a proxy, for example when combined
18535 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018536
Willy Tarreau1ede1da2015-05-07 16:06:18 +020018537urlp([<name>[,<delim>]]) : string
18538url_param([<name>[,<delim>]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020018539 This extracts the first occurrence of the parameter <name> in the query
18540 string, which begins after either '?' or <delim>, and which ends before '&',
Willy Tarreau1ede1da2015-05-07 16:06:18 +020018541 ';' or <delim>. The parameter name is case-sensitive. If no name is given,
18542 any parameter will match, and the first one will be returned. The result is
18543 a string corresponding to the value of the parameter <name> as presented in
18544 the request (no URL decoding is performed). This can be used for session
Willy Tarreau74ca5042013-06-11 23:12:07 +020018545 stickiness based on a client ID, to extract an application cookie passed as a
18546 URL parameter, or in ACLs to apply some checks. Note that the ACL version of
Willy Tarreau1ede1da2015-05-07 16:06:18 +020018547 this fetch iterates over multiple parameters and will iteratively report all
18548 parameters values if no name is given
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018549
Willy Tarreau74ca5042013-06-11 23:12:07 +020018550 ACL derivatives :
18551 urlp(<name>[,<delim>]) : exact string match
18552 urlp_beg(<name>[,<delim>]) : prefix match
18553 urlp_dir(<name>[,<delim>]) : subdir match
18554 urlp_dom(<name>[,<delim>]) : domain match
18555 urlp_end(<name>[,<delim>]) : suffix match
18556 urlp_len(<name>[,<delim>]) : length match
18557 urlp_reg(<name>[,<delim>]) : regex match
18558 urlp_sub(<name>[,<delim>]) : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018559
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018560
Willy Tarreau74ca5042013-06-11 23:12:07 +020018561 Example :
18562 # match http://example.com/foo?PHPSESSIONID=some_id
18563 stick on urlp(PHPSESSIONID)
18564 # match http://example.com/foo;JSESSIONID=some_id
18565 stick on urlp(JSESSIONID,;)
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018566
Jarno Huuskonen676f6222017-03-30 09:19:45 +030018567urlp_val([<name>[,<delim>]]) : integer
Willy Tarreau74ca5042013-06-11 23:12:07 +020018568 See "urlp" above. This one extracts the URL parameter <name> in the request
18569 and converts it to an integer value. This can be used for session stickiness
18570 based on a user ID for example, or with ACLs to match a page number or price.
Willy Tarreaua9fddca2012-07-31 07:51:48 +020018571
Dragan Dosen0070cd52016-06-16 12:19:49 +020018572url32 : integer
18573 This returns a 32-bit hash of the value obtained by concatenating the first
18574 Host header and the whole URL including parameters (not only the path part of
18575 the request, as in the "base32" fetch above). This is useful to track per-URL
18576 activity. A shorter hash is stored, saving a lot of memory. The output type
18577 is an unsigned integer.
18578
18579url32+src : binary
18580 This returns the concatenation of the "url32" fetch and the "src" fetch. The
18581 resulting type is of type binary, with a size of 8 or 20 bytes depending on
18582 the source address family. This can be used to track per-IP, per-URL counters.
18583
Christopher Faulet16032ab2020-04-30 11:30:00 +020018584
Christopher Faulete596d182020-05-05 17:46:34 +0200185857.3.7. Fetching samples for developers
Christopher Fauletd47941d2020-01-08 14:40:19 +010018586---------------------------------------
18587
18588This set of sample fetch methods is reserved to developers and must never be
18589used on a production environment, except on developer demand, for debugging
18590purposes. Moreover, no special care will be taken on backwards compatibility.
18591There is no warranty the following sample fetches will never change, be renamed
18592or simply removed. So be really careful if you should use one of them. To avoid
18593any ambiguity, these sample fetches are placed in the dedicated scope "internal",
18594for instance "internal.strm.is_htx".
18595
18596internal.htx.data : integer
18597 Returns the size in bytes used by data in the HTX message associated to a
18598 channel. The channel is chosen depending on the sample direction.
18599
18600internal.htx.free : integer
18601 Returns the free space (size - used) in bytes in the HTX message associated
18602 to a channel. The channel is chosen depending on the sample direction.
18603
18604internal.htx.free_data : integer
18605 Returns the free space for the data in bytes in the HTX message associated to
18606 a channel. The channel is chosen depending on the sample direction.
18607
18608internal.htx.has_eom : boolean
18609 Returns true if the HTX message associated to a channel contains an
18610 end-of-message block (EOM). Otherwise, it returns false. The channel is
18611 chosen depending on the sample direction.
18612
18613internal.htx.nbblks : integer
18614 Returns the number of blocks present in the HTX message associated to a
18615 channel. The channel is chosen depending on the sample direction.
18616
18617internal.htx.size : integer
18618 Returns the total size in bytes of the HTX message associated to a
18619 channel. The channel is chosen depending on the sample direction.
18620
18621internal.htx.used : integer
18622 Returns the total size used in bytes (data + metadata) in the HTX message
18623 associated to a channel. The channel is chosen depending on the sample
18624 direction.
18625
18626internal.htx_blk.size(<idx>) : integer
18627 Returns the size of the block at the position <idx> in the HTX message
18628 associated to a channel or 0 if it does not exist. The channel is chosen
18629 depending on the sample direction. <idx> may be any positive integer or one
18630 of the special value :
18631 * head : The oldest inserted block
18632 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050018633 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010018634
18635internal.htx_blk.type(<idx>) : string
18636 Returns the type of the block at the position <idx> in the HTX message
18637 associated to a channel or "HTX_BLK_UNUSED" if it does not exist. The channel
18638 is chosen depending on the sample direction. <idx> may be any positive
18639 integer or one of the special value :
18640 * head : The oldest inserted block
18641 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050018642 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010018643
18644internal.htx_blk.data(<idx>) : binary
18645 Returns the value of the DATA block at the position <idx> in the HTX message
18646 associated to a channel or an empty string if it does not exist or if it is
18647 not a DATA block. The channel is chosen depending on the sample direction.
18648 <idx> may be any positive integer or one of the special value :
18649
18650 * head : The oldest inserted block
18651 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050018652 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010018653
18654internal.htx_blk.hdrname(<idx>) : string
18655 Returns the header name of the HEADER block at the position <idx> in the HTX
18656 message associated to a channel or an empty string if it does not exist or if
18657 it is not an HEADER block. The channel is chosen depending on the sample
18658 direction. <idx> may be any positive integer or one of the special value :
18659
18660 * head : The oldest inserted block
18661 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050018662 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010018663
18664internal.htx_blk.hdrval(<idx>) : string
18665 Returns the header value of the HEADER block at the position <idx> in the HTX
18666 message associated to a channel or an empty string if it does not exist or if
18667 it is not an HEADER block. The channel is chosen depending on the sample
18668 direction. <idx> may be any positive integer or one of the special value :
18669
18670 * head : The oldest inserted block
18671 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050018672 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010018673
18674internal.htx_blk.start_line(<idx>) : string
18675 Returns the value of the REQ_SL or RES_SL block at the position <idx> in the
18676 HTX message associated to a channel or an empty string if it does not exist
18677 or if it is not a SL block. The channel is chosen depending on the sample
18678 direction. <idx> may be any positive integer or one of the special value :
18679
18680 * head : The oldest inserted block
18681 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050018682 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010018683
18684internal.strm.is_htx : boolean
18685 Returns true if the current stream is an HTX stream. It means the data in the
18686 channels buffers are stored using the internal HTX representation. Otherwise,
18687 it returns false.
18688
18689
Willy Tarreau74ca5042013-06-11 23:12:07 +0200186907.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018691---------------------
Willy Tarreauced27012008-01-17 20:35:34 +010018692
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018693Some predefined ACLs are hard-coded so that they do not have to be declared in
18694every frontend which needs them. They all have their names in upper case in
Patrick Mézard2382ad62010-05-09 10:43:32 +020018695order to avoid confusion. Their equivalence is provided below.
Willy Tarreauced27012008-01-17 20:35:34 +010018696
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018697ACL name Equivalent to Usage
18698---------------+-----------------------------+---------------------------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018699FALSE always_false never match
Willy Tarreau2492d5b2009-07-11 00:06:00 +020018700HTTP req_proto_http match if protocol is valid HTTP
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018701HTTP_1.0 req_ver 1.0 match HTTP version 1.0
18702HTTP_1.1 req_ver 1.1 match HTTP version 1.1
Willy Tarreaud63335a2010-02-26 12:56:52 +010018703HTTP_CONTENT hdr_val(content-length) gt 0 match an existing content-length
18704HTTP_URL_ABS url_reg ^[^/:]*:// match absolute URL with scheme
18705HTTP_URL_SLASH url_beg / match URL beginning with "/"
18706HTTP_URL_STAR url * match URL equal to "*"
18707LOCALHOST src 127.0.0.1/8 match connection from local host
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018708METH_CONNECT method CONNECT match HTTP CONNECT method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020018709METH_DELETE method DELETE match HTTP DELETE method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018710METH_GET method GET HEAD match HTTP GET or HEAD method
18711METH_HEAD method HEAD match HTTP HEAD method
18712METH_OPTIONS method OPTIONS match HTTP OPTIONS method
18713METH_POST method POST match HTTP POST method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020018714METH_PUT method PUT match HTTP PUT method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018715METH_TRACE method TRACE match HTTP TRACE method
Emeric Brunbede3d02009-06-30 17:54:00 +020018716RDP_COOKIE req_rdp_cookie_cnt gt 0 match presence of an RDP cookie
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018717REQ_CONTENT req_len gt 0 match data in the request buffer
Willy Tarreaud63335a2010-02-26 12:56:52 +010018718TRUE always_true always match
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018719WAIT_END wait_end wait for end of content analysis
18720---------------+-----------------------------+---------------------------------
Willy Tarreauced27012008-01-17 20:35:34 +010018721
Willy Tarreaub937b7e2010-01-12 15:27:54 +010018722
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200187238. Logging
18724----------
Willy Tarreau844e3c52008-01-11 16:28:18 +010018725
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018726One of HAProxy's strong points certainly lies is its precise logs. It probably
18727provides the finest level of information available for such a product, which is
18728very important for troubleshooting complex environments. Standard information
18729provided in logs include client ports, TCP/HTTP state timers, precise session
18730state at termination and precise termination cause, information about decisions
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010018731to direct traffic to a server, and of course the ability to capture arbitrary
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018732headers.
18733
18734In order to improve administrators reactivity, it offers a great transparency
18735about encountered problems, both internal and external, and it is possible to
18736send logs to different sources at the same time with different level filters :
18737
18738 - global process-level logs (system errors, start/stop, etc..)
18739 - per-instance system and internal errors (lack of resource, bugs, ...)
18740 - per-instance external troubles (servers up/down, max connections)
18741 - per-instance activity (client connections), either at the establishment or
18742 at the termination.
Davor Ocelice9ed2812017-12-25 17:49:28 +010018743 - per-request control of log-level, e.g.
Jim Freeman9e8714b2015-05-26 09:16:34 -060018744 http-request set-log-level silent if sensitive_request
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018745
18746The ability to distribute different levels of logs to different log servers
18747allow several production teams to interact and to fix their problems as soon
18748as possible. For example, the system team might monitor system-wide errors,
18749while the application team might be monitoring the up/down for their servers in
18750real time, and the security team might analyze the activity logs with one hour
18751delay.
18752
18753
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200187548.1. Log levels
18755---------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018756
Simon Hormandf791f52011-05-29 15:01:10 +090018757TCP and HTTP connections can be logged with information such as the date, time,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018758source IP address, destination address, connection duration, response times,
Simon Hormandf791f52011-05-29 15:01:10 +090018759HTTP request, HTTP return code, number of bytes transmitted, conditions
18760in which the session ended, and even exchanged cookies values. For example
18761track a particular user's problems. All messages may be sent to up to two
18762syslog servers. Check the "log" keyword in section 4.2 for more information
18763about log facilities.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018764
18765
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200187668.2. Log formats
18767----------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018768
William Lallemand48940402012-01-30 16:47:22 +010018769HAProxy supports 5 log formats. Several fields are common between these formats
Simon Hormandf791f52011-05-29 15:01:10 +090018770and will be detailed in the following sections. A few of them may vary
18771slightly with the configuration, due to indicators specific to certain
18772options. The supported formats are as follows :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018773
18774 - the default format, which is very basic and very rarely used. It only
18775 provides very basic information about the incoming connection at the moment
18776 it is accepted : source IP:port, destination IP:port, and frontend-name.
18777 This mode will eventually disappear so it will not be described to great
18778 extents.
18779
18780 - the TCP format, which is more advanced. This format is enabled when "option
18781 tcplog" is set on the frontend. HAProxy will then usually wait for the
18782 connection to terminate before logging. This format provides much richer
18783 information, such as timers, connection counts, queue size, etc... This
18784 format is recommended for pure TCP proxies.
18785
18786 - the HTTP format, which is the most advanced for HTTP proxying. This format
18787 is enabled when "option httplog" is set on the frontend. It provides the
18788 same information as the TCP format with some HTTP-specific fields such as
18789 the request, the status code, and captures of headers and cookies. This
18790 format is recommended for HTTP proxies.
18791
Emeric Brun3a058f32009-06-30 18:26:00 +020018792 - the CLF HTTP format, which is equivalent to the HTTP format, but with the
18793 fields arranged in the same order as the CLF format. In this mode, all
18794 timers, captures, flags, etc... appear one per field after the end of the
18795 common fields, in the same order they appear in the standard HTTP format.
