blob: b4a316ef4869ff41511502bd6bf3d50f6b508922 [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 Tarreau33205c22020-07-07 16:35:28 +02005 version 2.3
Willy Tarreau6a06a402007-07-15 20:15:28 +02006 willy tarreau
Christopher Faulet7b663c12021-07-07 18:26:59 +02007 2021/07/07
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
William Lallemand83972542020-11-18 10:41:24 +0100573.10. Log forwarding
Willy Tarreauc57f0e22009-05-10 13:12:33 +020058
594. Proxies
604.1. Proxy keywords matrix
614.2. Alphabetically sorted keywords reference
62
Davor Ocelice9ed2812017-12-25 17:49:28 +0100635. Bind and server options
Willy Tarreau086fbf52012-09-24 20:34:51 +0200645.1. Bind options
655.2. Server and default-server options
Baptiste Assmann1fa66662015-04-14 00:28:47 +0200665.3. Server DNS resolution
675.3.1. Global overview
685.3.2. The resolvers section
Willy Tarreauc57f0e22009-05-10 13:12:33 +020069
Julien Pivotto6ccee412019-11-27 15:49:54 +0100706. Cache
716.1. Limitation
726.2. Setup
736.2.1. Cache section
746.2.2. Proxy section
75
Willy Tarreau74ca5042013-06-11 23:12:07 +0200767. Using ACLs and fetching samples
777.1. ACL basics
787.1.1. Matching booleans
797.1.2. Matching integers
807.1.3. Matching strings
817.1.4. Matching regular expressions (regexes)
827.1.5. Matching arbitrary data blocks
837.1.6. Matching IPv4 and IPv6 addresses
847.2. Using ACLs to form conditions
857.3. Fetching samples
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200867.3.1. Converters
877.3.2. Fetching samples from internal states
887.3.3. Fetching samples at Layer 4
897.3.4. Fetching samples at Layer 5
907.3.5. Fetching samples from buffer contents (Layer 6)
917.3.6. Fetching HTTP samples (Layer 7)
Christopher Faulete596d182020-05-05 17:46:34 +0200927.3.7. Fetching samples for developers
Willy Tarreau74ca5042013-06-11 23:12:07 +0200937.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020094
958. Logging
968.1. Log levels
978.2. Log formats
988.2.1. Default log format
998.2.2. TCP log format
1008.2.3. HTTP log format
William Lallemand48940402012-01-30 16:47:22 +01001018.2.4. Custom log format
Willy Tarreau5f51e1a2012-12-03 18:40:10 +01001028.2.5. Error log format
Willy Tarreauc57f0e22009-05-10 13:12:33 +02001038.3. Advanced logging options
1048.3.1. Disabling logging of external tests
1058.3.2. Logging before waiting for the session to terminate
1068.3.3. Raising log level upon errors
1078.3.4. Disabling logging of successful connections
1088.4. Timing events
1098.5. Session state at disconnection
1108.6. Non-printable characters
1118.7. Capturing HTTP cookies
1128.8. Capturing HTTP headers
1138.9. Examples of logs
114
Christopher Fauletc3fe5332016-04-07 15:30:10 +02001159. Supported filters
1169.1. Trace
1179.2. HTTP compression
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +02001189.3. Stream Processing Offload Engine (SPOE)
Christopher Faulet99a17a22018-12-11 09:18:27 +01001199.4. Cache
Christopher Fauletb30b3102019-09-12 23:03:09 +02001209.5. fcgi-app
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200121
Christopher Fauletb30b3102019-09-12 23:03:09 +020012210. FastCGI applications
12310.1. Setup
12410.1.1. Fcgi-app section
12510.1.2. Proxy section
12610.1.3. Example
12710.2. Default parameters
12810.3. Limitations
129
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200130
1311. Quick reminder about HTTP
132----------------------------
133
Davor Ocelice9ed2812017-12-25 17:49:28 +0100134When HAProxy is running in HTTP mode, both the request and the response are
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200135fully analyzed and indexed, thus it becomes possible to build matching criteria
136on almost anything found in the contents.
137
138However, it is important to understand how HTTP requests and responses are
139formed, and how HAProxy decomposes them. It will then become easier to write
140correct rules and to debug existing configurations.
141
142
1431.1. The HTTP transaction model
144-------------------------------
145
146The HTTP protocol is transaction-driven. This means that each request will lead
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100147to one and only one response. Traditionally, a TCP connection is established
Davor Ocelice9ed2812017-12-25 17:49:28 +0100148from the client to the server, a request is sent by the client through the
149connection, the server responds, and the connection is closed. A new request
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200150will involve a new connection :
151
152 [CON1] [REQ1] ... [RESP1] [CLO1] [CON2] [REQ2] ... [RESP2] [CLO2] ...
153
154In this mode, called the "HTTP close" mode, there are as many connection
155establishments as there are HTTP transactions. Since the connection is closed
156by the server after the response, the client does not need to know the content
157length.
158
159Due to the transactional nature of the protocol, it was possible to improve it
160to avoid closing a connection between two subsequent transactions. In this mode
161however, it is mandatory that the server indicates the content length for each
162response so that the client does not wait indefinitely. For this, a special
163header is used: "Content-length". This mode is called the "keep-alive" mode :
164
165 [CON] [REQ1] ... [RESP1] [REQ2] ... [RESP2] [CLO] ...
166
167Its advantages are a reduced latency between transactions, and less processing
168power required on the server side. It is generally better than the close mode,
169but not always because the clients often limit their concurrent connections to
Patrick Mezard9ec2ec42010-06-12 17:02:45 +0200170a smaller value.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200171
Willy Tarreau95c4e142017-11-26 12:18:55 +0100172Another improvement in the communications is the pipelining mode. It still uses
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200173keep-alive, but the client does not wait for the first response to send the
174second request. This is useful for fetching large number of images composing a
175page :
176
177 [CON] [REQ1] [REQ2] ... [RESP1] [RESP2] [CLO] ...
178
179This can obviously have a tremendous benefit on performance because the network
180latency is eliminated between subsequent requests. Many HTTP agents do not
181correctly support pipelining since there is no way to associate a response with
182the corresponding request in HTTP. For this reason, it is mandatory for the
Cyril Bonté78caf842010-03-10 22:41:43 +0100183server to reply in the exact same order as the requests were received.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200184
Willy Tarreau95c4e142017-11-26 12:18:55 +0100185The next improvement is the multiplexed mode, as implemented in HTTP/2. This
186time, each transaction is assigned a single stream identifier, and all streams
187are multiplexed over an existing connection. Many requests can be sent in
188parallel by the client, and responses can arrive in any order since they also
189carry the stream identifier.
190
Willy Tarreau70dffda2014-01-30 03:07:23 +0100191By default HAProxy operates in keep-alive mode with regards to persistent
192connections: for each connection it processes each request and response, and
193leaves the connection idle on both sides between the end of a response and the
Willy Tarreau95c4e142017-11-26 12:18:55 +0100194start of a new request. When it receives HTTP/2 connections from a client, it
195processes all the requests in parallel and leaves the connection idling,
196waiting for new requests, just as if it was a keep-alive HTTP connection.
Patrick Mezard9ec2ec42010-06-12 17:02:45 +0200197
Christopher Faulet315b39c2018-09-21 16:26:19 +0200198HAProxy supports 4 connection modes :
Willy Tarreau70dffda2014-01-30 03:07:23 +0100199 - keep alive : all requests and responses are processed (default)
200 - tunnel : only the first request and response are processed,
Christopher Faulet6c9bbb22019-03-26 21:37:23 +0100201 everything else is forwarded with no analysis (deprecated).
Willy Tarreau70dffda2014-01-30 03:07:23 +0100202 - server close : the server-facing connection is closed after the response.
Christopher Faulet315b39c2018-09-21 16:26:19 +0200203 - close : the connection is actively closed after end of response.
Willy Tarreau70dffda2014-01-30 03:07:23 +0100204
Willy Tarreau95c4e142017-11-26 12:18:55 +0100205
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200206
2071.2. HTTP request
208-----------------
209
210First, let's consider this HTTP request :
211
212 Line Contents
Willy Tarreaud72758d2010-01-12 10:42:19 +0100213 number
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200214 1 GET /serv/login.php?lang=en&profile=2 HTTP/1.1
215 2 Host: www.mydomain.com
216 3 User-agent: my small browser
217 4 Accept: image/jpeg, image/gif
218 5 Accept: image/png
219
220
2211.2.1. The Request line
222-----------------------
223
224Line 1 is the "request line". It is always composed of 3 fields :
225
226 - a METHOD : GET
227 - a URI : /serv/login.php?lang=en&profile=2
228 - a version tag : HTTP/1.1
229
230All of them are delimited by what the standard calls LWS (linear white spaces),
231which are commonly spaces, but can also be tabs or line feeds/carriage returns
232followed by spaces/tabs. The method itself cannot contain any colon (':') and
233is limited to alphabetic letters. All those various combinations make it
234desirable that HAProxy performs the splitting itself rather than leaving it to
235the user to write a complex or inaccurate regular expression.
236
237The URI itself can have several forms :
238
239 - A "relative URI" :
240
241 /serv/login.php?lang=en&profile=2
242
243 It is a complete URL without the host part. This is generally what is
244 received by servers, reverse proxies and transparent proxies.
245
246 - An "absolute URI", also called a "URL" :
247
248 http://192.168.0.12:8080/serv/login.php?lang=en&profile=2
249
250 It is composed of a "scheme" (the protocol name followed by '://'), a host
251 name or address, optionally a colon (':') followed by a port number, then
252 a relative URI beginning at the first slash ('/') after the address part.
253 This is generally what proxies receive, but a server supporting HTTP/1.1
254 must accept this form too.
255
256 - a star ('*') : this form is only accepted in association with the OPTIONS
257 method and is not relayable. It is used to inquiry a next hop's
258 capabilities.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100259
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200260 - an address:port combination : 192.168.0.12:80
261 This is used with the CONNECT method, which is used to establish TCP
262 tunnels through HTTP proxies, generally for HTTPS, but sometimes for
263 other protocols too.
264
265In a relative URI, two sub-parts are identified. The part before the question
266mark is called the "path". It is typically the relative path to static objects
267on the server. The part after the question mark is called the "query string".
268It is mostly used with GET requests sent to dynamic scripts and is very
269specific to the language, framework or application in use.
270
Willy Tarreau95c4e142017-11-26 12:18:55 +0100271HTTP/2 doesn't convey a version information with the request, so the version is
Davor Ocelice9ed2812017-12-25 17:49:28 +0100272assumed to be the same as the one of the underlying protocol (i.e. "HTTP/2").
Willy Tarreau95c4e142017-11-26 12:18:55 +0100273
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200274
2751.2.2. The request headers
276--------------------------
277
278The headers start at the second line. They are composed of a name at the
279beginning of the line, immediately followed by a colon (':'). Traditionally,
280an LWS is added after the colon but that's not required. Then come the values.
281Multiple identical headers may be folded into one single line, delimiting the
282values with commas, provided that their order is respected. This is commonly
283encountered in the "Cookie:" field. A header may span over multiple lines if
284the subsequent lines begin with an LWS. In the example in 1.2, lines 4 and 5
285define a total of 3 values for the "Accept:" header.
286
Davor Ocelice9ed2812017-12-25 17:49:28 +0100287Contrary to a common misconception, header names are not case-sensitive, and
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200288their values are not either if they refer to other header names (such as the
Willy Tarreau95c4e142017-11-26 12:18:55 +0100289"Connection:" header). In HTTP/2, header names are always sent in lower case,
Willy Tarreau253c2512020-07-07 15:55:23 +0200290as can be seen when running in debug mode. Internally, all header names are
291normalized to lower case so that HTTP/1.x and HTTP/2 use the exact same
292representation, and they are sent as-is on the other side. This explains why an
293HTTP/1.x request typed with camel case is delivered in lower case.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200294
295The end of the headers is indicated by the first empty line. People often say
296that it's a double line feed, which is not exact, even if a double line feed
297is one valid form of empty line.
298
299Fortunately, HAProxy takes care of all these complex combinations when indexing
300headers, checking values and counting them, so there is no reason to worry
301about the way they could be written, but it is important not to accuse an
302application of being buggy if it does unusual, valid things.
303
304Important note:
Lukas Tribus23953682017-04-28 13:24:30 +0000305 As suggested by RFC7231, HAProxy normalizes headers by replacing line breaks
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200306 in the middle of headers by LWS in order to join multi-line headers. This
307 is necessary for proper analysis and helps less capable HTTP parsers to work
308 correctly and not to be fooled by such complex constructs.
309
310
3111.3. HTTP response
312------------------
313
314An HTTP response looks very much like an HTTP request. Both are called HTTP
315messages. Let's consider this HTTP response :
316
317 Line Contents
Willy Tarreaud72758d2010-01-12 10:42:19 +0100318 number
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200319 1 HTTP/1.1 200 OK
320 2 Content-length: 350
321 3 Content-Type: text/html
322
Willy Tarreau816b9792009-09-15 21:25:21 +0200323As a special case, HTTP supports so called "Informational responses" as status
324codes 1xx. These messages are special in that they don't convey any part of the
325response, they're just used as sort of a signaling message to ask a client to
Willy Tarreau5843d1a2010-02-01 15:13:32 +0100326continue to post its request for instance. In the case of a status 100 response
327the requested information will be carried by the next non-100 response message
328following the informational one. This implies that multiple responses may be
329sent to a single request, and that this only works when keep-alive is enabled
330(1xx messages are HTTP/1.1 only). HAProxy handles these messages and is able to
331correctly forward and skip them, and only process the next non-100 response. As
332such, these messages are neither logged nor transformed, unless explicitly
333state otherwise. Status 101 messages indicate that the protocol is changing
334over the same connection and that haproxy must switch to tunnel mode, just as
335if a CONNECT had occurred. Then the Upgrade header would contain additional
336information about the type of protocol the connection is switching to.
Willy Tarreau816b9792009-09-15 21:25:21 +0200337
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200338
Davor Ocelice9ed2812017-12-25 17:49:28 +01003391.3.1. The response line
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200340------------------------
341
342Line 1 is the "response line". It is always composed of 3 fields :
343
344 - a version tag : HTTP/1.1
345 - a status code : 200
346 - a reason : OK
347
348The status code is always 3-digit. The first digit indicates a general status :
Davor Ocelice9ed2812017-12-25 17:49:28 +0100349 - 1xx = informational message to be skipped (e.g. 100, 101)
350 - 2xx = OK, content is following (e.g. 200, 206)
351 - 3xx = OK, no content following (e.g. 302, 304)
352 - 4xx = error caused by the client (e.g. 401, 403, 404)
353 - 5xx = error caused by the server (e.g. 500, 502, 503)
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200354
Lukas Tribus23953682017-04-28 13:24:30 +0000355Please refer to RFC7231 for the detailed meaning of all such codes. The
Willy Tarreaud72758d2010-01-12 10:42:19 +0100356"reason" field is just a hint, but is not parsed by clients. Anything can be
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200357found there, but it's a common practice to respect the well-established
358messages. It can be composed of one or multiple words, such as "OK", "Found",
359or "Authentication Required".
360
Davor Ocelice9ed2812017-12-25 17:49:28 +0100361HAProxy may emit the following status codes by itself :
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200362
363 Code When / reason
364 200 access to stats page, and when replying to monitoring requests
365 301 when performing a redirection, depending on the configured code
366 302 when performing a redirection, depending on the configured code
367 303 when performing a redirection, depending on the configured code
Willy Tarreaub67fdc42013-03-29 19:28:11 +0100368 307 when performing a redirection, depending on the configured code
369 308 when performing a redirection, depending on the configured code
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200370 400 for an invalid or too large request
371 401 when an authentication is required to perform the action (when
372 accessing the stats page)
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200373 403 when a request is forbidden by a "http-request deny" rule
Florian Tham9205fea2020-01-08 13:35:30 +0100374 404 when the requested resource could not be found
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200375 408 when the request timeout strikes before the request is complete
Florian Tham272e29b2020-01-08 10:19:05 +0100376 410 when the requested resource is no longer available and will not
377 be available again
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200378 500 when haproxy encounters an unrecoverable internal error, such as a
379 memory allocation failure, which should never happen
380 502 when the server returns an empty, invalid or incomplete response, or
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200381 when an "http-response deny" rule blocks the response.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200382 503 when no server was available to handle the request, or in response to
383 monitoring requests which match the "monitor fail" condition
384 504 when the response timeout strikes before the server responds
385
386The error 4xx and 5xx codes above may be customized (see "errorloc" in section
3874.2).
388
389
3901.3.2. The response headers
391---------------------------
392
393Response headers work exactly like request headers, and as such, HAProxy uses
394the same parsing function for both. Please refer to paragraph 1.2.2 for more
395details.
396
397
3982. Configuring HAProxy
399----------------------
400
4012.1. Configuration file format
402------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +0200403
404HAProxy's configuration process involves 3 major sources of parameters :
405
406 - the arguments from the command-line, which always take precedence
Willy Tarreauc66de522020-11-25 19:58:20 +0100407 - the configuration file(s), whose format is described here
Thayne McCombsdab4ba62021-01-07 21:24:41 -0700408 - the running process's environment, in case some environment variables are
Willy Tarreauc66de522020-11-25 19:58:20 +0100409 explicitly referenced
Willy Tarreau6a06a402007-07-15 20:15:28 +0200410
Willy Tarreauc66de522020-11-25 19:58:20 +0100411The configuration file follows a fairly simple hierarchical format which obey
412a few basic rules:
Willy Tarreau0ba27502007-12-24 16:55:16 +0100413
Willy Tarreauc66de522020-11-25 19:58:20 +0100414 1. a configuration file is an ordered sequence of statements
415
416 2. a statement is a single non-empty line before any unprotected "#" (hash)
417
418 3. a line is a series of tokens or "words" delimited by unprotected spaces or
419 tab characters
420
421 4. the first word or sequence of words of a line is one of the keywords or
422 keyword sequences listed in this document
423
424 5. all other words are all arguments of the first one, some being well-known
425 keywords listed in this document, others being values, references to other
426 parts of the configuration, or expressions
427
428 6. certain keywords delimit a section inside which only a subset of keywords
429 are supported
430
431 7. a section ends at the end of a file or on a special keyword starting a new
432 section
433
434This is all that is needed to know to write a simple but reliable configuration
435generator, but this is not enough to reliably parse any configuration nor to
436figure how to deal with certain corner cases.
437
438First, there are a few consequences of the rules above. Rule 6 and 7 imply that
439the keywords used to define a new section are valid everywhere and cannot have
440a different meaning in a specific section. These keywords are always a single
441word (as opposed to a sequence of words), and traditionally the section that
442follows them is designated using the same name. For example when speaking about
443the "global section", it designates the section of configuration that follows
444the "global" keyword. This usage is used a lot in error messages to help locate
445the parts that need to be addressed.
446
447A number of sections create an internal object or configuration space, which
448requires to be distinguished from other ones. In this case they will take an
449extra word which will set the name of this particular section. For some of them
450the section name is mandatory. For example "frontend foo" will create a new
451section of type "frontend" named "foo". Usually a name is specific to its
452section and two sections of different types may use the same name, but this is
453not recommended as it tends to complexify configuration management.
454
455A direct consequence of rule 7 is that when multiple files are read at once,
456each of them must start with a new section, and the end of each file will end
457a section. A file cannot contain sub-sections nor end an existing section and
458start a new one.
459
460Rule 1 mentioned that ordering matters. Indeed, some keywords create directives
461that can be repeated multiple times to create ordered sequences of rules to be
462applied in a certain order. For example "tcp-request" can be used to alternate
463"accept" and "reject" rules on varying criteria. As such, a configuration file
464processor must always preserve a section's ordering when editing a file. The
465ordering of sections usually does not matter except for the global section
466which must be placed before other sections, but it may be repeated if needed.
467In addition, some automatic identifiers may automatically be assigned to some
468of the created objects (e.g. proxies), and by reordering sections, their
469identifiers will change. These ones appear in the statistics for example. As
470such, the configuration below will assign "foo" ID number 1 and "bar" ID number
4712, which will be swapped if the two sections are reversed:
472
473 listen foo
474 bind :80
475
476 listen bar
477 bind :81
478
479Another important point is that according to rules 2 and 3 above, empty lines,
480spaces, tabs, and comments following and unprotected "#" character are not part
481of the configuration as they are just used as delimiters. This implies that the
482following configurations are strictly equivalent:
483
484 global#this is the global section
485 daemon#daemonize
486 frontend foo
487 mode http # or tcp
488
489and:
490
491 global
492 daemon
493
494 # this is the public web frontend
495 frontend foo
496 mode http
497
498The common practice is to align to the left only the keyword that initiates a
499new section, and indent (i.e. prepend a tab character or a few spaces) all
500other keywords so that it's instantly visible that they belong to the same
501section (as done in the second example above). Placing comments before a new
502section helps the reader decide if it's the desired one. Leaving a blank line
503at the end of a section also visually helps spotting the end when editing it.
504
505Tabs are very convenient for indent but they do not copy-paste well. If spaces
506are used instead, it is recommended to avoid placing too many (2 to 4) so that
507editing in field doesn't become a burden with limited editors that do not
508support automatic indent.
509
510In the early days it used to be common to see arguments split at fixed tab
511positions because most keywords would not take more than two arguments. With
512modern versions featuring complex expressions this practice does not stand
513anymore, and is not recommended.
514
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200515
William Lallemandf9873ba2015-05-05 17:37:14 +02005162.2. Quoting and escaping
517-------------------------
518
Willy Tarreauc66de522020-11-25 19:58:20 +0100519In modern configurations, some arguments require the use of some characters
520that were previously considered as pure delimiters. In order to make this
521possible, HAProxy supports character escaping by prepending a backslash ('\')
522in front of the character to be escaped, weak quoting within double quotes
523('"') and strong quoting within single quotes ("'").
William Lallemandf9873ba2015-05-05 17:37:14 +0200524
Willy Tarreauc66de522020-11-25 19:58:20 +0100525This is pretty similar to what is done in a number of programming languages and
526very close to what is commonly encountered in Bourne shell. The principle is
527the following: while the configuration parser cuts the lines into words, it
528also takes care of quotes and backslashes to decide whether a character is a
529delimiter or is the raw representation of this character within the current
530word. The escape character is then removed, the quotes are removed, and the
531remaining word is used as-is as a keyword or argument for example.
William Lallemandf9873ba2015-05-05 17:37:14 +0200532
Willy Tarreauc66de522020-11-25 19:58:20 +0100533If a backslash is needed in a word, it must either be escaped using itself
534(i.e. double backslash) or be strongly quoted.
535
536Escaping outside quotes is achieved by preceding a special character by a
537backslash ('\'):
William Lallemandf9873ba2015-05-05 17:37:14 +0200538
539 \ to mark a space and differentiate it from a delimiter
540 \# to mark a hash and differentiate it from a comment
541 \\ to use a backslash
542 \' to use a single quote and differentiate it from strong quoting
543 \" to use a double quote and differentiate it from weak quoting
544
Willy Tarreauc66de522020-11-25 19:58:20 +0100545In addition, a few non-printable characters may be emitted using their usual
546C-language representation:
547
548 \n to insert a line feed (LF, character \x0a or ASCII 10 decimal)
549 \r to insert a carriage return (CR, character \x0d or ASCII 13 decimal)
550 \t to insert a tab (character \x09 or ASCII 9 decimal)
551 \xNN to insert character having ASCII code hex NN (e.g \x0a for LF).
552
553Weak quoting is achieved by surrounding double quotes ("") around the character
554or sequence of characters to protect. Weak quoting prevents the interpretation
555of:
William Lallemandf9873ba2015-05-05 17:37:14 +0200556
Willy Tarreauc66de522020-11-25 19:58:20 +0100557 space or tab as a word separator
William Lallemandf9873ba2015-05-05 17:37:14 +0200558 ' single quote as a strong quoting delimiter
559 # hash as a comment start
560
Willy Tarreauc66de522020-11-25 19:58:20 +0100561Weak quoting permits the interpretation of environment variables (which are not
562evaluated outside of quotes) by preceding them with a dollar sign ('$'). If a
563dollar character is needed inside double quotes, it must be escaped using a
564backslash.
William Lallemandb2f07452015-05-12 14:27:13 +0200565
Willy Tarreauc66de522020-11-25 19:58:20 +0100566Strong quoting is achieved by surrounding single quotes ('') around the
567character or sequence of characters to protect. Inside single quotes, nothing
568is interpreted, it's the efficient way to quote regular expressions.
William Lallemandf9873ba2015-05-05 17:37:14 +0200569
Willy Tarreauc66de522020-11-25 19:58:20 +0100570As a result, here is the matrix indicating how special characters can be
571entered in different contexts (unprintable characters are replaced with their
572name within angle brackets). Note that some characters that may only be
573represented escaped have no possible representation inside single quotes,
574hence the '-' there:
William Lallemandf9873ba2015-05-05 17:37:14 +0200575
Willy Tarreauc66de522020-11-25 19:58:20 +0100576 Character | Unquoted | Weakly quoted | Strongly quoted
577 -----------+---------------+-----------------------------+-----------------
578 <TAB> | \<TAB>, \x09 | "<TAB>", "\<TAB>", "\x09" | '<TAB>'
579 <LF> | \n, \x0a | "\n", "\x0a" | -
580 <CR> | \r, \x0d | "\r", "\x0d" | -
581 <SPC> | \<SPC>, \x20 | "<SPC>", "\<SPC>", "\x20" | '<SPC>'
582 " | \", \x22 | "\"", "\x22" | '"'
583 # | \#, \x23 | "#", "\#", "\x23" | '#'
584 $ | $, \$, \x24 | "\$", "\x24" | '$'
585 ' | \', \x27 | "'", "\'", "\x27" | -
586 \ | \\, \x5c | "\\", "\x5c" | '\'
William Lallemandf9873ba2015-05-05 17:37:14 +0200587
588 Example:
Willy Tarreauc66de522020-11-25 19:58:20 +0100589 # those are all strictly equivalent:
William Lallemandf9873ba2015-05-05 17:37:14 +0200590 log-format %{+Q}o\ %t\ %s\ %{-Q}r
591 log-format "%{+Q}o %t %s %{-Q}r"
592 log-format '%{+Q}o %t %s %{-Q}r'
593 log-format "%{+Q}o %t"' %s %{-Q}r'
594 log-format "%{+Q}o %t"' %s'\ %{-Q}r
595
Willy Tarreauc66de522020-11-25 19:58:20 +0100596There is one particular case where a second level of quoting or escaping may be
597necessary. Some keywords take arguments within parenthesis, sometimes delimited
598by commas. These arguments are commonly integers or predefined words, but when
599they are arbitrary strings, it may be required to perform a separate level of
600escaping to disambiguate the characters that belong to the argument from the
601characters that are used to delimit the arguments themselves. A pretty common
602case is the "regsub" converter. It takes a regular expression in argument, and
603if a closing parenthesis is needed inside, this one will require to have its
604own quotes.
605
606The keyword argument parser is exactly the same as the top-level one regarding
607quotes, except that is will not make special cases of backslashes. But what is
608not always obvious is that the delimitors used inside must first be escaped or
609quoted so that they are not resolved at the top level.
610
611Let's take this example making use of the "regsub" converter which takes 3
612arguments, one regular expression, one replacement string and one set of flags:
613
614 # replace all occurrences of "foo" with "blah" in the path:
615 http-request set-path %[path,regsub(foo,blah,g)]
616
617Here no special quoting was necessary. But if now we want to replace either
618"foo" or "bar" with "blah", we'll need the regular expression "(foo|bar)". We
619cannot write:
620
621 http-request set-path %[path,regsub((foo|bar),blah,g)]
622
623because we would like the string to cut like this:
624
625 http-request set-path %[path,regsub((foo|bar),blah,g)]
626 |---------|----|-|
627 arg1 _/ / /
628 arg2 __________/ /
629 arg3 ______________/
630
631but actually what is passed is a string between the opening and closing
632parenthesis then garbage:
633
634 http-request set-path %[path,regsub((foo|bar),blah,g)]
635 |--------|--------|
636 arg1=(foo|bar _/ /
637 trailing garbage _________/
638
639The obvious solution here seems to be that the closing parenthesis needs to be
640quoted, but alone this will not work, because as mentioned above, quotes are
641processed by the top-level parser which will resolve them before processing
642this word:
643
644 http-request set-path %[path,regsub("(foo|bar)",blah,g)]
645 ------------ -------- ----------------------------------
646 word1 word2 word3=%[path,regsub((foo|bar),blah,g)]
647
648So we didn't change anything for the argument parser at the second level which
649still sees a truncated regular expression as the only argument, and garbage at
650the end of the string. By escaping the quotes they will be passed unmodified to
651the second level:
652
653 http-request set-path %[path,regsub(\"(foo|bar)\",blah,g)]
654 ------------ -------- ------------------------------------
655 word1 word2 word3=%[path,regsub("(foo|bar)",blah,g)]
656 |---------||----|-|
657 arg1=(foo|bar) _/ / /
658 arg2=blah ___________/ /
659 arg3=g _______________/
660
661Another approch consists in using single quotes outside the whole string and
662double quotes inside (so that the double quotes are not stripped again):
663
664 http-request set-path '%[path,regsub("(foo|bar)",blah,g)]'
665 ------------ -------- ----------------------------------
666 word1 word2 word3=%[path,regsub("(foo|bar)",blah,g)]
667 |---------||----|-|
668 arg1=(foo|bar) _/ / /
669 arg2 ___________/ /
670 arg3 _______________/
671
672When using regular expressions, it can happen that the dollar ('$') character
673appears in the expression or that a backslash ('\') is used in the replacement
674string. In this case these ones will also be processed inside the double quotes
675thus single quotes are preferred (or double escaping). Example:
676
677 http-request set-path '%[path,regsub("^/(here)(/|$)","my/\1",g)]'
678 ------------ -------- -----------------------------------------
679 word1 word2 word3=%[path,regsub("^/(here)(/|$)","my/\1",g)]
680 |-------------| |-----||-|
681 arg1=(here)(/|$) _/ / /
682 arg2=my/\1 ________________/ /
683 arg3 ______________________/
684
685Remember that backslahes are not escape characters withing single quotes and
686that the whole word3 above is already protected against them using the single
687quotes. Conversely, if double quotes had been used around the whole expression,
688single the dollar character and the backslashes would have been resolved at top
689level, breaking the argument contents at the second level.
690
691When in doubt, simply do not use quotes anywhere, and start to place single or
692double quotes around arguments that require a comma or a closing parenthesis,
693and think about escaping these quotes using a backslash of the string contains
694a dollar or a backslash. Again, this is pretty similar to what is used under
695a Bourne shell when double-escaping a command passed to "eval". For API writers
696the best is probably to place escaped quotes around each and every argument,
697regardless of their contents. Users will probably find that using single quotes
698around the whole expression and double quotes around each argument provides
699more readable configurations.
William Lallemandf9873ba2015-05-05 17:37:14 +0200700
701
William Lallemandb2f07452015-05-12 14:27:13 +02007022.3. Environment variables
703--------------------------
704
705HAProxy's configuration supports environment variables. Those variables are
706interpreted only within double quotes. Variables are expanded during the
707configuration parsing. Variable names must be preceded by a dollar ("$") and
708optionally enclosed with braces ("{}") similarly to what is done in Bourne
709shell. Variable names can contain alphanumerical characters or the character
Amaury Denoyellefa41cb62020-10-01 14:32:35 +0200710underscore ("_") but should not start with a digit. If the variable contains a
711list of several values separated by spaces, it can be expanded as individual
712arguments by enclosing the variable with braces and appending the suffix '[*]'
713before the closing brace.
William Lallemandb2f07452015-05-12 14:27:13 +0200714
715 Example:
716
717 bind "fd@${FD_APP1}"
718
719 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
720
721 user "$HAPROXY_USER"
722
William Lallemand4d03e432019-06-14 15:35:37 +0200723Some variables are defined by HAProxy, they can be used in the configuration
724file, or could be inherited by a program (See 3.7. Programs):
William Lallemanddaf4cd22018-04-17 16:46:13 +0200725
William Lallemand4d03e432019-06-14 15:35:37 +0200726* HAPROXY_LOCALPEER: defined at the startup of the process which contains the
727 name of the local peer. (See "-L" in the management guide.)
728
729* HAPROXY_CFGFILES: list of the configuration files loaded by HAProxy,
730 separated by semicolons. Can be useful in the case you specified a
731 directory.
732
733* HAPROXY_MWORKER: In master-worker mode, this variable is set to 1.
734
John Roeslerfb2fce12019-07-10 15:45:51 -0500735* HAPROXY_CLI: configured listeners addresses of the stats socket for every
William Lallemand4d03e432019-06-14 15:35:37 +0200736 processes, separated by semicolons.
737
John Roeslerfb2fce12019-07-10 15:45:51 -0500738* HAPROXY_MASTER_CLI: In master-worker mode, listeners addresses of the master
William Lallemand4d03e432019-06-14 15:35:37 +0200739 CLI, separated by semicolons.
740
741See also "external-check command" for other variables.
William Lallemandb2f07452015-05-12 14:27:13 +0200742
7432.4. Time format
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200744----------------
745
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100746Some parameters involve values representing time, such as timeouts. These
Willy Tarreau0ba27502007-12-24 16:55:16 +0100747values are generally expressed in milliseconds (unless explicitly stated
748otherwise) but may be expressed in any other unit by suffixing the unit to the
749numeric value. It is important to consider this because it will not be repeated
750for every keyword. Supported units are :
751
752 - us : microseconds. 1 microsecond = 1/1000000 second
753 - ms : milliseconds. 1 millisecond = 1/1000 second. This is the default.
754 - s : seconds. 1s = 1000ms
755 - m : minutes. 1m = 60s = 60000ms
756 - h : hours. 1h = 60m = 3600s = 3600000ms
757 - d : days. 1d = 24h = 1440m = 86400s = 86400000ms
758
759
Lukas Tribusaa83a312017-03-21 09:25:09 +00007602.5. Examples
Patrick Mezard35da19c2010-06-12 17:02:47 +0200761-------------
762
763 # Simple configuration for an HTTP proxy listening on port 80 on all
764 # interfaces and forwarding requests to a single backend "servers" with a
765 # single server "server1" listening on 127.0.0.1:8000
766 global
767 daemon
768 maxconn 256
769
770 defaults
771 mode http
772 timeout connect 5000ms
773 timeout client 50000ms
774 timeout server 50000ms
775
776 frontend http-in
777 bind *:80
778 default_backend servers
779
780 backend servers
781 server server1 127.0.0.1:8000 maxconn 32
782
783
784 # The same configuration defined with a single listen block. Shorter but
785 # less expressive, especially in HTTP mode.
786 global
787 daemon
788 maxconn 256
789
790 defaults
791 mode http
792 timeout connect 5000ms
793 timeout client 50000ms
794 timeout server 50000ms
795
796 listen http-in
797 bind *:80
798 server server1 127.0.0.1:8000 maxconn 32
799
800
801Assuming haproxy is in $PATH, test these configurations in a shell with:
802
Willy Tarreauccb289d2010-12-11 20:19:38 +0100803 $ sudo haproxy -f configuration.conf -c
Patrick Mezard35da19c2010-06-12 17:02:47 +0200804
805
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008063. Global parameters
Willy Tarreau6a06a402007-07-15 20:15:28 +0200807--------------------
808
809Parameters in the "global" section are process-wide and often OS-specific. They
810are generally set once for all and do not need being changed once correct. Some
811of them have command-line equivalents.
812
813The following keywords are supported in the "global" section :
814
815 * Process management and security
Emeric Brunc8e8d122012-10-02 18:42:10 +0200816 - ca-base
Willy Tarreau6a06a402007-07-15 20:15:28 +0200817 - chroot
Emeric Brunc8e8d122012-10-02 18:42:10 +0200818 - crt-base
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200819 - cpu-map
Willy Tarreau6a06a402007-07-15 20:15:28 +0200820 - daemon
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200821 - description
822 - deviceatlas-json-file
823 - deviceatlas-log-level
824 - deviceatlas-separator
825 - deviceatlas-properties-cookie
Simon Horman98637e52014-06-20 12:30:16 +0900826 - external-check
Willy Tarreau6a06a402007-07-15 20:15:28 +0200827 - gid
828 - group
Cyril Bonté203ec5a2017-03-23 22:44:13 +0100829 - hard-stop-after
Christopher Faulet98fbe952019-07-22 16:18:24 +0200830 - h1-case-adjust
831 - h1-case-adjust-file
Willy Tarreaud96f1122019-12-03 07:07:36 +0100832 - insecure-fork-wanted
Willy Tarreaua45a8b52019-12-06 16:31:45 +0100833 - insecure-setuid-wanted
Emmanuel Hocdet70df7bf2019-01-04 11:08:20 +0100834 - issuers-chain-path
Dragan Dosen13cd54c2020-06-18 18:24:05 +0200835 - localpeer
Willy Tarreau6a06a402007-07-15 20:15:28 +0200836 - log
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200837 - log-tag
Joe Williamsdf5b38f2010-12-29 17:05:48 +0100838 - log-send-hostname
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200839 - lua-load
Tim Duesterhusdd74b5f2020-01-12 13:55:40 +0100840 - lua-prepend-path
William Lallemand27edc4b2019-05-07 17:49:33 +0200841 - mworker-max-reloads
Willy Tarreau6a06a402007-07-15 20:15:28 +0200842 - nbproc
Christopher Fauletbe0faa22017-08-29 15:37:10 +0200843 - nbthread
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200844 - node
Willy Tarreau6a06a402007-07-15 20:15:28 +0200845 - pidfile
Willy Tarreau119e50e2020-05-22 13:53:29 +0200846 - pp2-never-send-local
Willy Tarreau1d549722016-02-16 12:41:57 +0100847 - presetenv
848 - resetenv
Willy Tarreau6a06a402007-07-15 20:15:28 +0200849 - uid
850 - ulimit-n
851 - user
Willy Tarreau636848a2019-04-15 19:38:50 +0200852 - set-dumpable
Willy Tarreau1d549722016-02-16 12:41:57 +0100853 - setenv
Willy Tarreaufbee7132007-10-18 13:53:22 +0200854 - stats
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200855 - ssl-default-bind-ciphers
Dirkjan Bussink415150f2018-09-14 11:14:21 +0200856 - ssl-default-bind-ciphersuites
Jerome Magninb203ff62020-04-03 15:28:22 +0200857 - ssl-default-bind-curves
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200858 - ssl-default-bind-options
859 - ssl-default-server-ciphers
Dirkjan Bussink415150f2018-09-14 11:14:21 +0200860 - ssl-default-server-ciphersuites
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200861 - ssl-default-server-options
862 - ssl-dh-param-file
Emeric Brun850efd52014-01-29 12:24:34 +0100863 - ssl-server-verify
Emmanuel Hocdetc3b7e742020-04-22 11:06:19 +0200864 - ssl-skip-self-issued-ca
Willy Tarreauceb24bc2010-11-09 12:46:41 +0100865 - unix-bind
Willy Tarreau1d549722016-02-16 12:41:57 +0100866 - unsetenv
Thomas Holmesdb04f192015-05-18 13:21:39 +0100867 - 51degrees-data-file
868 - 51degrees-property-name-list
Dragan Dosen93b38d92015-06-29 16:43:25 +0200869 - 51degrees-property-separator
Dragan Dosenae6d39a2015-06-29 16:43:27 +0200870 - 51degrees-cache-size
Willy Tarreaub3cc9f22019-04-19 16:03:32 +0200871 - wurfl-data-file
872 - wurfl-information-list
873 - wurfl-information-list-separator
Willy Tarreaub3cc9f22019-04-19 16:03:32 +0200874 - wurfl-cache-size
William Dauchy0fec3ab2019-10-27 20:08:11 +0100875 - strict-limits
Willy Tarreaud72758d2010-01-12 10:42:19 +0100876
Willy Tarreau6a06a402007-07-15 20:15:28 +0200877 * Performance tuning
William Dauchy0a8824f2019-10-27 20:08:09 +0100878 - busy-polling
Willy Tarreau1746eec2014-04-25 10:46:47 +0200879 - max-spread-checks
Willy Tarreau6a06a402007-07-15 20:15:28 +0200880 - maxconn
Willy Tarreau81c25d02011-09-07 15:17:21 +0200881 - maxconnrate
William Lallemandd85f9172012-11-09 17:05:39 +0100882 - maxcomprate
William Lallemand072a2bf2012-11-20 17:01:01 +0100883 - maxcompcpuusage
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100884 - maxpipes
Willy Tarreau93e7c002013-10-07 18:51:07 +0200885 - maxsessrate
Willy Tarreau403edff2012-09-06 11:58:37 +0200886 - maxsslconn
Willy Tarreaue43d5322013-10-07 20:01:52 +0200887 - maxsslrate
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200888 - maxzlibmem
Willy Tarreau6a06a402007-07-15 20:15:28 +0200889 - noepoll
890 - nokqueue
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +0000891 - noevports
Willy Tarreau6a06a402007-07-15 20:15:28 +0200892 - nopoll
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100893 - nosplice
Jarno Huuskonen0e82b922014-04-12 18:22:19 +0300894 - nogetaddrinfo
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +0000895 - noreuseport
Willy Tarreau75c62c22018-11-22 11:02:09 +0100896 - profiling.tasks
Willy Tarreaufe255b72007-10-14 23:09:26 +0200897 - spread-checks
Baptiste Assmann5626f482015-08-23 10:00:10 +0200898 - server-state-base
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +0200899 - server-state-file
Grant Zhang872f9c22017-01-21 01:10:18 +0000900 - ssl-engine
Grant Zhangfa6c7ee2017-01-14 01:42:15 +0000901 - ssl-mode-async
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200902 - tune.buffers.limit
903 - tune.buffers.reserve
Willy Tarreau27a674e2009-08-17 07:23:33 +0200904 - tune.bufsize
Willy Tarreau43961d52010-10-04 20:39:20 +0200905 - tune.chksize
William Lallemandf3747832012-11-09 12:33:10 +0100906 - tune.comp.maxlevel
Willy Tarreaubc52bec2020-06-18 08:58:47 +0200907 - tune.fd.edge-triggered
Willy Tarreaufe20e5b2017-07-27 11:42:14 +0200908 - tune.h2.header-table-size
Willy Tarreaue6baec02017-07-27 11:45:11 +0200909 - tune.h2.initial-window-size
Willy Tarreau5242ef82017-07-27 11:47:28 +0200910 - tune.h2.max-concurrent-streams
Willy Tarreau193b8c62012-11-22 00:17:38 +0100911 - tune.http.cookielen
Stéphane Cottin23e9e932017-05-18 08:58:41 +0200912 - tune.http.logurilen
Willy Tarreauac1932d2011-10-24 19:14:41 +0200913 - tune.http.maxhdr
Willy Tarreau76cc6992020-07-01 18:49:24 +0200914 - tune.idle-pool.shared
Willy Tarreau7e312732014-02-12 16:35:14 +0100915 - tune.idletimer
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100916 - tune.lua.forced-yield
Willy Tarreau32f61e22015-03-18 17:54:59 +0100917 - tune.lua.maxmem
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100918 - tune.lua.session-timeout
919 - tune.lua.task-timeout
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +0200920 - tune.lua.service-timeout
Willy Tarreaua0250ba2008-01-06 11:22:57 +0100921 - tune.maxaccept
922 - tune.maxpollevents
Willy Tarreau27a674e2009-08-17 07:23:33 +0200923 - tune.maxrewrite
Willy Tarreauf3045d22015-04-29 16:24:50 +0200924 - tune.pattern.cache-size
Willy Tarreaubd9a0a72011-10-23 21:14:29 +0200925 - tune.pipesize
Willy Tarreaua8e2d972020-07-01 18:27:16 +0200926 - tune.pool-high-fd-ratio
927 - tune.pool-low-fd-ratio
Willy Tarreaue803de22010-01-21 17:43:04 +0100928 - tune.rcvbuf.client
929 - tune.rcvbuf.server
Willy Tarreaub22fc302015-12-14 12:04:35 +0100930 - tune.recv_enough
Olivier Houchard1599b802018-05-24 18:59:04 +0200931 - tune.runqueue-depth
Willy Tarreaue7723bd2020-06-24 11:11:02 +0200932 - tune.sched.low-latency
Willy Tarreaue803de22010-01-21 17:43:04 +0100933 - tune.sndbuf.client
934 - tune.sndbuf.server
Willy Tarreau6ec58db2012-11-16 16:32:15 +0100935 - tune.ssl.cachesize
William Lallemand7d42ef52020-07-06 11:41:30 +0200936 - tune.ssl.keylog
Willy Tarreaubfd59462013-02-21 07:46:09 +0100937 - tune.ssl.lifetime
Emeric Brun8dc60392014-05-09 13:52:00 +0200938 - tune.ssl.force-private-cache
Willy Tarreaubfd59462013-02-21 07:46:09 +0100939 - tune.ssl.maxrecord
Remi Gacognef46cd6e2014-06-12 14:58:40 +0200940 - tune.ssl.default-dh-param
Christopher Faulet31af49d2015-06-09 17:29:50 +0200941 - tune.ssl.ssl-ctx-cache-size
Thierry FOURNIER5bf77322017-02-25 12:45:22 +0100942 - tune.ssl.capture-cipherlist-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200943 - tune.vars.global-max-size
Christopher Fauletff2613e2016-11-09 11:36:17 +0100944 - tune.vars.proc-max-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200945 - tune.vars.reqres-max-size
946 - tune.vars.sess-max-size
947 - tune.vars.txn-max-size
William Lallemanda509e4c2012-11-07 16:54:34 +0100948 - tune.zlib.memlevel
949 - tune.zlib.windowsize
Willy Tarreaud72758d2010-01-12 10:42:19 +0100950
Willy Tarreau6a06a402007-07-15 20:15:28 +0200951 * Debugging
Willy Tarreau6a06a402007-07-15 20:15:28 +0200952 - quiet
Willy Tarreau3eb10b82020-04-15 16:42:39 +0200953 - zero-warning
Willy Tarreau6a06a402007-07-15 20:15:28 +0200954
955
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009563.1. Process management and security
Willy Tarreau6a06a402007-07-15 20:15:28 +0200957------------------------------------
958
Emeric Brunc8e8d122012-10-02 18:42:10 +0200959ca-base <dir>
960 Assigns a default directory to fetch SSL CA certificates and CRLs from when a
Emmanuel Hocdet842e94e2019-12-16 16:39:17 +0100961 relative path is used with "ca-file", "ca-verify-file" or "crl-file"
962 directives. Absolute locations specified in "ca-file", "ca-verify-file" and
963 "crl-file" prevail and ignore "ca-base".
Emeric Brunc8e8d122012-10-02 18:42:10 +0200964
Willy Tarreau6a06a402007-07-15 20:15:28 +0200965chroot <jail dir>
966 Changes current directory to <jail dir> and performs a chroot() there before
967 dropping privileges. This increases the security level in case an unknown
968 vulnerability would be exploited, since it would make it very hard for the
969 attacker to exploit the system. This only works when the process is started
970 with superuser privileges. It is important to ensure that <jail_dir> is both
Davor Ocelice9ed2812017-12-25 17:49:28 +0100971 empty and non-writable to anyone.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100972
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100973cpu-map [auto:]<process-set>[/<thread-set>] <cpu-set>...
974 On Linux 2.6 and above, it is possible to bind a process or a thread to a
975 specific CPU set. This means that the process or the thread will never run on
976 other CPUs. The "cpu-map" directive specifies CPU sets for process or thread
977 sets. The first argument is a process set, eventually followed by a thread
978 set. These sets have the format
979
980 all | odd | even | number[-[number]]
981
982 <number>> must be a number between 1 and 32 or 64, depending on the machine's
Davor Ocelice9ed2812017-12-25 17:49:28 +0100983 word size. Any process IDs above nbproc and any thread IDs above nbthread are
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100984 ignored. It is possible to specify a range with two such number delimited by
985 a dash ('-'). It also is possible to specify all processes at once using
Christopher Faulet1dcb9cb2017-11-22 10:24:40 +0100986 "all", only odd numbers using "odd" or even numbers using "even", just like
987 with the "bind-process" directive. The second and forthcoming arguments are
Davor Ocelice9ed2812017-12-25 17:49:28 +0100988 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 +0100989 range with two such numbers delimited by a dash ('-'). Multiple CPU numbers
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100990 or ranges may be specified, and the processes or threads will be allowed to
Davor Ocelice9ed2812017-12-25 17:49:28 +0100991 bind to all of them. Obviously, multiple "cpu-map" directives may be
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100992 specified. Each "cpu-map" directive will replace the previous ones when they
993 overlap. A thread will be bound on the intersection of its mapping and the
994 one of the process on which it is attached. If the intersection is null, no
995 specific binding will be set for the thread.
Willy Tarreaufc6c0322012-11-16 16:12:27 +0100996
Christopher Fauletff4121f2017-11-22 16:38:49 +0100997 Ranges can be partially defined. The higher bound can be omitted. In such
998 case, it is replaced by the corresponding maximum value, 32 or 64 depending
999 on the machine's word size.
1000
Christopher Faulet26028f62017-11-22 15:01:51 +01001001 The prefix "auto:" can be added before the process set to let HAProxy
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001002 automatically bind a process or a thread to a CPU by incrementing
1003 process/thread and CPU sets. To be valid, both sets must have the same
1004 size. No matter the declaration order of the CPU sets, it will be bound from
1005 the lowest to the highest bound. Having a process and a thread range with the
1006 "auto:" prefix is not supported. Only one range is supported, the other one
1007 must be a fixed number.
Christopher Faulet26028f62017-11-22 15:01:51 +01001008
1009 Examples:
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001010 cpu-map 1-4 0-3 # bind processes 1 to 4 on the first 4 CPUs
1011
1012 cpu-map 1/all 0-3 # bind all threads of the first process on the
1013 # first 4 CPUs
1014
1015 cpu-map 1- 0- # will be replaced by "cpu-map 1-64 0-63"
1016 # or "cpu-map 1-32 0-31" depending on the machine's
1017 # word size.
1018
Christopher Faulet26028f62017-11-22 15:01:51 +01001019 # all these lines bind the process 1 to the cpu 0, the process 2 to cpu 1
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001020 # and so on.
Christopher Faulet26028f62017-11-22 15:01:51 +01001021 cpu-map auto:1-4 0-3
1022 cpu-map auto:1-4 0-1 2-3
1023 cpu-map auto:1-4 3 2 1 0
1024
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001025 # all these lines bind the thread 1 to the cpu 0, the thread 2 to cpu 1
1026 # and so on.
1027 cpu-map auto:1/1-4 0-3
1028 cpu-map auto:1/1-4 0-1 2-3
1029 cpu-map auto:1/1-4 3 2 1 0
1030
Davor Ocelice9ed2812017-12-25 17:49:28 +01001031 # bind each process to exactly one CPU using all/odd/even keyword
Christopher Faulet26028f62017-11-22 15:01:51 +01001032 cpu-map auto:all 0-63
1033 cpu-map auto:even 0-31
1034 cpu-map auto:odd 32-63
1035
1036 # invalid cpu-map because process and CPU sets have different sizes.
1037 cpu-map auto:1-4 0 # invalid
1038 cpu-map auto:1 0-3 # invalid
1039
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001040 # invalid cpu-map because automatic binding is used with a process range
1041 # and a thread range.
1042 cpu-map auto:all/all 0 # invalid
1043 cpu-map auto:all/1-4 0 # invalid
1044 cpu-map auto:1-4/all 0 # invalid
1045
Emeric Brunc8e8d122012-10-02 18:42:10 +02001046crt-base <dir>
1047 Assigns a default directory to fetch SSL certificates from when a relative
William Dauchy238ea3b2020-01-11 13:09:12 +01001048 path is used with "crtfile" or "crt" directives. Absolute locations specified
1049 prevail and ignore "crt-base".
Emeric Brunc8e8d122012-10-02 18:42:10 +02001050
Willy Tarreau6a06a402007-07-15 20:15:28 +02001051daemon
1052 Makes the process fork into background. This is the recommended mode of
1053 operation. It is equivalent to the command line "-D" argument. It can be
Lukas Tribusf46bf952017-11-21 12:39:34 +01001054 disabled by the command line "-db" argument. This option is ignored in
1055 systemd mode.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001056
David Carlier8167f302015-06-01 13:50:06 +02001057deviceatlas-json-file <path>
1058 Sets the path of the DeviceAtlas JSON data file to be loaded by the API.
Davor Ocelice9ed2812017-12-25 17:49:28 +01001059 The path must be a valid JSON data file and accessible by HAProxy process.
David Carlier8167f302015-06-01 13:50:06 +02001060
1061deviceatlas-log-level <value>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001062 Sets the level of information returned by the API. This directive is
David Carlier8167f302015-06-01 13:50:06 +02001063 optional and set to 0 by default if not set.
1064
1065deviceatlas-separator <char>
1066 Sets the character separator for the API properties results. This directive
1067 is optional and set to | by default if not set.
1068
Cyril Bonté0306c4a2015-10-26 22:37:38 +01001069deviceatlas-properties-cookie <name>
Cyril Bonté307ee1e2015-09-28 23:16:06 +02001070 Sets the client cookie's name used for the detection if the DeviceAtlas
1071 Client-side component was used during the request. This directive is optional
1072 and set to DAPROPS by default if not set.
David Carlier29b3ca32015-09-25 14:09:21 +01001073
Simon Horman98637e52014-06-20 12:30:16 +09001074external-check
Willy Tarreaud96f1122019-12-03 07:07:36 +01001075 Allows the use of an external agent to perform health checks. This is
1076 disabled by default as a security precaution, and even when enabled, checks
Willy Tarreaua45a8b52019-12-06 16:31:45 +01001077 may still fail unless "insecure-fork-wanted" is enabled as well. If the
1078 program launched makes use of a setuid executable (it should really not),
1079 you may also need to set "insecure-setuid-wanted" in the global section.
1080 See "option external-check", and "insecure-fork-wanted", and
1081 "insecure-setuid-wanted".
Simon Horman98637e52014-06-20 12:30:16 +09001082
Willy Tarreau6a06a402007-07-15 20:15:28 +02001083gid <number>
Thayne McCombsdab4ba62021-01-07 21:24:41 -07001084 Changes the process's group ID to <number>. It is recommended that the group
Willy Tarreau6a06a402007-07-15 20:15:28 +02001085 ID is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
1086 be started with a user belonging to this group, or with superuser privileges.
Michael Schererab012dd2013-01-12 18:35:19 +01001087 Note that if haproxy is started from a user having supplementary groups, it
1088 will only be able to drop these groups if started with superuser privileges.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001089 See also "group" and "uid".
Willy Tarreaud72758d2010-01-12 10:42:19 +01001090
Willy Tarreau11770ce2019-12-03 08:29:22 +01001091group <group name>
1092 Similar to "gid" but uses the GID of group name <group name> from /etc/group.
1093 See also "gid" and "user".
1094
Cyril Bonté203ec5a2017-03-23 22:44:13 +01001095hard-stop-after <time>
1096 Defines the maximum time allowed to perform a clean soft-stop.
1097
1098 Arguments :
1099 <time> is the maximum time (by default in milliseconds) for which the
1100 instance will remain alive when a soft-stop is received via the
1101 SIGUSR1 signal.
1102
1103 This may be used to ensure that the instance will quit even if connections
1104 remain opened during a soft-stop (for example with long timeouts for a proxy
1105 in tcp mode). It applies both in TCP and HTTP mode.
1106
1107 Example:
1108 global
1109 hard-stop-after 30s
1110
Christopher Faulet98fbe952019-07-22 16:18:24 +02001111h1-case-adjust <from> <to>
1112 Defines the case adjustment to apply, when enabled, to the header name
1113 <from>, to change it to <to> before sending it to HTTP/1 clients or
1114 servers. <from> must be in lower case, and <from> and <to> must not differ
1115 except for their case. It may be repeated if several header names need to be
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05001116 adjusted. Duplicate entries are not allowed. If a lot of header names have to
Christopher Faulet98fbe952019-07-22 16:18:24 +02001117 be adjusted, it might be more convenient to use "h1-case-adjust-file".
1118 Please note that no transformation will be applied unless "option
1119 h1-case-adjust-bogus-client" or "option h1-case-adjust-bogus-server" is
1120 specified in a proxy.
1121
1122 There is no standard case for header names because, as stated in RFC7230,
1123 they are case-insensitive. So applications must handle them in a case-
1124 insensitive manner. But some bogus applications violate the standards and
1125 erroneously rely on the cases most commonly used by browsers. This problem
1126 becomes critical with HTTP/2 because all header names must be exchanged in
1127 lower case, and HAProxy follows the same convention. All header names are
1128 sent in lower case to clients and servers, regardless of the HTTP version.
1129
1130 Applications which fail to properly process requests or responses may require
1131 to temporarily use such workarounds to adjust header names sent to them for
1132 the time it takes the application to be fixed. Please note that an
1133 application which requires such workarounds might be vulnerable to content
1134 smuggling attacks and must absolutely be fixed.
1135
1136 Example:
1137 global
1138 h1-case-adjust content-length Content-Length
1139
1140 See "h1-case-adjust-file", "option h1-case-adjust-bogus-client" and
1141 "option h1-case-adjust-bogus-server".
1142
1143h1-case-adjust-file <hdrs-file>
1144 Defines a file containing a list of key/value pairs used to adjust the case
1145 of some header names before sending them to HTTP/1 clients or servers. The
1146 file <hdrs-file> must contain 2 header names per line. The first one must be
1147 in lower case and both must not differ except for their case. Lines which
1148 start with '#' are ignored, just like empty lines. Leading and trailing tabs
1149 and spaces are stripped. Duplicate entries are not allowed. Please note that
1150 no transformation will be applied unless "option h1-case-adjust-bogus-client"
1151 or "option h1-case-adjust-bogus-server" is specified in a proxy.
1152
1153 If this directive is repeated, only the last one will be processed. It is an
1154 alternative to the directive "h1-case-adjust" if a lot of header names need
1155 to be adjusted. Please read the risks associated with using this.
1156
1157 See "h1-case-adjust", "option h1-case-adjust-bogus-client" and
1158 "option h1-case-adjust-bogus-server".
1159
Willy Tarreaud96f1122019-12-03 07:07:36 +01001160insecure-fork-wanted
1161 By default haproxy tries hard to prevent any thread and process creation
1162 after it starts. Doing so is particularly important when using Lua files of
1163 uncertain origin, and when experimenting with development versions which may
1164 still contain bugs whose exploitability is uncertain. And generally speaking
1165 it's good hygiene to make sure that no unexpected background activity can be
1166 triggered by traffic. But this prevents external checks from working, and may
1167 break some very specific Lua scripts which actively rely on the ability to
1168 fork. This option is there to disable this protection. Note that it is a bad
1169 idea to disable it, as a vulnerability in a library or within haproxy itself
1170 will be easier to exploit once disabled. In addition, forking from Lua or
1171 anywhere else is not reliable as the forked process may randomly embed a lock
1172 set by another thread and never manage to finish an operation. As such it is
1173 highly recommended that this option is never used and that any workload
1174 requiring such a fork be reconsidered and moved to a safer solution (such as
1175 agents instead of external checks). This option supports the "no" prefix to
1176 disable it.
1177
Willy Tarreaua45a8b52019-12-06 16:31:45 +01001178insecure-setuid-wanted
1179 HAProxy doesn't need to call executables at run time (except when using
1180 external checks which are strongly recommended against), and is even expected
1181 to isolate itself into an empty chroot. As such, there basically is no valid
1182 reason to allow a setuid executable to be called without the user being fully
1183 aware of the risks. In a situation where haproxy would need to call external
1184 checks and/or disable chroot, exploiting a vulnerability in a library or in
1185 haproxy itself could lead to the execution of an external program. On Linux
1186 it is possible to lock the process so that any setuid bit present on such an
1187 executable is ignored. This significantly reduces the risk of privilege
1188 escalation in such a situation. This is what haproxy does by default. In case
1189 this causes a problem to an external check (for example one which would need
1190 the "ping" command), then it is possible to disable this protection by
1191 explicitly adding this directive in the global section. If enabled, it is
1192 possible to turn it back off by prefixing it with the "no" keyword.
1193
Emmanuel Hocdet70df7bf2019-01-04 11:08:20 +01001194issuers-chain-path <dir>
1195 Assigns a directory to load certificate chain for issuer completion. All
1196 files must be in PEM format. For certificates loaded with "crt" or "crt-list",
1197 if certificate chain is not included in PEM (also commonly known as
1198 intermediate certificate), haproxy will complete chain if the issuer of the
1199 certificate corresponds to the first certificate of the chain loaded with
1200 "issuers-chain-path".
1201 A "crt" file with PrivateKey+Certificate+IntermediateCA2+IntermediateCA1
1202 could be replaced with PrivateKey+Certificate. HAProxy will complete the
1203 chain if a file with IntermediateCA2+IntermediateCA1 is present in
1204 "issuers-chain-path" directory. All other certificates with the same issuer
1205 will share the chain in memory.
1206
Dragan Dosen13cd54c2020-06-18 18:24:05 +02001207localpeer <name>
1208 Sets the local instance's peer name. It will be ignored if the "-L"
1209 command line argument is specified or if used after "peers" section
1210 definitions. In such cases, a warning message will be emitted during
1211 the configuration parsing.
1212
1213 This option will also set the HAPROXY_LOCALPEER environment variable.
1214 See also "-L" in the management guide and "peers" section below.
1215
Jan Wagnerf2f5c4e2020-12-17 22:22:32 +01001216log <address> [len <length>] [format <format>] [sample <ranges>:<sample_size>]
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02001217 <facility> [max level [min level]]
Cyril Bonté3e954872018-03-20 23:30:27 +01001218 Adds a global syslog server. Several global servers can be defined. They
Davor Ocelice9ed2812017-12-25 17:49:28 +01001219 will receive logs for starts and exits, as well as all logs from proxies
Robert Tsai81ae1952007-12-05 10:47:29 +01001220 configured with "log global".
1221
1222 <address> can be one of:
1223
Willy Tarreau2769aa02007-12-27 18:26:09 +01001224 - An IPv4 address optionally followed by a colon and a UDP port. If
Robert Tsai81ae1952007-12-05 10:47:29 +01001225 no port is specified, 514 is used by default (the standard syslog
1226 port).