18796
William Lallemand48940402012-01-30 16:47:22 +010018797 - the custom log format, allows you to make your own log line.
18798
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018799Next sections will go deeper into details for each of these formats. Format
18800specification will be performed on a "field" basis. Unless stated otherwise, a
18801field is a portion of text delimited by any number of spaces. Since syslog
18802servers are susceptible of inserting fields at the beginning of a line, it is
18803always assumed that the first field is the one containing the process name and
18804identifier.
18805
18806Note : Since log lines may be quite long, the log examples in sections below
18807 might be broken into multiple lines. The example log lines will be
18808 prefixed with 3 closing angle brackets ('>>>') and each time a log is
18809 broken into multiple lines, each non-final line will end with a
18810 backslash ('\') and the next line will start indented by two characters.
18811
18812
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200188138.2.1. Default log format
18814-------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018815
18816This format is used when no specific option is set. The log is emitted as soon
18817as the connection is accepted. One should note that this currently is the only
18818format which logs the request's destination IP and ports.
18819
18820 Example :
18821 listen www
18822 mode http
18823 log global
18824 server srv1 127.0.0.1:8000
18825
18826 >>> Feb 6 12:12:09 localhost \
18827 haproxy[14385]: Connect from 10.0.1.2:33312 to 10.0.3.31:8012 \
18828 (www/HTTP)
18829
18830 Field Format Extract from the example above
18831 1 process_name '[' pid ']:' haproxy[14385]:
18832 2 'Connect from' Connect from
18833 3 source_ip ':' source_port 10.0.1.2:33312
18834 4 'to' to
18835 5 destination_ip ':' destination_port 10.0.3.31:8012
18836 6 '(' frontend_name '/' mode ')' (www/HTTP)
18837
18838Detailed fields description :
18839 - "source_ip" is the IP address of the client which initiated the connection.
18840 - "source_port" is the TCP port of the client which initiated the connection.
18841 - "destination_ip" is the IP address the client connected to.
18842 - "destination_port" is the TCP port the client connected to.
18843 - "frontend_name" is the name of the frontend (or listener) which received
18844 and processed the connection.
18845 - "mode is the mode the frontend is operating (TCP or HTTP).
18846
Willy Tarreauceb24bc2010-11-09 12:46:41 +010018847In case of a UNIX socket, the source and destination addresses are marked as
18848"unix:" and the ports reflect the internal ID of the socket which accepted the
18849connection (the same ID as reported in the stats).
18850
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018851It is advised not to use this deprecated format for newer installations as it
18852will eventually disappear.
18853
18854
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200188558.2.2. TCP log format
18856---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018857
18858The TCP format is used when "option tcplog" is specified in the frontend, and
18859is the recommended format for pure TCP proxies. It provides a lot of precious
18860information for troubleshooting. Since this format includes timers and byte
18861counts, the log is normally emitted at the end of the session. It can be
18862emitted earlier if "option logasap" is specified, which makes sense in most
18863environments with long sessions such as remote terminals. Sessions which match
18864the "monitor" rules are never logged. It is also possible not to emit logs for
18865sessions for which no data were exchanged between the client and the server, by
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020018866specifying "option dontlognull" in the frontend. Successful connections will
18867not be logged if "option dontlog-normal" is specified in the frontend. A few
18868fields may slightly vary depending on some configuration options, those are
18869marked with a star ('*') after the field name below.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018870
18871 Example :
18872 frontend fnt
18873 mode tcp
18874 option tcplog
18875 log global
18876 default_backend bck
18877
18878 backend bck
18879 server srv1 127.0.0.1:8000
18880
18881 >>> Feb 6 12:12:56 localhost \
18882 haproxy[14387]: 10.0.1.2:33313 [06/Feb/2009:12:12:51.443] fnt \
18883 bck/srv1 0/0/5007 212 -- 0/0/0/0/3 0/0
18884
18885 Field Format Extract from the example above
18886 1 process_name '[' pid ']:' haproxy[14387]:
18887 2 client_ip ':' client_port 10.0.1.2:33313
18888 3 '[' accept_date ']' [06/Feb/2009:12:12:51.443]
18889 4 frontend_name fnt
18890 5 backend_name '/' server_name bck/srv1
18891 6 Tw '/' Tc '/' Tt* 0/0/5007
18892 7 bytes_read* 212
18893 8 termination_state --
18894 9 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 0/0/0/0/3
18895 10 srv_queue '/' backend_queue 0/0
18896
18897Detailed fields description :
18898 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010018899 connection to haproxy. If the connection was accepted on a UNIX socket
18900 instead, the IP address would be replaced with the word "unix". Note that
18901 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010018902 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010018903 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010018904 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018905
18906 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010018907 If the connection was accepted on a UNIX socket instead, the port would be
18908 replaced with the ID of the accepting socket, which is also reported in the
18909 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018910
18911 - "accept_date" is the exact date when the connection was received by haproxy
18912 (which might be very slightly different from the date observed on the
18913 network if there was some queuing in the system's backlog). This is usually
Willy Tarreau590a0512018-09-05 11:56:48 +020018914 the same date which may appear in any upstream firewall's log. When used in
18915 HTTP mode, the accept_date field will be reset to the first moment the
18916 connection is ready to receive a new request (end of previous response for
18917 HTTP/1, immediately after previous request for HTTP/2).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018918
18919 - "frontend_name" is the name of the frontend (or listener) which received
18920 and processed the connection.
18921
18922 - "backend_name" is the name of the backend (or listener) which was selected
18923 to manage the connection to the server. This will be the same as the
18924 frontend if no switching rule has been applied, which is common for TCP
18925 applications.
18926
18927 - "server_name" is the name of the last server to which the connection was
18928 sent, which might differ from the first one if there were connection errors
18929 and a redispatch occurred. Note that this server belongs to the backend
18930 which processed the request. If the connection was aborted before reaching
18931 a server, "<NOSRV>" is indicated instead of a server name.
18932
18933 - "Tw" is the total time in milliseconds spent waiting in the various queues.
18934 It can be "-1" if the connection was aborted before reaching the queue.
18935 See "Timers" below for more details.
18936
18937 - "Tc" is the total time in milliseconds spent waiting for the connection to
18938 establish to the final server, including retries. It can be "-1" if the
18939 connection was aborted before a connection could be established. See
18940 "Timers" below for more details.
18941
18942 - "Tt" is the total time in milliseconds elapsed between the accept and the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030018943 last close. It covers all possible processing. There is one exception, if
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018944 "option logasap" was specified, then the time counting stops at the moment
18945 the log is emitted. In this case, a '+' sign is prepended before the value,
18946 indicating that the final one will be larger. See "Timers" below for more
18947 details.
18948
18949 - "bytes_read" is the total number of bytes transmitted from the server to
18950 the client when the log is emitted. If "option logasap" is specified, the
18951 this value will be prefixed with a '+' sign indicating that the final one
18952 may be larger. Please note that this value is a 64-bit counter, so log
18953 analysis tools must be able to handle it without overflowing.
18954
18955 - "termination_state" is the condition the session was in when the session
18956 ended. This indicates the session state, which side caused the end of
18957 session to happen, and for what reason (timeout, error, ...). The normal
18958 flags should be "--", indicating the session was closed by either end with
18959 no data remaining in buffers. See below "Session state at disconnection"
18960 for more details.
18961
18962 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040018963 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018964 limits have been reached. For instance, if actconn is close to 512 when
18965 multiple connection errors occur, chances are high that the system limits
18966 the process to use a maximum of 1024 file descriptors and that all of them
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018967 are used. See section 3 "Global parameters" to find how to tune the system.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018968
18969 - "feconn" is the total number of concurrent connections on the frontend when
18970 the session was logged. It is useful to estimate the amount of resource
18971 required to sustain high loads, and to detect when the frontend's "maxconn"
18972 has been reached. Most often when this value increases by huge jumps, it is
18973 because there is congestion on the backend servers, but sometimes it can be
18974 caused by a denial of service attack.
18975
18976 - "beconn" is the total number of concurrent connections handled by the
18977 backend when the session was logged. It includes the total number of
18978 concurrent connections active on servers as well as the number of
18979 connections pending in queues. It is useful to estimate the amount of
18980 additional servers needed to support high loads for a given application.
18981 Most often when this value increases by huge jumps, it is because there is
18982 congestion on the backend servers, but sometimes it can be caused by a
18983 denial of service attack.
18984
18985 - "srv_conn" is the total number of concurrent connections still active on
18986 the server when the session was logged. It can never exceed the server's
18987 configured "maxconn" parameter. If this value is very often close or equal
18988 to the server's "maxconn", it means that traffic regulation is involved a
18989 lot, meaning that either the server's maxconn value is too low, or that
18990 there aren't enough servers to process the load with an optimal response
18991 time. When only one of the server's "srv_conn" is high, it usually means
18992 that this server has some trouble causing the connections to take longer to
18993 be processed than on other servers.
18994
18995 - "retries" is the number of connection retries experienced by this session
18996 when trying to connect to the server. It must normally be zero, unless a
18997 server is being stopped at the same moment the connection was attempted.
18998 Frequent retries generally indicate either a network problem between
18999 haproxy and the server, or a misconfigured system backlog on the server
19000 preventing new connections from being queued. This field may optionally be
19001 prefixed with a '+' sign, indicating that the session has experienced a
19002 redispatch after the maximal retry count has been reached on the initial
19003 server. In this case, the server name appearing in the log is the one the
19004 connection was redispatched to, and not the first one, though both may
19005 sometimes be the same in case of hashing for instance. So as a general rule
19006 of thumb, when a '+' is present in front of the retry count, this count
19007 should not be attributed to the logged server.
19008
19009 - "srv_queue" is the total number of requests which were processed before
19010 this one in the server queue. It is zero when the request has not gone
19011 through the server queue. It makes it possible to estimate the approximate
19012 server's response time by dividing the time spent in queue by the number of
19013 requests in the queue. It is worth noting that if a session experiences a
19014 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010019015 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019016 backend queue unless a redispatch occurs.
19017
19018 - "backend_queue" is the total number of requests which were processed before
19019 this one in the backend's global queue. It is zero when the request has not
19020 gone through the global queue. It makes it possible to estimate the average
19021 queue length, which easily translates into a number of missing servers when
19022 divided by a server's "maxconn" parameter. It is worth noting that if a
19023 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010019024 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019025 through both the server queue and the backend queue unless a redispatch
19026 occurs.
19027
19028
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200190298.2.3. HTTP log format
19030----------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019031
19032The HTTP format is the most complete and the best suited for HTTP proxies. It
19033is enabled by when "option httplog" is specified in the frontend. It provides
19034the same level of information as the TCP format with additional features which
19035are specific to the HTTP protocol. Just like the TCP format, the log is usually
19036emitted at the end of the session, unless "option logasap" is specified, which
19037generally only makes sense for download sites. A session which matches the
19038"monitor" rules will never logged. It is also possible not to log sessions for
19039which no data were sent by the client by specifying "option dontlognull" in the
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020019040frontend. Successful connections will not be logged if "option dontlog-normal"
19041is specified in the frontend.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019042
19043Most fields are shared with the TCP log, some being different. A few fields may
19044slightly vary depending on some configuration options. Those ones are marked
19045with a star ('*') after the field name below.
19046
19047 Example :
19048 frontend http-in
19049 mode http
19050 option httplog
19051 log global
19052 default_backend bck
19053
19054 backend static
19055 server srv1 127.0.0.1:8000
19056
19057 >>> Feb 6 12:14:14 localhost \
19058 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
19059 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 +010019060 {} "GET /index.html HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019061
19062 Field Format Extract from the example above
19063 1 process_name '[' pid ']:' haproxy[14389]:
19064 2 client_ip ':' client_port 10.0.1.2:33317
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019065 3 '[' request_date ']' [06/Feb/2009:12:14:14.655]
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019066 4 frontend_name http-in
19067 5 backend_name '/' server_name static/srv1
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019068 6 TR '/' Tw '/' Tc '/' Tr '/' Ta* 10/0/30/69/109
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019069 7 status_code 200
19070 8 bytes_read* 2750
19071 9 captured_request_cookie -
19072 10 captured_response_cookie -
19073 11 termination_state ----
19074 12 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 1/1/1/1/0
19075 13 srv_queue '/' backend_queue 0/0
19076 14 '{' captured_request_headers* '}' {haproxy.1wt.eu}
19077 15 '{' captured_response_headers* '}' {}
19078 16 '"' http_request '"' "GET /index.html HTTP/1.1"
Willy Tarreaud72758d2010-01-12 10:42:19 +010019079
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019080Detailed fields description :
19081 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010019082 connection to haproxy. If the connection was accepted on a UNIX socket
19083 instead, the IP address would be replaced with the word "unix". Note that
19084 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010019085 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010019086 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010019087 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019088
19089 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010019090 If the connection was accepted on a UNIX socket instead, the port would be
19091 replaced with the ID of the accepting socket, which is also reported in the
19092 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019093
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019094 - "request_date" is the exact date when the first byte of the HTTP request
19095 was received by haproxy (log field %tr).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019096
19097 - "frontend_name" is the name of the frontend (or listener) which received
19098 and processed the connection.
19099
19100 - "backend_name" is the name of the backend (or listener) which was selected
19101 to manage the connection to the server. This will be the same as the
19102 frontend if no switching rule has been applied.
19103
19104 - "server_name" is the name of the last server to which the connection was
19105 sent, which might differ from the first one if there were connection errors
19106 and a redispatch occurred. Note that this server belongs to the backend
19107 which processed the request. If the request was aborted before reaching a
19108 server, "<NOSRV>" is indicated instead of a server name. If the request was
19109 intercepted by the stats subsystem, "<STATS>" is indicated instead.