1227
David du Colombier24bb5f52011-03-17 10:40:23 +01001228 - An IPv6 address followed by a colon and optionally a UDP port. If
1229 no port is specified, 514 is used by default (the standard syslog
1230 port).
1231
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01001232 - A filesystem path to a datagram UNIX domain socket, keeping in mind
Robert Tsai81ae1952007-12-05 10:47:29 +01001233 considerations for chroot (be sure the path is accessible inside
1234 the chroot) and uid/gid (be sure the path is appropriately
Davor Ocelice9ed2812017-12-25 17:49:28 +01001235 writable).
Robert Tsai81ae1952007-12-05 10:47:29 +01001236
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01001237 - A file descriptor number in the form "fd@<number>", which may point
1238 to a pipe, terminal, or socket. In this case unbuffered logs are used
1239 and one writev() call per log is performed. This is a bit expensive
1240 but acceptable for most workloads. Messages sent this way will not be
1241 truncated but may be dropped, in which case the DroppedLogs counter
1242 will be incremented. The writev() call is atomic even on pipes for
1243 messages up to PIPE_BUF size, which POSIX recommends to be at least
1244 512 and which is 4096 bytes on most modern operating systems. Any
1245 larger message may be interleaved with messages from other processes.
1246 Exceptionally for debugging purposes the file descriptor may also be
1247 directed to a file, but doing so will significantly slow haproxy down
1248 as non-blocking calls will be ignored. Also there will be no way to
1249 purge nor rotate this file without restarting the process. Note that
1250 the configured syslog format is preserved, so the output is suitable
Willy Tarreauc1b06452018-11-12 11:57:56 +01001251 for use with a TCP syslog server. See also the "short" and "raw"
1252 format below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01001253
1254 - "stdout" / "stderr", which are respectively aliases for "fd@1" and
1255 "fd@2", see above.
1256
Willy Tarreauc046d162019-08-30 15:24:59 +02001257 - A ring buffer in the form "ring@<name>", which will correspond to an
1258 in-memory ring buffer accessible over the CLI using the "show events"
1259 command, which will also list existing rings and their sizes. Such
1260 buffers are lost on reload or restart but when used as a complement
1261 this can help troubleshooting by having the logs instantly available.
1262
William Lallemandb2f07452015-05-12 14:27:13 +02001263 You may want to reference some environment variables in the address
1264 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01001265
Willy Tarreau18324f52014-06-27 18:10:07 +02001266 <length> is an optional maximum line length. Log lines larger than this value
1267 will be truncated before being sent. The reason is that syslog
1268 servers act differently on log line length. All servers support the
1269 default value of 1024, but some servers simply drop larger lines
1270 while others do log them. If a server supports long lines, it may
1271 make sense to set this value here in order to avoid truncating long
1272 lines. Similarly, if a server drops long lines, it is preferable to
1273 truncate them before sending them. Accepted values are 80 to 65535
1274 inclusive. The default value of 1024 is generally fine for all
1275 standard usages. Some specific cases of long captures or
Davor Ocelice9ed2812017-12-25 17:49:28 +01001276 JSON-formatted logs may require larger values. You may also need to
1277 increase "tune.http.logurilen" if your request URIs are truncated.
Willy Tarreau18324f52014-06-27 18:10:07 +02001278
Dragan Dosen7ad31542015-09-28 17:16:47 +02001279 <format> is the log format used when generating syslog messages. It may be
1280 one of the following :
1281
Emeric Bruna0338b92020-11-27 16:24:34 +01001282 local Analog to rfc3164 syslog message format except that hostname
1283 field is stripped. This is the default.
1284 Note: option "log-send-hostname" switches the default to
1285 rfc3164.
1286
1287 rfc3164 The RFC3164 syslog message format.
Dragan Dosen7ad31542015-09-28 17:16:47 +02001288 (https://tools.ietf.org/html/rfc3164)
1289
1290 rfc5424 The RFC5424 syslog message format.
1291 (https://tools.ietf.org/html/rfc5424)
1292
Emeric Brun54648852020-07-06 15:54:06 +02001293 priority A message containing only a level plus syslog facility between
1294 angle brackets such as '<63>', followed by the text. The PID,
1295 date, time, process name and system name are omitted. This is
1296 designed to be used with a local log server.
1297
Willy Tarreaue8746a02018-11-12 08:45:00 +01001298 short A message containing only a level between angle brackets such as
1299 '<3>', followed by the text. The PID, date, time, process name
1300 and system name are omitted. This is designed to be used with a
1301 local log server. This format is compatible with what the systemd
1302 logger consumes.
1303
Emeric Brun54648852020-07-06 15:54:06 +02001304 timed A message containing only a level between angle brackets such as
1305 '<3>', followed by ISO date and by the text. The PID, process
1306 name and system name are omitted. This is designed to be
1307 used with a local log server.
1308
1309 iso A message containing only the ISO date, followed by the text.
1310 The PID, process name and system name are omitted. This is
1311 designed to be used with a local log server.
1312
Willy Tarreauc1b06452018-11-12 11:57:56 +01001313 raw A message containing only the text. The level, PID, date, time,
1314 process name and system name are omitted. This is designed to be
1315 used in containers or during development, where the severity only
1316 depends on the file descriptor used (stdout/stderr).
1317
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02001318 <ranges> A list of comma-separated ranges to identify the logs to sample.
1319 This is used to balance the load of the logs to send to the log
1320 server. The limits of the ranges cannot be null. They are numbered
1321 from 1. The size or period (in number of logs) of the sample must be
1322 set with <sample_size> parameter.
1323
1324 <sample_size>
1325 The size of the sample in number of logs to consider when balancing
1326 their logging loads. It is used to balance the load of the logs to
1327 send to the syslog server. This size must be greater or equal to the
1328 maximum of the high limits of the ranges.
1329 (see also <ranges> parameter).
1330
Robert Tsai81ae1952007-12-05 10:47:29 +01001331 <facility> must be one of the 24 standard syslog facilities :
Willy Tarreau6a06a402007-07-15 20:15:28 +02001332
Willy Tarreaue8746a02018-11-12 08:45:00 +01001333 kern user mail daemon auth syslog lpr news
1334 uucp cron auth2 ftp ntp audit alert cron2
1335 local0 local1 local2 local3 local4 local5 local6 local7
1336
Willy Tarreauc1b06452018-11-12 11:57:56 +01001337 Note that the facility is ignored for the "short" and "raw"
1338 formats, but still required as a positional field. It is
1339 recommended to use "daemon" in this case to make it clear that
1340 it's only supposed to be used locally.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001341
1342 An optional level can be specified to filter outgoing messages. By default,
Willy Tarreauf7edefa2009-05-10 17:20:05 +02001343 all messages are sent. If a maximum level is specified, only messages with a
1344 severity at least as important as this level will be sent. An optional minimum
1345 level can be specified. If it is set, logs emitted with a more severe level
1346 than this one will be capped to this level. This is used to avoid sending
1347 "emerg" messages on all terminals on some default syslog configurations.
1348 Eight levels are known :
Willy Tarreau6a06a402007-07-15 20:15:28 +02001349
Cyril Bontédc4d9032012-04-08 21:57:39 +02001350 emerg alert crit err warning notice info debug
Willy Tarreau6a06a402007-07-15 20:15:28 +02001351
Joe Williamsdf5b38f2010-12-29 17:05:48 +01001352log-send-hostname [<string>]
1353 Sets the hostname field in the syslog header. If optional "string" parameter
1354 is set the header is set to the string contents, otherwise uses the hostname
1355 of the system. Generally used if one is not relaying logs through an
1356 intermediate syslog server or for simply customizing the hostname printed in
1357 the logs.
1358
Kevinm48936af2010-12-22 16:08:21 +00001359log-tag <string>
1360 Sets the tag field in the syslog header to this string. It defaults to the
1361 program name as launched from the command line, which usually is "haproxy".
1362 Sometimes it can be useful to differentiate between multiple processes
Willy Tarreau094af4e2015-01-07 15:03:42 +01001363 running on the same host. See also the per-proxy "log-tag" directive.
Kevinm48936af2010-12-22 16:08:21 +00001364
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001365lua-load <file>
1366 This global directive loads and executes a Lua file. This directive can be
1367 used multiple times.
1368
Tim Duesterhusdd74b5f2020-01-12 13:55:40 +01001369lua-prepend-path <string> [<type>]
1370 Prepends the given string followed by a semicolon to Lua's package.<type>
1371 variable.
1372 <type> must either be "path" or "cpath". If <type> is not given it defaults
1373 to "path".
1374
1375 Lua's paths are semicolon delimited lists of patterns that specify how the
1376 `require` function attempts to find the source file of a library. Question
1377 marks (?) within a pattern will be replaced by module name. The path is
1378 evaluated left to right. This implies that paths that are prepended later
1379 will be checked earlier.
1380
1381 As an example by specifying the following path:
1382
1383 lua-prepend-path /usr/share/haproxy-lua/?/init.lua
1384 lua-prepend-path /usr/share/haproxy-lua/?.lua
1385
1386 When `require "example"` is being called Lua will first attempt to load the
1387 /usr/share/haproxy-lua/example.lua script, if that does not exist the
1388 /usr/share/haproxy-lua/example/init.lua will be attempted and the default
1389 paths if that does not exist either.
1390
1391 See https://www.lua.org/pil/8.1.html for the details within the Lua
1392 documentation.
1393
William Lallemand4cfede82017-11-24 22:02:34 +01001394master-worker [no-exit-on-failure]
William Lallemande202b1e2017-06-01 17:38:56 +02001395 Master-worker mode. It is equivalent to the command line "-W" argument.
1396 This mode will launch a "master" which will monitor the "workers". Using
1397 this mode, you can reload HAProxy directly by sending a SIGUSR2 signal to
Davor Ocelice9ed2812017-12-25 17:49:28 +01001398 the master. The master-worker mode is compatible either with the foreground
William Lallemande202b1e2017-06-01 17:38:56 +02001399 or daemon mode. It is recommended to use this mode with multiprocess and
1400 systemd.
William Lallemand4cfede82017-11-24 22:02:34 +01001401 By default, if a worker exits with a bad return code, in the case of a
1402 segfault for example, all workers will be killed, and the master will leave.
1403 It is convenient to combine this behavior with Restart=on-failure in a
1404 systemd unit file in order to relaunch the whole process. If you don't want
1405 this behavior, you must use the keyword "no-exit-on-failure".
William Lallemande202b1e2017-06-01 17:38:56 +02001406
William Lallemand4cfede82017-11-24 22:02:34 +01001407 See also "-W" in the management guide.
William Lallemande202b1e2017-06-01 17:38:56 +02001408
William Lallemand27edc4b2019-05-07 17:49:33 +02001409mworker-max-reloads <number>
1410 In master-worker mode, this option limits the number of time a worker can
John Roeslerfb2fce12019-07-10 15:45:51 -05001411 survive to a reload. If the worker did not leave after a reload, once its
William Lallemand27edc4b2019-05-07 17:49:33 +02001412 number of reloads is greater than this number, the worker will receive a
1413 SIGTERM. This option helps to keep under control the number of workers.
1414 See also "show proc" in the Management Guide.
1415
Willy Tarreauf42d7942020-10-20 11:54:49 +02001416nbproc <number> (deprecated)
Willy Tarreau6a06a402007-07-15 20:15:28 +02001417 Creates <number> processes when going daemon. This requires the "daemon"
1418 mode. By default, only one process is created, which is the recommended mode
1419 of operation. For systems limited to small sets of file descriptors per
Willy Tarreau149ab772019-01-26 14:27:06 +01001420 process, it may be needed to fork multiple daemons. When set to a value
1421 larger than 1, threads are automatically disabled. USING MULTIPLE PROCESSES
Willy Tarreauf42d7942020-10-20 11:54:49 +02001422 IS HARDER TO DEBUG AND IS REALLY DISCOURAGED. This directive is deprecated
1423 and scheduled for removal in 2.5. Please use "nbthread" instead. See also
1424 "daemon" and "nbthread".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001425
Christopher Fauletbe0faa22017-08-29 15:37:10 +02001426nbthread <number>
1427 This setting is only available when support for threads was built in. It
Willy Tarreau26f6ae12019-02-02 12:56:15 +01001428 makes haproxy run on <number> threads. This is exclusive with "nbproc". While
1429 "nbproc" historically used to be the only way to use multiple processors, it
1430 also involved a number of shortcomings related to the lack of synchronization
1431 between processes (health-checks, peers, stick-tables, stats, ...) which do
1432 not affect threads. As such, any modern configuration is strongly encouraged
Willy Tarreau149ab772019-01-26 14:27:06 +01001433 to migrate away from "nbproc" to "nbthread". "nbthread" also works when
1434 HAProxy is started in foreground. On some platforms supporting CPU affinity,
1435 when nbproc is not used, the default "nbthread" value is automatically set to
1436 the number of CPUs the process is bound to upon startup. This means that the
1437 thread count can easily be adjusted from the calling process using commands
1438 like "taskset" or "cpuset". Otherwise, this value defaults to 1. The default
1439 value is reported in the output of "haproxy -vv". See also "nbproc".
Christopher Fauletbe0faa22017-08-29 15:37:10 +02001440
Willy Tarreau6a06a402007-07-15 20:15:28 +02001441pidfile <pidfile>
MIZUTA Takeshic32f3942020-08-26 13:46:19 +09001442 Writes PIDs of all daemons into file <pidfile> when daemon mode or writes PID
1443 of master process into file <pidfile> when master-worker mode. This option is
1444 equivalent to the "-p" command line argument. The file must be accessible to
1445 the user starting the process. See also "daemon" and "master-worker".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001446
Willy Tarreau119e50e2020-05-22 13:53:29 +02001447pp2-never-send-local
1448 A bug in the PROXY protocol v2 implementation was present in HAProxy up to
1449 version 2.1, causing it to emit a PROXY command instead of a LOCAL command
1450 for health checks. This is particularly minor but confuses some servers'
1451 logs. Sadly, the bug was discovered very late and revealed that some servers
1452 which possibly only tested their PROXY protocol implementation against
1453 HAProxy fail to properly handle the LOCAL command, and permanently remain in
1454 the "down" state when HAProxy checks them. When this happens, it is possible
1455 to enable this global option to revert to the older (bogus) behavior for the
1456 time it takes to contact the affected components' vendors and get them fixed.
1457 This option is disabled by default and acts on all servers having the
1458 "send-proxy-v2" statement.
1459
Willy Tarreau1d549722016-02-16 12:41:57 +01001460presetenv <name> <value>
1461 Sets environment variable <name> to value <value>. If the variable exists, it
1462 is NOT overwritten. The changes immediately take effect so that the next line
1463 in the configuration file sees the new value. See also "setenv", "resetenv",
1464 and "unsetenv".
1465
1466resetenv [<name> ...]
1467 Removes all environment variables except the ones specified in argument. It
1468 allows to use a clean controlled environment before setting new values with
1469 setenv or unsetenv. Please note that some internal functions may make use of
1470 some environment variables, such as time manipulation functions, but also
1471 OpenSSL or even external checks. This must be used with extreme care and only
1472 after complete validation. The changes immediately take effect so that the
1473 next line in the configuration file sees the new environment. See also
1474 "setenv", "presetenv", and "unsetenv".
1475
Christopher Fauletff4121f2017-11-22 16:38:49 +01001476stats bind-process [ all | odd | even | <process_num>[-[process_num>]] ] ...
Willy Tarreau35b7b162012-10-22 23:17:18 +02001477 Limits the stats socket to a certain set of processes numbers. By default the
1478 stats socket is bound to all processes, causing a warning to be emitted when
1479 nbproc is greater than 1 because there is no way to select the target process
1480 when connecting. However, by using this setting, it becomes possible to pin
1481 the stats socket to a specific set of processes, typically the first one. The
1482 warning will automatically be disabled when this setting is used, whatever
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001483 the number of processes used. The maximum process ID depends on the machine's
Christopher Fauletff4121f2017-11-22 16:38:49 +01001484 word size (32 or 64). Ranges can be partially defined. The higher bound can
1485 be omitted. In such case, it is replaced by the corresponding maximum
1486 value. A better option consists in using the "process" setting of the "stats
1487 socket" line to force the process on each line.
Willy Tarreau35b7b162012-10-22 23:17:18 +02001488
Baptiste Assmann5626f482015-08-23 10:00:10 +02001489server-state-base <directory>
1490 Specifies the directory prefix to be prepended in front of all servers state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001491 file names which do not start with a '/'. See also "server-state-file",
1492 "load-server-state-from-file" and "server-state-file-name".
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +02001493
1494server-state-file <file>
1495 Specifies the path to the file containing state of servers. If the path starts
1496 with a slash ('/'), it is considered absolute, otherwise it is considered
1497 relative to the directory specified using "server-state-base" (if set) or to
1498 the current directory. Before reloading HAProxy, it is possible to save the
1499 servers' current state using the stats command "show servers state". The
1500 output of this command must be written in the file pointed by <file>. When
1501 starting up, before handling traffic, HAProxy will read, load and apply state
1502 for each server found in the file and available in its current running
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001503 configuration. See also "server-state-base" and "show servers state",
1504 "load-server-state-from-file" and "server-state-file-name"
Baptiste Assmann5626f482015-08-23 10:00:10 +02001505
Willy Tarreau1d549722016-02-16 12:41:57 +01001506setenv <name> <value>
1507 Sets environment variable <name> to value <value>. If the variable exists, it
1508 is overwritten. The changes immediately take effect so that the next line in
1509 the configuration file sees the new value. See also "presetenv", "resetenv",
1510 and "unsetenv".
1511
Willy Tarreau636848a2019-04-15 19:38:50 +02001512set-dumpable
1513 This option is better left disabled by default and enabled only upon a
William Dauchyec730982019-10-27 20:08:10 +01001514 developer's request. If it has been enabled, it may still be forcibly
1515 disabled by prefixing it with the "no" keyword. It has no impact on
1516 performance nor stability but will try hard to re-enable core dumps that were
1517 possibly disabled by file size limitations (ulimit -f), core size limitations
1518 (ulimit -c), or "dumpability" of a process after changing its UID/GID (such
1519 as /proc/sys/fs/suid_dumpable on Linux). Core dumps might still be limited by
1520 the current directory's permissions (check what directory the file is started
1521 from), the chroot directory's permission (it may be needed to temporarily
1522 disable the chroot directive or to move it to a dedicated writable location),
1523 or any other system-specific constraint. For example, some Linux flavours are
1524 notorious for replacing the default core file with a path to an executable
1525 not even installed on the system (check /proc/sys/kernel/core_pattern). Often,
1526 simply writing "core", "core.%p" or "/var/log/core/core.%p" addresses the
1527 issue. When trying to enable this option waiting for a rare issue to
1528 re-appear, it's often a good idea to first try to obtain such a dump by
1529 issuing, for example, "kill -11" to the haproxy process and verify that it
1530 leaves a core where expected when dying.
Willy Tarreau636848a2019-04-15 19:38:50 +02001531
Willy Tarreau610f04b2014-02-13 11:36:41 +01001532ssl-default-bind-ciphers <ciphers>
1533 This setting is only available when support for OpenSSL was built in. It sets
1534 the default string describing the list of cipher algorithms ("cipher suite")
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +00001535 that are negotiated during the SSL/TLS handshake up to TLSv1.2 for all
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001536 "bind" lines which do not explicitly define theirs. The format of the string
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001537 is defined in "man 1 ciphers" from OpenSSL man pages. For background
1538 information and recommendations see e.g.
1539 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
1540 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
1541 cipher configuration, please check the "ssl-default-bind-ciphersuites" keyword.
1542 Please check the "bind" keyword for more information.
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001543
1544ssl-default-bind-ciphersuites <ciphersuites>
1545 This setting is only available when support for OpenSSL was built in and
1546 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default string
1547 describing the list of cipher algorithms ("cipher suite") that are negotiated
1548 during the TLSv1.3 handshake for all "bind" lines which do not explicitly define
1549 theirs. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001550 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites". For
1551 cipher configuration for TLSv1.2 and earlier, please check the
1552 "ssl-default-bind-ciphers" keyword. Please check the "bind" keyword for more
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001553 information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001554
Jerome Magninb203ff62020-04-03 15:28:22 +02001555ssl-default-bind-curves <curves>
1556 This setting is only available when support for OpenSSL was built in. It sets
1557 the default string describing the list of elliptic curves algorithms ("curve
1558 suite") that are negotiated during the SSL/TLS handshake with ECDHE. The format
1559 of the string is a colon-delimited list of curve name.
1560 Please check the "bind" keyword for more information.
1561
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001562ssl-default-bind-options [<option>]...
1563 This setting is only available when support for OpenSSL was built in. It sets
1564 default ssl-options to force on all "bind" lines. Please check the "bind"
1565 keyword to see available options.
1566
1567 Example:
1568 global
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +02001569 ssl-default-bind-options ssl-min-ver TLSv1.0 no-tls-tickets
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001570
Willy Tarreau610f04b2014-02-13 11:36:41 +01001571ssl-default-server-ciphers <ciphers>
1572 This setting is only available when support for OpenSSL was built in. It
1573 sets the default string describing the list of cipher algorithms that are
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +00001574 negotiated during the SSL/TLS handshake up to TLSv1.2 with the server,
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001575 for all "server" lines which do not explicitly define theirs. The format of
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001576 the string is defined in "man 1 ciphers" from OpenSSL man pages. For background
1577 information and recommendations see e.g.
1578 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
1579 (https://mozilla.github.io/server-side-tls/ssl-config-generator/).
1580 For TLSv1.3 cipher configuration, please check the
1581 "ssl-default-server-ciphersuites" keyword. Please check the "server" keyword
1582 for more information.
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001583
1584ssl-default-server-ciphersuites <ciphersuites>
1585 This setting is only available when support for OpenSSL was built in and
1586 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default
1587 string describing the list of cipher algorithms that are negotiated during
1588 the TLSv1.3 handshake with the server, for all "server" lines which do not
1589 explicitly define theirs. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001590 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites". For
1591 cipher configuration for TLSv1.2 and earlier, please check the
1592 "ssl-default-server-ciphers" keyword. Please check the "server" keyword for
1593 more information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001594
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001595ssl-default-server-options [<option>]...
1596 This setting is only available when support for OpenSSL was built in. It sets
1597 default ssl-options to force on all "server" lines. Please check the "server"
1598 keyword to see available options.
1599
Remi Gacogne47783ef2015-05-29 15:53:22 +02001600ssl-dh-param-file <file>
1601 This setting is only available when support for OpenSSL was built in. It sets
1602 the default DH parameters that are used during the SSL/TLS handshake when
1603 ephemeral Diffie-Hellman (DHE) key exchange is used, for all "bind" lines
Davor Ocelice9ed2812017-12-25 17:49:28 +01001604 which do not explicitly define theirs. It will be overridden by custom DH
Remi Gacogne47783ef2015-05-29 15:53:22 +02001605 parameters found in a bind certificate file if any. If custom DH parameters
Cyril Bonté307ee1e2015-09-28 23:16:06 +02001606 are not specified either by using ssl-dh-param-file or by setting them
1607 directly in the certificate file, pre-generated DH parameters of the size
1608 specified by tune.ssl.default-dh-param will be used. Custom parameters are
1609 known to be more secure and therefore their use is recommended.
Remi Gacogne47783ef2015-05-29 15:53:22 +02001610 Custom DH parameters may be generated by using the OpenSSL command
1611 "openssl dhparam <size>", where size should be at least 2048, as 1024-bit DH
1612 parameters should not be considered secure anymore.
1613
William Lallemand8e8581e2020-10-20 17:36:46 +02001614ssl-load-extra-del-ext
1615 This setting allows to configure the way HAProxy does the lookup for the
1616 extra SSL files. By default HAProxy adds a new extension to the filename.
William Lallemand089c1382020-10-23 17:35:12 +02001617 (ex: with "foobar.crt" load "foobar.crt.key"). With this option enabled,
William Lallemand8e8581e2020-10-20 17:36:46 +02001618 HAProxy removes the extension before adding the new one (ex: with
William Lallemand089c1382020-10-23 17:35:12 +02001619 "foobar.crt" load "foobar.key").
1620
1621 Your crt file must have a ".crt" extension for this option to work.
William Lallemand8e8581e2020-10-20 17:36:46 +02001622
1623 This option is not compatible with bundle extensions (.ecdsa, .rsa. .dsa)
1624 and won't try to remove them.
1625
1626 This option is disabled by default. See also "ssl-load-extra-files".
1627
William Lallemand4c5adbf2020-02-24 14:23:22 +01001628ssl-load-extra-files <none|all|bundle|sctl|ocsp|issuer|key>*
William Lallemand3af48e72020-02-03 17:15:52 +01001629 This setting alters the way HAProxy will look for unspecified files during
Jerome Magnin587be9c2020-09-07 11:55:57 +02001630 the loading of the SSL certificates associated to "bind" lines. It does not
1631 apply to certificates used for client authentication on "server" lines.
William Lallemand3af48e72020-02-03 17:15:52 +01001632
1633 By default, HAProxy discovers automatically a lot of files not specified in
1634 the configuration, and you may want to disable this behavior if you want to
1635 optimize the startup time.
1636
1637 "none": Only load the files specified in the configuration. Don't try to load
1638 a certificate bundle if the file does not exist. In the case of a directory,
1639 it won't try to bundle the certificates if they have the same basename.
1640
1641 "all": This is the default behavior, it will try to load everything,
William Lallemand4c5adbf2020-02-24 14:23:22 +01001642 bundles, sctl, ocsp, issuer, key.
William Lallemand3af48e72020-02-03 17:15:52 +01001643
1644 "bundle": When a file specified in the configuration does not exist, HAProxy
William Lallemandf9ff3ec2020-10-02 17:57:44 +02001645 will try to load a "cert bundle".
1646
1647 Starting from HAProxy 2.3, the bundles are not loaded in the same OpenSSL
1648 certificate store, instead it will loads each certificate in a separate
1649 store which is equivalent to declaring multiple "crt". OpenSSL 1.1.1 is
1650 required to achieve this. Which means that bundles are now used only for
1651 backward compatibility and are not mandatory anymore to do an hybrid RSA/ECC
1652 bind configuration..
1653
1654 To associate these PEM files into a "cert bundle" that is recognized by
1655 haproxy, they must be named in the following way: All PEM files that are to
1656 be bundled must have the same base name, with a suffix indicating the key
1657 type. Currently, three suffixes are supported: rsa, dsa and ecdsa. For
1658 example, if www.example.com has two PEM files, an RSA file and an ECDSA
1659 file, they must be named: "example.pem.rsa" and "example.pem.ecdsa". The
1660 first part of the filename is arbitrary; only the suffix matters. To load
1661 this bundle into haproxy, specify the base name only:
1662
1663 Example : bind :8443 ssl crt example.pem
1664
1665 Note that the suffix is not given to haproxy; this tells haproxy to look for
1666 a cert bundle.
1667
1668 HAProxy will load all PEM files in the bundle as if they were configured
1669 separately in several "crt".
1670
1671 The bundle loading does not have an impact anymore on the directory loading
1672 since files are loading separately.
1673
1674 On the CLI, bundles are seen as separate files, and the bundle extension is
1675 required to commit them.
1676
William Dauchy57dd6f12020-10-06 15:22:37 +02001677 OCSP files (.ocsp), issuer files (.issuer), Certificate Transparency (.sctl)
William Lallemandf9ff3ec2020-10-02 17:57:44 +02001678 as well as private keys (.key) are supported with multi-cert bundling.
William Lallemand3af48e72020-02-03 17:15:52 +01001679
1680 "sctl": Try to load "<basename>.sctl" for each crt keyword.
1681
1682 "ocsp": Try to load "<basename>.ocsp" for each crt keyword.
1683
1684 "issuer": Try to load "<basename>.issuer" if the issuer of the OCSP file is
1685 not provided in the PEM file.
1686
William Lallemand4c5adbf2020-02-24 14:23:22 +01001687 "key": If the private key was not provided by the PEM file, try to load a
1688 file "<basename>.key" containing a private key.
1689
William Lallemand3af48e72020-02-03 17:15:52 +01001690 The default behavior is "all".
1691
1692 Example:
1693 ssl-load-extra-files bundle sctl
1694 ssl-load-extra-files sctl ocsp issuer
1695 ssl-load-extra-files none
1696
1697 See also: "crt", section 5.1 about bind options.
1698
Emeric Brun850efd52014-01-29 12:24:34 +01001699ssl-server-verify [none|required]
1700 The default behavior for SSL verify on servers side. If specified to 'none',
1701 servers certificates are not verified. The default is 'required' except if
1702 forced using cmdline option '-dV'.
1703
Emmanuel Hocdetc3b7e742020-04-22 11:06:19 +02001704ssl-skip-self-issued-ca
Daniel Corbett67a82712020-07-06 23:01:19 -04001705 Self issued CA, aka x509 root CA, is the anchor for chain validation: as a
Emmanuel Hocdetc3b7e742020-04-22 11:06:19 +02001706 server is useless to send it, client must have it. Standard configuration
1707 need to not include such CA in PEM file. This option allows you to keep such
1708 CA in PEM file without sending it to the client. Use case is to provide
1709 issuer for ocsp without the need for '.issuer' file and be able to share it
1710 with 'issuers-chain-path'. This concerns all certificates without intermediate
1711 certificates. It's useless for BoringSSL, .issuer is ignored because ocsp
William Lallemand9a1d8392020-08-10 17:28:23 +02001712 bits does not need it. Requires at least OpenSSL 1.0.2.
Emmanuel Hocdetc3b7e742020-04-22 11:06:19 +02001713
Willy Tarreauabb175f2012-09-24 12:43:26 +02001714stats socket [<address:port>|<path>] [param*]
1715 Binds a UNIX socket to <path> or a TCPv4/v6 address to <address:port>.
1716 Connections to this socket will return various statistics outputs and even
1717 allow some commands to be issued to change some runtime settings. Please
Willy Tarreau1af20c72017-06-23 16:01:14 +02001718 consult section 9.3 "Unix Socket commands" of Management Guide for more
Kevin Decherf949c7202015-10-13 23:26:44 +02001719 details.
Willy Tarreau6162db22009-10-10 17:13:00 +02001720
Willy Tarreauabb175f2012-09-24 12:43:26 +02001721 All parameters supported by "bind" lines are supported, for instance to
1722 restrict access to some users or their access rights. Please consult
1723 section 5.1 for more information.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001724
1725stats timeout <timeout, in milliseconds>
1726 The default timeout on the stats socket is set to 10 seconds. It is possible
1727 to change this value with "stats timeout". The value must be passed in
Willy Tarreaubefdff12007-12-02 22:27:38 +01001728 milliseconds, or be suffixed by a time unit among { us, ms, s, m, h, d }.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001729
1730stats maxconn <connections>
1731 By default, the stats socket is limited to 10 concurrent connections. It is
1732 possible to change this value with "stats maxconn".
1733
Willy Tarreau6a06a402007-07-15 20:15:28 +02001734uid <number>
Thayne McCombsdab4ba62021-01-07 21:24:41 -07001735 Changes the process's user ID to <number>. It is recommended that the user ID
Willy Tarreau6a06a402007-07-15 20:15:28 +02001736 is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
1737 be started with superuser privileges in order to be able to switch to another
1738 one. See also "gid" and "user".
1739
1740ulimit-n <number>
1741 Sets the maximum number of per-process file-descriptors to <number>. By
1742 default, it is automatically computed, so it is recommended not to use this
1743 option.
1744
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001745unix-bind [ prefix <prefix> ] [ mode <mode> ] [ user <user> ] [ uid <uid> ]
1746 [ group <group> ] [ gid <gid> ]
1747
1748 Fixes common settings to UNIX listening sockets declared in "bind" statements.
1749 This is mainly used to simplify declaration of those UNIX sockets and reduce
1750 the risk of errors, since those settings are most commonly required but are
1751 also process-specific. The <prefix> setting can be used to force all socket
1752 path to be relative to that directory. This might be needed to access another
1753 component's chroot. Note that those paths are resolved before haproxy chroots
1754 itself, so they are absolute. The <mode>, <user>, <uid>, <group> and <gid>
1755 all have the same meaning as their homonyms used by the "bind" statement. If
1756 both are specified, the "bind" statement has priority, meaning that the
1757 "unix-bind" settings may be seen as process-wide default settings.
1758
Willy Tarreau1d549722016-02-16 12:41:57 +01001759unsetenv [<name> ...]
1760 Removes environment variables specified in arguments. This can be useful to
1761 hide some sensitive information that are occasionally inherited from the
1762 user's environment during some operations. Variables which did not exist are
1763 silently ignored so that after the operation, it is certain that none of
1764 these variables remain. The changes immediately take effect so that the next
1765 line in the configuration file will not see these variables. See also
1766 "setenv", "presetenv", and "resetenv".
1767
Willy Tarreau6a06a402007-07-15 20:15:28 +02001768user <user name>
1769 Similar to "uid" but uses the UID of user name <user name> from /etc/passwd.
1770 See also "uid" and "group".
1771
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02001772node <name>
1773 Only letters, digits, hyphen and underscore are allowed, like in DNS names.
1774
1775 This statement is useful in HA configurations where two or more processes or
1776 servers share the same IP address. By setting a different node-name on all
1777 nodes, it becomes easy to immediately spot what server is handling the
1778 traffic.
1779
1780description <text>
1781 Add a text that describes the instance.
1782
1783 Please note that it is required to escape certain characters (# for example)
1784 and this text is inserted into a html page so you should avoid using
1785 "<" and ">" characters.
1786
Thomas Holmesdb04f192015-05-18 13:21:39 +0100178751degrees-data-file <file path>
1788 The path of the 51Degrees data file to provide device detection services. The
Davor Ocelice9ed2812017-12-25 17:49:28 +01001789 file should be unzipped and accessible by HAProxy with relevant permissions.
Thomas Holmesdb04f192015-05-18 13:21:39 +01001790
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001791 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001792 compiled with USE_51DEGREES.
1793
Ben Shillitof25e8e52016-12-02 14:25:37 +0000179451degrees-property-name-list [<string> ...]
Thomas Holmesdb04f192015-05-18 13:21:39 +01001795 A list of 51Degrees property names to be load from the dataset. A full list
1796 of names is available on the 51Degrees website:
1797 https://51degrees.com/resources/property-dictionary
1798
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001799 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001800 compiled with USE_51DEGREES.
1801
Dragan Dosen93b38d92015-06-29 16:43:25 +0200180251degrees-property-separator <char>
Thomas Holmesdb04f192015-05-18 13:21:39 +01001803 A char that will be appended to every property value in a response header
1804 containing 51Degrees results. If not set that will be set as ','.
1805
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001806 Please note that this option is only available when haproxy has been
1807 compiled with USE_51DEGREES.
1808
180951degrees-cache-size <number>
1810 Sets the size of the 51Degrees converter cache to <number> entries. This
1811 is an LRU cache which reminds previous device detections and their results.
1812 By default, this cache is disabled.
1813
1814 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001815 compiled with USE_51DEGREES.
1816
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001817wurfl-data-file <file path>
1818 The path of the WURFL data file to provide device detection services. The
1819 file should be accessible by HAProxy with relevant permissions.
1820
1821 Please note that this option is only available when haproxy has been compiled
1822 with USE_WURFL=1.
1823
1824wurfl-information-list [<capability>]*
1825 A space-delimited list of WURFL capabilities, virtual capabilities, property
1826 names we plan to use in injected headers. A full list of capability and
1827 virtual capability names is available on the Scientiamobile website :
1828
1829 https://www.scientiamobile.com/wurflCapability
1830
1831 Valid WURFL properties are:
1832 - wurfl_id Contains the device ID of the matched device.
1833
1834 - wurfl_root_id Contains the device root ID of the matched
1835 device.
1836
1837 - wurfl_isdevroot Tells if the matched device is a root device.
1838 Possible values are "TRUE" or "FALSE".
1839
1840 - wurfl_useragent The original useragent coming with this
1841 particular web request.
1842
1843 - wurfl_api_version Contains a string representing the currently
1844 used Libwurfl API version.
1845
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001846 - wurfl_info A string containing information on the parsed
1847 wurfl.xml and its full path.
1848
1849 - wurfl_last_load_time Contains the UNIX timestamp of the last time
1850 WURFL has been loaded successfully.
1851
1852 - wurfl_normalized_useragent The normalized useragent.
1853
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001854 Please note that this option is only available when haproxy has been compiled
1855 with USE_WURFL=1.
1856
1857wurfl-information-list-separator <char>
1858 A char that will be used to separate values in a response header containing
1859 WURFL results. If not set that a comma (',') will be used by default.
1860
1861 Please note that this option is only available when haproxy has been compiled
1862 with USE_WURFL=1.
1863
1864wurfl-patch-file [<file path>]
1865 A list of WURFL patch file paths. Note that patches are loaded during startup
1866 thus before the chroot.
1867
1868 Please note that this option is only available when haproxy has been compiled
1869 with USE_WURFL=1.
1870
paulborilebad132c2019-04-18 11:57:04 +02001871wurfl-cache-size <size>
1872 Sets the WURFL Useragent cache size. For faster lookups, already processed user
1873 agents are kept in a LRU cache :
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001874 - "0" : no cache is used.
paulborilebad132c2019-04-18 11:57:04 +02001875 - <size> : size of lru cache in elements.
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001876
1877 Please note that this option is only available when haproxy has been compiled
1878 with USE_WURFL=1.
1879
William Dauchy0fec3ab2019-10-27 20:08:11 +01001880strict-limits
William Dauchya5194602020-03-28 19:29:58 +01001881 Makes process fail at startup when a setrlimit fails. Haproxy tries to set the
1882 best setrlimit according to what has been calculated. If it fails, it will
1883 emit a warning. This option is here to guarantee an explicit failure of
1884 haproxy when those limits fail. It is enabled by default. It may still be
1885 forcibly disabled by prefixing it with the "no" keyword.
William Dauchy0fec3ab2019-10-27 20:08:11 +01001886
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018873.2. Performance tuning
Willy Tarreau6a06a402007-07-15 20:15:28 +02001888-----------------------
1889
Willy Tarreaubeb859a2018-11-22 18:07:59 +01001890busy-polling
1891 In some situations, especially when dealing with low latency on processors
1892 supporting a variable frequency or when running inside virtual machines, each
1893 time the process waits for an I/O using the poller, the processor goes back
1894 to sleep or is offered to another VM for a long time, and it causes
1895 excessively high latencies. This option provides a solution preventing the
1896 processor from sleeping by always using a null timeout on the pollers. This
1897 results in a significant latency reduction (30 to 100 microseconds observed)
1898 at the expense of a risk to overheat the processor. It may even be used with
1899 threads, in which case improperly bound threads may heavily conflict,
1900 resulting in a worse performance and high values for the CPU stolen fields
1901 in "show info" output, indicating which threads are misconfigured. It is
1902 important not to let the process run on the same processor as the network
1903 interrupts when this option is used. It is also better to avoid using it on
1904 multiple CPU threads sharing the same core. This option is disabled by
1905 default. If it has been enabled, it may still be forcibly disabled by
1906 prefixing it with the "no" keyword. It is ignored by the "select" and
1907 "poll" pollers.
1908
William Dauchy3894d972019-12-28 15:36:02 +01001909 This option is automatically disabled on old processes in the context of
1910 seamless reload; it avoids too much cpu conflicts when multiple processes
1911 stay around for some time waiting for the end of their current connections.
1912
Willy Tarreau1746eec2014-04-25 10:46:47 +02001913max-spread-checks <delay in milliseconds>
1914 By default, haproxy tries to spread the start of health checks across the
1915 smallest health check interval of all the servers in a farm. The principle is
1916 to avoid hammering services running on the same server. But when using large
1917 check intervals (10 seconds or more), the last servers in the farm take some
1918 time before starting to be tested, which can be a problem. This parameter is
1919 used to enforce an upper bound on delay between the first and the last check,
1920 even if the servers' check intervals are larger. When servers run with
1921 shorter intervals, their intervals will be respected though.
1922
Willy Tarreau6a06a402007-07-15 20:15:28 +02001923maxconn <number>
1924 Sets the maximum per-process number of concurrent connections to <number>. It
1925 is equivalent to the command-line argument "-n". Proxies will stop accepting
1926 connections when this limit is reached. The "ulimit-n" parameter is
Willy Tarreau8274e102014-06-19 15:31:25 +02001927 automatically adjusted according to this value. See also "ulimit-n". Note:
1928 the "select" poller cannot reliably use more than 1024 file descriptors on
1929 some platforms. If your platform only supports select and reports "select
1930 FAILED" on startup, you need to reduce maxconn until it works (slightly
Willy Tarreaub28f3442019-03-04 08:13:43 +01001931 below 500 in general). If this value is not set, it will automatically be
1932 calculated based on the current file descriptors limit reported by the
1933 "ulimit -n" command, possibly reduced to a lower value if a memory limit
1934 is enforced, based on the buffer size, memory allocated to compression, SSL
1935 cache size, and use or not of SSL and the associated maxsslconn (which can
1936 also be automatic).
Willy Tarreau6a06a402007-07-15 20:15:28 +02001937
Willy Tarreau81c25d02011-09-07 15:17:21 +02001938maxconnrate <number>
1939 Sets the maximum per-process number of connections per second to <number>.
1940 Proxies will stop accepting connections when this limit is reached. It can be
1941 used to limit the global capacity regardless of each frontend capacity. It is
1942 important to note that this can only be used as a service protection measure,
1943 as there will not necessarily be a fair share between frontends when the
1944 limit is reached, so it's a good idea to also limit each frontend to some
1945 value close to its expected share. Also, lowering tune.maxaccept can improve
1946 fairness.
1947
William Lallemandd85f9172012-11-09 17:05:39 +01001948maxcomprate <number>
1949 Sets the maximum per-process input compression rate to <number> kilobytes
Davor Ocelice9ed2812017-12-25 17:49:28 +01001950 per second. For each session, if the maximum is reached, the compression
William Lallemandd85f9172012-11-09 17:05:39 +01001951 level will be decreased during the session. If the maximum is reached at the
1952 beginning of a session, the session will not compress at all. If the maximum
1953 is not reached, the compression level will be increased up to
Davor Ocelice9ed2812017-12-25 17:49:28 +01001954 tune.comp.maxlevel. A value of zero means there is no limit, this is the
William Lallemandd85f9172012-11-09 17:05:39 +01001955 default value.
1956
William Lallemand072a2bf2012-11-20 17:01:01 +01001957maxcompcpuusage <number>
1958 Sets the maximum CPU usage HAProxy can reach before stopping the compression
1959 for new requests or decreasing the compression level of current requests.
1960 It works like 'maxcomprate' but measures CPU usage instead of incoming data
1961 bandwidth. The value is expressed in percent of the CPU used by haproxy. In
1962 case of multiple processes (nbproc > 1), each process manages its individual
1963 usage. A value of 100 disable the limit. The default value is 100. Setting
1964 a lower value will prevent the compression work from slowing the whole
1965 process down and from introducing high latencies.
1966
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001967maxpipes <number>
1968 Sets the maximum per-process number of pipes to <number>. Currently, pipes
1969 are only used by kernel-based tcp splicing. Since a pipe contains two file
1970 descriptors, the "ulimit-n" value will be increased accordingly. The default
1971 value is maxconn/4, which seems to be more than enough for most heavy usages.
1972 The splice code dynamically allocates and releases pipes, and can fall back
1973 to standard copy, so setting this value too low may only impact performance.
1974
Willy Tarreau93e7c002013-10-07 18:51:07 +02001975maxsessrate <number>
1976 Sets the maximum per-process number of sessions per second to <number>.
1977 Proxies will stop accepting connections when this limit is reached. It can be
1978 used to limit the global capacity regardless of each frontend capacity. It is
1979 important to note that this can only be used as a service protection measure,
1980 as there will not necessarily be a fair share between frontends when the
1981 limit is reached, so it's a good idea to also limit each frontend to some
1982 value close to its expected share. Also, lowering tune.maxaccept can improve
1983 fairness.
1984
Willy Tarreau403edff2012-09-06 11:58:37 +02001985maxsslconn <number>
1986 Sets the maximum per-process number of concurrent SSL connections to
1987 <number>. By default there is no SSL-specific limit, which means that the
1988 global maxconn setting will apply to all connections. Setting this limit
1989 avoids having openssl use too much memory and crash when malloc returns NULL
1990 (since it unfortunately does not reliably check for such conditions). Note
1991 that the limit applies both to incoming and outgoing connections, so one
1992 connection which is deciphered then ciphered accounts for 2 SSL connections.
Willy Tarreaud0256482015-01-15 21:45:22 +01001993 If this value is not set, but a memory limit is enforced, this value will be
1994 automatically computed based on the memory limit, maxconn, the buffer size,
1995 memory allocated to compression, SSL cache size, and use of SSL in either
1996 frontends, backends or both. If neither maxconn nor maxsslconn are specified
1997 when there is a memory limit, haproxy will automatically adjust these values
1998 so that 100% of the connections can be made over SSL with no risk, and will
1999 consider the sides where it is enabled (frontend, backend, both).
Willy Tarreau403edff2012-09-06 11:58:37 +02002000
Willy Tarreaue43d5322013-10-07 20:01:52 +02002001maxsslrate <number>
2002 Sets the maximum per-process number of SSL sessions per second to <number>.
2003 SSL listeners will stop accepting connections when this limit is reached. It
2004 can be used to limit the global SSL CPU usage regardless of each frontend
2005 capacity. It is important to note that this can only be used as a service
2006 protection measure, as there will not necessarily be a fair share between
2007 frontends when the limit is reached, so it's a good idea to also limit each
2008 frontend to some value close to its expected share. It is also important to
2009 note that the sessions are accounted before they enter the SSL stack and not
2010 after, which also protects the stack against bad handshakes. Also, lowering
2011 tune.maxaccept can improve fairness.
2012
William Lallemand9d5f5482012-11-07 16:12:57 +01002013maxzlibmem <number>
2014 Sets the maximum amount of RAM in megabytes per process usable by the zlib.
2015 When the maximum amount is reached, future sessions will not compress as long
2016 as RAM is unavailable. When sets to 0, there is no limit.
William Lallemande3a7d992012-11-20 11:25:20 +01002017 The default value is 0. The value is available in bytes on the UNIX socket
2018 with "show info" on the line "MaxZlibMemUsage", the memory used by zlib is
2019 "ZlibMemUsage" in bytes.
2020
Willy Tarreau6a06a402007-07-15 20:15:28 +02002021noepoll
2022 Disables the use of the "epoll" event polling system on Linux. It is
2023 equivalent to the command-line argument "-de". The next polling system
Willy Tarreaue9f49e72012-11-11 17:42:00 +01002024 used will generally be "poll". See also "nopoll".
Willy Tarreau6a06a402007-07-15 20:15:28 +02002025
2026nokqueue
2027 Disables the use of the "kqueue" event polling system on BSD. It is
2028 equivalent to the command-line argument "-dk". The next polling system
2029 used will generally be "poll". See also "nopoll".
2030
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +00002031noevports
2032 Disables the use of the event ports event polling system on SunOS systems
2033 derived from Solaris 10 and later. It is equivalent to the command-line
2034 argument "-dv". The next polling system used will generally be "poll". See
2035 also "nopoll".
2036
Willy Tarreau6a06a402007-07-15 20:15:28 +02002037nopoll
2038 Disables the use of the "poll" event polling system. It is equivalent to the
2039 command-line argument "-dp". The next polling system used will be "select".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002040 It should never be needed to disable "poll" since it's available on all
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +00002041 platforms supported by HAProxy. See also "nokqueue", "noepoll" and
2042 "noevports".
Willy Tarreau6a06a402007-07-15 20:15:28 +02002043
Willy Tarreauff4f82d2009-02-06 11:28:13 +01002044nosplice
2045 Disables the use of kernel tcp splicing between sockets on Linux. It is
Davor Ocelice9ed2812017-12-25 17:49:28 +01002046 equivalent to the command line argument "-dS". Data will then be copied
Willy Tarreauff4f82d2009-02-06 11:28:13 +01002047 using conventional and more portable recv/send calls. Kernel tcp splicing is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002048 limited to some very recent instances of kernel 2.6. Most versions between
Willy Tarreauff4f82d2009-02-06 11:28:13 +01002049 2.6.25 and 2.6.28 are buggy and will forward corrupted data, so they must not
2050 be used. This option makes it easier to globally disable kernel splicing in
2051 case of doubt. See also "option splice-auto", "option splice-request" and
2052 "option splice-response".
2053
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002054nogetaddrinfo
2055 Disables the use of getaddrinfo(3) for name resolving. It is equivalent to
2056 the command line argument "-dG". Deprecated gethostbyname(3) will be used.
2057
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +00002058noreuseport
2059 Disables the use of SO_REUSEPORT - see socket(7). It is equivalent to the
2060 command line argument "-dR".
2061
Willy Tarreaud2d33482019-04-25 17:09:07 +02002062profiling.tasks { auto | on | off }
2063 Enables ('on') or disables ('off') per-task CPU profiling. When set to 'auto'
2064 the profiling automatically turns on a thread when it starts to suffer from
2065 an average latency of 1000 microseconds or higher as reported in the
2066 "avg_loop_us" activity field, and automatically turns off when the latency
John Roeslerfb2fce12019-07-10 15:45:51 -05002067 returns below 990 microseconds (this value is an average over the last 1024
Willy Tarreaud2d33482019-04-25 17:09:07 +02002068 loops so it does not vary quickly and tends to significantly smooth short
2069 spikes). It may also spontaneously trigger from time to time on overloaded
2070 systems, containers, or virtual machines, or when the system swaps (which
2071 must absolutely never happen on a load balancer).
2072
2073 CPU profiling per task can be very convenient to report where the time is
2074 spent and which requests have what effect on which other request. Enabling
2075 it will typically affect the overall's performance by less than 1%, thus it
2076 is recommended to leave it to the default 'auto' value so that it only
2077 operates when a problem is identified. This feature requires a system
Willy Tarreau75c62c22018-11-22 11:02:09 +01002078 supporting the clock_gettime(2) syscall with clock identifiers
2079 CLOCK_MONOTONIC and CLOCK_THREAD_CPUTIME_ID, otherwise the reported time will
2080 be zero. This option may be changed at run time using "set profiling" on the
2081 CLI.
2082
Willy Tarreaufe255b72007-10-14 23:09:26 +02002083spread-checks <0..50, in percent>
Simon Hormand60d6912013-11-25 10:46:36 +09002084 Sometimes it is desirable to avoid sending agent and health checks to
2085 servers at exact intervals, for instance when many logical servers are
2086 located on the same physical server. With the help of this parameter, it
2087 becomes possible to add some randomness in the check interval between 0
2088 and +/- 50%. A value between 2 and 5 seems to show good results. The
2089 default value remains at 0.
Willy Tarreaufe255b72007-10-14 23:09:26 +02002090
Davor Ocelice9ed2812017-12-25 17:49:28 +01002091ssl-engine <name> [algo <comma-separated list of algorithms>]
Grant Zhang872f9c22017-01-21 01:10:18 +00002092 Sets the OpenSSL engine to <name>. List of valid values for <name> may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01002093 obtained using the command "openssl engine". This statement may be used
Grant Zhang872f9c22017-01-21 01:10:18 +00002094 multiple times, it will simply enable multiple crypto engines. Referencing an
2095 unsupported engine will prevent haproxy from starting. Note that many engines
2096 will lead to lower HTTPS performance than pure software with recent
2097 processors. The optional command "algo" sets the default algorithms an ENGINE
2098 will supply using the OPENSSL function ENGINE_set_default_string(). A value
Davor Ocelice9ed2812017-12-25 17:49:28 +01002099 of "ALL" uses the engine for all cryptographic operations. If no list of
2100 algo is specified then the value of "ALL" is used. A comma-separated list
Grant Zhang872f9c22017-01-21 01:10:18 +00002101 of different algorithms may be specified, including: RSA, DSA, DH, EC, RAND,
2102 CIPHERS, DIGESTS, PKEY, PKEY_CRYPTO, PKEY_ASN1. This is the same format that
2103 openssl configuration file uses:
2104 https://www.openssl.org/docs/man1.0.2/apps/config.html
2105
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00002106ssl-mode-async
2107 Adds SSL_MODE_ASYNC mode to the SSL context. This enables asynchronous TLS
Emeric Brun3854e012017-05-17 20:42:48 +02002108 I/O operations if asynchronous capable SSL engines are used. The current
Emeric Brunb5e42a82017-06-06 12:35:14 +00002109 implementation supports a maximum of 32 engines. The Openssl ASYNC API
2110 doesn't support moving read/write buffers and is not compliant with
2111 haproxy's buffer management. So the asynchronous mode is disabled on
John Roeslerfb2fce12019-07-10 15:45:51 -05002112 read/write operations (it is only enabled during initial and renegotiation
Emeric Brunb5e42a82017-06-06 12:35:14 +00002113 handshakes).
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00002114
Willy Tarreau33cb0652014-12-23 22:52:37 +01002115tune.buffers.limit <number>
2116 Sets a hard limit on the number of buffers which may be allocated per process.
2117 The default value is zero which means unlimited. The minimum non-zero value
2118 will always be greater than "tune.buffers.reserve" and should ideally always
2119 be about twice as large. Forcing this value can be particularly useful to
2120 limit the amount of memory a process may take, while retaining a sane
Davor Ocelice9ed2812017-12-25 17:49:28 +01002121 behavior. When this limit is reached, sessions which need a buffer wait for
Willy Tarreau33cb0652014-12-23 22:52:37 +01002122 another one to be released by another session. Since buffers are dynamically
2123 allocated and released, the waiting time is very short and not perceptible
2124 provided that limits remain reasonable. In fact sometimes reducing the limit
2125 may even increase performance by increasing the CPU cache's efficiency. Tests
2126 have shown good results on average HTTP traffic with a limit to 1/10 of the
2127 expected global maxconn setting, which also significantly reduces memory
2128 usage. The memory savings come from the fact that a number of connections
2129 will not allocate 2*tune.bufsize. It is best not to touch this value unless
2130 advised to do so by an haproxy core developer.
2131
Willy Tarreau1058ae72014-12-23 22:40:40 +01002132tune.buffers.reserve <number>
2133 Sets the number of buffers which are pre-allocated and reserved for use only
2134 during memory shortage conditions resulting in failed memory allocations. The
2135 minimum value is 2 and is also the default. There is no reason a user would
2136 want to change this value, it's mostly aimed at haproxy core developers.
2137
Willy Tarreau27a674e2009-08-17 07:23:33 +02002138tune.bufsize <number>
2139 Sets the buffer size to this size (in bytes). Lower values allow more
2140 sessions to coexist in the same amount of RAM, and higher values allow some
2141 applications with very large cookies to work. The default value is 16384 and
2142 can be changed at build time. It is strongly recommended not to change this
2143 from the default value, as very low values will break some services such as
2144 statistics, and values larger than default size will increase memory usage,
2145 possibly causing the system to run out of memory. At least the global maxconn
Willy Tarreau45a66cc2017-11-24 11:28:00 +01002146 parameter should be decreased by the same factor as this one is increased. In
2147 addition, use of HTTP/2 mandates that this value must be 16384 or more. If an
2148 HTTP request is larger than (tune.bufsize - tune.maxrewrite), haproxy will
Dmitry Sivachenkof6f4f7b2012-10-21 18:10:25 +04002149 return HTTP 400 (Bad Request) error. Similarly if an HTTP response is larger
Willy Tarreauc77d3642018-12-12 06:19:42 +01002150 than this size, haproxy will return HTTP 502 (Bad Gateway). Note that the
2151 value set using this parameter will automatically be rounded up to the next
2152 multiple of 8 on 32-bit machines and 16 on 64-bit machines.
Willy Tarreau27a674e2009-08-17 07:23:33 +02002153
Christopher Faulet7151a122020-11-25 17:20:57 +01002154tune.chksize <number> (deprecated)
2155 This option is deprecated and ignored.
Willy Tarreau43961d52010-10-04 20:39:20 +02002156
William Lallemandf3747832012-11-09 12:33:10 +01002157tune.comp.maxlevel <number>
2158 Sets the maximum compression level. The compression level affects CPU
2159 usage during compression. This value affects CPU usage during compression.
2160 Each session using compression initializes the compression algorithm with
2161 this value. The default value is 1.
2162
Willy Tarreauc299e1e2019-02-27 11:35:12 +01002163tune.fail-alloc
2164 If compiled with DEBUG_FAIL_ALLOC, gives the percentage of chances an
2165 allocation attempt fails. Must be between 0 (no failure) and 100 (no
2166 success). This is useful to debug and make sure memory failures are handled
2167 gracefully.
2168
Willy Tarreaubc52bec2020-06-18 08:58:47 +02002169tune.fd.edge-triggered { on | off } [ EXPERIMENTAL ]
2170 Enables ('on') or disables ('off') the edge-triggered polling mode for FDs
2171 that support it. This is currently only support with epoll. It may noticeably
2172 reduce the number of epoll_ctl() calls and slightly improve performance in
2173 certain scenarios. This is still experimental, it may result in frozen
2174 connections if bugs are still present, and is disabled by default.
2175
Willy Tarreaufe20e5b2017-07-27 11:42:14 +02002176tune.h2.header-table-size <number>
2177 Sets the HTTP/2 dynamic header table size. It defaults to 4096 bytes and
2178 cannot be larger than 65536 bytes. A larger value may help certain clients
2179 send more compact requests, depending on their capabilities. This amount of
2180 memory is consumed for each HTTP/2 connection. It is recommended not to
2181 change it.
2182
Willy Tarreaue6baec02017-07-27 11:45:11 +02002183tune.h2.initial-window-size <number>
2184 Sets the HTTP/2 initial window size, which is the number of bytes the client
Davor Ocelice9ed2812017-12-25 17:49:28 +01002185 can upload before waiting for an acknowledgment from haproxy. This setting
2186 only affects payload contents (i.e. the body of POST requests), not headers.
Willy Tarreaue6baec02017-07-27 11:45:11 +02002187 The default value is 65535, which roughly allows up to 5 Mbps of upload
2188 bandwidth per client over a network showing a 100 ms ping time, or 500 Mbps
2189 over a 1-ms local network. It can make sense to increase this value to allow
2190 faster uploads, or to reduce it to increase fairness when dealing with many
2191 clients. It doesn't affect resource usage.
2192
Willy Tarreau5242ef82017-07-27 11:47:28 +02002193tune.h2.max-concurrent-streams <number>
2194 Sets the HTTP/2 maximum number of concurrent streams per connection (ie the
2195 number of outstanding requests on a single connection). The default value is
2196 100. A larger one may slightly improve page load time for complex sites when
2197 visited over high latency networks, but increases the amount of resources a
2198 single client may allocate. A value of zero disables the limit so a single
2199 client may create as many streams as allocatable by haproxy. It is highly
2200 recommended not to change this value.
2201
Willy Tarreaua24b35c2019-02-21 13:24:36 +01002202tune.h2.max-frame-size <number>
2203 Sets the HTTP/2 maximum frame size that haproxy announces it is willing to
2204 receive to its peers. The default value is the largest between 16384 and the
2205 buffer size (tune.bufsize). In any case, haproxy will not announce support
2206 for frame sizes larger than buffers. The main purpose of this setting is to
2207 allow to limit the maximum frame size setting when using large buffers. Too
2208 large frame sizes might have performance impact or cause some peers to
2209 misbehave. It is highly recommended not to change this value.
2210
Willy Tarreau193b8c62012-11-22 00:17:38 +01002211tune.http.cookielen <number>
2212 Sets the maximum length of captured cookies. This is the maximum value that
2213 the "capture cookie xxx len yyy" will be allowed to take, and any upper value
2214 will automatically be truncated to this one. It is important not to set too
2215 high a value because all cookie captures still allocate this size whatever
2216 their configured value (they share a same pool). This value is per request
2217 per response, so the memory allocated is twice this value per connection.
2218 When not specified, the limit is set to 63 characters. It is recommended not
2219 to change this value.
2220
Stéphane Cottin23e9e932017-05-18 08:58:41 +02002221tune.http.logurilen <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01002222 Sets the maximum length of request URI in logs. This prevents truncating long
2223 request URIs with valuable query strings in log lines. This is not related
Stéphane Cottin23e9e932017-05-18 08:58:41 +02002224 to syslog limits. If you increase this limit, you may also increase the
Davor Ocelice9ed2812017-12-25 17:49:28 +01002225 'log ... len yyy' parameter. Your syslog daemon may also need specific
Stéphane Cottin23e9e932017-05-18 08:58:41 +02002226 configuration directives too.
2227 The default value is 1024.
2228
Willy Tarreauac1932d2011-10-24 19:14:41 +02002229tune.http.maxhdr <number>
2230 Sets the maximum number of headers in a request. When a request comes with a
2231 number of headers greater than this value (including the first line), it is
2232 rejected with a "400 Bad Request" status code. Similarly, too large responses
2233 are blocked with "502 Bad Gateway". The default value is 101, which is enough
2234 for all usages, considering that the widely deployed Apache server uses the
2235 same limit. It can be useful to push this limit further to temporarily allow
Christopher Faulet50174f32017-06-21 16:31:35 +02002236 a buggy application to work by the time it gets fixed. The accepted range is
2237 1..32767. Keep in mind that each new header consumes 32bits of memory for
2238 each session, so don't push this limit too high.
Willy Tarreauac1932d2011-10-24 19:14:41 +02002239
Willy Tarreau76cc6992020-07-01 18:49:24 +02002240tune.idle-pool.shared { on | off }
2241 Enables ('on') or disables ('off') sharing of idle connection pools between
2242 threads for a same server. The default is to share them between threads in
2243 order to minimize the number of persistent connections to a server, and to
2244 optimize the connection reuse rate. But to help with debugging or when
2245 suspecting a bug in HAProxy around connection reuse, it can be convenient to
2246 forcefully disable this idle pool sharing between multiple threads, and force
Willy Tarreau37e07892021-02-19 11:45:22 +01002247 this option to "off". The default is on. It is strongly recommended against
2248 disabling this option without setting a conservative value on "pool-low-conn"
2249 for all servers relying on connection reuse to achieve a high performance
2250 level, otherwise connections might be closed very often as the thread count
2251 increases.