19110
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019111 - "TR" is the total time in milliseconds spent waiting for a full HTTP
19112 request from the client (not counting body) after the first byte was
19113 received. It can be "-1" if the connection was aborted before a complete
John Roeslerfb2fce12019-07-10 15:45:51 -050019114 request could be received or a bad request was received. It should
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019115 always be very small because a request generally fits in one single packet.
19116 Large times here generally indicate network issues between the client and
Willy Tarreau590a0512018-09-05 11:56:48 +020019117 haproxy or requests being typed by hand. See section 8.4 "Timing Events"
19118 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019119
19120 - "Tw" is the total time in milliseconds spent waiting in the various queues.
19121 It can be "-1" if the connection was aborted before reaching the queue.
Willy Tarreau590a0512018-09-05 11:56:48 +020019122 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019123
19124 - "Tc" is the total time in milliseconds spent waiting for the connection to
19125 establish to the final server, including retries. It can be "-1" if the
Willy Tarreau590a0512018-09-05 11:56:48 +020019126 request was aborted before a connection could be established. See section
19127 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019128
19129 - "Tr" is the total time in milliseconds spent waiting for the server to send
19130 a full HTTP response, not counting data. It can be "-1" if the request was
19131 aborted before a complete response could be received. It generally matches
19132 the server's processing time for the request, though it may be altered by
19133 the amount of data sent by the client to the server. Large times here on
Willy Tarreau590a0512018-09-05 11:56:48 +020019134 "GET" requests generally indicate an overloaded server. See section 8.4
19135 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019136
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019137 - "Ta" is the time the request remained active in haproxy, which is the total
19138 time in milliseconds elapsed between the first byte of the request was
19139 received and the last byte of response was sent. It covers all possible
19140 processing except the handshake (see Th) and idle time (see Ti). There is
19141 one exception, if "option logasap" was specified, then the time counting
19142 stops at the moment the log is emitted. In this case, a '+' sign is
19143 prepended before the value, indicating that the final one will be larger.
Willy Tarreau590a0512018-09-05 11:56:48 +020019144 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019145
19146 - "status_code" is the HTTP status code returned to the client. This status
19147 is generally set by the server, but it might also be set by haproxy when
19148 the server cannot be reached or when its response is blocked by haproxy.
19149
19150 - "bytes_read" is the total number of bytes transmitted to the client when
19151 the log is emitted. This does include HTTP headers. If "option logasap" is
John Roeslerfb2fce12019-07-10 15:45:51 -050019152 specified, this value will be prefixed with a '+' sign indicating that
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019153 the final one may be larger. Please note that this value is a 64-bit
19154 counter, so log analysis tools must be able to handle it without
19155 overflowing.
19156
19157 - "captured_request_cookie" is an optional "name=value" entry indicating that
19158 the client had this cookie in the request. The cookie name and its maximum
19159 length are defined by the "capture cookie" statement in the frontend
19160 configuration. The field is a single dash ('-') when the option is not
19161 set. Only one cookie may be captured, it is generally used to track session
19162 ID exchanges between a client and a server to detect session crossing
19163 between clients due to application bugs. For more details, please consult
19164 the section "Capturing HTTP headers and cookies" below.
19165
19166 - "captured_response_cookie" is an optional "name=value" entry indicating
19167 that the server has returned a cookie with its response. The cookie name
19168 and its maximum length are defined by the "capture cookie" statement in the
19169 frontend configuration. The field is a single dash ('-') when the option is
19170 not set. Only one cookie may be captured, it is generally used to track
19171 session ID exchanges between a client and a server to detect session
19172 crossing between clients due to application bugs. For more details, please
19173 consult the section "Capturing HTTP headers and cookies" below.
19174
19175 - "termination_state" is the condition the session was in when the session
19176 ended. This indicates the session state, which side caused the end of
19177 session to happen, for what reason (timeout, error, ...), just like in TCP
19178 logs, and information about persistence operations on cookies in the last
19179 two characters. The normal flags should begin with "--", indicating the
19180 session was closed by either end with no data remaining in buffers. See
19181 below "Session state at disconnection" for more details.
19182
19183 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040019184 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019185 limits have been reached. For instance, if actconn is close to 512 or 1024
19186 when multiple connection errors occur, chances are high that the system
19187 limits the process to use a maximum of 1024 file descriptors and that all
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019188 of them are used. See section 3 "Global parameters" to find how to tune the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019189 system.
19190
19191 - "feconn" is the total number of concurrent connections on the frontend when
19192 the session was logged. It is useful to estimate the amount of resource
19193 required to sustain high loads, and to detect when the frontend's "maxconn"
19194 has been reached. Most often when this value increases by huge jumps, it is
19195 because there is congestion on the backend servers, but sometimes it can be
19196 caused by a denial of service attack.
19197
19198 - "beconn" is the total number of concurrent connections handled by the
19199 backend when the session was logged. It includes the total number of
19200 concurrent connections active on servers as well as the number of
19201 connections pending in queues. It is useful to estimate the amount of
19202 additional servers needed to support high loads for a given application.
19203 Most often when this value increases by huge jumps, it is because there is
19204 congestion on the backend servers, but sometimes it can be caused by a
19205 denial of service attack.
19206
19207 - "srv_conn" is the total number of concurrent connections still active on
19208 the server when the session was logged. It can never exceed the server's
19209 configured "maxconn" parameter. If this value is very often close or equal
19210 to the server's "maxconn", it means that traffic regulation is involved a
19211 lot, meaning that either the server's maxconn value is too low, or that
19212 there aren't enough servers to process the load with an optimal response
19213 time. When only one of the server's "srv_conn" is high, it usually means
19214 that this server has some trouble causing the requests to take longer to be
19215 processed than on other servers.
19216
19217 - "retries" is the number of connection retries experienced by this session
19218 when trying to connect to the server. It must normally be zero, unless a
19219 server is being stopped at the same moment the connection was attempted.
19220 Frequent retries generally indicate either a network problem between
19221 haproxy and the server, or a misconfigured system backlog on the server
19222 preventing new connections from being queued. This field may optionally be
19223 prefixed with a '+' sign, indicating that the session has experienced a
19224 redispatch after the maximal retry count has been reached on the initial
19225 server. In this case, the server name appearing in the log is the one the
19226 connection was redispatched to, and not the first one, though both may
19227 sometimes be the same in case of hashing for instance. So as a general rule
19228 of thumb, when a '+' is present in front of the retry count, this count
19229 should not be attributed to the logged server.
19230
19231 - "srv_queue" is the total number of requests which were processed before
19232 this one in the server queue. It is zero when the request has not gone
19233 through the server queue. It makes it possible to estimate the approximate
19234 server's response time by dividing the time spent in queue by the number of
19235 requests in the queue. It is worth noting that if a session experiences a
19236 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010019237 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019238 backend queue unless a redispatch occurs.
19239
19240 - "backend_queue" is the total number of requests which were processed before
19241 this one in the backend's global queue. It is zero when the request has not
19242 gone through the global queue. It makes it possible to estimate the average
19243 queue length, which easily translates into a number of missing servers when
19244 divided by a server's "maxconn" parameter. It is worth noting that if a
19245 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010019246 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019247 through both the server queue and the backend queue unless a redispatch
19248 occurs.
19249
19250 - "captured_request_headers" is a list of headers captured in the request due
19251 to the presence of the "capture request header" statement in the frontend.
19252 Multiple headers can be captured, they will be delimited by a vertical bar
19253 ('|'). When no capture is enabled, the braces do not appear, causing a
19254 shift of remaining fields. It is important to note that this field may
19255 contain spaces, and that using it requires a smarter log parser than when
19256 it's not used. Please consult the section "Capturing HTTP headers and
19257 cookies" below for more details.
19258
19259 - "captured_response_headers" is a list of headers captured in the response
19260 due to the presence of the "capture response header" statement in the
19261 frontend. Multiple headers can be captured, they will be delimited by a
19262 vertical bar ('|'). When no capture is enabled, the braces do not appear,
19263 causing a shift of remaining fields. It is important to note that this
19264 field may contain spaces, and that using it requires a smarter log parser
19265 than when it's not used. Please consult the section "Capturing HTTP headers
19266 and cookies" below for more details.
19267
19268 - "http_request" is the complete HTTP request line, including the method,
19269 request and HTTP version string. Non-printable characters are encoded (see
19270 below the section "Non-printable characters"). This is always the last
19271 field, and it is always delimited by quotes and is the only one which can
19272 contain quotes. If new fields are added to the log format, they will be
19273 added before this field. This field might be truncated if the request is
19274 huge and does not fit in the standard syslog buffer (1024 characters). This
19275 is the reason why this field must always remain the last one.
19276
19277
Cyril Bontédc4d9032012-04-08 21:57:39 +0200192788.2.4. Custom log format
19279------------------------
William Lallemand48940402012-01-30 16:47:22 +010019280
Willy Tarreau2beef582012-12-20 17:22:52 +010019281The directive log-format allows you to customize the logs in http mode and tcp
William Lallemandbddd4fd2012-02-27 11:23:10 +010019282mode. It takes a string as argument.
William Lallemand48940402012-01-30 16:47:22 +010019283
Davor Ocelice9ed2812017-12-25 17:49:28 +010019284HAProxy understands some log format variables. % precedes log format variables.
William Lallemand48940402012-01-30 16:47:22 +010019285Variables can take arguments using braces ('{}'), and multiple arguments are
19286separated by commas within the braces. Flags may be added or removed by
19287prefixing them with a '+' or '-' sign.
19288
19289Special variable "%o" may be used to propagate its flags to all other
19290variables on the same format string. This is particularly handy with quoted
Dragan Dosen835b9212016-02-12 13:23:03 +010019291("Q") and escaped ("E") string formats.
William Lallemand48940402012-01-30 16:47:22 +010019292
Willy Tarreauc8368452012-12-21 00:09:23 +010019293If a variable is named between square brackets ('[' .. ']') then it is used
Willy Tarreaube722a22014-06-13 16:31:59 +020019294as a sample expression rule (see section 7.3). This it useful to add some
Willy Tarreauc8368452012-12-21 00:09:23 +010019295less common information such as the client's SSL certificate's DN, or to log
19296the key that would be used to store an entry into a stick table.
19297
Dragan Dosen1e3b16f2020-06-23 18:16:44 +020019298Note: spaces must be escaped. In configuration directives "log-format",
19299"log-format-sd" and "unique-id-format", spaces are considered as
19300delimiters and are merged. In order to emit a verbatim '%', it must be
19301preceded by another '%' resulting in '%%'.
William Lallemand48940402012-01-30 16:47:22 +010019302
Dragan Dosen835b9212016-02-12 13:23:03 +010019303Note: when using the RFC5424 syslog message format, the characters '"',
19304'\' and ']' inside PARAM-VALUE should be escaped with '\' as prefix (see
19305https://tools.ietf.org/html/rfc5424#section-6.3.3 for more details). In
19306such cases, the use of the flag "E" should be considered.