Willy Tarreau76cc6992020-07-01 18:49:24 +02002252
Willy Tarreau7e312732014-02-12 16:35:14 +01002253tune.idletimer <timeout>
2254 Sets the duration after which haproxy will consider that an empty buffer is
2255 probably associated with an idle stream. This is used to optimally adjust
2256 some packet sizes while forwarding large and small data alternatively. The
2257 decision to use splice() or to send large buffers in SSL is modulated by this
2258 parameter. The value is in milliseconds between 0 and 65535. A value of zero
2259 means that haproxy will not try to detect idle streams. The default is 1000,
Davor Ocelice9ed2812017-12-25 17:49:28 +01002260 which seems to correctly detect end user pauses (e.g. read a page before
John Roeslerfb2fce12019-07-10 15:45:51 -05002261 clicking). There should be no reason for changing this value. Please check
Willy Tarreau7e312732014-02-12 16:35:14 +01002262 tune.ssl.maxrecord below.
2263
Willy Tarreau7ac908b2019-02-27 12:02:18 +01002264tune.listener.multi-queue { on | off }
2265 Enables ('on') or disables ('off') the listener's multi-queue accept which
2266 spreads the incoming traffic to all threads a "bind" line is allowed to run
2267 on instead of taking them for itself. This provides a smoother traffic
2268 distribution and scales much better, especially in environments where threads
2269 may be unevenly loaded due to external activity (network interrupts colliding
2270 with one thread for example). This option is enabled by default, but it may
2271 be forcefully disabled for troubleshooting or for situations where it is
2272 estimated that the operating system already provides a good enough
2273 distribution and connections are extremely short-lived.
2274
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002275tune.lua.forced-yield <number>
2276 This directive forces the Lua engine to execute a yield each <number> of
Tim Düsterhus4896c442016-11-29 02:15:19 +01002277 instructions executed. This permits interrupting a long script and allows the
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002278 HAProxy scheduler to process other tasks like accepting connections or
2279 forwarding traffic. The default value is 10000 instructions. If HAProxy often
Davor Ocelice9ed2812017-12-25 17:49:28 +01002280 executes some Lua code but more responsiveness is required, this value can be
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002281 lowered. If the Lua code is quite long and its result is absolutely required
2282 to process the data, the <number> can be increased.
2283
Willy Tarreau32f61e22015-03-18 17:54:59 +01002284tune.lua.maxmem
2285 Sets the maximum amount of RAM in megabytes per process usable by Lua. By
2286 default it is zero which means unlimited. It is important to set a limit to
2287 ensure that a bug in a script will not result in the system running out of
2288 memory.
2289
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002290tune.lua.session-timeout <timeout>
2291 This is the execution timeout for the Lua sessions. This is useful for
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02002292 preventing infinite loops or spending too much time in Lua. This timeout
2293 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01002294 not taken in account. The default timeout is 4s.
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002295
2296tune.lua.task-timeout <timeout>
2297 Purpose is the same as "tune.lua.session-timeout", but this timeout is
2298 dedicated to the tasks. By default, this timeout isn't set because a task may
2299 remain alive during of the lifetime of HAProxy. For example, a task used to
2300 check servers.
2301
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02002302tune.lua.service-timeout <timeout>
2303 This is the execution timeout for the Lua services. This is useful for
2304 preventing infinite loops or spending too much time in Lua. This timeout
2305 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01002306 not taken in account. The default timeout is 4s.
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02002307
Willy Tarreaua0250ba2008-01-06 11:22:57 +01002308tune.maxaccept <number>
Willy Tarreau16a21472012-11-19 12:39:59 +01002309 Sets the maximum number of consecutive connections a process may accept in a
2310 row before switching to other work. In single process mode, higher numbers
Willy Tarreau8fa7f222021-02-19 15:50:27 +01002311 used to give better performance at high connection rates, though this is not
2312 the case anymore with the multi-queue. This value applies individually to
2313 each listener, so that the number of processes a listener is bound to is
2314 taken into account. This value defaults to 4 which showed best results. If a
2315 significantly higher value was inherited from an ancient config, it might be
2316 worth removing it as it will both increase performance and lower response
2317 time. In multi-process mode, it is divided by twice the number of processes
2318 the listener is bound to. Setting this value to -1 completely disables the
2319 limitation. It should normally not be needed to tweak this value.
Willy Tarreaua0250ba2008-01-06 11:22:57 +01002320
2321tune.maxpollevents <number>
2322 Sets the maximum amount of events that can be processed at once in a call to
2323 the polling system. The default value is adapted to the operating system. It
2324 has been noticed that reducing it below 200 tends to slightly decrease
2325 latency at the expense of network bandwidth, and increasing it above 200
2326 tends to trade latency for slightly increased bandwidth.
2327
Willy Tarreau27a674e2009-08-17 07:23:33 +02002328tune.maxrewrite <number>
2329 Sets the reserved buffer space to this size in bytes. The reserved space is
2330 used for header rewriting or appending. The first reads on sockets will never
2331 fill more than bufsize-maxrewrite. Historically it has defaulted to half of
2332 bufsize, though that does not make much sense since there are rarely large
2333 numbers of headers to add. Setting it too high prevents processing of large
2334 requests or responses. Setting it too low prevents addition of new headers
2335 to already large requests or to POST requests. It is generally wise to set it
2336 to about 1024. It is automatically readjusted to half of bufsize if it is
2337 larger than that. This means you don't have to worry about it when changing
2338 bufsize.
2339
Willy Tarreauf3045d22015-04-29 16:24:50 +02002340tune.pattern.cache-size <number>
2341 Sets the size of the pattern lookup cache to <number> entries. This is an LRU
2342 cache which reminds previous lookups and their results. It is used by ACLs
2343 and maps on slow pattern lookups, namely the ones using the "sub", "reg",
2344 "dir", "dom", "end", "bin" match methods as well as the case-insensitive
2345 strings. It applies to pattern expressions which means that it will be able
2346 to memorize the result of a lookup among all the patterns specified on a
2347 configuration line (including all those loaded from files). It automatically
2348 invalidates entries which are updated using HTTP actions or on the CLI. The
2349 default cache size is set to 10000 entries, which limits its footprint to
Willy Tarreau403bfbb2019-10-23 06:59:31 +02002350 about 5 MB per process/thread on 32-bit systems and 8 MB per process/thread
2351 on 64-bit systems, as caches are thread/process local. There is a very low
Willy Tarreauf3045d22015-04-29 16:24:50 +02002352 risk of collision in this cache, which is in the order of the size of the
2353 cache divided by 2^64. Typically, at 10000 requests per second with the
2354 default cache size of 10000 entries, there's 1% chance that a brute force
2355 attack could cause a single collision after 60 years, or 0.1% after 6 years.
2356 This is considered much lower than the risk of a memory corruption caused by
2357 aging components. If this is not acceptable, the cache can be disabled by
2358 setting this parameter to 0.
2359
Willy Tarreaubd9a0a72011-10-23 21:14:29 +02002360tune.pipesize <number>
2361 Sets the kernel pipe buffer size to this size (in bytes). By default, pipes
2362 are the default size for the system. But sometimes when using TCP splicing,
2363 it can improve performance to increase pipe sizes, especially if it is
2364 suspected that pipes are not filled and that many calls to splice() are
2365 performed. This has an impact on the kernel's memory footprint, so this must
2366 not be changed if impacts are not understood.
2367
Olivier Houchard88698d92019-04-16 19:07:22 +02002368tune.pool-high-fd-ratio <number>
2369 This setting sets the max number of file descriptors (in percentage) used by
2370 haproxy globally against the maximum number of file descriptors haproxy can
2371 use before we start killing idle connections when we can't reuse a connection
2372 and we have to create a new one. The default is 25 (one quarter of the file
2373 descriptor will mean that roughly half of the maximum front connections can
2374 keep an idle connection behind, anything beyond this probably doesn't make
John Roeslerfb2fce12019-07-10 15:45:51 -05002375 much sense in the general case when targeting connection reuse).
Olivier Houchard88698d92019-04-16 19:07:22 +02002376
Willy Tarreau83ca3052020-07-01 18:30:16 +02002377tune.pool-low-fd-ratio <number>
2378 This setting sets the max number of file descriptors (in percentage) used by
2379 haproxy globally against the maximum number of file descriptors haproxy can
2380 use before we stop putting connection into the idle pool for reuse. The
2381 default is 20.
2382
Willy Tarreaue803de22010-01-21 17:43:04 +01002383tune.rcvbuf.client <number>
2384tune.rcvbuf.server <number>
2385 Forces the kernel socket receive buffer size on the client or the server side
2386 to the specified value in bytes. This value applies to all TCP/HTTP frontends
2387 and backends. It should normally never be set, and the default size (0) lets
John Roeslerfb2fce12019-07-10 15:45:51 -05002388 the kernel auto-tune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01002389 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01002390 order to save kernel memory by preventing it from buffering too large amounts
2391 of received data. Lower values will significantly increase CPU usage though.
2392
Willy Tarreaub22fc302015-12-14 12:04:35 +01002393tune.recv_enough <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01002394 HAProxy uses some hints to detect that a short read indicates the end of the
Willy Tarreaub22fc302015-12-14 12:04:35 +01002395 socket buffers. One of them is that a read returns more than <recv_enough>
2396 bytes, which defaults to 10136 (7 segments of 1448 each). This default value
2397 may be changed by this setting to better deal with workloads involving lots
2398 of short messages such as telnet or SSH sessions.
2399
Olivier Houchard1599b802018-05-24 18:59:04 +02002400tune.runqueue-depth <number>
John Roeslerfb2fce12019-07-10 15:45:51 -05002401 Sets the maximum amount of task that can be processed at once when running
Willy Tarreau20076a02021-03-10 11:06:26 +01002402 tasks. The default value depends on the number of threads but sits between 35
2403 and 280, which tend to show the highest request rates and lowest latencies.
2404 Increasing it may incur latency when dealing with I/Os, making it too small
2405 can incur extra overhead. Higher thread counts benefit from lower values.
2406 When experimenting with much larger values, it may be useful to also enable
2407 tune.sched.low-latency and possibly tune.fd.edge-triggered to limit the
2408 maximum latency to the lowest possible.
Willy Tarreaue7723bd2020-06-24 11:11:02 +02002409
2410tune.sched.low-latency { on | off }
2411 Enables ('on') or disables ('off') the low-latency task scheduler. By default
2412 haproxy processes tasks from several classes one class at a time as this is
2413 the most efficient. But when running with large values of tune.runqueue-depth
2414 this can have a measurable effect on request or connection latency. When this
2415 low-latency setting is enabled, tasks of lower priority classes will always
2416 be executed before other ones if they exist. This will permit to lower the
2417 maximum latency experienced by new requests or connections in the middle of
2418 massive traffic, at the expense of a higher impact on this large traffic.
2419 For regular usage it is better to leave this off. The default value is off.
Olivier Houchard1599b802018-05-24 18:59:04 +02002420
Willy Tarreaue803de22010-01-21 17:43:04 +01002421tune.sndbuf.client <number>
2422tune.sndbuf.server <number>
2423 Forces the kernel socket send buffer size on the client or the server side to
2424 the specified value in bytes. This value applies to all TCP/HTTP frontends
2425 and backends. It should normally never be set, and the default size (0) lets
John Roeslerfb2fce12019-07-10 15:45:51 -05002426 the kernel auto-tune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01002427 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01002428 order to save kernel memory by preventing it from buffering too large amounts
2429 of received data. Lower values will significantly increase CPU usage though.
2430 Another use case is to prevent write timeouts with extremely slow clients due
2431 to the kernel waiting for a large part of the buffer to be read before
2432 notifying haproxy again.
2433
Willy Tarreau6ec58db2012-11-16 16:32:15 +01002434tune.ssl.cachesize <number>
Emeric Brunaf9619d2012-11-28 18:47:52 +01002435 Sets the size of the global SSL session cache, in a number of blocks. A block
William Dauchydede64a2021-02-12 15:58:46 +01002436 is large enough to contain an encoded session without peer certificate. An
2437 encoded session with peer certificate is stored in multiple blocks depending
2438 on the size of the peer certificate. A block uses approximately 200 bytes of
2439 memory (based on `sizeof(struct sh_ssl_sess_hdr) + SHSESS_BLOCK_MIN_SIZE`
2440 calculation used for `shctx_init` function). The default value may be forced
2441 at build time, otherwise defaults to 20000. When the cache is full, the most
2442 idle entries are purged and reassigned. Higher values reduce the occurrence
2443 of such a purge, hence the number of CPU-intensive SSL handshakes by ensuring
2444 that all users keep their session as long as possible. All entries are
2445 pre-allocated upon startup and are shared between all processes if "nbproc"
2446 is greater than 1. Setting this value to 0 disables the SSL session cache.
Willy Tarreau6ec58db2012-11-16 16:32:15 +01002447
Emeric Brun8dc60392014-05-09 13:52:00 +02002448tune.ssl.force-private-cache
Lukas Tribus27935782018-10-01 02:00:16 +02002449 This option disables SSL session cache sharing between all processes. It
Emeric Brun8dc60392014-05-09 13:52:00 +02002450 should normally not be used since it will force many renegotiations due to
2451 clients hitting a random process. But it may be required on some operating
2452 systems where none of the SSL cache synchronization method may be used. In
2453 this case, adding a first layer of hash-based load balancing before the SSL
2454 layer might limit the impact of the lack of session sharing.
2455
William Lallemand7d42ef52020-07-06 11:41:30 +02002456tune.ssl.keylog { on | off }
2457 This option activates the logging of the TLS keys. It should be used with
2458 care as it will consume more memory per SSL session and could decrease
2459 performances. This is disabled by default.
2460
2461 These sample fetches should be used to generate the SSLKEYLOGFILE that is
2462 required to decipher traffic with wireshark.
2463
2464 https://developer.mozilla.org/en-US/docs/Mozilla/Projects/NSS/Key_Log_Format
2465
2466 The SSLKEYLOG is a series of lines which are formatted this way:
2467
2468 <Label> <space> <ClientRandom> <space> <Secret>
2469
2470 The ClientRandom is provided by the %[ssl_fc_client_random,hex] sample
2471 fetch, the secret and the Label could be find in the array below. You need
2472 to generate a SSLKEYLOGFILE with all the labels in this array.
2473
2474 The following sample fetches are hexadecimal strings and does not need to be
2475 converted.
2476
2477 SSLKEYLOGFILE Label | Sample fetches for the Secrets
2478 --------------------------------|-----------------------------------------
2479 CLIENT_EARLY_TRAFFIC_SECRET | %[ssl_fc_client_early_traffic_secret]
2480 CLIENT_HANDSHAKE_TRAFFIC_SECRET | %[ssl_fc_client_handshake_traffic_secret]
2481 SERVER_HANDSHAKE_TRAFFIC_SECRET | %[ssl_fc_server_handshake_traffic_secret]
2482 CLIENT_TRAFFIC_SECRET_0 | %[ssl_fc_client_traffic_secret_0]
2483 SERVER_TRAFFIC_SECRET_0 | %[ssl_fc_server_traffic_secret_0]
William Lallemandd742b6c2020-07-07 10:14:56 +02002484 EXPORTER_SECRET | %[ssl_fc_exporter_secret]
2485 EARLY_EXPORTER_SECRET | %[ssl_fc_early_exporter_secret]
William Lallemand7d42ef52020-07-06 11:41:30 +02002486
2487 This is only available with OpenSSL 1.1.1, and useful with TLS1.3 session.
2488
2489 If you want to generate the content of a SSLKEYLOGFILE with TLS < 1.3, you
2490 only need this line:
2491
2492 "CLIENT_RANDOM %[ssl_fc_client_random,hex] %[ssl_fc_session_key,hex]"
2493
Emeric Brun4f65bff2012-11-16 15:11:00 +01002494tune.ssl.lifetime <timeout>
2495 Sets how long a cached SSL session may remain valid. This time is expressed
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002496 in seconds and defaults to 300 (5 min). It is important to understand that it
Emeric Brun4f65bff2012-11-16 15:11:00 +01002497 does not guarantee that sessions will last that long, because if the cache is
2498 full, the longest idle sessions will be purged despite their configured
2499 lifetime. The real usefulness of this setting is to prevent sessions from
2500 being used for too long.
2501
Willy Tarreaubfd59462013-02-21 07:46:09 +01002502tune.ssl.maxrecord <number>
2503 Sets the maximum amount of bytes passed to SSL_write() at a time. Default
2504 value 0 means there is no limit. Over SSL/TLS, the client can decipher the
2505 data only once it has received a full record. With large records, it means
2506 that clients might have to download up to 16kB of data before starting to
2507 process them. Limiting the value can improve page load times on browsers
2508 located over high latency or low bandwidth networks. It is suggested to find
2509 optimal values which fit into 1 or 2 TCP segments (generally 1448 bytes over
2510 Ethernet with TCP timestamps enabled, or 1460 when timestamps are disabled),
2511 keeping in mind that SSL/TLS add some overhead. Typical values of 1419 and
2512 2859 gave good results during tests. Use "strace -e trace=write" to find the
Davor Ocelice9ed2812017-12-25 17:49:28 +01002513 best value. HAProxy will automatically switch to this setting after an idle
Willy Tarreau7e312732014-02-12 16:35:14 +01002514 stream has been detected (see tune.idletimer above).
Willy Tarreaubfd59462013-02-21 07:46:09 +01002515
Remi Gacognef46cd6e2014-06-12 14:58:40 +02002516tune.ssl.default-dh-param <number>
2517 Sets the maximum size of the Diffie-Hellman parameters used for generating
2518 the ephemeral/temporary Diffie-Hellman key in case of DHE key exchange. The
2519 final size will try to match the size of the server's RSA (or DSA) key (e.g,
2520 a 2048 bits temporary DH key for a 2048 bits RSA key), but will not exceed
Willy Tarreau3ba77d22020-05-08 09:31:18 +02002521 this maximum value. Default value if 2048. Only 1024 or higher values are
Remi Gacognef46cd6e2014-06-12 14:58:40 +02002522 allowed. Higher values will increase the CPU load, and values greater than
2523 1024 bits are not supported by Java 7 and earlier clients. This value is not
Remi Gacogne47783ef2015-05-29 15:53:22 +02002524 used if static Diffie-Hellman parameters are supplied either directly
2525 in the certificate file or by using the ssl-dh-param-file parameter.
Remi Gacognef46cd6e2014-06-12 14:58:40 +02002526
Christopher Faulet31af49d2015-06-09 17:29:50 +02002527tune.ssl.ssl-ctx-cache-size <number>
2528 Sets the size of the cache used to store generated certificates to <number>
2529 entries. This is a LRU cache. Because generating a SSL certificate
2530 dynamically is expensive, they are cached. The default cache size is set to
2531 1000 entries.
2532
Thierry FOURNIER5bf77322017-02-25 12:45:22 +01002533tune.ssl.capture-cipherlist-size <number>
2534 Sets the maximum size of the buffer used for capturing client-hello cipher
2535 list. If the value is 0 (default value) the capture is disabled, otherwise
2536 a buffer is allocated for each SSL/TLS connection.
2537
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002538tune.vars.global-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01002539tune.vars.proc-max-size <size>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002540tune.vars.reqres-max-size <size>
2541tune.vars.sess-max-size <size>
2542tune.vars.txn-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01002543 These five tunes help to manage the maximum amount of memory used by the
2544 variables system. "global" limits the overall amount of memory available for
2545 all scopes. "proc" limits the memory for the process scope, "sess" limits the
2546 memory for the session scope, "txn" for the transaction scope, and "reqres"
2547 limits the memory for each request or response processing.
2548 Memory accounting is hierarchical, meaning more coarse grained limits include
2549 the finer grained ones: "proc" includes "sess", "sess" includes "txn", and
2550 "txn" includes "reqres".
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002551
Daniel Schneller0b547052016-03-21 20:46:57 +01002552 For example, when "tune.vars.sess-max-size" is limited to 100,
2553 "tune.vars.txn-max-size" and "tune.vars.reqres-max-size" cannot exceed
2554 100 either. If we create a variable "txn.var" that contains 100 bytes,
2555 all available space is consumed.
2556 Notice that exceeding the limits at runtime will not result in an error
2557 message, but values might be cut off or corrupted. So make sure to accurately
2558 plan for the amount of space needed to store all your variables.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002559
William Lallemanda509e4c2012-11-07 16:54:34 +01002560tune.zlib.memlevel <number>
2561 Sets the memLevel parameter in zlib initialization for each session. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002562 defines how much memory should be allocated for the internal compression
William Lallemanda509e4c2012-11-07 16:54:34 +01002563 state. A value of 1 uses minimum memory but is slow and reduces compression
Davor Ocelice9ed2812017-12-25 17:49:28 +01002564 ratio, a value of 9 uses maximum memory for optimal speed. Can be a value
William Lallemanda509e4c2012-11-07 16:54:34 +01002565 between 1 and 9. The default value is 8.
2566
2567tune.zlib.windowsize <number>
2568 Sets the window size (the size of the history buffer) as a parameter of the
2569 zlib initialization for each session. Larger values of this parameter result
Davor Ocelice9ed2812017-12-25 17:49:28 +01002570 in better compression at the expense of memory usage. Can be a value between
2571 8 and 15. The default value is 15.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002572
Willy Tarreauc57f0e22009-05-10 13:12:33 +020025733.3. Debugging
2574--------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02002575
Willy Tarreau6a06a402007-07-15 20:15:28 +02002576quiet
2577 Do not display any message during startup. It is equivalent to the command-
2578 line argument "-q".
2579
Willy Tarreau3eb10b82020-04-15 16:42:39 +02002580zero-warning
2581 When this option is set, haproxy will refuse to start if any warning was
2582 emitted while processing the configuration. It is highly recommended to set
2583 this option on configurations that are not changed often, as it helps detect
2584 subtle mistakes and keep the configuration clean and forward-compatible. Note
2585 that "haproxy -c" will also report errors in such a case. This option is
2586 equivalent to command line argument "-dW".
2587
Emeric Brunf099e792010-09-27 12:05:28 +02002588
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010025893.4. Userlists
2590--------------
2591It is possible to control access to frontend/backend/listen sections or to
2592http stats by allowing only authenticated and authorized users. To do this,
2593it is required to create at least one userlist and to define users.
2594
2595userlist <listname>
Cyril Bonté78caf842010-03-10 22:41:43 +01002596 Creates new userlist with name <listname>. Many independent userlists can be
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002597 used to store authentication & authorization data for independent customers.
2598
2599group <groupname> [users <user>,<user>,(...)]
Cyril Bonté78caf842010-03-10 22:41:43 +01002600 Adds group <groupname> to the current userlist. It is also possible to
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002601 attach users to this group by using a comma separated list of names
2602 proceeded by "users" keyword.
2603
Cyril Bontéf0c60612010-02-06 14:44:47 +01002604user <username> [password|insecure-password <password>]
2605 [groups <group>,<group>,(...)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002606 Adds user <username> to the current userlist. Both secure (encrypted) and
2607 insecure (unencrypted) passwords can be used. Encrypted passwords are
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01002608 evaluated using the crypt(3) function, so depending on the system's
2609 capabilities, different algorithms are supported. For example, modern Glibc
2610 based Linux systems support MD5, SHA-256, SHA-512, and, of course, the
2611 classic DES-based method of encrypting passwords.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002612
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01002613 Attention: Be aware that using encrypted passwords might cause significantly
2614 increased CPU usage, depending on the number of requests, and the algorithm
2615 used. For any of the hashed variants, the password for each request must
2616 be processed through the chosen algorithm, before it can be compared to the
2617 value specified in the config file. Most current algorithms are deliberately
2618 designed to be expensive to compute to achieve resistance against brute
2619 force attacks. They do not simply salt/hash the clear text password once,
2620 but thousands of times. This can quickly become a major factor in haproxy's
2621 overall CPU consumption!
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002622
2623 Example:
Cyril Bontéf0c60612010-02-06 14:44:47 +01002624 userlist L1
2625 group G1 users tiger,scott
2626 group G2 users xdb,scott
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002627
Cyril Bontéf0c60612010-02-06 14:44:47 +01002628 user tiger password $6$k6y3o.eP$JlKBx9za9667qe4(...)xHSwRv6J.C0/D7cV91
2629 user scott insecure-password elgato
2630 user xdb insecure-password hello
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002631
Cyril Bontéf0c60612010-02-06 14:44:47 +01002632 userlist L2
2633 group G1
2634 group G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002635
Cyril Bontéf0c60612010-02-06 14:44:47 +01002636 user tiger password $6$k6y3o.eP$JlKBx(...)xHSwRv6J.C0/D7cV91 groups G1
2637 user scott insecure-password elgato groups G1,G2
2638 user xdb insecure-password hello groups G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002639
2640 Please note that both lists are functionally identical.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002641
Emeric Brunf099e792010-09-27 12:05:28 +02002642
26433.5. Peers
Cyril Bontédc4d9032012-04-08 21:57:39 +02002644----------
Emeric Brun94900952015-06-11 18:25:54 +02002645It is possible to propagate entries of any data-types in stick-tables between
2646several haproxy instances over TCP connections in a multi-master fashion. Each
2647instance pushes its local updates and insertions to remote peers. The pushed
2648values overwrite remote ones without aggregation. Interrupted exchanges are
2649automatically detected and recovered from the last known point.
2650In addition, during a soft restart, the old process connects to the new one
2651using such a TCP connection to push all its entries before the new process
2652tries to connect to other peers. That ensures very fast replication during a
2653reload, it typically takes a fraction of a second even for large tables.
2654Note that Server IDs are used to identify servers remotely, so it is important
2655that configurations look similar or at least that the same IDs are forced on
2656each server on all participants.
Emeric Brunf099e792010-09-27 12:05:28 +02002657
2658peers <peersect>
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002659 Creates a new peer list with name <peersect>. It is an independent section,
Emeric Brunf099e792010-09-27 12:05:28 +02002660 which is referenced by one or more stick-tables.
2661
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002662bind [<address>]:<port_range> [, ...] [param*]
2663 Defines the binding parameters of the local peer of this "peers" section.
2664 Such lines are not supported with "peer" line in the same "peers" section.
2665
Willy Tarreau77e4bd12015-05-01 20:02:17 +02002666disabled
2667 Disables a peers section. It disables both listening and any synchronization
2668 related to this section. This is provided to disable synchronization of stick
2669 tables without having to comment out all "peers" references.
2670
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002671default-bind [param*]
2672 Defines the binding parameters for the local peer, excepted its address.
2673
2674default-server [param*]
2675 Change default options for a server in a "peers" section.
2676
2677 Arguments:
2678 <param*> is a list of parameters for this server. The "default-server"
2679 keyword accepts an important number of options and has a complete
2680 section dedicated to it. Please refer to section 5 for more
2681 details.
2682
2683
2684 See also: "server" and section 5 about server options
2685
Willy Tarreau77e4bd12015-05-01 20:02:17 +02002686enable
2687 This re-enables a disabled peers section which was previously disabled.
2688
Jan Wagnerf2f5c4e2020-12-17 22:22:32 +01002689log <address> [len <length>] [format <format>] [sample <ranges>:<sample_size>]
Frédéric Lécailleb6f759b2019-11-05 09:57:45 +01002690 <facility> [<level> [<minlevel>]]
2691 "peers" sections support the same "log" keyword as for the proxies to
2692 log information about the "peers" listener. See "log" option for proxies for
2693 more details.
2694
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002695peer <peername> <ip>:<port> [param*]
Emeric Brunf099e792010-09-27 12:05:28 +02002696 Defines a peer inside a peers section.
2697 If <peername> is set to the local peer name (by default hostname, or forced
Dragan Dosen13cd54c2020-06-18 18:24:05 +02002698 using "-L" command line option or "localpeer" global configuration setting),
2699 haproxy will listen for incoming remote peer connection on <ip>:<port>.
2700 Otherwise, <ip>:<port> defines where to connect to in order to join the
2701 remote peer, and <peername> is used at the protocol level to identify and
2702 validate the remote peer on the server side.
Emeric Brunf099e792010-09-27 12:05:28 +02002703
2704 During a soft restart, local peer <ip>:<port> is used by the old instance to
2705 connect the new one and initiate a complete replication (teaching process).
2706
2707 It is strongly recommended to have the exact same peers declaration on all
Dragan Dosen13cd54c2020-06-18 18:24:05 +02002708 peers and to only rely on the "-L" command line argument or the "localpeer"
2709 global configuration setting to change the local peer name. This makes it
2710 easier to maintain coherent configuration files across all peers.
Emeric Brunf099e792010-09-27 12:05:28 +02002711
William Lallemandb2f07452015-05-12 14:27:13 +02002712 You may want to reference some environment variables in the address
2713 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01002714
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002715 Note: "peer" keyword may transparently be replaced by "server" keyword (see
2716 "server" keyword explanation below).
2717
2718server <peername> [<ip>:<port>] [param*]
Michael Prokop4438c602019-05-24 10:25:45 +02002719 As previously mentioned, "peer" keyword may be replaced by "server" keyword
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002720 with a support for all "server" parameters found in 5.2 paragraph.
2721 If the underlying peer is local, <ip>:<port> parameters must not be present.
2722 These parameters must be provided on a "bind" line (see "bind" keyword
2723 of this "peers" section).
2724 Some of these parameters are irrelevant for "peers" sections.
2725
2726
Cyril Bontédc4d9032012-04-08 21:57:39 +02002727 Example:
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002728 # The old way.
Emeric Brunf099e792010-09-27 12:05:28 +02002729 peers mypeers
Willy Tarreauf7b30a92010-12-06 22:59:17 +01002730 peer haproxy1 192.168.0.1:1024
2731 peer haproxy2 192.168.0.2:1024
2732 peer haproxy3 10.2.0.1:1024
Emeric Brunf099e792010-09-27 12:05:28 +02002733
2734 backend mybackend
2735 mode tcp
2736 balance roundrobin
2737 stick-table type ip size 20k peers mypeers
2738 stick on src
2739
Willy Tarreauf7b30a92010-12-06 22:59:17 +01002740 server srv1 192.168.0.30:80
2741 server srv2 192.168.0.31:80
Emeric Brunf099e792010-09-27 12:05:28 +02002742
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002743 Example:
2744 peers mypeers
2745 bind 127.0.0.11:10001 ssl crt mycerts/pem
2746 default-server ssl verify none
2747 server hostA 127.0.0.10:10000
2748 server hostB #local peer
Emeric Brunf099e792010-09-27 12:05:28 +02002749
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01002750
2751table <tablename> type {ip | integer | string [len <length>] | binary [len <length>]}
2752 size <size> [expire <expire>] [nopurge] [store <data_type>]*
2753
2754 Configure a stickiness table for the current section. This line is parsed
2755 exactly the same way as the "stick-table" keyword in others section, except
John Roeslerfb2fce12019-07-10 15:45:51 -05002756 for the "peers" argument which is not required here and with an additional
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01002757 mandatory first parameter to designate the stick-table. Contrary to others
2758 sections, there may be several "table" lines in "peers" sections (see also
2759 "stick-table" keyword).
2760
2761 Also be aware of the fact that "peers" sections have their own stick-table
2762 namespaces to avoid collisions between stick-table names identical in
2763 different "peers" section. This is internally handled prepending the "peers"
2764 sections names to the name of the stick-tables followed by a '/' character.
2765 If somewhere else in the configuration file you have to refer to such
2766 stick-tables declared in "peers" sections you must use the prefixed version
2767 of the stick-table name as follows:
2768
2769 peers mypeers
2770 peer A ...
2771 peer B ...
2772 table t1 ...
2773
2774 frontend fe1
2775 tcp-request content track-sc0 src table mypeers/t1
2776
2777 This is also this prefixed version of the stick-table names which must be
2778 used to refer to stick-tables through the CLI.
2779
2780 About "peers" protocol, as only "peers" belonging to the same section may
2781 communicate with each others, there is no need to do such a distinction.
2782 Several "peers" sections may declare stick-tables with the same name.
2783 This is shorter version of the stick-table name which is sent over the network.
2784 There is only a '/' character as prefix to avoid stick-table name collisions between
2785 stick-tables declared as backends and stick-table declared in "peers" sections
2786 as follows in this weird but supported configuration:
2787
2788 peers mypeers
2789 peer A ...
2790 peer B ...
2791 table t1 type string size 10m store gpc0
2792
2793 backend t1
2794 stick-table type string size 10m store gpc0 peers mypeers
2795
Daniel Corbett67a82712020-07-06 23:01:19 -04002796 Here "t1" table declared in "mypeers" section has "mypeers/t1" as global name.
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01002797 "t1" table declared as a backend as "t1" as global name. But at peer protocol
2798 level the former table is named "/t1", the latter is again named "t1".
2799
Simon Horman51a1cf62015-02-03 13:00:44 +090028003.6. Mailers
2801------------
2802It is possible to send email alerts when the state of servers changes.
2803If configured email alerts are sent to each mailer that is configured
2804in a mailers section. Email is sent to mailers using SMTP.
2805
Pieter Baauw386a1272015-08-16 15:26:24 +02002806mailers <mailersect>
Simon Horman51a1cf62015-02-03 13:00:44 +09002807 Creates a new mailer list with the name <mailersect>. It is an
2808 independent section which is referenced by one or more proxies.
2809
2810mailer <mailername> <ip>:<port>
2811 Defines a mailer inside a mailers section.
2812
2813 Example:
2814 mailers mymailers
2815 mailer smtp1 192.168.0.1:587
2816 mailer smtp2 192.168.0.2:587
2817
2818 backend mybackend
2819 mode tcp
2820 balance roundrobin
2821
2822 email-alert mailers mymailers
2823 email-alert from test1@horms.org
2824 email-alert to test2@horms.org
2825
2826 server srv1 192.168.0.30:80
2827 server srv2 192.168.0.31:80
2828
Pieter Baauw235fcfc2016-02-13 15:33:40 +01002829timeout mail <time>
2830 Defines the time available for a mail/connection to be made and send to
2831 the mail-server. If not defined the default value is 10 seconds. To allow
2832 for at least two SYN-ACK packets to be send during initial TCP handshake it
2833 is advised to keep this value above 4 seconds.
2834
2835 Example:
2836 mailers mymailers
2837 timeout mail 20s
2838 mailer smtp1 192.168.0.1:587
Simon Horman51a1cf62015-02-03 13:00:44 +09002839
William Lallemandc9515522019-06-12 16:32:11 +020028403.7. Programs
2841-------------
2842In master-worker mode, it is possible to launch external binaries with the
2843master, these processes are called programs. These programs are launched and
2844managed the same way as the workers.
2845
2846During a reload of HAProxy, those processes are dealing with the same
2847sequence as a worker:
2848
2849 - the master is re-executed
2850 - the master sends a SIGUSR1 signal to the program
2851 - if "option start-on-reload" is not disabled, the master launches a new
2852 instance of the program
2853
2854During a stop, or restart, a SIGTERM is sent to the programs.
2855
2856program <name>
2857 This is a new program section, this section will create an instance <name>
2858 which is visible in "show proc" on the master CLI. (See "9.4. Master CLI" in
2859 the management guide).
2860
2861command <command> [arguments*]
2862 Define the command to start with optional arguments. The command is looked
2863 up in the current PATH if it does not include an absolute path. This is a
2864 mandatory option of the program section. Arguments containing spaces must
2865 be enclosed in quotes or double quotes or be prefixed by a backslash.
2866
Andrew Heberle97236962019-07-12 11:50:26 +08002867user <user name>
2868 Changes the executed command user ID to the <user name> from /etc/passwd.
2869 See also "group".
2870
2871group <group name>
2872 Changes the executed command group ID to the <group name> from /etc/group.
2873 See also "user".
2874
William Lallemandc9515522019-06-12 16:32:11 +02002875option start-on-reload
2876no option start-on-reload
2877 Start (or not) a new instance of the program upon a reload of the master.
2878 The default is to start a new instance. This option may only be used in a
2879 program section.
2880
2881
Christopher Faulet76edc0f2020-01-13 15:52:01 +010028823.8. HTTP-errors
2883----------------
2884
2885It is possible to globally declare several groups of HTTP errors, to be
2886imported afterwards in any proxy section. Same group may be referenced at
2887several places and can be fully or partially imported.
2888
2889http-errors <name>
2890 Create a new http-errors group with the name <name>. It is an independent
2891 section that may be referenced by one or more proxies using its name.
2892
2893errorfile <code> <file>
2894 Associate a file contents to an HTTP error code
2895
2896 Arguments :
2897 <code> is the HTTP status code. Currently, HAProxy is capable of
Christopher Faulet612f2ea2020-05-27 09:57:28 +02002898 generating codes 200, 400, 401, 403, 404, 405, 407, 408, 410,
2899 425, 429, 500, 502, 503, and 504.
Christopher Faulet76edc0f2020-01-13 15:52:01 +01002900
2901 <file> designates a file containing the full HTTP response. It is
2902 recommended to follow the common practice of appending ".http" to
2903 the filename so that people do not confuse the response with HTML
2904 error pages, and to use absolute paths, since files are read
2905 before any chroot is performed.
2906
2907 Please referrers to "errorfile" keyword in section 4 for details.
2908
2909 Example:
2910 http-errors website-1
2911 errorfile 400 /etc/haproxy/errorfiles/site1/400.http
2912 errorfile 404 /etc/haproxy/errorfiles/site1/404.http
2913 errorfile 408 /dev/null # work around Chrome pre-connect bug
2914
2915 http-errors website-2
2916 errorfile 400 /etc/haproxy/errorfiles/site2/400.http
2917 errorfile 404 /etc/haproxy/errorfiles/site2/404.http
2918 errorfile 408 /dev/null # work around Chrome pre-connect bug
2919
Emeric Brun99c453d2020-05-25 15:01:04 +020029203.9. Rings
2921----------
2922
2923It is possible to globally declare ring-buffers, to be used as target for log
2924servers or traces.
2925
2926ring <ringname>
2927 Creates a new ring-buffer with name <ringname>.
2928
2929description <text>
Daniel Corbett67a82712020-07-06 23:01:19 -04002930 The description is an optional description string of the ring. It will
Emeric Brun99c453d2020-05-25 15:01:04 +02002931 appear on CLI. By default, <name> is reused to fill this field.
2932
2933format <format>
2934 Format used to store events into the ring buffer.
2935
2936 Arguments:
2937 <format> is the log format used when generating syslog messages. It may be
2938 one of the following :
2939
2940 iso A message containing only the ISO date, followed by the text.
2941 The PID, process name and system name are omitted. This is
2942 designed to be used with a local log server.
2943
Emeric Bruna0338b92020-11-27 16:24:34 +01002944 local Analog to rfc3164 syslog message format except that hostname
2945 field is stripped. This is the default.
2946 Note: option "log-send-hostname" switches the default to
2947 rfc3164.
2948
Emeric Brun99c453d2020-05-25 15:01:04 +02002949 raw A message containing only the text. The level, PID, date, time,
2950 process name and system name are omitted. This is designed to be
2951 used in containers or during development, where the severity
2952 only depends on the file descriptor used (stdout/stderr). This
2953 is the default.
2954
Emeric Bruna0338b92020-11-27 16:24:34 +01002955 rfc3164 The RFC3164 syslog message format.
Emeric Brun99c453d2020-05-25 15:01:04 +02002956 (https://tools.ietf.org/html/rfc3164)
2957
2958 rfc5424 The RFC5424 syslog message format.
2959 (https://tools.ietf.org/html/rfc5424)
2960
2961 short A message containing only a level between angle brackets such as
2962 '<3>', followed by the text. The PID, date, time, process name
2963 and system name are omitted. This is designed to be used with a
2964 local log server. This format is compatible with what the systemd
2965 logger consumes.
2966
Emeric Brun54648852020-07-06 15:54:06 +02002967 priority A message containing only a level plus syslog facility between angle
2968 brackets such as '<63>', followed by the text. The PID, date, time,
2969 process name and system name are omitted. This is designed to be used
2970 with a local log server.
2971
Emeric Brun99c453d2020-05-25 15:01:04 +02002972 timed A message containing only a level between angle brackets such as
2973 '<3>', followed by ISO date and by the text. The PID, process
2974 name and system name are omitted. This is designed to be
2975 used with a local log server.
2976
2977maxlen <length>
2978 The maximum length of an event message stored into the ring,
2979 including formatted header. If an event message is longer than
2980 <length>, it will be truncated to this length.
2981
Emeric Brun494c5052020-05-28 11:13:15 +02002982server <name> <address> [param*]
2983 Used to configure a syslog tcp server to forward messages from ring buffer.
2984 This supports for all "server" parameters found in 5.2 paragraph. Some of
2985 these parameters are irrelevant for "ring" sections. Important point: there
2986 is little reason to add more than one server to a ring, because all servers
2987 will receive the exact same copy of the ring contents, and as such the ring
2988 will progress at the speed of the slowest server. If one server does not
2989 respond, it will prevent old messages from being purged and may block new
2990 messages from being inserted into the ring. The proper way to send messages
2991 to multiple servers is to use one distinct ring per log server, not to
Emeric Brun97556472020-05-30 01:42:45 +02002992 attach multiple servers to the same ring. Note that specific server directive
2993 "log-proto" is used to set the protocol used to send messages.
Emeric Brun494c5052020-05-28 11:13:15 +02002994
Emeric Brun99c453d2020-05-25 15:01:04 +02002995size <size>
2996 This is the optional size in bytes for the ring-buffer. Default value is
2997 set to BUFSIZE.
2998
Emeric Brun494c5052020-05-28 11:13:15 +02002999timeout connect <timeout>
3000 Set the maximum time to wait for a connection attempt to a server to succeed.
3001
3002 Arguments :
3003 <timeout> is the timeout value specified in milliseconds by default, but
3004 can be in any other unit if the number is suffixed by the unit,
3005 as explained at the top of this document.
3006
3007timeout server <timeout>
3008 Set the maximum time for pending data staying into output buffer.
3009
3010 Arguments :
3011 <timeout> is the timeout value specified in milliseconds by default, but
3012 can be in any other unit if the number is suffixed by the unit,
3013 as explained at the top of this document.
3014
Emeric Brun99c453d2020-05-25 15:01:04 +02003015 Example:
3016 global
3017 log ring@myring local7
3018
3019 ring myring
3020 description "My local buffer"
3021 format rfc3164
3022 maxlen 1200
3023 size 32764
Emeric Brun494c5052020-05-28 11:13:15 +02003024 timeout connect 5s
3025 timeout server 10s
Emeric Brun97556472020-05-30 01:42:45 +02003026 server mysyslogsrv 127.0.0.1:6514 log-proto octet-count
Emeric Brun99c453d2020-05-25 15:01:04 +02003027
Emeric Brun12941c82020-07-07 14:19:42 +020030283.10. Log forwarding
3029-------------------
3030
3031It is possible to declare one or multiple log forwarding section,
3032haproxy will forward all received log messages to a log servers list.
3033
3034log-forward <name>
3035 Creates a new log forwarder proxy identified as <name>.
3036
Emeric Bruncbb7bf72020-10-05 14:39:35 +02003037backlog <conns>
3038 Give hints to the system about the approximate listen backlog desired size
3039 on connections accept.
3040
3041bind <addr> [param*]
3042 Used to configure a stream log listener to receive messages to forward.
Emeric Brunda46c1c2020-10-08 08:39:02 +02003043 This supports the "bind" parameters found in 5.1 paragraph including
3044 those about ssl but some statements such as "alpn" may be irrelevant for
3045 syslog protocol over TCP.
3046 Those listeners support both "Octet Counting" and "Non-Transparent-Framing"
Emeric Bruncbb7bf72020-10-05 14:39:35 +02003047 modes as defined in rfc-6587.
3048
Willy Tarreau76aaa7f2020-09-16 15:07:22 +02003049dgram-bind <addr> [param*]
Emeric Bruncbb7bf72020-10-05 14:39:35 +02003050 Used to configure a datagram log listener to receive messages to forward.
3051 Addresses must be in IPv4 or IPv6 form,followed by a port. This supports
3052 for some of the "bind" parameters found in 5.1 paragraph among which
3053 "interface", "namespace" or "transparent", the other ones being
Willy Tarreau26ff5da2020-09-16 15:22:19 +02003054 silently ignored as irrelevant for UDP/syslog case.
Emeric Brun12941c82020-07-07 14:19:42 +02003055
3056log global
Jan Wagnerf2f5c4e2020-12-17 22:22:32 +01003057log <address> [len <length>] [format <format>] [sample <ranges>:<sample_size>]
Emeric Brun12941c82020-07-07 14:19:42 +02003058 <facility> [<level> [<minlevel>]]
3059 Used to configure target log servers. See more details on proxies
3060 documentation.
3061 If no format specified, haproxy tries to keep the incoming log format.
3062 Configured facility is ignored, except if incoming message does not
3063 present a facility but one is mandatory on the outgoing format.
3064 If there is no timestamp available in the input format, but the field
3065 exists in output format, haproxy will use the local date.
3066
3067 Example:
3068 global
3069 log stderr format iso local7
3070
3071 ring myring
3072 description "My local buffer"
3073 format rfc5424
3074 maxlen 1200
3075 size 32764
3076 timeout connect 5s
3077 timeout server 10s
3078 # syslog tcp server
3079 server mysyslogsrv 127.0.0.1:514 log-proto octet-count
3080
3081 log-forward sylog-loadb
Emeric Bruncbb7bf72020-10-05 14:39:35 +02003082 dgram-bind 127.0.0.1:1514
3083 bind 127.0.0.1:1514
Emeric Brun12941c82020-07-07 14:19:42 +02003084 # all messages on stderr
3085 log global
3086 # all messages on local tcp syslog server
3087 log ring@myring local0
3088 # load balance messages on 4 udp syslog servers
3089 log 127.0.0.1:10001 sample 1:4 local0
3090 log 127.0.0.1:10002 sample 2:4 local0
3091 log 127.0.0.1:10003 sample 3:4 local0
3092 log 127.0.0.1:10004 sample 4:4 local0
Christopher Faulet76edc0f2020-01-13 15:52:01 +01003093
Emeric Bruncbb7bf72020-10-05 14:39:35 +02003094maxconn <conns>
3095 Fix the maximum number of concurrent connections on a log forwarder.
3096 10 is the default.
3097
3098timeout client <timeout>
3099 Set the maximum inactivity time on the client side.
3100
Willy Tarreauc57f0e22009-05-10 13:12:33 +020031014. Proxies
Willy Tarreau6a06a402007-07-15 20:15:28 +02003102----------
Willy Tarreau0ba27502007-12-24 16:55:16 +01003103
Willy Tarreau6a06a402007-07-15 20:15:28 +02003104Proxy configuration can be located in a set of sections :
William Lallemand6e62fb62015-04-28 16:55:23 +02003105 - defaults [<name>]
Willy Tarreau6a06a402007-07-15 20:15:28 +02003106 - frontend <name>
3107 - backend <name>
3108 - listen <name>
3109
3110A "defaults" section sets default parameters for all other sections following
3111its declaration. Those default parameters are reset by the next "defaults"
3112section. See below for the list of parameters which can be set in a "defaults"
Willy Tarreau0ba27502007-12-24 16:55:16 +01003113section. The name is optional but its use is encouraged for better readability.
Willy Tarreau6a06a402007-07-15 20:15:28 +02003114
3115A "frontend" section describes a set of listening sockets accepting client
3116connections.
3117
3118A "backend" section describes a set of servers to which the proxy will connect
3119to forward incoming connections.
3120
3121A "listen" section defines a complete proxy with its frontend and backend
3122parts combined in one section. It is generally useful for TCP-only traffic.
3123
Willy Tarreau0ba27502007-12-24 16:55:16 +01003124All proxy names must be formed from upper and lower case letters, digits,
3125'-' (dash), '_' (underscore) , '.' (dot) and ':' (colon). ACL names are
3126case-sensitive, which means that "www" and "WWW" are two different proxies.
3127
3128Historically, all proxy names could overlap, it just caused troubles in the
3129logs. Since the introduction of content switching, it is mandatory that two
3130proxies with overlapping capabilities (frontend/backend) have different names.
3131However, it is still permitted that a frontend and a backend share the same
3132name, as this configuration seems to be commonly encountered.
3133
3134Right now, two major proxy modes are supported : "tcp", also known as layer 4,
3135and "http", also known as layer 7. In layer 4 mode, HAProxy simply forwards
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003136bidirectional traffic between two sides. In layer 7 mode, HAProxy analyzes the
Willy Tarreau0ba27502007-12-24 16:55:16 +01003137protocol, and can interact with it by allowing, blocking, switching, adding,
3138modifying, or removing arbitrary contents in requests or responses, based on
3139arbitrary criteria.
3140
Willy Tarreau70dffda2014-01-30 03:07:23 +01003141In HTTP mode, the processing applied to requests and responses flowing over
3142a connection depends in the combination of the frontend's HTTP options and
Julien Pivotto21ad3152019-12-10 13:11:17 +01003143the backend's. HAProxy supports 3 connection modes :
Willy Tarreau70dffda2014-01-30 03:07:23 +01003144
3145 - KAL : keep alive ("option http-keep-alive") which is the default mode : all
3146 requests and responses are processed, and connections remain open but idle
3147 between responses and new requests.
3148
Willy Tarreau70dffda2014-01-30 03:07:23 +01003149 - SCL: server close ("option http-server-close") : the server-facing
3150 connection is closed after the end of the response is received, but the
3151 client-facing connection remains open.
3152
Christopher Faulet315b39c2018-09-21 16:26:19 +02003153 - CLO: close ("option httpclose"): the connection is closed after the end of
3154 the response and "Connection: close" appended in both directions.
Willy Tarreau70dffda2014-01-30 03:07:23 +01003155
3156The effective mode that will be applied to a connection passing through a
3157frontend and a backend can be determined by both proxy modes according to the
3158following matrix, but in short, the modes are symmetric, keep-alive is the
Christopher Faulet315b39c2018-09-21 16:26:19 +02003159weakest option and close is the strongest.
Willy Tarreau70dffda2014-01-30 03:07:23 +01003160
Christopher Faulet315b39c2018-09-21 16:26:19 +02003161 Backend mode
Willy Tarreau70dffda2014-01-30 03:07:23 +01003162
Christopher Faulet315b39c2018-09-21 16:26:19 +02003163 | KAL | SCL | CLO
3164 ----+-----+-----+----
3165 KAL | KAL | SCL | CLO
3166 ----+-----+-----+----
Christopher Faulet315b39c2018-09-21 16:26:19 +02003167 mode SCL | SCL | SCL | CLO
3168 ----+-----+-----+----
3169 CLO | CLO | CLO | CLO
Willy Tarreau70dffda2014-01-30 03:07:23 +01003170
Willy Tarreau0ba27502007-12-24 16:55:16 +01003171
Willy Tarreau70dffda2014-01-30 03:07:23 +01003172
Willy Tarreauc57f0e22009-05-10 13:12:33 +020031734.1. Proxy keywords matrix
3174--------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01003175
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003176The following list of keywords is supported. Most of them may only be used in a
3177limited set of section types. Some of them are marked as "deprecated" because
3178they are inherited from an old syntax which may be confusing or functionally
3179limited, and there are new recommended keywords to replace them. Keywords
Davor Ocelice9ed2812017-12-25 17:49:28 +01003180marked with "(*)" can be optionally inverted using the "no" prefix, e.g. "no
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003181option contstats". This makes sense when the option has been enabled by default
Willy Tarreau3842f002009-06-14 11:39:52 +02003182and must be disabled for a specific instance. Such options may also be prefixed
3183with "default" in order to restore default settings regardless of what has been
3184specified in a previous "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003185
Willy Tarreau6a06a402007-07-15 20:15:28 +02003186
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003187 keyword defaults frontend listen backend
3188------------------------------------+----------+----------+---------+---------
3189acl - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003190backlog X X X -
3191balance X - X X
3192bind - X X -
3193bind-process X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003194capture cookie - X X -
3195capture request header - X X -
3196capture response header - X X -
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09003197clitcpka-cnt X X X -
3198clitcpka-idle X X X -
3199clitcpka-intvl X X X -
William Lallemand82fe75c2012-10-23 10:25:10 +02003200compression X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003201cookie X - X X
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02003202declare capture - X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003203default-server X - X X
3204default_backend X X X -
3205description - X X X
3206disabled X X X X
3207dispatch - - X X
Simon Horman51a1cf62015-02-03 13:00:44 +09003208email-alert from X X X X
Simon Horman64e34162015-02-06 11:11:57 +09003209email-alert level X X X X
Simon Horman51a1cf62015-02-03 13:00:44 +09003210email-alert mailers X X X X
3211email-alert myhostname X X X X
3212email-alert to X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003213enabled X X X X
3214errorfile X X X X
Christopher Faulet76edc0f2020-01-13 15:52:01 +01003215errorfiles X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003216errorloc X X X X
3217errorloc302 X X X X
3218-- keyword -------------------------- defaults - frontend - listen -- backend -
3219errorloc303 X X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01003220force-persist - - X X
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003221filter - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003222fullconn X - X X
3223grace X X X X
3224hash-type X - X X
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01003225http-after-response - X X X
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02003226http-check comment X - X X
Christopher Faulete5870d82020-04-15 11:32:03 +02003227http-check connect X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003228http-check disable-on-404 X - X X
Christopher Faulete5870d82020-04-15 11:32:03 +02003229http-check expect X - X X
Peter Gervai8912ae62020-06-11 18:26:36 +02003230http-check send X - X X
Willy Tarreau7ab6aff2010-10-12 06:30:16 +02003231http-check send-state X - X X
Christopher Faulete5870d82020-04-15 11:32:03 +02003232http-check set-var X - X X
3233http-check unset-var X - X X
Christopher Faulet3b967c12020-05-15 15:47:44 +02003234http-error X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003235http-request - X X X
Willy Tarreaue365c0b2013-06-11 16:06:12 +02003236http-response - X X X
Willy Tarreau30631952015-08-06 15:05:24 +02003237http-reuse X - X X
Baptiste Assmann2c42ef52013-10-09 21:57:02 +02003238http-send-name-header - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003239id - X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01003240ignore-persist - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02003241load-server-state-from-file X - X X
William Lallemand0f99e342011-10-12 17:50:54 +02003242log (*) X X X X
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01003243log-format X X X -
Dragan Dosen7ad31542015-09-28 17:16:47 +02003244log-format-sd X X X -
Willy Tarreau094af4e2015-01-07 15:03:42 +01003245log-tag X X X X
Willy Tarreauc35362a2014-04-25 13:58:37 +02003246max-keep-alive-queue X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003247maxconn X X X -
3248mode X X X X
3249monitor fail - X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003250monitor-uri X X X -
3251option abortonclose (*) X - X X
3252option accept-invalid-http-request (*) X X X -
3253option accept-invalid-http-response (*) X - X X
3254option allbackups (*) X - X X
3255option checkcache (*) X - X X
3256option clitcpka (*) X X X -
3257option contstats (*) X X X -
Christopher Faulet89aed322020-06-02 17:33:56 +02003258option disable-h2-upgrade (*) X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003259option dontlog-normal (*) X X X -
3260option dontlognull (*) X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003261-- keyword -------------------------- defaults - frontend - listen -- backend -
3262option forwardfor X X X X
Christopher Faulet98fbe952019-07-22 16:18:24 +02003263option h1-case-adjust-bogus-client (*) X X X -
3264option h1-case-adjust-bogus-server (*) X - X X
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02003265option http-buffer-request (*) X X X X
Willy Tarreau82649f92015-05-01 22:40:51 +02003266option http-ignore-probes (*) X X X -
Willy Tarreau16bfb022010-01-16 19:48:41 +01003267option http-keep-alive (*) X X X X
Willy Tarreau96e31212011-05-30 18:10:30 +02003268option http-no-delay (*) X X X X
Christopher Faulet98db9762018-09-21 10:25:19 +02003269option http-pretend-keepalive (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003270option http-server-close (*) X X X X
3271option http-use-proxy-header (*) X X X -
3272option httpchk X - X X
3273option httpclose (*) X X X X
Freddy Spierenburge88b7732019-03-25 14:35:17 +01003274option httplog X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003275option http_proxy (*) X X X X
Jamie Gloudon801a0a32012-08-25 00:18:33 -04003276option independent-streams (*) X X X X
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02003277option ldap-check X - X X
Simon Horman98637e52014-06-20 12:30:16 +09003278option external-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003279option log-health-checks (*) X - X X
3280option log-separate-errors (*) X X X -
3281option logasap (*) X X X -
3282option mysql-check X - X X
3283option nolinger (*) X X X X
3284option originalto X X X X
3285option persist (*) X - X X
Baptiste Assmann809e22a2015-10-12 20:22:55 +02003286option pgsql-check X - X X
3287option prefer-last-server (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003288option redispatch (*) X - X X
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02003289option redis-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003290option smtpchk X - X X
3291option socket-stats (*) X X X -
3292option splice-auto (*) X X X X
3293option splice-request (*) X X X X
3294option splice-response (*) X X X X
Christopher Fauletba7bc162016-11-07 21:07:38 +01003295option spop-check - - - X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003296option srvtcpka (*) X - X X
3297option ssl-hello-chk X - X X
3298-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreaued179852013-12-16 01:07:00 +01003299option tcp-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003300option tcp-smart-accept (*) X X X -
3301option tcp-smart-connect (*) X - X X
3302option tcpka X X X X
3303option tcplog X X X X
3304option transparent (*) X - X X
Simon Horman98637e52014-06-20 12:30:16 +09003305external-check command X - X X
3306external-check path X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003307persist rdp-cookie X - X X
3308rate-limit sessions X X X -
3309redirect - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003310-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003311retries X - X X
Olivier Houcharda254a372019-04-05 15:30:12 +02003312retry-on X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003313server - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02003314server-state-file-name X - X X
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02003315server-template - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003316source X - X X
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09003317srvtcpka-cnt X - X X
3318srvtcpka-idle X - X X
3319srvtcpka-intvl X - X X
Baptiste Assmann5a549212015-10-12 20:30:24 +02003320stats admin - X X X
3321stats auth X X X X
3322stats enable X X X X
3323stats hide-version X X X X
3324stats http-request - X X X
3325stats realm X X X X
3326stats refresh X X X X
3327stats scope X X X X
3328stats show-desc X X X X
3329stats show-legends X X X X
3330stats show-node X X X X
3331stats uri X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003332-- keyword -------------------------- defaults - frontend - listen -- backend -
3333stick match - - X X
3334stick on - - X X
3335stick store-request - - X X
Willy Tarreaud8dc99f2011-07-01 11:33:25 +02003336stick store-response - - X X
Adam Spiers68af3c12017-04-06 16:31:39 +01003337stick-table - X X X
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02003338tcp-check comment X - X X
Christopher Faulet404f9192020-04-09 23:13:54 +02003339tcp-check connect X - X X
3340tcp-check expect X - X X
3341tcp-check send X - X X
Christopher Fauletb50b3e62020-05-05 18:43:43 +02003342tcp-check send-lf X - X X
Christopher Faulet404f9192020-04-09 23:13:54 +02003343tcp-check send-binary X - X X
Christopher Fauletb50b3e62020-05-05 18:43:43 +02003344tcp-check send-binary-lf X - X X
Christopher Faulet404f9192020-04-09 23:13:54 +02003345tcp-check set-var X - X X
3346tcp-check unset-var X - X X
Willy Tarreaue9656522010-08-17 15:40:09 +02003347tcp-request connection - X X -
3348tcp-request content - X X X
Willy Tarreaua56235c2010-09-14 11:31:36 +02003349tcp-request inspect-delay - X X X
Willy Tarreau4f614292016-10-21 17:49:36 +02003350tcp-request session - X X -
Emeric Brun0a3b67f2010-09-24 15:34:53 +02003351tcp-response content - - X X
3352tcp-response inspect-delay - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003353timeout check X - X X
3354timeout client X X X -
Willy Tarreau05cdd962014-05-10 14:30:07 +02003355timeout client-fin X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003356timeout connect X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003357timeout http-keep-alive X X X X
3358timeout http-request X X X X
3359timeout queue X - X X
3360timeout server X - X X
Willy Tarreau05cdd962014-05-10 14:30:07 +02003361timeout server-fin X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003362timeout tarpit X X X X
Willy Tarreauce887fd2012-05-12 12:50:00 +02003363timeout tunnel X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003364transparent (deprecated) X - X X
William Lallemanda73203e2012-03-12 12:48:57 +01003365unique-id-format X X X -
3366unique-id-header X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003367use_backend - X X -
Christopher Fauletb30b3102019-09-12 23:03:09 +02003368use-fcgi-app - - X X
Willy Tarreau4a5cade2012-04-05 21:09:48 +02003369use-server - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003370------------------------------------+----------+----------+---------+---------
3371 keyword defaults frontend listen backend
Willy Tarreau6a06a402007-07-15 20:15:28 +02003372
Willy Tarreau0ba27502007-12-24 16:55:16 +01003373
Willy Tarreauc57f0e22009-05-10 13:12:33 +020033744.2. Alphabetically sorted keywords reference
3375---------------------------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01003376
3377This section provides a description of each keyword and its usage.
3378
3379
3380acl <aclname> <criterion> [flags] [operator] <value> ...
3381 Declare or complete an access list.
3382 May be used in sections : defaults | frontend | listen | backend
3383 no | yes | yes | yes
3384 Example:
3385 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
3386 acl invalid_src src_port 0:1023
3387 acl local_dst hdr(host) -i localhost
3388
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003389 See section 7 about ACL usage.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003390
3391
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01003392backlog <conns>
3393 Give hints to the system about the approximate listen backlog desired size
3394 May be used in sections : defaults | frontend | listen | backend
3395 yes | yes | yes | no
3396 Arguments :
3397 <conns> is the number of pending connections. Depending on the operating
3398 system, it may represent the number of already acknowledged
Cyril Bontédc4d9032012-04-08 21:57:39 +02003399 connections, of non-acknowledged ones, or both.
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01003400
3401 In order to protect against SYN flood attacks, one solution is to increase
3402 the system's SYN backlog size. Depending on the system, sometimes it is just
3403 tunable via a system parameter, sometimes it is not adjustable at all, and
3404 sometimes the system relies on hints given by the application at the time of
3405 the listen() syscall. By default, HAProxy passes the frontend's maxconn value
3406 to the listen() syscall. On systems which can make use of this value, it can
3407 sometimes be useful to be able to specify a different value, hence this
3408 backlog parameter.