19307
William Lallemand48940402012-01-30 16:47:22 +010019308Flags are :
19309 * Q: quote a string
Jamie Gloudonaaa21002012-08-25 00:18:33 -040019310 * X: hexadecimal representation (IPs, Ports, %Ts, %rt, %pid)
Dragan Dosen835b9212016-02-12 13:23:03 +010019311 * E: escape characters '"', '\' and ']' in a string with '\' as prefix
19312 (intended purpose is for the RFC5424 structured-data log formats)
William Lallemand48940402012-01-30 16:47:22 +010019313
19314 Example:
19315
19316 log-format %T\ %t\ Some\ Text
19317 log-format %{+Q}o\ %t\ %s\ %{-Q}r
19318
Dragan Dosen835b9212016-02-12 13:23:03 +010019319 log-format-sd %{+Q,+E}o\ [exampleSDID@1234\ header=%[capture.req.hdr(0)]]
19320
William Lallemand48940402012-01-30 16:47:22 +010019321At the moment, the default HTTP format is defined this way :
19322
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019323 log-format "%ci:%cp [%tr] %ft %b/%s %TR/%Tw/%Tc/%Tr/%Ta %ST %B %CC \
19324 %CS %tsc %ac/%fc/%bc/%sc/%rc %sq/%bq %hr %hs %{+Q}r"
William Lallemand48940402012-01-30 16:47:22 +010019325
William Lallemandbddd4fd2012-02-27 11:23:10 +010019326the default CLF format is defined this way :
William Lallemand48940402012-01-30 16:47:22 +010019327
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019328 log-format "%{+Q}o %{-Q}ci - - [%trg] %r %ST %B \"\" \"\" %cp \
19329 %ms %ft %b %s %TR %Tw %Tc %Tr %Ta %tsc %ac %fc \
19330 %bc %sc %rc %sq %bq %CC %CS %hrl %hsl"
William Lallemand48940402012-01-30 16:47:22 +010019331
William Lallemandbddd4fd2012-02-27 11:23:10 +010019332and the default TCP format is defined this way :
19333
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019334 log-format "%ci:%cp [%t] %ft %b/%s %Tw/%Tc/%Tt %B %ts \
19335 %ac/%fc/%bc/%sc/%rc %sq/%bq"
William Lallemandbddd4fd2012-02-27 11:23:10 +010019336
William Lallemand48940402012-01-30 16:47:22 +010019337Please refer to the table below for currently defined variables :
19338
William Lallemandbddd4fd2012-02-27 11:23:10 +010019339 +---+------+-----------------------------------------------+-------------+
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020019340 | R | var | field name (8.2.2 and 8.2.3 for description) | type |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019341 +---+------+-----------------------------------------------+-------------+
19342 | | %o | special variable, apply flags on all next var | |
19343 +---+------+-----------------------------------------------+-------------+
Willy Tarreau2beef582012-12-20 17:22:52 +010019344 | | %B | bytes_read (from server to client) | numeric |
19345 | H | %CC | captured_request_cookie | string |
19346 | H | %CS | captured_response_cookie | string |
William Lallemand5f232402012-04-05 18:02:55 +020019347 | | %H | hostname | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000019348 | H | %HM | HTTP method (ex: POST) | string |
19349 | H | %HP | HTTP request URI without query string (path) | string |
Andrew Hayworthe63ac872015-07-31 16:14:16 +000019350 | H | %HQ | HTTP request URI query string (ex: ?bar=baz) | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000019351 | H | %HU | HTTP request URI (ex: /foo?bar=baz) | string |
19352 | H | %HV | HTTP version (ex: HTTP/1.0) | string |
William Lallemanda73203e2012-03-12 12:48:57 +010019353 | | %ID | unique-id | string |
Willy Tarreau4bf99632014-06-13 12:21:40 +020019354 | | %ST | status_code | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020019355 | | %T | gmt_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019356 | | %Ta | Active time of the request (from TR to end) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019357 | | %Tc | Tc | numeric |
Willy Tarreau27b639d2016-05-17 17:55:27 +020019358 | | %Td | Td = Tt - (Tq + Tw + Tc + Tr) | numeric |
Yuxans Yao4e25b012012-10-19 10:36:09 +080019359 | | %Tl | local_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019360 | | %Th | connection handshake time (SSL, PROXY proto) | numeric |
19361 | H | %Ti | idle time before the HTTP request | numeric |
19362 | H | %Tq | Th + Ti + TR | numeric |
19363 | H | %TR | time to receive the full request from 1st byte| numeric |
19364 | H | %Tr | Tr (response time) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020019365 | | %Ts | timestamp | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019366 | | %Tt | Tt | numeric |
Damien Claisse57c8eb92020-04-28 12:09:19 +000019367 | | %Tu | Tu | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019368 | | %Tw | Tw | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010019369 | | %U | bytes_uploaded (from client to server) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019370 | | %ac | actconn | numeric |
19371 | | %b | backend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010019372 | | %bc | beconn (backend concurrent connections) | numeric |
19373 | | %bi | backend_source_ip (connecting address) | IP |
19374 | | %bp | backend_source_port (connecting address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019375 | | %bq | backend_queue | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010019376 | | %ci | client_ip (accepted address) | IP |
19377 | | %cp | client_port (accepted address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019378 | | %f | frontend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010019379 | | %fc | feconn (frontend concurrent connections) | numeric |
19380 | | %fi | frontend_ip (accepting address) | IP |
19381 | | %fp | frontend_port (accepting address) | numeric |
Willy Tarreau773d65f2012-10-12 14:56:11 +020019382 | | %ft | frontend_name_transport ('~' suffix for SSL) | string |
Willy Tarreau7346acb2014-08-28 15:03:15 +020019383 | | %lc | frontend_log_counter | numeric |
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020019384 | | %hr | captured_request_headers default style | string |
19385 | | %hrl | captured_request_headers CLF style | string list |
19386 | | %hs | captured_response_headers default style | string |
19387 | | %hsl | captured_response_headers CLF style | string list |
Willy Tarreau812c88e2015-08-09 10:56:35 +020019388 | | %ms | accept date milliseconds (left-padded with 0) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020019389 | | %pid | PID | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020019390 | H | %r | http_request | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019391 | | %rc | retries | numeric |
Willy Tarreau1f0da242014-01-25 11:01:50 +010019392 | | %rt | request_counter (HTTP req or TCP session) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019393 | | %s | server_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010019394 | | %sc | srv_conn (server concurrent connections) | numeric |
19395 | | %si | server_IP (target address) | IP |
19396 | | %sp | server_port (target address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019397 | | %sq | srv_queue | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020019398 | S | %sslc| ssl_ciphers (ex: AES-SHA) | string |
19399 | S | %sslv| ssl_version (ex: TLSv1) | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010019400 | | %t | date_time (with millisecond resolution) | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019401 | H | %tr | date_time of HTTP request | date |
19402 | H | %trg | gmt_date_time of start of HTTP request | date |
Jens Bissinger15c64ff2018-08-23 14:11:27 +020019403 | H | %trl | local_date_time of start of HTTP request | date |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019404 | | %ts | termination_state | string |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020019405 | H | %tsc | termination_state with cookie status | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019406 +---+------+-----------------------------------------------+-------------+
William Lallemand48940402012-01-30 16:47:22 +010019407
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020019408 R = Restrictions : H = mode http only ; S = SSL only
William Lallemand48940402012-01-30 16:47:22 +010019409
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010019410
194118.2.5. Error log format
19412-----------------------
19413
19414When an incoming connection fails due to an SSL handshake or an invalid PROXY
19415protocol header, haproxy will log the event using a shorter, fixed line format.
19416By default, logs are emitted at the LOG_INFO level, unless the option
19417"log-separate-errors" is set in the backend, in which case the LOG_ERR level
Davor Ocelice9ed2812017-12-25 17:49:28 +010019418will be used. Connections on which no data are exchanged (e.g. probes) are not
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010019419logged if the "dontlognull" option is set.
19420
19421The format looks like this :
19422
19423 >>> Dec 3 18:27:14 localhost \
19424 haproxy[6103]: 127.0.0.1:56059 [03/Dec/2012:17:35:10.380] frt/f1: \
19425 Connection error during SSL handshake
19426
19427 Field Format Extract from the example above
19428 1 process_name '[' pid ']:' haproxy[6103]:
19429 2 client_ip ':' client_port 127.0.0.1:56059
19430 3 '[' accept_date ']' [03/Dec/2012:17:35:10.380]
19431 4 frontend_name "/" bind_name ":" frt/f1:
19432 5 message Connection error during SSL handshake
19433
19434These fields just provide minimal information to help debugging connection
19435failures.
19436
19437
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200194388.3. Advanced logging options
19439-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019440
19441Some advanced logging options are often looked for but are not easy to find out
19442just by looking at the various options. Here is an entry point for the few
19443options which can enable better logging. Please refer to the keywords reference
19444for more information about their usage.
19445
19446
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200194478.3.1. Disabling logging of external tests
19448------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019449
19450It is quite common to have some monitoring tools perform health checks on
19451haproxy. Sometimes it will be a layer 3 load-balancer such as LVS or any
19452commercial load-balancer, and sometimes it will simply be a more complete
19453monitoring system such as Nagios. When the tests are very frequent, users often
19454ask how to disable logging for those checks. There are three possibilities :
19455
19456 - if connections come from everywhere and are just TCP probes, it is often
19457 desired to simply disable logging of connections without data exchange, by
19458 setting "option dontlognull" in the frontend. It also disables logging of
19459 port scans, which may or may not be desired.
19460
Willy Tarreau9e9919d2020-10-14 15:55:23 +020019461 - it is possible to use the "http-request set-log-level silent" action using
19462 a variety of conditions (source networks, paths, user-agents, etc).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019463
19464 - if the tests are performed on a known URI, use "monitor-uri" to declare
19465 this URI as dedicated to monitoring. Any host sending this request will
19466 only get the result of a health-check, and the request will not be logged.
19467
19468
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200194698.3.2. Logging before waiting for the session to terminate
19470----------------------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019471
19472The problem with logging at end of connection is that you have no clue about
19473what is happening during very long sessions, such as remote terminal sessions
19474or large file downloads. This problem can be worked around by specifying
Davor Ocelice9ed2812017-12-25 17:49:28 +010019475"option logasap" in the frontend. HAProxy will then log as soon as possible,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019476just before data transfer begins. This means that in case of TCP, it will still
19477log the connection status to the server, and in case of HTTP, it will log just
19478after processing the server headers. In this case, the number of bytes reported
19479is the number of header bytes sent to the client. In order to avoid confusion
19480with normal logs, the total time field and the number of bytes are prefixed
19481with a '+' sign which means that real numbers are certainly larger.
19482
19483
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200194848.3.3. Raising log level upon errors
19485------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020019486
19487Sometimes it is more convenient to separate normal traffic from errors logs,
19488for instance in order to ease error monitoring from log files. When the option
19489"log-separate-errors" is used, connections which experience errors, timeouts,
19490retries, redispatches or HTTP status codes 5xx will see their syslog level
19491raised from "info" to "err". This will help a syslog daemon store the log in
19492a separate file. It is very important to keep the errors in the normal traffic
19493file too, so that log ordering is not altered. You should also be careful if
19494you already have configured your syslog daemon to store all logs higher than
19495"notice" in an "admin" file, because the "err" level is higher than "notice".
19496
19497
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200194988.3.4. Disabling logging of successful connections
19499--------------------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020019500
19501Although this may sound strange at first, some large sites have to deal with
19502multiple thousands of logs per second and are experiencing difficulties keeping
19503them intact for a long time or detecting errors within them. If the option
19504"dontlog-normal" is set on the frontend, all normal connections will not be
19505logged. In this regard, a normal connection is defined as one without any
19506error, timeout, retry nor redispatch. In HTTP, the status code is checked too,
19507and a response with a status 5xx is not considered normal and will be logged
19508too. Of course, doing is is really discouraged as it will remove most of the
19509useful information from the logs. Do this only if you have no other
19510alternative.
19511
19512
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200195138.4. Timing events
19514------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019515
19516Timers provide a great help in troubleshooting network problems. All values are
19517reported in milliseconds (ms). These timers should be used in conjunction with
19518the session termination flags. In TCP mode with "option tcplog" set on the
19519frontend, 3 control points are reported under the form "Tw/Tc/Tt", and in HTTP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019520mode, 5 control points are reported under the form "TR/Tw/Tc/Tr/Ta". In
19521addition, three other measures are provided, "Th", "Ti", and "Tq".
19522
Guillaume de Lafondf27cddc2016-12-23 17:32:43 +010019523Timings events in HTTP mode:
19524
19525 first request 2nd request
19526 |<-------------------------------->|<-------------- ...
19527 t tr t tr ...
19528 ---|----|----|----|----|----|----|----|----|--
19529 : Th Ti TR Tw Tc Tr Td : Ti ...
19530 :<---- Tq ---->: :
19531 :<-------------- Tt -------------->:
Damien Claisse57c8eb92020-04-28 12:09:19 +000019532 :<-- -----Tu--------------->:
Guillaume de Lafondf27cddc2016-12-23 17:32:43 +010019533 :<--------- Ta --------->:
19534
19535Timings events in TCP mode:
19536
19537 TCP session
19538 |<----------------->|
19539 t t
19540 ---|----|----|----|----|---
19541 | Th Tw Tc Td |
19542 |<------ Tt ------->|
19543
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019544 - Th: total time to accept tcp connection and execute handshakes for low level
Davor Ocelice9ed2812017-12-25 17:49:28 +010019545 protocols. Currently, these protocols are proxy-protocol and SSL. This may
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019546 only happen once during the whole connection's lifetime. A large time here
19547 may indicate that the client only pre-established the connection without
19548 speaking, that it is experiencing network issues preventing it from
Davor Ocelice9ed2812017-12-25 17:49:28 +010019549 completing a handshake in a reasonable time (e.g. MTU issues), or that an
Willy Tarreau590a0512018-09-05 11:56:48 +020019550 SSL handshake was very expensive to compute. Please note that this time is
19551 reported only before the first request, so it is safe to average it over
19552 all request to calculate the amortized value. The second and subsequent
19553 request will always report zero here.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019554
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019555 - Ti: is the idle time before the HTTP request (HTTP mode only). This timer
19556 counts between the end of the handshakes and the first byte of the HTTP
19557 request. When dealing with a second request in keep-alive mode, it starts
Willy Tarreau590a0512018-09-05 11:56:48 +020019558 to count after the end of the transmission the previous response. When a
19559 multiplexed protocol such as HTTP/2 is used, it starts to count immediately
19560 after the previous request. Some browsers pre-establish connections to a
19561 server in order to reduce the latency of a future request, and keep them
19562 pending until they need it. This delay will be reported as the idle time. A
19563 value of -1 indicates that nothing was received on the connection.
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019564
19565 - TR: total time to get the client request (HTTP mode only). It's the time
19566 elapsed between the first bytes received and the moment the proxy received
19567 the empty line marking the end of the HTTP headers. The value "-1"
19568 indicates that the end of headers has never been seen. This happens when
19569 the client closes prematurely or times out. This time is usually very short
19570 since most requests fit in a single packet. A large time may indicate a
19571 request typed by hand during a test.
19572
19573 - Tq: total time to get the client request from the accept date or since the
19574 emission of the last byte of the previous response (HTTP mode only). It's
Davor Ocelice9ed2812017-12-25 17:49:28 +010019575 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 +020019576 returns -1 as well. This timer used to be very useful before the arrival of
19577 HTTP keep-alive and browsers' pre-connect feature. It's recommended to drop
19578 it in favor of TR nowadays, as the idle time adds a lot of noise to the
19579 reports.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019580
19581 - Tw: total time spent in the queues waiting for a connection slot. It
19582 accounts for backend queue as well as the server queues, and depends on the
19583 queue size, and the time needed for the server to complete previous
19584 requests. The value "-1" means that the request was killed before reaching
19585 the queue, which is generally what happens with invalid or denied requests.
19586
19587 - Tc: total time to establish the TCP connection to the server. It's the time
19588 elapsed between the moment the proxy sent the connection request, and the
19589 moment it was acknowledged by the server, or between the TCP SYN packet and
19590 the matching SYN/ACK packet in return. The value "-1" means that the
19591 connection never established.
19592
19593 - Tr: server response time (HTTP mode only). It's the time elapsed between
19594 the moment the TCP connection was established to the server and the moment
19595 the server sent its complete response headers. It purely shows its request
19596 processing time, without the network overhead due to the data transmission.
19597 It is worth noting that when the client has data to send to the server, for
19598 instance during a POST request, the time already runs, and this can distort
19599 apparent response time. For this reason, it's generally wise not to trust
19600 too much this field for POST requests initiated from clients behind an
19601 untrusted network. A value of "-1" here means that the last the response
19602 header (empty line) was never seen, most likely because the server timeout
19603 stroke before the server managed to process the request.
19604
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019605 - Ta: total active time for the HTTP request, between the moment the proxy
19606 received the first byte of the request header and the emission of the last
19607 byte of the response body. The exception is when the "logasap" option is
19608 specified. In this case, it only equals (TR+Tw+Tc+Tr), and is prefixed with
19609 a '+' sign. From this field, we can deduce "Td", the data transmission time,
19610 by subtracting other timers when valid :
19611
19612 Td = Ta - (TR + Tw + Tc + Tr)
19613
19614 Timers with "-1" values have to be excluded from this equation. Note that
19615 "Ta" can never be negative.
19616
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019617 - Tt: total session duration time, between the moment the proxy accepted it
19618 and the moment both ends were closed. The exception is when the "logasap"
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019619 option is specified. In this case, it only equals (Th+Ti+TR+Tw+Tc+Tr), and
19620 is prefixed with a '+' sign. From this field, we can deduce "Td", the data
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030019621 transmission time, by subtracting other timers when valid :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019622
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019623 Td = Tt - (Th + Ti + TR + Tw + Tc + Tr)
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019624
19625 Timers with "-1" values have to be excluded from this equation. In TCP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019626 mode, "Ti", "Tq" and "Tr" have to be excluded too. Note that "Tt" can never
19627 be negative and that for HTTP, Tt is simply equal to (Th+Ti+Ta).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019628
Damien Claisse57c8eb92020-04-28 12:09:19 +000019629 - Tu: total estimated time as seen from client, between the moment the proxy
19630 accepted it and the moment both ends were closed, without idle time.
19631 This is useful to roughly measure end-to-end time as a user would see it,
19632 without idle time pollution from keep-alive time between requests. This
19633 timer in only an estimation of time seen by user as it assumes network
19634 latency is the same in both directions. The exception is when the "logasap"
19635 option is specified. In this case, it only equals (Th+TR+Tw+Tc+Tr), and is
19636 prefixed with a '+' sign.
19637
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019638These timers provide precious indications on trouble causes. Since the TCP
19639protocol defines retransmit delays of 3, 6, 12... seconds, we know for sure
19640that timers close to multiples of 3s are nearly always related to lost packets
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019641due to network problems (wires, negotiation, congestion). Moreover, if "Ta" or
19642"Tt" is close to a timeout value specified in the configuration, it often means
19643that a session has been aborted on timeout.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019644
19645Most common cases :
19646
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019647 - If "Th" or "Ti" are close to 3000, a packet has probably been lost between
19648 the client and the proxy. This is very rare on local networks but might
19649 happen when clients are on far remote networks and send large requests. It
19650 may happen that values larger than usual appear here without any network
19651 cause. Sometimes, during an attack or just after a resource starvation has
19652 ended, haproxy may accept thousands of connections in a few milliseconds.
19653 The time spent accepting these connections will inevitably slightly delay
19654 processing of other connections, and it can happen that request times in the
19655 order of a few tens of milliseconds are measured after a few thousands of
19656 new connections have been accepted at once. Using one of the keep-alive
19657 modes may display larger idle times since "Ti" measures the time spent
Patrick Mezard105faca2010-06-12 17:02:46 +020019658 waiting for additional requests.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019659
19660 - If "Tc" is close to 3000, a packet has probably been lost between the
19661 server and the proxy during the server connection phase. This value should
19662 always be very low, such as 1 ms on local networks and less than a few tens
19663 of ms on remote networks.
19664
Willy Tarreau55165fe2009-05-10 12:02:55 +020019665 - If "Tr" is nearly always lower than 3000 except some rare values which seem
19666 to be the average majored by 3000, there are probably some packets lost
19667 between the proxy and the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019668
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019669 - If "Ta" is large even for small byte counts, it generally is because
19670 neither the client nor the server decides to close the connection while
19671 haproxy is running in tunnel mode and both have agreed on a keep-alive
19672 connection mode. In order to solve this issue, it will be needed to specify
19673 one of the HTTP options to manipulate keep-alive or close options on either
19674 the frontend or the backend. Having the smallest possible 'Ta' or 'Tt' is
19675 important when connection regulation is used with the "maxconn" option on
19676 the servers, since no new connection will be sent to the server until
19677 another one is released.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019678
19679Other noticeable HTTP log cases ('xx' means any value to be ignored) :
19680
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019681 TR/Tw/Tc/Tr/+Ta The "option logasap" is present on the frontend and the log
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019682 was emitted before the data phase. All the timers are valid
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019683 except "Ta" which is shorter than reality.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019684
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019685 -1/xx/xx/xx/Ta The client was not able to send a complete request in time
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019686 or it aborted too early. Check the session termination flags
19687 then "timeout http-request" and "timeout client" settings.
19688
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019689 TR/-1/xx/xx/Ta It was not possible to process the request, maybe because
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019690 servers were out of order, because the request was invalid
19691 or forbidden by ACL rules. Check the session termination
19692 flags.
19693
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019694 TR/Tw/-1/xx/Ta The connection could not establish on the server. Either it
19695 actively refused it or it timed out after Ta-(TR+Tw) ms.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019696 Check the session termination flags, then check the
19697 "timeout connect" setting. Note that the tarpit action might
19698 return similar-looking patterns, with "Tw" equal to the time
19699 the client connection was maintained open.
19700
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019701 TR/Tw/Tc/-1/Ta The server has accepted the connection but did not return
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030019702 a complete response in time, or it closed its connection
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019703 unexpectedly after Ta-(TR+Tw+Tc) ms. Check the session
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019704 termination flags, then check the "timeout server" setting.
19705
19706
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200197078.5. Session state at disconnection
19708-----------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019709
19710TCP and HTTP logs provide a session termination indicator in the
19711"termination_state" field, just before the number of active connections. It is
197122-characters long in TCP mode, and is extended to 4 characters in HTTP mode,
19713each of which has a special meaning :
19714
19715 - On the first character, a code reporting the first event which caused the
19716 session to terminate :
19717
19718 C : the TCP session was unexpectedly aborted by the client.
19719
19720 S : the TCP session was unexpectedly aborted by the server, or the
19721 server explicitly refused it.
19722
19723 P : the session was prematurely aborted by the proxy, because of a
19724 connection limit enforcement, because a DENY filter was matched,
19725 because of a security check which detected and blocked a dangerous
19726 error in server response which might have caused information leak
Davor Ocelice9ed2812017-12-25 17:49:28 +010019727 (e.g. cacheable cookie).
Willy Tarreau570f2212013-06-10 16:42:09 +020019728
19729 L : the session was locally processed by haproxy and was not passed to
19730 a server. This is what happens for stats and redirects.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019731
19732 R : a resource on the proxy has been exhausted (memory, sockets, source
19733 ports, ...). Usually, this appears during the connection phase, and
19734 system logs should contain a copy of the precise error. If this
19735 happens, it must be considered as a very serious anomaly which
19736 should be fixed as soon as possible by any means.
19737
19738 I : an internal error was identified by the proxy during a self-check.
19739 This should NEVER happen, and you are encouraged to report any log
19740 containing this, because this would almost certainly be a bug. It
19741 would be wise to preventively restart the process after such an
19742 event too, in case it would be caused by memory corruption.
19743
Simon Horman752dc4a2011-06-21 14:34:59 +090019744 D : the session was killed by haproxy because the server was detected
19745 as down and was configured to kill all connections when going down.
19746
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070019747 U : the session was killed by haproxy on this backup server because an
19748 active server was detected as up and was configured to kill all
19749 backup connections when going up.
19750
Willy Tarreaua2a64e92011-09-07 23:01:56 +020019751 K : the session was actively killed by an admin operating on haproxy.
19752
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019753 c : the client-side timeout expired while waiting for the client to
19754 send or receive data.
19755
19756 s : the server-side timeout expired while waiting for the server to
19757 send or receive data.
19758
19759 - : normal session completion, both the client and the server closed
19760 with nothing left in the buffers.
19761
19762 - on the second character, the TCP or HTTP session state when it was closed :
19763
Willy Tarreauf7b30a92010-12-06 22:59:17 +010019764 R : the proxy was waiting for a complete, valid REQUEST from the client
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019765 (HTTP mode only). Nothing was sent to any server.
19766
19767 Q : the proxy was waiting in the QUEUE for a connection slot. This can
19768 only happen when servers have a 'maxconn' parameter set. It can
19769 also happen in the global queue after a redispatch consecutive to
19770 a failed attempt to connect to a dying server. If no redispatch is
19771 reported, then no connection attempt was made to any server.
19772
19773 C : the proxy was waiting for the CONNECTION to establish on the
19774 server. The server might at most have noticed a connection attempt.
19775
19776 H : the proxy was waiting for complete, valid response HEADERS from the
19777 server (HTTP only).
19778
19779 D : the session was in the DATA phase.
19780
19781 L : the proxy was still transmitting LAST data to the client while the
19782 server had already finished. This one is very rare as it can only
19783 happen when the client dies while receiving the last packets.
19784
19785 T : the request was tarpitted. It has been held open with the client
19786 during the whole "timeout tarpit" duration or until the client
19787 closed, both of which will be reported in the "Tw" timer.
19788
19789 - : normal session completion after end of data transfer.
19790
19791 - the third character tells whether the persistence cookie was provided by
19792 the client (only in HTTP mode) :
19793
19794 N : the client provided NO cookie. This is usually the case for new
19795 visitors, so counting the number of occurrences of this flag in the
19796 logs generally indicate a valid trend for the site frequentation.
19797
19798 I : the client provided an INVALID cookie matching no known server.
19799 This might be caused by a recent configuration change, mixed
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +020019800 cookies between HTTP/HTTPS sites, persistence conditionally
19801 ignored, or an attack.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019802
19803 D : the client provided a cookie designating a server which was DOWN,
19804 so either "option persist" was used and the client was sent to
19805 this server, or it was not set and the client was redispatched to
19806 another server.
19807
Willy Tarreau996a92c2010-10-13 19:30:47 +020019808 V : the client provided a VALID cookie, and was sent to the associated
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019809 server.
19810
Willy Tarreau996a92c2010-10-13 19:30:47 +020019811 E : the client provided a valid cookie, but with a last date which was
19812 older than what is allowed by the "maxidle" cookie parameter, so
19813 the cookie is consider EXPIRED and is ignored. The request will be
19814 redispatched just as if there was no cookie.
19815
19816 O : the client provided a valid cookie, but with a first date which was
19817 older than what is allowed by the "maxlife" cookie parameter, so
19818 the cookie is consider too OLD and is ignored. The request will be
19819 redispatched just as if there was no cookie.
19820
Willy Tarreauc89ccb62012-04-05 21:18:22 +020019821 U : a cookie was present but was not used to select the server because
19822 some other server selection mechanism was used instead (typically a
19823 "use-server" rule).
19824
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019825 - : does not apply (no cookie set in configuration).
19826
19827 - the last character reports what operations were performed on the persistence
19828 cookie returned by the server (only in HTTP mode) :
19829
19830 N : NO cookie was provided by the server, and none was inserted either.
19831
19832 I : no cookie was provided by the server, and the proxy INSERTED one.
19833 Note that in "cookie insert" mode, if the server provides a cookie,
19834 it will still be overwritten and reported as "I" here.
19835
Willy Tarreau996a92c2010-10-13 19:30:47 +020019836 U : the proxy UPDATED the last date in the cookie that was presented by
19837 the client. This can only happen in insert mode with "maxidle". It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030019838 happens every time there is activity at a different date than the
Willy Tarreau996a92c2010-10-13 19:30:47 +020019839 date indicated in the cookie. If any other change happens, such as
19840 a redispatch, then the cookie will be marked as inserted instead.
19841
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019842 P : a cookie was PROVIDED by the server and transmitted as-is.
19843
19844 R : the cookie provided by the server was REWRITTEN by the proxy, which
19845 happens in "cookie rewrite" or "cookie prefix" modes.
19846
19847 D : the cookie provided by the server was DELETED by the proxy.
19848
19849 - : does not apply (no cookie set in configuration).
19850
Willy Tarreau996a92c2010-10-13 19:30:47 +020019851The combination of the two first flags gives a lot of information about what
19852was happening when the session terminated, and why it did terminate. It can be
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019853helpful to detect server saturation, network troubles, local system resource
19854starvation, attacks, etc...
19855
19856The most common termination flags combinations are indicated below. They are
19857alphabetically sorted, with the lowercase set just after the upper case for
19858easier finding and understanding.
19859
19860 Flags Reason
19861
19862 -- Normal termination.
19863
19864 CC The client aborted before the connection could be established to the
19865 server. This can happen when haproxy tries to connect to a recently
19866 dead (or unchecked) server, and the client aborts while haproxy is
19867 waiting for the server to respond or for "timeout connect" to expire.