3409
3410 On Linux 2.4, the parameter is ignored by the system. On Linux 2.6, it is
3411 used as a hint and the system accepts up to the smallest greater power of
3412 two, and never more than some limits (usually 32768).
3413
3414 See also : "maxconn" and the target operating system's tuning guide.
3415
3416
Willy Tarreau0ba27502007-12-24 16:55:16 +01003417balance <algorithm> [ <arguments> ]
Willy Tarreau226071e2014-04-10 11:55:45 +02003418balance url_param <param> [check_post]
Willy Tarreau0ba27502007-12-24 16:55:16 +01003419 Define the load balancing algorithm to be used in a backend.
3420 May be used in sections : defaults | frontend | listen | backend
3421 yes | no | yes | yes
3422 Arguments :
3423 <algorithm> is the algorithm used to select a server when doing load
3424 balancing. This only applies when no persistence information
3425 is available, or when a connection is redispatched to another
3426 server. <algorithm> may be one of the following :
3427
3428 roundrobin Each server is used in turns, according to their weights.
3429 This is the smoothest and fairest algorithm when the server's
3430 processing time remains equally distributed. This algorithm
3431 is dynamic, which means that server weights may be adjusted
Willy Tarreau9757a382009-10-03 12:56:50 +02003432 on the fly for slow starts for instance. It is limited by
Godbacha34bdc02013-07-22 07:44:53 +08003433 design to 4095 active servers per backend. Note that in some
Willy Tarreau9757a382009-10-03 12:56:50 +02003434 large farms, when a server becomes up after having been down
3435 for a very short time, it may sometimes take a few hundreds
3436 requests for it to be re-integrated into the farm and start
3437 receiving traffic. This is normal, though very rare. It is
3438 indicated here in case you would have the chance to observe
3439 it, so that you don't worry.
3440
3441 static-rr Each server is used in turns, according to their weights.
3442 This algorithm is as similar to roundrobin except that it is
3443 static, which means that changing a server's weight on the
3444 fly will have no effect. On the other hand, it has no design
3445 limitation on the number of servers, and when a server goes
3446 up, it is always immediately reintroduced into the farm, once
3447 the full map is recomputed. It also uses slightly less CPU to
3448 run (around -1%).
Willy Tarreau0ba27502007-12-24 16:55:16 +01003449
Willy Tarreau2d2a7f82008-03-17 12:07:56 +01003450 leastconn The server with the lowest number of connections receives the
3451 connection. Round-robin is performed within groups of servers
3452 of the same load to ensure that all servers will be used. Use
3453 of this algorithm is recommended where very long sessions are
3454 expected, such as LDAP, SQL, TSE, etc... but is not very well
3455 suited for protocols using short sessions such as HTTP. This
3456 algorithm is dynamic, which means that server weights may be
Willy Tarreau8c855f62020-10-22 17:41:45 +02003457 adjusted on the fly for slow starts for instance. It will
3458 also consider the number of queued connections in addition to
3459 the established ones in order to minimize queuing.
Willy Tarreau2d2a7f82008-03-17 12:07:56 +01003460
Willy Tarreauf09c6602012-02-13 17:12:08 +01003461 first The first server with available connection slots receives the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03003462 connection. The servers are chosen from the lowest numeric
Willy Tarreauf09c6602012-02-13 17:12:08 +01003463 identifier to the highest (see server parameter "id"), which
3464 defaults to the server's position in the farm. Once a server
Willy Tarreau64559c52012-04-07 09:08:45 +02003465 reaches its maxconn value, the next server is used. It does
Willy Tarreauf09c6602012-02-13 17:12:08 +01003466 not make sense to use this algorithm without setting maxconn.
3467 The purpose of this algorithm is to always use the smallest
3468 number of servers so that extra servers can be powered off
3469 during non-intensive hours. This algorithm ignores the server
3470 weight, and brings more benefit to long session such as RDP
Willy Tarreau64559c52012-04-07 09:08:45 +02003471 or IMAP than HTTP, though it can be useful there too. In
3472 order to use this algorithm efficiently, it is recommended
3473 that a cloud controller regularly checks server usage to turn
3474 them off when unused, and regularly checks backend queue to
3475 turn new servers on when the queue inflates. Alternatively,
3476 using "http-check send-state" may inform servers on the load.
Willy Tarreauf09c6602012-02-13 17:12:08 +01003477
Willy Tarreau0ba27502007-12-24 16:55:16 +01003478 source The source IP address is hashed and divided by the total
3479 weight of the running servers to designate which server will
3480 receive the request. This ensures that the same client IP
3481 address will always reach the same server as long as no
3482 server goes down or up. If the hash result changes due to the
3483 number of running servers changing, many clients will be
3484 directed to a different server. This algorithm is generally
3485 used in TCP mode where no cookie may be inserted. It may also
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003486 be used on the Internet to provide a best-effort stickiness
Willy Tarreau0ba27502007-12-24 16:55:16 +01003487 to clients which refuse session cookies. This algorithm is
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003488 static by default, which means that changing a server's
3489 weight on the fly will have no effect, but this can be
3490 changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003491
Oskar Stolc8dc41842012-05-19 10:19:54 +01003492 uri This algorithm hashes either the left part of the URI (before
3493 the question mark) or the whole URI (if the "whole" parameter
3494 is present) and divides the hash value by the total weight of
3495 the running servers. The result designates which server will
3496 receive the request. This ensures that the same URI will
3497 always be directed to the same server as long as no server
3498 goes up or down. This is used with proxy caches and
3499 anti-virus proxies in order to maximize the cache hit rate.
3500 Note that this algorithm may only be used in an HTTP backend.
3501 This algorithm is static by default, which means that
3502 changing a server's weight on the fly will have no effect,
3503 but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003504
Oskar Stolc8dc41842012-05-19 10:19:54 +01003505 This algorithm supports two optional parameters "len" and
Marek Majkowski9c30fc12008-04-27 23:25:55 +02003506 "depth", both followed by a positive integer number. These
3507 options may be helpful when it is needed to balance servers
3508 based on the beginning of the URI only. The "len" parameter
3509 indicates that the algorithm should only consider that many
3510 characters at the beginning of the URI to compute the hash.
3511 Note that having "len" set to 1 rarely makes sense since most
3512 URIs start with a leading "/".
3513
3514 The "depth" parameter indicates the maximum directory depth
3515 to be used to compute the hash. One level is counted for each
3516 slash in the request. If both parameters are specified, the
3517 evaluation stops when either is reached.
3518
Willy Tarreau57a37412020-09-23 08:56:29 +02003519 A "path-only" parameter indicates that the hashing key starts
3520 at the first '/' of the path. This can be used to ignore the
3521 authority part of absolute URIs, and to make sure that HTTP/1
3522 and HTTP/2 URIs will provide the same hash.
3523
Willy Tarreau0ba27502007-12-24 16:55:16 +01003524 url_param The URL parameter specified in argument will be looked up in
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003525 the query string of each HTTP GET request.
3526
3527 If the modifier "check_post" is used, then an HTTP POST
Cyril Bontédc4d9032012-04-08 21:57:39 +02003528 request entity will be searched for the parameter argument,
3529 when it is not found in a query string after a question mark
Willy Tarreau226071e2014-04-10 11:55:45 +02003530 ('?') in the URL. The message body will only start to be
3531 analyzed once either the advertised amount of data has been
3532 received or the request buffer is full. In the unlikely event
3533 that chunked encoding is used, only the first chunk is
Cyril Bontédc4d9032012-04-08 21:57:39 +02003534 scanned. Parameter values separated by a chunk boundary, may
Willy Tarreau226071e2014-04-10 11:55:45 +02003535 be randomly balanced if at all. This keyword used to support
3536 an optional <max_wait> parameter which is now ignored.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003537
3538 If the parameter is found followed by an equal sign ('=') and
3539 a value, then the value is hashed and divided by the total
3540 weight of the running servers. The result designates which
3541 server will receive the request.
3542
3543 This is used to track user identifiers in requests and ensure
3544 that a same user ID will always be sent to the same server as
3545 long as no server goes up or down. If no value is found or if
3546 the parameter is not found, then a round robin algorithm is
3547 applied. Note that this algorithm may only be used in an HTTP
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003548 backend. This algorithm is static by default, which means
3549 that changing a server's weight on the fly will have no
3550 effect, but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003551
Cyril Bontédc4d9032012-04-08 21:57:39 +02003552 hdr(<name>) The HTTP header <name> will be looked up in each HTTP
3553 request. Just as with the equivalent ACL 'hdr()' function,
3554 the header name in parenthesis is not case sensitive. If the
3555 header is absent or if it does not contain any value, the
3556 roundrobin algorithm is applied instead.
Benoitaffb4812009-03-25 13:02:10 +01003557
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003558 An optional 'use_domain_only' parameter is available, for
Benoitaffb4812009-03-25 13:02:10 +01003559 reducing the hash algorithm to the main domain part with some
3560 specific headers such as 'Host'. For instance, in the Host
3561 value "haproxy.1wt.eu", only "1wt" will be considered.
3562
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003563 This algorithm is static by default, which means that
3564 changing a server's weight on the fly will have no effect,
3565 but this can be changed using "hash-type".
3566
Willy Tarreau21c741a2019-01-14 18:14:27 +01003567 random
3568 random(<draws>)
3569 A random number will be used as the key for the consistent
Willy Tarreau760e81d2018-05-03 07:20:40 +02003570 hashing function. This means that the servers' weights are
3571 respected, dynamic weight changes immediately take effect, as
3572 well as new server additions. Random load balancing can be
3573 useful with large farms or when servers are frequently added
Willy Tarreau21c741a2019-01-14 18:14:27 +01003574 or removed as it may avoid the hammering effect that could
3575 result from roundrobin or leastconn in this situation. The
3576 hash-balance-factor directive can be used to further improve
3577 fairness of the load balancing, especially in situations
3578 where servers show highly variable response times. When an
3579 argument <draws> is present, it must be an integer value one
3580 or greater, indicating the number of draws before selecting
3581 the least loaded of these servers. It was indeed demonstrated
3582 that picking the least loaded of two servers is enough to
3583 significantly improve the fairness of the algorithm, by
3584 always avoiding to pick the most loaded server within a farm
3585 and getting rid of any bias that could be induced by the
3586 unfair distribution of the consistent list. Higher values N
3587 will take away N-1 of the highest loaded servers at the
3588 expense of performance. With very high values, the algorithm
3589 will converge towards the leastconn's result but much slower.
3590 The default value is 2, which generally shows very good
3591 distribution and performance. This algorithm is also known as
3592 the Power of Two Random Choices and is described here :
3593 http://www.eecs.harvard.edu/~michaelm/postscripts/handbook2001.pdf
Willy Tarreau760e81d2018-05-03 07:20:40 +02003594
Emeric Brun736aa232009-06-30 17:56:00 +02003595 rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02003596 rdp-cookie(<name>)
Emeric Brun736aa232009-06-30 17:56:00 +02003597 The RDP cookie <name> (or "mstshash" if omitted) will be
3598 looked up and hashed for each incoming TCP request. Just as
3599 with the equivalent ACL 'req_rdp_cookie()' function, the name
3600 is not case-sensitive. This mechanism is useful as a degraded
3601 persistence mode, as it makes it possible to always send the
3602 same user (or the same session ID) to the same server. If the
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003603 cookie is not found, the normal roundrobin algorithm is
Emeric Brun736aa232009-06-30 17:56:00 +02003604 used instead.
3605
3606 Note that for this to work, the frontend must ensure that an
3607 RDP cookie is already present in the request buffer. For this
3608 you must use 'tcp-request content accept' rule combined with
3609 a 'req_rdp_cookie_cnt' ACL.
3610
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003611 This algorithm is static by default, which means that
3612 changing a server's weight on the fly will have no effect,
3613 but this can be changed using "hash-type".
3614
Cyril Bontédc4d9032012-04-08 21:57:39 +02003615 See also the rdp_cookie pattern fetch function.
Simon Hormanab814e02011-06-24 14:50:20 +09003616
Willy Tarreau0ba27502007-12-24 16:55:16 +01003617 <arguments> is an optional list of arguments which may be needed by some
Marek Majkowski9c30fc12008-04-27 23:25:55 +02003618 algorithms. Right now, only "url_param" and "uri" support an
3619 optional argument.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003620
Willy Tarreau3cd9af22009-03-15 14:06:41 +01003621 The load balancing algorithm of a backend is set to roundrobin when no other
3622 algorithm, mode nor option have been set. The algorithm may only be set once
3623 for each backend.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003624
Lukas Tribus80512b12018-10-27 20:07:40 +02003625 With authentication schemes that require the same connection like NTLM, URI
John Roeslerfb2fce12019-07-10 15:45:51 -05003626 based algorithms must not be used, as they would cause subsequent requests
Lukas Tribus80512b12018-10-27 20:07:40 +02003627 to be routed to different backend servers, breaking the invalid assumptions
3628 NTLM relies on.
3629
Willy Tarreau0ba27502007-12-24 16:55:16 +01003630 Examples :
3631 balance roundrobin
3632 balance url_param userid
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003633 balance url_param session_id check_post 64
Benoitaffb4812009-03-25 13:02:10 +01003634 balance hdr(User-Agent)
3635 balance hdr(host)
3636 balance hdr(Host) use_domain_only
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003637
3638 Note: the following caveats and limitations on using the "check_post"
3639 extension with "url_param" must be considered :
3640
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003641 - all POST requests are eligible for consideration, because there is no way
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003642 to determine if the parameters will be found in the body or entity which
3643 may contain binary data. Therefore another method may be required to
3644 restrict consideration of POST requests that have no URL parameters in
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02003645 the body. (see acl http_end)
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003646
3647 - using a <max_wait> value larger than the request buffer size does not
3648 make sense and is useless. The buffer size is set at build time, and
3649 defaults to 16 kB.
3650
3651 - Content-Encoding is not supported, the parameter search will probably
3652 fail; and load balancing will fall back to Round Robin.
3653
3654 - Expect: 100-continue is not supported, load balancing will fall back to
3655 Round Robin.
3656
Lukas Tribus23953682017-04-28 13:24:30 +00003657 - Transfer-Encoding (RFC7230 3.3.1) is only supported in the first chunk.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003658 If the entire parameter value is not present in the first chunk, the
3659 selection of server is undefined (actually, defined by how little
3660 actually appeared in the first chunk).
3661
3662 - This feature does not support generation of a 100, 411 or 501 response.
3663
3664 - In some cases, requesting "check_post" MAY attempt to scan the entire
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003665 contents of a message body. Scanning normally terminates when linear
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003666 white space or control characters are found, indicating the end of what
3667 might be a URL parameter list. This is probably not a concern with SGML
3668 type message bodies.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003669
Willy Tarreau294d0f02015-08-10 19:40:12 +02003670 See also : "dispatch", "cookie", "transparent", "hash-type" and "http_proxy".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003671
3672
Willy Tarreaub6205fd2012-09-24 12:27:33 +02003673bind [<address>]:<port_range> [, ...] [param*]
3674bind /<path> [, ...] [param*]
Willy Tarreau0ba27502007-12-24 16:55:16 +01003675 Define one or several listening addresses and/or ports in a frontend.
3676 May be used in sections : defaults | frontend | listen | backend
3677 no | yes | yes | no
3678 Arguments :
Willy Tarreaub1e52e82008-01-13 14:49:51 +01003679 <address> is optional and can be a host name, an IPv4 address, an IPv6
3680 address, or '*'. It designates the address the frontend will
3681 listen on. If unset, all IPv4 addresses of the system will be
3682 listened on. The same will apply for '*' or the system's
David du Colombier9c938da2011-03-17 10:40:27 +01003683 special address "0.0.0.0". The IPv6 equivalent is '::'.
Willy Tarreau24709282013-03-10 21:32:12 +01003684 Optionally, an address family prefix may be used before the
3685 address to force the family regardless of the address format,
3686 which can be useful to specify a path to a unix socket with
3687 no slash ('/'). Currently supported prefixes are :
3688 - 'ipv4@' -> address is always IPv4
3689 - 'ipv6@' -> address is always IPv6
Emeric Brun3835c0d2020-07-07 09:46:09 +02003690 - 'udp@' -> address is resolved as IPv4 or IPv6 and
Emeric Brun12941c82020-07-07 14:19:42 +02003691 protocol UDP is used. Currently those listeners are
3692 supported only in log-forward sections.
Emeric Brun3835c0d2020-07-07 09:46:09 +02003693 - 'udp4@' -> address is always IPv4 and protocol UDP
Emeric Brun12941c82020-07-07 14:19:42 +02003694 is used. Currently those listeners are supported
3695 only in log-forward sections.
Emeric Brun3835c0d2020-07-07 09:46:09 +02003696 - 'udp6@' -> address is always IPv6 and protocol UDP
Emeric Brun12941c82020-07-07 14:19:42 +02003697 is used. Currently those listeners are supported
3698 only in log-forward sections.
Willy Tarreau24709282013-03-10 21:32:12 +01003699 - 'unix@' -> address is a path to a local unix socket
Willy Tarreau70f72e02014-07-08 00:37:50 +02003700 - 'abns@' -> address is in abstract namespace (Linux only).
3701 Note: since abstract sockets are not "rebindable", they
3702 do not cope well with multi-process mode during
3703 soft-restart, so it is better to avoid them if
3704 nbproc is greater than 1. The effect is that if the
3705 new process fails to start, only one of the old ones
3706 will be able to rebind to the socket.
Willy Tarreau40aa0702013-03-10 23:51:38 +01003707 - 'fd@<n>' -> use file descriptor <n> inherited from the
3708 parent. The fd must be bound and may or may not already
3709 be listening.
William Lallemand2fe7dd02018-09-11 16:51:29 +02003710 - 'sockpair@<n>'-> like fd@ but you must use the fd of a
3711 connected unix socket or of a socketpair. The bind waits
3712 to receive a FD over the unix socket and uses it as if it
3713 was the FD of an accept(). Should be used carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02003714 You may want to reference some environment variables in the
3715 address parameter, see section 2.3 about environment
3716 variables.
Willy Tarreaub1e52e82008-01-13 14:49:51 +01003717
Willy Tarreauc5011ca2010-03-22 11:53:56 +01003718 <port_range> is either a unique TCP port, or a port range for which the
3719 proxy will accept connections for the IP address specified
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003720 above. The port is mandatory for TCP listeners. Note that in
3721 the case of an IPv6 address, the port is always the number
3722 after the last colon (':'). A range can either be :
Willy Tarreauc5011ca2010-03-22 11:53:56 +01003723 - a numerical port (ex: '80')
3724 - a dash-delimited ports range explicitly stating the lower
3725 and upper bounds (ex: '2000-2100') which are included in
3726 the range.
3727
3728 Particular care must be taken against port ranges, because
3729 every <address:port> couple consumes one socket (= a file
3730 descriptor), so it's easy to consume lots of descriptors
3731 with a simple range, and to run out of sockets. Also, each
3732 <address:port> couple must be used only once among all
3733 instances running on a same system. Please note that binding
3734 to ports lower than 1024 generally require particular
Jamie Gloudon801a0a32012-08-25 00:18:33 -04003735 privileges to start the program, which are independent of
Willy Tarreauc5011ca2010-03-22 11:53:56 +01003736 the 'uid' parameter.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003737
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003738 <path> is a UNIX socket path beginning with a slash ('/'). This is
Davor Ocelice9ed2812017-12-25 17:49:28 +01003739 alternative to the TCP listening port. HAProxy will then
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003740 receive UNIX connections on the socket located at this place.
3741 The path must begin with a slash and by default is absolute.
3742 It can be relative to the prefix defined by "unix-bind" in
3743 the global section. Note that the total length of the prefix
3744 followed by the socket path cannot exceed some system limits
3745 for UNIX sockets, which commonly are set to 107 characters.
3746
Willy Tarreaub6205fd2012-09-24 12:27:33 +02003747 <param*> is a list of parameters common to all sockets declared on the
3748 same line. These numerous parameters depend on OS and build
3749 options and have a complete section dedicated to them. Please
3750 refer to section 5 to for more details.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02003751
Willy Tarreau0ba27502007-12-24 16:55:16 +01003752 It is possible to specify a list of address:port combinations delimited by
3753 commas. The frontend will then listen on all of these addresses. There is no
3754 fixed limit to the number of addresses and ports which can be listened on in
3755 a frontend, as well as there is no limit to the number of "bind" statements
3756 in a frontend.
3757
3758 Example :
3759 listen http_proxy
3760 bind :80,:443
3761 bind 10.0.0.1:10080,10.0.0.1:10443
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003762 bind /var/run/ssl-frontend.sock user root mode 600 accept-proxy
Willy Tarreau0ba27502007-12-24 16:55:16 +01003763
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02003764 listen http_https_proxy
3765 bind :80
Cyril Bonté0d44fc62012-10-09 22:45:33 +02003766 bind :443 ssl crt /etc/haproxy/site.pem
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02003767
Willy Tarreau24709282013-03-10 21:32:12 +01003768 listen http_https_proxy_explicit
3769 bind ipv6@:80
3770 bind ipv4@public_ssl:443 ssl crt /etc/haproxy/site.pem
3771 bind unix@ssl-frontend.sock user root mode 600 accept-proxy
3772
Willy Tarreaudad36a32013-03-11 01:20:04 +01003773 listen external_bind_app1
William Lallemandb2f07452015-05-12 14:27:13 +02003774 bind "fd@${FD_APP1}"
Willy Tarreaudad36a32013-03-11 01:20:04 +01003775
Willy Tarreau55dcaf62015-09-27 15:03:15 +02003776 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
3777 sun_path length is used for the address length. Some other programs
3778 such as socat use the string length only by default. Pass the option
3779 ",unix-tightsocklen=0" to any abstract socket definition in socat to
3780 make it compatible with HAProxy's.
3781
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003782 See also : "source", "option forwardfor", "unix-bind" and the PROXY protocol
Willy Tarreaub6205fd2012-09-24 12:27:33 +02003783 documentation, and section 5 about bind options.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003784
3785
Christopher Fauletff4121f2017-11-22 16:38:49 +01003786bind-process [ all | odd | even | <process_num>[-[<process_num>]] ] ...
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003787 Limit visibility of an instance to a certain set of processes numbers.
3788 May be used in sections : defaults | frontend | listen | backend
3789 yes | yes | yes | yes
3790 Arguments :
3791 all All process will see this instance. This is the default. It
3792 may be used to override a default value.
3793
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003794 odd This instance will be enabled on processes 1,3,5,...63. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003795 option may be combined with other numbers.
3796
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003797 even This instance will be enabled on processes 2,4,6,...64. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003798 option may be combined with other numbers. Do not use it
3799 with less than 2 processes otherwise some instances might be
3800 missing from all processes.
3801
Christopher Fauletff4121f2017-11-22 16:38:49 +01003802 process_num The instance will be enabled on this process number or range,
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003803 whose values must all be between 1 and 32 or 64 depending on
Christopher Fauletff4121f2017-11-22 16:38:49 +01003804 the machine's word size. Ranges can be partially defined. The
3805 higher bound can be omitted. In such case, it is replaced by
3806 the corresponding maximum value. If a proxy is bound to
3807 process numbers greater than the configured global.nbproc, it
3808 will either be forced to process #1 if a single process was
Willy Tarreau102df612014-05-07 23:56:38 +02003809 specified, or to all processes otherwise.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003810
3811 This keyword limits binding of certain instances to certain processes. This
3812 is useful in order not to have too many processes listening to the same
3813 ports. For instance, on a dual-core machine, it might make sense to set
3814 'nbproc 2' in the global section, then distributes the listeners among 'odd'
3815 and 'even' instances.
3816
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003817 At the moment, it is not possible to reference more than 32 or 64 processes
3818 using this keyword, but this should be more than enough for most setups.
3819 Please note that 'all' really means all processes regardless of the machine's
3820 word size, and is not limited to the first 32 or 64.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003821
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02003822 Each "bind" line may further be limited to a subset of the proxy's processes,
3823 please consult the "process" bind keyword in section 5.1.
3824
Willy Tarreaub369a042014-09-16 13:21:03 +02003825 When a frontend has no explicit "bind-process" line, it tries to bind to all
3826 the processes referenced by its "bind" lines. That means that frontends can
3827 easily adapt to their listeners' processes.
3828
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003829 If some backends are referenced by frontends bound to other processes, the
3830 backend automatically inherits the frontend's processes.
3831
3832 Example :
3833 listen app_ip1
3834 bind 10.0.0.1:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02003835 bind-process odd
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003836
3837 listen app_ip2
3838 bind 10.0.0.2:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02003839 bind-process even
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003840
3841 listen management
3842 bind 10.0.0.3:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02003843 bind-process 1 2 3 4
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003844
Willy Tarreau110ecc12012-11-15 17:50:01 +01003845 listen management
3846 bind 10.0.0.4:80
3847 bind-process 1-4
3848
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02003849 See also : "nbproc" in global section, and "process" in section 5.1.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003850
3851
Willy Tarreau0ba27502007-12-24 16:55:16 +01003852capture cookie <name> len <length>
3853 Capture and log a cookie in the request and in the response.
3854 May be used in sections : defaults | frontend | listen | backend
3855 no | yes | yes | no
3856 Arguments :
3857 <name> is the beginning of the name of the cookie to capture. In order
3858 to match the exact name, simply suffix the name with an equal
3859 sign ('='). The full name will appear in the logs, which is
3860 useful with application servers which adjust both the cookie name
Davor Ocelice9ed2812017-12-25 17:49:28 +01003861 and value (e.g. ASPSESSIONXXX).
Willy Tarreau0ba27502007-12-24 16:55:16 +01003862
3863 <length> is the maximum number of characters to report in the logs, which
3864 include the cookie name, the equal sign and the value, all in the
3865 standard "name=value" form. The string will be truncated on the
3866 right if it exceeds <length>.
3867
3868 Only the first cookie is captured. Both the "cookie" request headers and the
3869 "set-cookie" response headers are monitored. This is particularly useful to
3870 check for application bugs causing session crossing or stealing between
3871 users, because generally the user's cookies can only change on a login page.
3872
3873 When the cookie was not presented by the client, the associated log column
3874 will report "-". When a request does not cause a cookie to be assigned by the
3875 server, a "-" is reported in the response column.
3876
3877 The capture is performed in the frontend only because it is necessary that
3878 the log format does not change for a given frontend depending on the
3879 backends. This may change in the future. Note that there can be only one
Willy Tarreau193b8c62012-11-22 00:17:38 +01003880 "capture cookie" statement in a frontend. The maximum capture length is set
3881 by the global "tune.http.cookielen" setting and defaults to 63 characters. It
3882 is not possible to specify a capture in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003883
3884 Example:
3885 capture cookie ASPSESSION len 32
3886
3887 See also : "capture request header", "capture response header" as well as
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003888 section 8 about logging.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003889
3890
3891capture request header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01003892 Capture and log the last occurrence of the specified request header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003893 May be used in sections : defaults | frontend | listen | backend
3894 no | yes | yes | no
3895 Arguments :
3896 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003897 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01003898 appear in the requests, with the first letter of each word in
3899 upper case. The header name will not appear in the logs, only the
3900 value is reported, but the position in the logs is respected.
3901
3902 <length> is the maximum number of characters to extract from the value and
3903 report in the logs. The string will be truncated on the right if
3904 it exceeds <length>.
3905
Willy Tarreau4460d032012-11-21 23:37:37 +01003906 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01003907 value will be added to the logs between braces ('{}'). If multiple headers
3908 are captured, they will be delimited by a vertical bar ('|') and will appear
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003909 in the same order they were declared in the configuration. Non-existent
3910 headers will be logged just as an empty string. Common uses for request
3911 header captures include the "Host" field in virtual hosting environments, the
3912 "Content-length" when uploads are supported, "User-agent" to quickly
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003913 differentiate between real users and robots, and "X-Forwarded-For" in proxied
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003914 environments to find where the request came from.
3915
3916 Note that when capturing headers such as "User-agent", some spaces may be
3917 logged, making the log analysis more difficult. Thus be careful about what
3918 you log if you know your log parser is not smart enough to rely on the
3919 braces.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003920
Willy Tarreau0900abb2012-11-22 00:21:46 +01003921 There is no limit to the number of captured request headers nor to their
3922 length, though it is wise to keep them low to limit memory usage per session.
3923 In order to keep log format consistent for a same frontend, header captures
3924 can only be declared in a frontend. It is not possible to specify a capture
3925 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003926
3927 Example:
3928 capture request header Host len 15
3929 capture request header X-Forwarded-For len 15
Cyril Bontéd1b0f7c2015-10-26 22:37:39 +01003930 capture request header Referer len 15
Willy Tarreau0ba27502007-12-24 16:55:16 +01003931
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003932 See also : "capture cookie", "capture response header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01003933 about logging.
3934
3935
3936capture response header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01003937 Capture and log the last occurrence of the specified response header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003938 May be used in sections : defaults | frontend | listen | backend
3939 no | yes | yes | no
3940 Arguments :
3941 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003942 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01003943 appear in the response, with the first letter of each word in
3944 upper case. The header name will not appear in the logs, only the
3945 value is reported, but the position in the logs is respected.
3946
3947 <length> is the maximum number of characters to extract from the value and
3948 report in the logs. The string will be truncated on the right if
3949 it exceeds <length>.
3950
Willy Tarreau4460d032012-11-21 23:37:37 +01003951 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01003952 result will be added to the logs between braces ('{}') after the captured
3953 request headers. If multiple headers are captured, they will be delimited by
3954 a vertical bar ('|') and will appear in the same order they were declared in
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003955 the configuration. Non-existent headers will be logged just as an empty
3956 string. Common uses for response header captures include the "Content-length"
3957 header which indicates how many bytes are expected to be returned, the
3958 "Location" header to track redirections.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003959
Willy Tarreau0900abb2012-11-22 00:21:46 +01003960 There is no limit to the number of captured response headers nor to their
3961 length, though it is wise to keep them low to limit memory usage per session.
3962 In order to keep log format consistent for a same frontend, header captures
3963 can only be declared in a frontend. It is not possible to specify a capture
3964 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003965
3966 Example:
3967 capture response header Content-length len 9
3968 capture response header Location len 15
3969
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003970 See also : "capture cookie", "capture request header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01003971 about logging.
3972
3973
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09003974clitcpka-cnt <count>
3975 Sets the maximum number of keepalive probes TCP should send before dropping
3976 the connection on the client side.
3977 May be used in sections : defaults | frontend | listen | backend
3978 yes | yes | yes | no
3979 Arguments :
3980 <count> is the maximum number of keepalive probes.
3981
3982 This keyword corresponds to the socket option TCP_KEEPCNT. If this keyword
3983 is not specified, system-wide TCP parameter (tcp_keepalive_probes) is used.
Willy Tarreau52543212020-07-09 05:58:51 +02003984 The availability of this setting depends on the operating system. It is
3985 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09003986
3987 See also : "option clitcpka", "clitcpka-idle", "clitcpka-intvl".
3988
3989
3990clitcpka-idle <timeout>
3991 Sets the time the connection needs to remain idle before TCP starts sending
3992 keepalive probes, if enabled the sending of TCP keepalive packets on the
3993 client side.
3994 May be used in sections : defaults | frontend | listen | backend
3995 yes | yes | yes | no
3996 Arguments :
3997 <timeout> is the time the connection needs to remain idle before TCP starts
3998 sending keepalive probes. It is specified in seconds by default,
3999 but can be in any other unit if the number is suffixed by the
4000 unit, as explained at the top of this document.