19868
19869 CD The client unexpectedly aborted during data transfer. This can be
19870 caused by a browser crash, by an intermediate equipment between the
19871 client and haproxy which decided to actively break the connection,
19872 by network routing issues between the client and haproxy, or by a
19873 keep-alive session between the server and the client terminated first
19874 by the client.
Willy Tarreaud72758d2010-01-12 10:42:19 +010019875
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019876 cD The client did not send nor acknowledge any data for as long as the
19877 "timeout client" delay. This is often caused by network failures on
Cyril Bontédc4d9032012-04-08 21:57:39 +020019878 the client side, or the client simply leaving the net uncleanly.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019879
19880 CH The client aborted while waiting for the server to start responding.
19881 It might be the server taking too long to respond or the client
19882 clicking the 'Stop' button too fast.
19883
19884 cH The "timeout client" stroke while waiting for client data during a
19885 POST request. This is sometimes caused by too large TCP MSS values
19886 for PPPoE networks which cannot transport full-sized packets. It can
19887 also happen when client timeout is smaller than server timeout and
19888 the server takes too long to respond.
19889
19890 CQ The client aborted while its session was queued, waiting for a server
19891 with enough empty slots to accept it. It might be that either all the
19892 servers were saturated or that the assigned server was taking too
19893 long a time to respond.
19894
19895 CR The client aborted before sending a full HTTP request. Most likely
19896 the request was typed by hand using a telnet client, and aborted
19897 too early. The HTTP status code is likely a 400 here. Sometimes this
19898 might also be caused by an IDS killing the connection between haproxy
Willy Tarreau0f228a02015-05-01 15:37:53 +020019899 and the client. "option http-ignore-probes" can be used to ignore
19900 connections without any data transfer.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019901
19902 cR The "timeout http-request" stroke before the client sent a full HTTP
19903 request. This is sometimes caused by too large TCP MSS values on the
19904 client side for PPPoE networks which cannot transport full-sized
19905 packets, or by clients sending requests by hand and not typing fast
19906 enough, or forgetting to enter the empty line at the end of the
Willy Tarreau2705a612014-05-23 17:38:34 +020019907 request. The HTTP status code is likely a 408 here. Note: recently,
Willy Tarreau0f228a02015-05-01 15:37:53 +020019908 some browsers started to implement a "pre-connect" feature consisting
19909 in speculatively connecting to some recently visited web sites just
19910 in case the user would like to visit them. This results in many
19911 connections being established to web sites, which end up in 408
19912 Request Timeout if the timeout strikes first, or 400 Bad Request when
19913 the browser decides to close them first. These ones pollute the log
19914 and feed the error counters. Some versions of some browsers have even
19915 been reported to display the error code. It is possible to work
Davor Ocelice9ed2812017-12-25 17:49:28 +010019916 around the undesirable effects of this behavior by adding "option
Willy Tarreau0f228a02015-05-01 15:37:53 +020019917 http-ignore-probes" in the frontend, resulting in connections with
19918 zero data transfer to be totally ignored. This will definitely hide
19919 the errors of people experiencing connectivity issues though.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019920
19921 CT The client aborted while its session was tarpitted. It is important to
19922 check if this happens on valid requests, in order to be sure that no
Willy Tarreau55165fe2009-05-10 12:02:55 +020019923 wrong tarpit rules have been written. If a lot of them happen, it
19924 might make sense to lower the "timeout tarpit" value to something
19925 closer to the average reported "Tw" timer, in order not to consume
19926 resources for just a few attackers.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019927
Willy Tarreau570f2212013-06-10 16:42:09 +020019928 LR The request was intercepted and locally handled by haproxy. Generally
19929 it means that this was a redirect or a stats request.
19930
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010019931 SC The server or an equipment between it and haproxy explicitly refused
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019932 the TCP connection (the proxy received a TCP RST or an ICMP message
19933 in return). Under some circumstances, it can also be the network
Davor Ocelice9ed2812017-12-25 17:49:28 +010019934 stack telling the proxy that the server is unreachable (e.g. no route,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019935 or no ARP response on local network). When this happens in HTTP mode,
19936 the status code is likely a 502 or 503 here.
19937
19938 sC The "timeout connect" stroke before a connection to the server could
19939 complete. When this happens in HTTP mode, the status code is likely a
19940 503 or 504 here.
19941
19942 SD The connection to the server died with an error during the data
19943 transfer. This usually means that haproxy has received an RST from
19944 the server or an ICMP message from an intermediate equipment while
19945 exchanging data with the server. This can be caused by a server crash
19946 or by a network issue on an intermediate equipment.
19947
19948 sD The server did not send nor acknowledge any data for as long as the
19949 "timeout server" setting during the data phase. This is often caused
Davor Ocelice9ed2812017-12-25 17:49:28 +010019950 by too short timeouts on L4 equipment before the server (firewalls,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019951 load-balancers, ...), as well as keep-alive sessions maintained
19952 between the client and the server expiring first on haproxy.
19953
19954 SH The server aborted before sending its full HTTP response headers, or
19955 it crashed while processing the request. Since a server aborting at
19956 this moment is very rare, it would be wise to inspect its logs to
19957 control whether it crashed and why. The logged request may indicate a
19958 small set of faulty requests, demonstrating bugs in the application.
19959 Sometimes this might also be caused by an IDS killing the connection
19960 between haproxy and the server.
19961
19962 sH The "timeout server" stroke before the server could return its
19963 response headers. This is the most common anomaly, indicating too
19964 long transactions, probably caused by server or database saturation.
19965 The immediate workaround consists in increasing the "timeout server"
19966 setting, but it is important to keep in mind that the user experience
19967 will suffer from these long response times. The only long term
19968 solution is to fix the application.
19969
19970 sQ The session spent too much time in queue and has been expired. See
19971 the "timeout queue" and "timeout connect" settings to find out how to
19972 fix this if it happens too often. If it often happens massively in
19973 short periods, it may indicate general problems on the affected
19974 servers due to I/O or database congestion, or saturation caused by
19975 external attacks.
19976
19977 PC The proxy refused to establish a connection to the server because the
19978 process' socket limit has been reached while attempting to connect.
Cyril Bontédc4d9032012-04-08 21:57:39 +020019979 The global "maxconn" parameter may be increased in the configuration
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019980 so that it does not happen anymore. This status is very rare and
19981 might happen when the global "ulimit-n" parameter is forced by hand.
19982
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010019983 PD The proxy blocked an incorrectly formatted chunked encoded message in
19984 a request or a response, after the server has emitted its headers. In
19985 most cases, this will indicate an invalid message from the server to
Davor Ocelice9ed2812017-12-25 17:49:28 +010019986 the client. HAProxy supports chunk sizes of up to 2GB - 1 (2147483647
Willy Tarreauf3a3e132013-08-31 08:16:26 +020019987 bytes). Any larger size will be considered as an error.
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010019988
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019989 PH The proxy blocked the server's response, because it was invalid,
19990 incomplete, dangerous (cache control), or matched a security filter.
19991 In any case, an HTTP 502 error is sent to the client. One possible
19992 cause for this error is an invalid syntax in an HTTP header name
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010019993 containing unauthorized characters. It is also possible but quite
19994 rare, that the proxy blocked a chunked-encoding request from the
19995 client due to an invalid syntax, before the server responded. In this
19996 case, an HTTP 400 error is sent to the client and reported in the
19997 logs.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019998
19999 PR The proxy blocked the client's HTTP request, either because of an
20000 invalid HTTP syntax, in which case it returned an HTTP 400 error to
20001 the client, or because a deny filter matched, in which case it
20002 returned an HTTP 403 error.
20003
20004 PT The proxy blocked the client's request and has tarpitted its
20005 connection before returning it a 500 server error. Nothing was sent
20006 to the server. The connection was maintained open for as long as
20007 reported by the "Tw" timer field.
20008
20009 RC A local resource has been exhausted (memory, sockets, source ports)
20010 preventing the connection to the server from establishing. The error
20011 logs will tell precisely what was missing. This is very rare and can
20012 only be solved by proper system tuning.
20013
Willy Tarreau996a92c2010-10-13 19:30:47 +020020014The combination of the two last flags gives a lot of information about how
20015persistence was handled by the client, the server and by haproxy. This is very
20016important to troubleshoot disconnections, when users complain they have to
20017re-authenticate. The commonly encountered flags are :
20018
20019 -- Persistence cookie is not enabled.
20020
20021 NN No cookie was provided by the client, none was inserted in the
20022 response. For instance, this can be in insert mode with "postonly"
20023 set on a GET request.
20024
20025 II A cookie designating an invalid server was provided by the client,
20026 a valid one was inserted in the response. This typically happens when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040020027 a "server" entry is removed from the configuration, since its cookie
Willy Tarreau996a92c2010-10-13 19:30:47 +020020028 value can be presented by a client when no other server knows it.
20029
20030 NI No cookie was provided by the client, one was inserted in the
20031 response. This typically happens for first requests from every user
20032 in "insert" mode, which makes it an easy way to count real users.
20033
20034 VN A cookie was provided by the client, none was inserted in the
20035 response. This happens for most responses for which the client has
20036 already got a cookie.
20037
20038 VU A cookie was provided by the client, with a last visit date which is
20039 not completely up-to-date, so an updated cookie was provided in
20040 response. This can also happen if there was no date at all, or if
20041 there was a date but the "maxidle" parameter was not set, so that the
20042 cookie can be switched to unlimited time.
20043
20044 EI A cookie was provided by the client, with a last visit date which is
20045 too old for the "maxidle" parameter, so the cookie was ignored and a
20046 new cookie was inserted in the response.
20047
20048 OI A cookie was provided by the client, with a first visit date which is
20049 too old for the "maxlife" parameter, so the cookie was ignored and a
20050 new cookie was inserted in the response.
20051
20052 DI The server designated by the cookie was down, a new server was
20053 selected and a new cookie was emitted in the response.
20054
20055 VI The server designated by the cookie was not marked dead but could not
20056 be reached. A redispatch happened and selected another one, which was
20057 then advertised in the response.
20058
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020059
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200200608.6. Non-printable characters
20061-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020062
20063In order not to cause trouble to log analysis tools or terminals during log
20064consulting, non-printable characters are not sent as-is into log files, but are
20065converted to the two-digits hexadecimal representation of their ASCII code,
20066prefixed by the character '#'. The only characters that can be logged without
20067being escaped are comprised between 32 and 126 (inclusive). Obviously, the
20068escape character '#' itself is also encoded to avoid any ambiguity ("#23"). It
20069is the same for the character '"' which becomes "#22", as well as '{', '|' and
20070'}' when logging headers.
20071
20072Note that the space character (' ') is not encoded in headers, which can cause
20073issues for tools relying on space count to locate fields. A typical header
20074containing spaces is "User-Agent".
20075
20076Last, it has been observed that some syslog daemons such as syslog-ng escape
20077the quote ('"') with a backslash ('\'). The reverse operation can safely be
20078performed since no quote may appear anywhere else in the logs.
20079
20080
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200200818.7. Capturing HTTP cookies
20082---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020083
20084Cookie capture simplifies the tracking a complete user session. This can be
20085achieved using the "capture cookie" statement in the frontend. Please refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020086section 4.2 for more details. Only one cookie can be captured, and the same
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020087cookie will simultaneously be checked in the request ("Cookie:" header) and in
20088the response ("Set-Cookie:" header). The respective values will be reported in
20089the HTTP logs at the "captured_request_cookie" and "captured_response_cookie"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020090locations (see section 8.2.3 about HTTP log format). When either cookie is
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020091not seen, a dash ('-') replaces the value. This way, it's easy to detect when a
20092user switches to a new session for example, because the server will reassign it
20093a new cookie. It is also possible to detect if a server unexpectedly sets a
20094wrong cookie to a client, leading to session crossing.
20095
20096 Examples :
20097 # capture the first cookie whose name starts with "ASPSESSION"
20098 capture cookie ASPSESSION len 32
20099
20100 # capture the first cookie whose name is exactly "vgnvisitor"
20101 capture cookie vgnvisitor= len 32
20102
20103
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200201048.8. Capturing HTTP headers
20105---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020106
20107Header captures are useful to track unique request identifiers set by an upper
20108proxy, virtual host names, user-agents, POST content-length, referrers, etc. In
20109the response, one can search for information about the response length, how the
20110server asked the cache to behave, or an object location during a redirection.
20111
20112Header captures are performed using the "capture request header" and "capture
20113response header" statements in the frontend. Please consult their definition in
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020114section 4.2 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020115
20116It is possible to include both request headers and response headers at the same
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010020117time. Non-existent headers are logged as empty strings, and if one header
20118appears more than once, only its last occurrence will be logged. Request headers
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020119are grouped within braces '{' and '}' in the same order as they were declared,
20120and delimited with a vertical bar '|' without any space. Response headers
20121follow the same representation, but are displayed after a space following the
20122request headers block. These blocks are displayed just before the HTTP request
20123in the logs.
20124
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020020125As a special case, it is possible to specify an HTTP header capture in a TCP
20126frontend. The purpose is to enable logging of headers which will be parsed in
20127an HTTP backend if the request is then switched to this HTTP backend.
20128
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020129 Example :
20130 # This instance chains to the outgoing proxy
20131 listen proxy-out
20132 mode http
20133 option httplog
20134 option logasap
20135 log global
20136 server cache1 192.168.1.1:3128
20137
20138 # log the name of the virtual server
20139 capture request header Host len 20
20140
20141 # log the amount of data uploaded during a POST
20142 capture request header Content-Length len 10
20143
20144 # log the beginning of the referrer
20145 capture request header Referer len 20
20146
20147 # server name (useful for outgoing proxies only)
20148 capture response header Server len 20
20149
20150 # logging the content-length is useful with "option logasap"
20151 capture response header Content-Length len 10
20152
Davor Ocelice9ed2812017-12-25 17:49:28 +010020153 # log the expected cache behavior on the response
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020154 capture response header Cache-Control len 8
20155
20156 # the Via header will report the next proxy's name
20157 capture response header Via len 20
20158
20159 # log the URL location during a redirection
20160 capture response header Location len 20
20161
20162 >>> Aug 9 20:26:09 localhost \
20163 haproxy[2022]: 127.0.0.1:34014 [09/Aug/2004:20:26:09] proxy-out \
20164 proxy-out/cache1 0/0/0/162/+162 200 +350 - - ---- 0/0/0/0/0 0/0 \
20165 {fr.adserver.yahoo.co||http://fr.f416.mail.} {|864|private||} \
20166 "GET http://fr.adserver.yahoo.com/"
20167
20168 >>> Aug 9 20:30:46 localhost \
20169 haproxy[2022]: 127.0.0.1:34020 [09/Aug/2004:20:30:46] proxy-out \
20170 proxy-out/cache1 0/0/0/182/+182 200 +279 - - ---- 0/0/0/0/0 0/0 \
20171 {w.ods.org||} {Formilux/0.1.8|3495|||} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010020172 "GET http://trafic.1wt.eu/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020173
20174 >>> Aug 9 20:30:46 localhost \
20175 haproxy[2022]: 127.0.0.1:34028 [09/Aug/2004:20:30:46] proxy-out \
20176 proxy-out/cache1 0/0/2/126/+128 301 +223 - - ---- 0/0/0/0/0 0/0 \
20177 {www.sytadin.equipement.gouv.fr||http://trafic.1wt.eu/} \
20178 {Apache|230|||http://www.sytadin.} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010020179 "GET http://www.sytadin.equipement.gouv.fr/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020180
20181
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200201828.9. Examples of logs
20183---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020184
20185These are real-world examples of logs accompanied with an explanation. Some of
20186them have been made up by hand. The syslog part has been removed for better
20187reading. Their sole purpose is to explain how to decipher them.
20188
20189 >>> haproxy[674]: 127.0.0.1:33318 [15/Oct/2003:08:31:57.130] px-http \
20190 px-http/srv1 6559/0/7/147/6723 200 243 - - ---- 5/3/3/1/0 0/0 \
20191 "HEAD / HTTP/1.0"
20192
20193 => long request (6.5s) entered by hand through 'telnet'. The server replied
20194 in 147 ms, and the session ended normally ('----')
20195
20196 >>> haproxy[674]: 127.0.0.1:33319 [15/Oct/2003:08:31:57.149] px-http \
20197 px-http/srv1 6559/1230/7/147/6870 200 243 - - ---- 324/239/239/99/0 \
20198 0/9 "HEAD / HTTP/1.0"
20199
20200 => Idem, but the request was queued in the global queue behind 9 other
20201 requests, and waited there for 1230 ms.
20202
20203 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.654] px-http \
20204 px-http/srv1 9/0/7/14/+30 200 +243 - - ---- 3/3/3/1/0 0/0 \
20205 "GET /image.iso HTTP/1.0"
20206
20207 => request for a long data transfer. The "logasap" option was specified, so
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010020208 the log was produced just before transferring data. The server replied in
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020209 14 ms, 243 bytes of headers were sent to the client, and total time from
20210 accept to first data byte is 30 ms.
20211
20212 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.925] px-http \
20213 px-http/srv1 9/0/7/14/30 502 243 - - PH-- 3/2/2/0/0 0/0 \
20214 "GET /cgi-bin/bug.cgi? HTTP/1.0"
20215
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020020216 => the proxy blocked a server response either because of an "http-response
20217 deny" rule, or because the response was improperly formatted and not
20218 HTTP-compliant, or because it blocked sensitive information which risked
20219 being cached. In this case, the response is replaced with a "502 bad
20220 gateway". The flags ("PH--") tell us that it was haproxy who decided to
20221 return the 502 and not the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020222
20223 >>> haproxy[18113]: 127.0.0.1:34548 [15/Oct/2003:15:18:55.798] px-http \
Willy Tarreaud72758d2010-01-12 10:42:19 +010020224 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 +010020225
20226 => the client never completed its request and aborted itself ("C---") after
20227 8.5s, while the proxy was waiting for the request headers ("-R--").
20228 Nothing was sent to any server.
20229
20230 >>> haproxy[18113]: 127.0.0.1:34549 [15/Oct/2003:15:19:06.103] px-http \
20231 px-http/<NOSRV> -1/-1/-1/-1/50001 408 0 - - cR-- 2/2/2/0/0 0/0 ""
20232
20233 => The client never completed its request, which was aborted by the
20234 time-out ("c---") after 50s, while the proxy was waiting for the request
Davor Ocelice9ed2812017-12-25 17:49:28 +010020235 headers ("-R--"). Nothing was sent to any server, but the proxy could
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020236 send a 408 return code to the client.
20237
20238 >>> haproxy[18989]: 127.0.0.1:34550 [15/Oct/2003:15:24:28.312] px-tcp \
20239 px-tcp/srv1 0/0/5007 0 cD 0/0/0/0/0 0/0
20240
20241 => This log was produced with "option tcplog". The client timed out after
20242 5 seconds ("c----").
20243
20244 >>> haproxy[18989]: 10.0.0.1:34552 [15/Oct/2003:15:26:31.462] px-http \
20245 px-http/srv1 3183/-1/-1/-1/11215 503 0 - - SC-- 205/202/202/115/3 \
Willy Tarreaud72758d2010-01-12 10:42:19 +010020246 0/0 "HEAD / HTTP/1.0"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020247
20248 => The request took 3s to complete (probably a network problem), and the
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020249 connection to the server failed ('SC--') after 4 attempts of 2 seconds
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020250 (config says 'retries 3'), and no redispatch (otherwise we would have
20251 seen "/+3"). Status code 503 was returned to the client. There were 115
20252 connections on this server, 202 connections on this proxy, and 205 on
20253 the global process. It is possible that the server refused the
20254 connection because of too many already established.
Willy Tarreau844e3c52008-01-11 16:28:18 +010020255
Willy Tarreau52b2d222011-09-07 23:48:48 +020020256
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200202579. Supported filters
20258--------------------
20259
20260Here are listed officially supported filters with the list of parameters they
20261accept. Depending on compile options, some of these filters might be
20262unavailable. The list of available filters is reported in haproxy -vv.
20263
20264See also : "filter"
20265
202669.1. Trace
20267----------
20268
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010020269filter trace [name <name>] [random-parsing] [random-forwarding] [hexdump]
Christopher Fauletc3fe5332016-04-07 15:30:10 +020020270
20271 Arguments:
20272 <name> is an arbitrary name that will be reported in
20273 messages. If no name is provided, "TRACE" is used.
20274
20275 <random-parsing> enables the random parsing of data exchanged between
20276 the client and the server. By default, this filter
20277 parses all available data. With this parameter, it
20278 only parses a random amount of the available data.
20279
Davor Ocelice9ed2812017-12-25 17:49:28 +010020280 <random-forwarding> enables the random forwarding of parsed data. By
Christopher Fauletc3fe5332016-04-07 15:30:10 +020020281 default, this filter forwards all previously parsed
20282 data. With this parameter, it only forwards a random
20283 amount of the parsed data.
20284
Davor Ocelice9ed2812017-12-25 17:49:28 +010020285 <hexdump> dumps all forwarded data to the server and the client.
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010020286
Christopher Fauletc3fe5332016-04-07 15:30:10 +020020287This filter can be used as a base to develop new filters. It defines all
20288callbacks and print a message on the standard error stream (stderr) with useful
20289information for all of them. It may be useful to debug the activity of other
20290filters or, quite simply, HAProxy's activity.
20291
20292Using <random-parsing> and/or <random-forwarding> parameters is a good way to
20293tests the behavior of a filter that parses data exchanged between a client and
20294a server by adding some latencies in the processing.
20295
20296
202979.2. HTTP compression
20298---------------------
20299
20300filter compression
20301
20302The HTTP compression has been moved in a filter in HAProxy 1.7. "compression"
20303keyword must still be used to enable and configure the HTTP compression. And
Christopher Fauletb30b3102019-09-12 23:03:09 +020020304when no other filter is used, it is enough. When used with the cache or the
20305fcgi-app enabled, it is also enough. In this case, the compression is always
20306done after the response is stored in the cache. But it is mandatory to
20307explicitly use a filter line to enable the HTTP compression when at least one
20308filter other than the cache or the fcgi-app is used for the same
20309listener/frontend/backend. This is important to know the filters evaluation
20310order.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020020311
Christopher Fauletb30b3102019-09-12 23:03:09 +020020312See also : "compression", section 9.4 about the cache filter and section 9.5
20313 about the fcgi-app filter.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020020314
20315
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +0200203169.3. Stream Processing Offload Engine (SPOE)
20317--------------------------------------------
20318
20319filter spoe [engine <name>] config <file>
20320
20321 Arguments :
20322
20323 <name> is the engine name that will be used to find the right scope in
20324 the configuration file. If not provided, all the file will be
20325 parsed.
20326
20327 <file> is the path of the engine configuration file. This file can
20328 contain configuration of several engines. In this case, each
20329 part must be placed in its own scope.
20330
20331The Stream Processing Offload Engine (SPOE) is a filter communicating with
20332external components. It allows the offload of some specifics processing on the
Davor Ocelice9ed2812017-12-25 17:49:28 +010020333streams in tiered applications. These external components and information
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020020334exchanged with them are configured in dedicated files, for the main part. It
20335also requires dedicated backends, defined in HAProxy configuration.
20336
20337SPOE communicates with external components using an in-house binary protocol,
20338the Stream Processing Offload Protocol (SPOP).
20339
Tim Düsterhus4896c442016-11-29 02:15:19 +010020340For all information about the SPOE configuration and the SPOP specification, see
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020020341"doc/SPOE.txt".
20342
Christopher Faulet99a17a22018-12-11 09:18:27 +0100203439.4. Cache
20344----------
20345
20346filter cache <name>
20347
20348 Arguments :
20349
20350 <name> is name of the cache section this filter will use.
20351
20352The cache uses a filter to store cacheable responses. The HTTP rules
20353"cache-store" and "cache-use" must be used to define how and when to use a
John Roeslerfb2fce12019-07-10 15:45:51 -050020354cache. By default the corresponding filter is implicitly defined. And when no
Christopher Fauletb30b3102019-09-12 23:03:09 +020020355other filters than fcgi-app or compression are used, it is enough. In such
20356case, the compression filter is always evaluated after the cache filter. But it
20357is mandatory to explicitly use a filter line to use a cache when at least one
20358filter other than the compression or the fcgi-app is used for the same
Christopher Faulet27d93c32018-12-15 22:32:02 +010020359listener/frontend/backend. This is important to know the filters evaluation
20360order.
Christopher Faulet99a17a22018-12-11 09:18:27 +010020361
Christopher Fauletb30b3102019-09-12 23:03:09 +020020362See also : section 9.2 about the compression filter, section 9.5 about the
20363 fcgi-app filter and section 6 about cache.
20364
20365
203669.5. Fcgi-app
20367-------------
20368
Daniel Corbett67a82712020-07-06 23:01:19 -040020369filter fcgi-app <name>
Christopher Fauletb30b3102019-09-12 23:03:09 +020020370
20371 Arguments :
20372
20373 <name> is name of the fcgi-app section this filter will use.
20374
20375The FastCGI application uses a filter to evaluate all custom parameters on the
20376request path, and to process the headers on the response path. the <name> must
20377reference an existing fcgi-app section. The directive "use-fcgi-app" should be
20378used to define the application to use. By default the corresponding filter is
20379implicitly defined. And when no other filters than cache or compression are
20380used, it is enough. But it is mandatory to explicitly use a filter line to a
20381fcgi-app when at least one filter other than the compression or the cache is
20382used for the same backend. This is important to know the filters evaluation
20383order.
20384
20385See also: "use-fcgi-app", section 9.2 about the compression filter, section 9.4
20386 about the cache filter and section 10 about FastCGI application.
20387
20388
2038910. FastCGI applications
20390-------------------------
20391
20392HAProxy is able to send HTTP requests to Responder FastCGI applications. This
20393feature was added in HAProxy 2.1. To do so, servers must be configured to use
20394the FastCGI protocol (using the keyword "proto fcgi" on the server line) and a
20395FastCGI application must be configured and used by the backend managing these
20396servers (using the keyword "use-fcgi-app" into the proxy section). Several
20397FastCGI applications may be defined, but only one can be used at a time by a
20398backend.
20399
20400HAProxy implements all features of the FastCGI specification for Responder
20401application. Especially it is able to multiplex several requests on a simple
20402connection.
20403
2040410.1. Setup
20405-----------
20406
2040710.1.1. Fcgi-app section
20408--------------------------
20409
20410fcgi-app <name>
20411 Declare a FastCGI application named <name>. To be valid, at least the
20412 document root must be defined.
20413
20414acl <aclname> <criterion> [flags] [operator] <value> ...
20415 Declare or complete an access list.
20416
20417 See "acl" keyword in section 4.2 and section 7 about ACL usage for
20418 details. ACLs defined for a FastCGI application are private. They cannot be
20419 used by any other application or by any proxy. In the same way, ACLs defined
20420 in any other section are not usable by a FastCGI application. However,
20421 Pre-defined ACLs are available.
20422
20423docroot <path>
20424 Define the document root on the remote host. <path> will be used to build
20425 the default value of FastCGI parameters SCRIPT_FILENAME and
20426 PATH_TRANSLATED. It is a mandatory setting.
20427
20428index <script-name>
20429 Define the script name that will be appended after an URI that ends with a
20430 slash ("/") to set the default value of the FastCGI parameter SCRIPT_NAME. It
20431 is an optional setting.
20432
20433 Example :
20434 index index.php
20435
20436log-stderr global
20437log-stderr <address> [len <length>] [format <format>]
20438 [sample <ranges>:<smp_size>] <facility> [<level> [<minlevel>]]
20439 Enable logging of STDERR messages reported by the FastCGI application.
20440
20441 See "log" keyword in section 4.2 for details. It is an optional setting. By
20442 default STDERR messages are ignored.
20443
20444pass-header <name> [ { if | unless } <condition> ]
20445 Specify the name of a request header which will be passed to the FastCGI
20446 application. It may optionally be followed by an ACL-based condition, in
20447 which case it will only be evaluated if the condition is true.
20448
20449 Most request headers are already available to the FastCGI application,
20450 prefixed with "HTTP_". Thus, this directive is only required to pass headers
20451 that are purposefully omitted. Currently, the headers "Authorization",
20452 "Proxy-Authorization" and hop-by-hop headers are omitted.
20453
20454 Note that the headers "Content-type" and "Content-length" are never passed to
20455 the FastCGI application because they are already converted into parameters.
20456
20457path-info <regex>
Christopher Faulet28cb3662020-02-14 14:47:37 +010020458 Define a regular expression to extract the script-name and the path-info from
Christopher Faulet6c57f2d2020-02-14 16:55:52 +010020459 the URL-decoded path. Thus, <regex> may have two captures: the first one to
20460 capture the script name and the second one to capture the path-info. The
20461 first one is mandatory, the second one is optional. This way, it is possible
20462 to extract the script-name from the path ignoring the path-info. It is an
20463 optional setting. If it is not defined, no matching is performed on the
20464 path. and the FastCGI parameters PATH_INFO and PATH_TRANSLATED are not
20465 filled.
Christopher Faulet28cb3662020-02-14 14:47:37 +010020466
20467 For security reason, when this regular expression is defined, the newline and
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050020468 the null characters are forbidden from the path, once URL-decoded. The reason
Christopher Faulet28cb3662020-02-14 14:47:37 +010020469 to such limitation is because otherwise the matching always fails (due to a
20470 limitation one the way regular expression are executed in HAProxy). So if one
20471 of these two characters is found in the URL-decoded path, an error is
20472 returned to the client. The principle of least astonishment is applied here.
Christopher Fauletb30b3102019-09-12 23:03:09 +020020473
20474 Example :
Christopher Faulet6c57f2d2020-02-14 16:55:52 +010020475 path-info ^(/.+\.php)(/.*)?$ # both script-name and path-info may be set
20476 path-info ^(/.+\.php) # the path-info is ignored
Christopher Fauletb30b3102019-09-12 23:03:09 +020020477
20478option get-values
20479no option get-values
20480 Enable or disable the retrieve of variables about connection management.
20481
Daniel Corbett67a82712020-07-06 23:01:19 -040020482 HAProxy is able to send the record FCGI_GET_VALUES on connection
Christopher Fauletb30b3102019-09-12 23:03:09 +020020483 establishment to retrieve the value for following variables:
20484
20485 * FCGI_MAX_REQS The maximum number of concurrent requests this
20486 application will accept.
20487
William Lallemand93e548e2019-09-30 13:54:02 +020020488 * FCGI_MPXS_CONNS "0" if this application does not multiplex connections,
20489 "1" otherwise.
Christopher Fauletb30b3102019-09-12 23:03:09 +020020490
20491 Some FastCGI applications does not support this feature. Some others close
Ilya Shipitsin11057a32020-06-21 21:18:27 +050020492 the connection immediately after sending their response. So, by default, this
Christopher Fauletb30b3102019-09-12 23:03:09 +020020493 option is disabled.
20494
20495 Note that the maximum number of concurrent requests accepted by a FastCGI
20496 application is a connection variable. It only limits the number of streams
20497 per connection. If the global load must be limited on the application, the
20498 server parameters "maxconn" and "pool-max-conn" must be set. In addition, if
20499 an application does not support connection multiplexing, the maximum number
20500 of concurrent requests is automatically set to 1.
20501
20502option keep-conn
20503no option keep-conn
20504 Instruct the FastCGI application to keep the connection open or not after
20505 sending a response.
20506
20507 If disabled, the FastCGI application closes the connection after responding
20508 to this request. By default, this option is enabled.
20509
20510option max-reqs <reqs>
20511 Define the maximum number of concurrent requests this application will
20512 accept.
20513
20514 This option may be overwritten if the variable FCGI_MAX_REQS is retrieved
20515 during connection establishment. Furthermore, if the application does not
20516 support connection multiplexing, this option will be ignored. By default set
20517 to 1.
20518
20519option mpxs-conns
20520no option mpxs-conns
20521 Enable or disable the support of connection multiplexing.
20522
20523 This option may be overwritten if the variable FCGI_MPXS_CONNS is retrieved
20524 during connection establishment. It is disabled by default.
20525
20526set-param <name> <fmt> [ { if | unless } <condition> ]
20527 Set a FastCGI parameter that should be passed to this application. Its
20528 value, defined by <fmt> must follows the log-format rules (see section 8.2.4
20529 "Custom Log format"). It may optionally be followed by an ACL-based
20530 condition, in which case it will only be evaluated if the condition is true.
20531
20532 With this directive, it is possible to overwrite the value of default FastCGI
20533 parameters. If the value is evaluated to an empty string, the rule is
20534 ignored. These directives are evaluated in their declaration order.
20535
20536 Example :
20537 # PHP only, required if PHP was built with --enable-force-cgi-redirect
20538 set-param REDIRECT_STATUS 200
20539
20540 set-param PHP_AUTH_DIGEST %[req.hdr(Authorization)]
20541
20542
2054310.1.2. Proxy section
20544---------------------
20545
20546use-fcgi-app <name>
20547 Define the FastCGI application to use for the backend.
20548
20549 Arguments :
20550 <name> is the name of the FastCGI application to use.
20551
20552 This keyword is only available for HTTP proxies with the backend capability
20553 and with at least one FastCGI server. However, FastCGI servers can be mixed
20554 with HTTP servers. But except there is a good reason to do so, it is not
20555 recommended (see section 10.3 about the limitations for details). Only one
20556 application may be defined at a time per backend.
20557
20558 Note that, once a FastCGI application is referenced for a backend, depending
20559 on the configuration some processing may be done even if the request is not
20560 sent to a FastCGI server. Rules to set parameters or pass headers to an
20561 application are evaluated.
20562
20563
2056410.1.3. Example
20565---------------
20566
20567 frontend front-http
20568 mode http
20569 bind *:80
20570 bind *:
20571
20572 use_backend back-dynamic if { path_reg ^/.+\.php(/.*)?$ }
20573 default_backend back-static
20574
20575 backend back-static
20576 mode http
20577 server www A.B.C.D:80
20578
20579 backend back-dynamic
20580 mode http
20581 use-fcgi-app php-fpm
20582 server php-fpm A.B.C.D:9000 proto fcgi
20583
20584 fcgi-app php-fpm
20585 log-stderr global
20586 option keep-conn
20587
20588 docroot /var/www/my-app
20589 index index.php
20590 path-info ^(/.+\.php)(/.*)?$
20591
20592
2059310.2. Default parameters
20594------------------------
20595
20596A Responder FastCGI application has the same purpose as a CGI/1.1 program. In
20597the CGI/1.1 specification (RFC3875), several variables must be passed to the
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050020598script. So HAProxy set them and some others commonly used by FastCGI
Christopher Fauletb30b3102019-09-12 23:03:09 +020020599applications. All these variables may be overwritten, with caution though.
20600
20601 +-------------------+-----------------------------------------------------+
20602 | AUTH_TYPE | Identifies the mechanism, if any, used by HAProxy |
20603 | | to authenticate the user. Concretely, only the |
20604 | | BASIC authentication mechanism is supported. |
20605 | | |
20606 +-------------------+-----------------------------------------------------+
20607 | CONTENT_LENGTH | Contains the size of the message-body attached to |
20608 | | the request. It means only requests with a known |
20609 | | size are considered as valid and sent to the |
20610 | | application. |
20611 | | |
20612 +-------------------+-----------------------------------------------------+
20613 | CONTENT_TYPE | Contains the type of the message-body attached to |
20614 | | the request. It may not be set. |
20615 | | |
20616 +-------------------+-----------------------------------------------------+
20617 | DOCUMENT_ROOT | Contains the document root on the remote host under |
20618 | | which the script should be executed, as defined in |
20619 | | the application's configuration. |
20620 | | |
20621 +-------------------+-----------------------------------------------------+
20622 | GATEWAY_INTERFACE | Contains the dialect of CGI being used by HAProxy |
20623 | | to communicate with the FastCGI application. |
20624 | | Concretely, it is set to "CGI/1.1". |
20625 | | |
20626 +-------------------+-----------------------------------------------------+
20627 | PATH_INFO | Contains the portion of the URI path hierarchy |
20628 | | following the part that identifies the script |
20629 | | itself. To be set, the directive "path-info" must |
20630 | | be defined. |
20631 | | |
20632 +-------------------+-----------------------------------------------------+
20633 | PATH_TRANSLATED | If PATH_INFO is set, it is its translated version. |
20634 | | It is the concatenation of DOCUMENT_ROOT and |
20635 | | PATH_INFO. If PATH_INFO is not set, this parameters |
20636 | | is not set too. |
20637 | | |
20638 +-------------------+-----------------------------------------------------+
20639 | QUERY_STRING | Contains the request's query string. It may not be |
20640 | | set. |
20641 | | |
20642 +-------------------+-----------------------------------------------------+
20643 | REMOTE_ADDR | Contains the network address of the client sending |
20644 | | the request. |
20645 | | |
20646 +-------------------+-----------------------------------------------------+
20647 | REMOTE_USER | Contains the user identification string supplied by |
20648 | | client as part of user authentication. |
20649 | | |
20650 +-------------------+-----------------------------------------------------+
20651 | REQUEST_METHOD | Contains the method which should be used by the |
20652 | | script to process the request. |
20653 | | |
20654 +-------------------+-----------------------------------------------------+
20655 | REQUEST_URI | Contains the request's URI. |
20656 | | |
20657 +-------------------+-----------------------------------------------------+
20658 | SCRIPT_FILENAME | Contains the absolute pathname of the script. it is |
20659 | | the concatenation of DOCUMENT_ROOT and SCRIPT_NAME. |
20660 | | |
20661 +-------------------+-----------------------------------------------------+
20662 | SCRIPT_NAME | Contains the name of the script. If the directive |
20663 | | "path-info" is defined, it is the first part of the |
20664 | | URI path hierarchy, ending with the script name. |
20665 | | Otherwise, it is the entire URI path. |
20666 | | |
20667 +-------------------+-----------------------------------------------------+
20668 | SERVER_NAME | Contains the name of the server host to which the |
20669 | | client request is directed. It is the value of the |
20670 | | header "Host", if defined. Otherwise, the |
20671 | | destination address of the connection on the client |
20672 | | side. |
20673 | | |
20674 +-------------------+-----------------------------------------------------+
20675 | SERVER_PORT | Contains the destination TCP port of the connection |
20676 | | on the client side, which is the port the client |
20677 | | connected to. |
20678 | | |
20679 +-------------------+-----------------------------------------------------+
20680 | SERVER_PROTOCOL | Contains the request's protocol. |
20681 | | |
20682 +-------------------+-----------------------------------------------------+
20683 | HTTPS | Set to a non-empty value ("on") if the script was |
20684 | | queried through the HTTPS protocol. |
20685 | | |
20686 +-------------------+-----------------------------------------------------+
20687
20688
2068910.3. Limitations
20690------------------
20691
20692The current implementation have some limitations. The first one is about the
20693way some request headers are hidden to the FastCGI applications. This happens
20694during the headers analysis, on the backend side, before the connection
20695establishment. At this stage, HAProxy know the backend is using a FastCGI
20696application but it don't know if the request will be routed to a FastCGI server
20697or not. But to hide request headers, it simply removes them from the HTX
20698message. So, if the request is finally routed to an HTTP server, it never see
20699these headers. For this reason, it is not recommended to mix FastCGI servers
20700and HTTP servers under the same backend.
20701
20702Similarly, the rules "set-param" and "pass-header" are evaluated during the
20703request headers analysis. So the evaluation is always performed, even if the
20704requests is finally forwarded to an HTTP server.
20705
20706About the rules "set-param", when a rule is applied, a pseudo header is added
20707into the HTX message. So, the same way than for HTTP header rewrites, it may
20708fail if the buffer is full. The rules "set-param" will compete with
20709"http-request" ones.
20710
20711Finally, all FastCGI params and HTTP headers are sent into a unique record
20712FCGI_PARAM. Encoding of this record must be done in one pass, otherwise a
20713processing error is returned. It means the record FCGI_PARAM, once encoded,
20714must not exceeds the size of a buffer. However, there is no reserve to respect
20715here.
William Lallemand86d0df02017-11-24 21:36:45 +010020716
Willy Tarreau0ba27502007-12-24 16:55:16 +010020717/*
20718 * Local variables:
20719 * fill-column: 79
20720 * End:
20721 */