4001
4002 This keyword corresponds to the socket option TCP_KEEPIDLE. If this keyword
4003 is not specified, system-wide TCP parameter (tcp_keepalive_time) is used.
Willy Tarreau52543212020-07-09 05:58:51 +02004004 The availability of this setting depends on the operating system. It is
4005 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09004006
4007 See also : "option clitcpka", "clitcpka-cnt", "clitcpka-intvl".
4008
4009
4010clitcpka-intvl <timeout>
4011 Sets the time between individual keepalive probes on the client side.
4012 May be used in sections : defaults | frontend | listen | backend
4013 yes | yes | yes | no
4014 Arguments :
4015 <timeout> is the time between individual keepalive probes. It is specified
4016 in seconds by default, but can be in any other unit if the number
4017 is suffixed by the unit, as explained at the top of this
4018 document.
4019
4020 This keyword corresponds to the socket option TCP_KEEPINTVL. If this keyword
4021 is not specified, system-wide TCP parameter (tcp_keepalive_intvl) is used.
Willy Tarreau52543212020-07-09 05:58:51 +02004022 The availability of this setting depends on the operating system. It is
4023 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09004024
4025 See also : "option clitcpka", "clitcpka-cnt", "clitcpka-idle".
4026
4027
Cyril Bonté316a8cf2012-11-11 13:38:27 +01004028compression algo <algorithm> ...
4029compression type <mime type> ...
Willy Tarreau70737d12012-10-27 00:34:28 +02004030compression offload
William Lallemand82fe75c2012-10-23 10:25:10 +02004031 Enable HTTP compression.
4032 May be used in sections : defaults | frontend | listen | backend
4033 yes | yes | yes | yes
4034 Arguments :
Cyril Bonté316a8cf2012-11-11 13:38:27 +01004035 algo is followed by the list of supported compression algorithms.
4036 type is followed by the list of MIME types that will be compressed.
4037 offload makes haproxy work as a compression offloader only (see notes).
4038
4039 The currently supported algorithms are :
Willy Tarreauc91840a2015-03-28 17:00:39 +01004040 identity this is mostly for debugging, and it was useful for developing
4041 the compression feature. Identity does not apply any change on
4042 data.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01004043
Willy Tarreauc91840a2015-03-28 17:00:39 +01004044 gzip applies gzip compression. This setting is only available when
Baptiste Assmannf085d632015-12-21 17:57:32 +01004045 support for zlib or libslz was built in.
Willy Tarreauc91840a2015-03-28 17:00:39 +01004046
4047 deflate same as "gzip", but with deflate algorithm and zlib format.
4048 Note that this algorithm has ambiguous support on many
4049 browsers and no support at all from recent ones. It is
4050 strongly recommended not to use it for anything else than
4051 experimentation. This setting is only available when support
Baptiste Assmannf085d632015-12-21 17:57:32 +01004052 for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01004053
Willy Tarreauc91840a2015-03-28 17:00:39 +01004054 raw-deflate same as "deflate" without the zlib wrapper, and used as an
4055 alternative when the browser wants "deflate". All major
4056 browsers understand it and despite violating the standards,
4057 it is known to work better than "deflate", at least on MSIE
4058 and some versions of Safari. Do not use it in conjunction
4059 with "deflate", use either one or the other since both react
4060 to the same Accept-Encoding token. This setting is only
Baptiste Assmannf085d632015-12-21 17:57:32 +01004061 available when support for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01004062
Dmitry Sivachenko87c208b2012-11-22 20:03:26 +04004063 Compression will be activated depending on the Accept-Encoding request
Cyril Bonté316a8cf2012-11-11 13:38:27 +01004064 header. With identity, it does not take care of that header.
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04004065 If backend servers support HTTP compression, these directives
4066 will be no-op: haproxy will see the compressed response and will not
4067 compress again. If backend servers do not support HTTP compression and
4068 there is Accept-Encoding header in request, haproxy will compress the
4069 matching response.
Willy Tarreau70737d12012-10-27 00:34:28 +02004070
4071 The "offload" setting makes haproxy remove the Accept-Encoding header to
4072 prevent backend servers from compressing responses. It is strongly
4073 recommended not to do this because this means that all the compression work
4074 will be done on the single point where haproxy is located. However in some
4075 deployment scenarios, haproxy may be installed in front of a buggy gateway
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04004076 with broken HTTP compression implementation which can't be turned off.
4077 In that case haproxy can be used to prevent that gateway from emitting
4078 invalid payloads. In this case, simply removing the header in the
4079 configuration does not work because it applies before the header is parsed,
4080 so that prevents haproxy from compressing. The "offload" setting should
Willy Tarreauffea9fd2014-07-12 16:37:02 +02004081 then be used for such scenarios. Note: for now, the "offload" setting is
4082 ignored when set in a defaults section.
William Lallemand82fe75c2012-10-23 10:25:10 +02004083
William Lallemand05097442012-11-20 12:14:28 +01004084 Compression is disabled when:
Baptiste Assmann650d53d2013-01-05 15:44:44 +01004085 * the request does not advertise a supported compression algorithm in the
4086 "Accept-Encoding" header
Julien Pivotto7f314c92021-03-29 12:41:40 +02004087 * the response message is not HTTP/1.1 or above
Tim Duesterhusbb48c9a2019-01-30 23:46:04 +01004088 * HTTP status code is not one of 200, 201, 202, or 203
Baptiste Assmann650d53d2013-01-05 15:44:44 +01004089 * response contain neither a "Content-Length" header nor a
4090 "Transfer-Encoding" whose last value is "chunked"
4091 * response contains a "Content-Type" header whose first value starts with
4092 "multipart"
4093 * the response contains the "no-transform" value in the "Cache-control"
4094 header
4095 * User-Agent matches "Mozilla/4" unless it is MSIE 6 with XP SP2, or MSIE 7
4096 and later
4097 * The response contains a "Content-Encoding" header, indicating that the
4098 response is already compressed (see compression offload)
Tim Duesterhusbb48c9a2019-01-30 23:46:04 +01004099 * The response contains an invalid "ETag" header or multiple ETag headers
William Lallemand05097442012-11-20 12:14:28 +01004100
Tim Duesterhusb229f012019-01-29 16:38:56 +01004101 Note: The compression does not emit the Warning header.
William Lallemand05097442012-11-20 12:14:28 +01004102
William Lallemand82fe75c2012-10-23 10:25:10 +02004103 Examples :
4104 compression algo gzip
4105 compression type text/html text/plain
Willy Tarreau0ba27502007-12-24 16:55:16 +01004106
Christopher Fauletc3fe5332016-04-07 15:30:10 +02004107
Willy Tarreau55165fe2009-05-10 12:02:55 +02004108cookie <name> [ rewrite | insert | prefix ] [ indirect ] [ nocache ]
Willy Tarreau4992dd22012-05-31 21:02:17 +02004109 [ postonly ] [ preserve ] [ httponly ] [ secure ]
4110 [ domain <domain> ]* [ maxidle <idle> ] [ maxlife <life> ]
Christopher Faulet2f533902020-01-21 11:06:48 +01004111 [ dynamic ] [ attr <value> ]*
Willy Tarreau0ba27502007-12-24 16:55:16 +01004112 Enable cookie-based persistence in a backend.
4113 May be used in sections : defaults | frontend | listen | backend
4114 yes | no | yes | yes
4115 Arguments :
4116 <name> is the name of the cookie which will be monitored, modified or
4117 inserted in order to bring persistence. This cookie is sent to
4118 the client via a "Set-Cookie" header in the response, and is
4119 brought back by the client in a "Cookie" header in all requests.
4120 Special care should be taken to choose a name which does not
4121 conflict with any likely application cookie. Also, if the same
Davor Ocelice9ed2812017-12-25 17:49:28 +01004122 backends are subject to be used by the same clients (e.g.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004123 HTTP/HTTPS), care should be taken to use different cookie names
4124 between all backends if persistence between them is not desired.
4125
4126 rewrite This keyword indicates that the cookie will be provided by the
4127 server and that haproxy will have to modify its value to set the
4128 server's identifier in it. This mode is handy when the management
4129 of complex combinations of "Set-cookie" and "Cache-control"
4130 headers is left to the application. The application can then
4131 decide whether or not it is appropriate to emit a persistence
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01004132 cookie. Since all responses should be monitored, this mode
4133 doesn't work in HTTP tunnel mode. Unless the application
Davor Ocelice9ed2812017-12-25 17:49:28 +01004134 behavior is very complex and/or broken, it is advised not to
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01004135 start with this mode for new deployments. This keyword is
4136 incompatible with "insert" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01004137
4138 insert This keyword indicates that the persistence cookie will have to
Willy Tarreaua79094d2010-08-31 22:54:15 +02004139 be inserted by haproxy in server responses if the client did not
Willy Tarreauba4c5be2010-10-23 12:46:42 +02004140
Willy Tarreaua79094d2010-08-31 22:54:15 +02004141 already have a cookie that would have permitted it to access this
Willy Tarreauba4c5be2010-10-23 12:46:42 +02004142 server. When used without the "preserve" option, if the server
Michael Prokop4438c602019-05-24 10:25:45 +02004143 emits a cookie with the same name, it will be removed before
Davor Ocelice9ed2812017-12-25 17:49:28 +01004144 processing. For this reason, this mode can be used to upgrade
Willy Tarreauba4c5be2010-10-23 12:46:42 +02004145 existing configurations running in the "rewrite" mode. The cookie
4146 will only be a session cookie and will not be stored on the
4147 client's disk. By default, unless the "indirect" option is added,
4148 the server will see the cookies emitted by the client. Due to
4149 caching effects, it is generally wise to add the "nocache" or
4150 "postonly" keywords (see below). The "insert" keyword is not
4151 compatible with "rewrite" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01004152
4153 prefix This keyword indicates that instead of relying on a dedicated
4154 cookie for the persistence, an existing one will be completed.
4155 This may be needed in some specific environments where the client
4156 does not support more than one single cookie and the application
4157 already needs it. In this case, whenever the server sets a cookie
4158 named <name>, it will be prefixed with the server's identifier
4159 and a delimiter. The prefix will be removed from all client
4160 requests so that the server still finds the cookie it emitted.
4161 Since all requests and responses are subject to being modified,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01004162 this mode doesn't work with tunnel mode. The "prefix" keyword is
Willy Tarreau37229df2011-10-17 12:24:55 +02004163 not compatible with "rewrite" and "insert". Note: it is highly
4164 recommended not to use "indirect" with "prefix", otherwise server
4165 cookie updates would not be sent to clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004166
Willy Tarreaua79094d2010-08-31 22:54:15 +02004167 indirect When this option is specified, no cookie will be emitted to a
4168 client which already has a valid one for the server which has
4169 processed the request. If the server sets such a cookie itself,
Willy Tarreauba4c5be2010-10-23 12:46:42 +02004170 it will be removed, unless the "preserve" option is also set. In
4171 "insert" mode, this will additionally remove cookies from the
4172 requests transmitted to the server, making the persistence
4173 mechanism totally transparent from an application point of view.
Willy Tarreau37229df2011-10-17 12:24:55 +02004174 Note: it is highly recommended not to use "indirect" with
4175 "prefix", otherwise server cookie updates would not be sent to
4176 clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004177
4178 nocache This option is recommended in conjunction with the insert mode
4179 when there is a cache between the client and HAProxy, as it
4180 ensures that a cacheable response will be tagged non-cacheable if
4181 a cookie needs to be inserted. This is important because if all
4182 persistence cookies are added on a cacheable home page for
4183 instance, then all customers will then fetch the page from an
4184 outer cache and will all share the same persistence cookie,
4185 leading to one server receiving much more traffic than others.
4186 See also the "insert" and "postonly" options.
4187
4188 postonly This option ensures that cookie insertion will only be performed
4189 on responses to POST requests. It is an alternative to the
4190 "nocache" option, because POST responses are not cacheable, so
4191 this ensures that the persistence cookie will never get cached.
4192 Since most sites do not need any sort of persistence before the
4193 first POST which generally is a login request, this is a very
4194 efficient method to optimize caching without risking to find a
4195 persistence cookie in the cache.
4196 See also the "insert" and "nocache" options.
4197
Willy Tarreauba4c5be2010-10-23 12:46:42 +02004198 preserve This option may only be used with "insert" and/or "indirect". It
4199 allows the server to emit the persistence cookie itself. In this
4200 case, if a cookie is found in the response, haproxy will leave it
4201 untouched. This is useful in order to end persistence after a
4202 logout request for instance. For this, the server just has to
Davor Ocelice9ed2812017-12-25 17:49:28 +01004203 emit a cookie with an invalid value (e.g. empty) or with a date in
Willy Tarreauba4c5be2010-10-23 12:46:42 +02004204 the past. By combining this mechanism with the "disable-on-404"
4205 check option, it is possible to perform a completely graceful
4206 shutdown because users will definitely leave the server after
4207 they logout.
4208
Willy Tarreau4992dd22012-05-31 21:02:17 +02004209 httponly This option tells haproxy to add an "HttpOnly" cookie attribute
4210 when a cookie is inserted. This attribute is used so that a
4211 user agent doesn't share the cookie with non-HTTP components.
4212 Please check RFC6265 for more information on this attribute.
4213
4214 secure This option tells haproxy to add a "Secure" cookie attribute when
4215 a cookie is inserted. This attribute is used so that a user agent
4216 never emits this cookie over non-secure channels, which means
4217 that a cookie learned with this flag will be presented only over
4218 SSL/TLS connections. Please check RFC6265 for more information on
4219 this attribute.
4220
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02004221 domain This option allows to specify the domain at which a cookie is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004222 inserted. It requires exactly one parameter: a valid domain
Willy Tarreau68a897b2009-12-03 23:28:34 +01004223 name. If the domain begins with a dot, the browser is allowed to
4224 use it for any host ending with that name. It is also possible to
4225 specify several domain names by invoking this option multiple
4226 times. Some browsers might have small limits on the number of
4227 domains, so be careful when doing that. For the record, sending
4228 10 domains to MSIE 6 or Firefox 2 works as expected.
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02004229
Willy Tarreau996a92c2010-10-13 19:30:47 +02004230 maxidle This option allows inserted cookies to be ignored after some idle
4231 time. It only works with insert-mode cookies. When a cookie is
4232 sent to the client, the date this cookie was emitted is sent too.
4233 Upon further presentations of this cookie, if the date is older
4234 than the delay indicated by the parameter (in seconds), it will
4235 be ignored. Otherwise, it will be refreshed if needed when the
4236 response is sent to the client. This is particularly useful to
4237 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01004238 too long on the same server (e.g. after a farm size change). When
Willy Tarreau996a92c2010-10-13 19:30:47 +02004239 this option is set and a cookie has no date, it is always
4240 accepted, but gets refreshed in the response. This maintains the
4241 ability for admins to access their sites. Cookies that have a
4242 date in the future further than 24 hours are ignored. Doing so
4243 lets admins fix timezone issues without risking kicking users off
4244 the site.
4245
4246 maxlife This option allows inserted cookies to be ignored after some life
4247 time, whether they're in use or not. It only works with insert
4248 mode cookies. When a cookie is first sent to the client, the date
4249 this cookie was emitted is sent too. Upon further presentations
4250 of this cookie, if the date is older than the delay indicated by
4251 the parameter (in seconds), it will be ignored. If the cookie in
4252 the request has no date, it is accepted and a date will be set.
4253 Cookies that have a date in the future further than 24 hours are
4254 ignored. Doing so lets admins fix timezone issues without risking
4255 kicking users off the site. Contrary to maxidle, this value is
4256 not refreshed, only the first visit date counts. Both maxidle and
4257 maxlife may be used at the time. This is particularly useful to
4258 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01004259 too long on the same server (e.g. after a farm size change). This
Willy Tarreau996a92c2010-10-13 19:30:47 +02004260 is stronger than the maxidle method in that it forces a
4261 redispatch after some absolute delay.
4262
Olivier Houchard4e694042017-03-14 20:01:29 +01004263 dynamic Activate dynamic cookies. When used, a session cookie is
4264 dynamically created for each server, based on the IP and port
4265 of the server, and a secret key, specified in the
4266 "dynamic-cookie-key" backend directive.
4267 The cookie will be regenerated each time the IP address change,
4268 and is only generated for IPv4/IPv6.
4269
Christopher Faulet2f533902020-01-21 11:06:48 +01004270 attr This option tells haproxy to add an extra attribute when a
4271 cookie is inserted. The attribute value can contain any
4272 characters except control ones or ";". This option may be
4273 repeated.
4274
Willy Tarreau0ba27502007-12-24 16:55:16 +01004275 There can be only one persistence cookie per HTTP backend, and it can be
4276 declared in a defaults section. The value of the cookie will be the value
4277 indicated after the "cookie" keyword in a "server" statement. If no cookie
4278 is declared for a given server, the cookie is not set.
Willy Tarreau6a06a402007-07-15 20:15:28 +02004279
Willy Tarreau0ba27502007-12-24 16:55:16 +01004280 Examples :
4281 cookie JSESSIONID prefix
4282 cookie SRV insert indirect nocache
4283 cookie SRV insert postonly indirect
Willy Tarreau996a92c2010-10-13 19:30:47 +02004284 cookie SRV insert indirect nocache maxidle 30m maxlife 8h
Willy Tarreau0ba27502007-12-24 16:55:16 +01004285
Willy Tarreau294d0f02015-08-10 19:40:12 +02004286 See also : "balance source", "capture cookie", "server" and "ignore-persist".
Willy Tarreau0ba27502007-12-24 16:55:16 +01004287
Willy Tarreau983e01e2010-01-11 18:42:06 +01004288
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02004289declare capture [ request | response ] len <length>
4290 Declares a capture slot.
4291 May be used in sections : defaults | frontend | listen | backend
4292 no | yes | yes | no
4293 Arguments:
4294 <length> is the length allowed for the capture.
4295
4296 This declaration is only available in the frontend or listen section, but the
4297 reserved slot can be used in the backends. The "request" keyword allocates a
4298 capture slot for use in the request, and "response" allocates a capture slot
4299 for use in the response.
4300
4301 See also: "capture-req", "capture-res" (sample converters),
Baptiste Assmann5ac425c2015-10-21 23:13:46 +02004302 "capture.req.hdr", "capture.res.hdr" (sample fetches),
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02004303 "http-request capture" and "http-response capture".
4304
4305
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01004306default-server [param*]
4307 Change default options for a server in a backend
4308 May be used in sections : defaults | frontend | listen | backend
4309 yes | no | yes | yes
4310 Arguments:
Willy Tarreau983e01e2010-01-11 18:42:06 +01004311 <param*> is a list of parameters for this server. The "default-server"
4312 keyword accepts an important number of options and has a complete
4313 section dedicated to it. Please refer to section 5 for more
4314 details.
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01004315
Willy Tarreau983e01e2010-01-11 18:42:06 +01004316 Example :
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01004317 default-server inter 1000 weight 13
4318
4319 See also: "server" and section 5 about server options
Willy Tarreau0ba27502007-12-24 16:55:16 +01004320
Willy Tarreau983e01e2010-01-11 18:42:06 +01004321
Willy Tarreau0ba27502007-12-24 16:55:16 +01004322default_backend <backend>
4323 Specify the backend to use when no "use_backend" rule has been matched.
4324 May be used in sections : defaults | frontend | listen | backend
4325 yes | yes | yes | no
4326 Arguments :
4327 <backend> is the name of the backend to use.
4328
4329 When doing content-switching between frontend and backends using the
4330 "use_backend" keyword, it is often useful to indicate which backend will be
4331 used when no rule has matched. It generally is the dynamic backend which
4332 will catch all undetermined requests.
4333
Willy Tarreau0ba27502007-12-24 16:55:16 +01004334 Example :
4335
4336 use_backend dynamic if url_dyn
4337 use_backend static if url_css url_img extension_img
4338 default_backend dynamic
4339
Willy Tarreau98d04852015-05-26 12:18:29 +02004340 See also : "use_backend"
Willy Tarreau2769aa02007-12-27 18:26:09 +01004341
Willy Tarreau0ba27502007-12-24 16:55:16 +01004342
Baptiste Assmann27f51342013-10-09 06:51:49 +02004343description <string>
4344 Describe a listen, frontend or backend.
4345 May be used in sections : defaults | frontend | listen | backend
4346 no | yes | yes | yes
4347 Arguments : string
4348
4349 Allows to add a sentence to describe the related object in the HAProxy HTML
4350 stats page. The description will be printed on the right of the object name
4351 it describes.
4352 No need to backslash spaces in the <string> arguments.
4353
4354
Willy Tarreau0ba27502007-12-24 16:55:16 +01004355disabled
4356 Disable a proxy, frontend or backend.
4357 May be used in sections : defaults | frontend | listen | backend
4358 yes | yes | yes | yes
4359 Arguments : none
4360
4361 The "disabled" keyword is used to disable an instance, mainly in order to
4362 liberate a listening port or to temporarily disable a service. The instance
4363 will still be created and its configuration will be checked, but it will be
4364 created in the "stopped" state and will appear as such in the statistics. It
4365 will not receive any traffic nor will it send any health-checks or logs. It
4366 is possible to disable many instances at once by adding the "disabled"
4367 keyword in a "defaults" section.
4368
4369 See also : "enabled"
4370
4371
Willy Tarreau5ce94572010-06-07 14:35:41 +02004372dispatch <address>:<port>
4373 Set a default server address
4374 May be used in sections : defaults | frontend | listen | backend
4375 no | no | yes | yes
Cyril Bonté108cf6e2012-04-21 23:30:29 +02004376 Arguments :
Willy Tarreau5ce94572010-06-07 14:35:41 +02004377
4378 <address> is the IPv4 address of the default server. Alternatively, a
4379 resolvable hostname is supported, but this name will be resolved
4380 during start-up.
4381
4382 <ports> is a mandatory port specification. All connections will be sent
4383 to this port, and it is not permitted to use port offsets as is
4384 possible with normal servers.
4385
Willy Tarreau787aed52011-04-15 06:45:37 +02004386 The "dispatch" keyword designates a default server for use when no other
Willy Tarreau5ce94572010-06-07 14:35:41 +02004387 server can take the connection. In the past it was used to forward non
4388 persistent connections to an auxiliary load balancer. Due to its simple
4389 syntax, it has also been used for simple TCP relays. It is recommended not to
4390 use it for more clarity, and to use the "server" directive instead.
4391
4392 See also : "server"
4393
Olivier Houchard4e694042017-03-14 20:01:29 +01004394
4395dynamic-cookie-key <string>
4396 Set the dynamic cookie secret key for a backend.
4397 May be used in sections : defaults | frontend | listen | backend
4398 yes | no | yes | yes
4399 Arguments : The secret key to be used.
4400
4401 When dynamic cookies are enabled (see the "dynamic" directive for cookie),
Davor Ocelice9ed2812017-12-25 17:49:28 +01004402 a dynamic cookie is created for each server (unless one is explicitly
Olivier Houchard4e694042017-03-14 20:01:29 +01004403 specified on the "server" line), using a hash of the IP address of the
4404 server, the TCP port, and the secret key.
Davor Ocelice9ed2812017-12-25 17:49:28 +01004405 That way, we can ensure session persistence across multiple load-balancers,
Olivier Houchard4e694042017-03-14 20:01:29 +01004406 even if servers are dynamically added or removed.
Willy Tarreau5ce94572010-06-07 14:35:41 +02004407
Willy Tarreau0ba27502007-12-24 16:55:16 +01004408enabled
4409 Enable a proxy, frontend or backend.
4410 May be used in sections : defaults | frontend | listen | backend
4411 yes | yes | yes | yes
4412 Arguments : none
4413
4414 The "enabled" keyword is used to explicitly enable an instance, when the
4415 defaults has been set to "disabled". This is very rarely used.
4416
4417 See also : "disabled"
4418
4419
4420errorfile <code> <file>
4421 Return a file contents instead of errors generated by HAProxy
4422 May be used in sections : defaults | frontend | listen | backend
4423 yes | yes | yes | yes
4424 Arguments :
4425 <code> is the HTTP status code. Currently, HAProxy is capable of
Christopher Faulet612f2ea2020-05-27 09:57:28 +02004426 generating codes 200, 400, 401, 403, 404, 405, 407, 408, 410,
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02004427 413, 425, 429, 500, 502, 503, and 504.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004428
4429 <file> designates a file containing the full HTTP response. It is
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01004430 recommended to follow the common practice of appending ".http" to
Willy Tarreau0ba27502007-12-24 16:55:16 +01004431 the filename so that people do not confuse the response with HTML
Willy Tarreau59140a22009-02-22 12:02:30 +01004432 error pages, and to use absolute paths, since files are read
4433 before any chroot is performed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004434
4435 It is important to understand that this keyword is not meant to rewrite
4436 errors returned by the server, but errors detected and returned by HAProxy.
4437 This is why the list of supported errors is limited to a small set.
4438
Willy Tarreauae94d4d2011-05-11 16:28:49 +02004439 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
4440
Christopher Faulet70170672020-05-18 17:42:48 +02004441 The files are parsed when HAProxy starts and must be valid according to the
4442 HTTP specification. They should not exceed the configured buffer size
4443 (BUFSIZE), which generally is 16 kB, otherwise an internal error will be
4444 returned. It is also wise not to put any reference to local contents
4445 (e.g. images) in order to avoid loops between the client and HAProxy when all
4446 servers are down, causing an error to be returned instead of an
4447 image. Finally, The response cannot exceed (tune.bufsize - tune.maxrewrite)
4448 so that "http-after-response" rules still have room to operate (see
4449 "tune.maxrewrite").
Willy Tarreau59140a22009-02-22 12:02:30 +01004450
Willy Tarreau0ba27502007-12-24 16:55:16 +01004451 The files are read at the same time as the configuration and kept in memory.
4452 For this reason, the errors continue to be returned even when the process is
4453 chrooted, and no file change is considered while the process is running. A
Willy Tarreauc27debf2008-01-06 08:57:02 +01004454 simple method for developing those files consists in associating them to the
Willy Tarreau0ba27502007-12-24 16:55:16 +01004455 403 status code and interrogating a blocked URL.
4456
Christopher Faulet3b967c12020-05-15 15:47:44 +02004457 See also : "http-error", "errorloc", "errorloc302", "errorloc303"
Willy Tarreau0ba27502007-12-24 16:55:16 +01004458
Willy Tarreau59140a22009-02-22 12:02:30 +01004459 Example :
4460 errorfile 400 /etc/haproxy/errorfiles/400badreq.http
Willy Tarreau989222a2016-01-15 10:26:26 +01004461 errorfile 408 /dev/null # work around Chrome pre-connect bug
Willy Tarreau59140a22009-02-22 12:02:30 +01004462 errorfile 403 /etc/haproxy/errorfiles/403forbid.http
4463 errorfile 503 /etc/haproxy/errorfiles/503sorry.http
4464
Willy Tarreau2769aa02007-12-27 18:26:09 +01004465
Christopher Faulet76edc0f2020-01-13 15:52:01 +01004466errorfiles <name> [<code> ...]
4467 Import, fully or partially, the error files defined in the <name> http-errors
4468 section.
4469 May be used in sections : defaults | frontend | listen | backend
4470 yes | yes | yes | yes
4471 Arguments :
4472 <name> is the name of an existing http-errors section.
4473
4474 <code> is a HTTP status code. Several status code may be listed.
Christopher Faulet612f2ea2020-05-27 09:57:28 +02004475 Currently, HAProxy is capable of generating codes 200, 400, 401,
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02004476 403, 404, 405, 407, 408, 410, 413, 425, 429, 500, 502, 503, and 504.
Christopher Faulet76edc0f2020-01-13 15:52:01 +01004477
4478 Errors defined in the http-errors section with the name <name> are imported
4479 in the current proxy. If no status code is specified, all error files of the
4480 http-errors section are imported. Otherwise, only error files associated to
4481 the listed status code are imported. Those error files override the already
4482 defined custom errors for the proxy. And they may be overridden by following
Daniel Corbett67a82712020-07-06 23:01:19 -04004483 ones. Functionally, it is exactly the same as declaring all error files by
Christopher Faulet76edc0f2020-01-13 15:52:01 +01004484 hand using "errorfile" directives.
4485
Christopher Faulet3b967c12020-05-15 15:47:44 +02004486 See also : "http-error", "errorfile", "errorloc", "errorloc302" ,
4487 "errorloc303" and section 3.8 about http-errors.
Christopher Faulet76edc0f2020-01-13 15:52:01 +01004488
4489 Example :
4490 errorfiles generic
4491 errorfiles site-1 403 404
4492
4493
Willy Tarreau2769aa02007-12-27 18:26:09 +01004494errorloc <code> <url>
4495errorloc302 <code> <url>
4496 Return an HTTP redirection to a URL instead of errors generated by HAProxy
4497 May be used in sections : defaults | frontend | listen | backend
4498 yes | yes | yes | yes
4499 Arguments :
4500 <code> is the HTTP status code. Currently, HAProxy is capable of
Christopher Faulet612f2ea2020-05-27 09:57:28 +02004501 generating codes 200, 400, 401, 403, 404, 405, 407, 408, 410,
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02004502 413, 425, 429, 500, 502, 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01004503
4504 <url> it is the exact contents of the "Location" header. It may contain
4505 either a relative URI to an error page hosted on the same site,
4506 or an absolute URI designating an error page on another site.
4507 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01004508 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01004509
4510 It is important to understand that this keyword is not meant to rewrite
4511 errors returned by the server, but errors detected and returned by HAProxy.
4512 This is why the list of supported errors is limited to a small set.
4513
Willy Tarreauae94d4d2011-05-11 16:28:49 +02004514 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
4515
Willy Tarreau2769aa02007-12-27 18:26:09 +01004516 Note that both keyword return the HTTP 302 status code, which tells the
4517 client to fetch the designated URL using the same HTTP method. This can be
4518 quite problematic in case of non-GET methods such as POST, because the URL
4519 sent to the client might not be allowed for something other than GET. To
Willy Tarreau989222a2016-01-15 10:26:26 +01004520 work around this problem, please use "errorloc303" which send the HTTP 303
Willy Tarreau2769aa02007-12-27 18:26:09 +01004521 status code, indicating to the client that the URL must be fetched with a GET
4522 request.
4523
Christopher Faulet3b967c12020-05-15 15:47:44 +02004524 See also : "http-error", "errorfile", "errorloc303"
Willy Tarreau2769aa02007-12-27 18:26:09 +01004525
4526
4527errorloc303 <code> <url>
4528 Return an HTTP redirection to a URL instead of errors generated by HAProxy
4529 May be used in sections : defaults | frontend | listen | backend
4530 yes | yes | yes | yes
4531 Arguments :
4532 <code> is the HTTP status code. Currently, HAProxy is capable of
Christopher Faulet612f2ea2020-05-27 09:57:28 +02004533 generating codes 200, 400, 401, 403, 404, 405, 407, 408, 410,
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02004534 413, 425, 429, 500, 502, 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01004535
4536 <url> it is the exact contents of the "Location" header. It may contain
4537 either a relative URI to an error page hosted on the same site,
4538 or an absolute URI designating an error page on another site.
4539 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01004540 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01004541
4542 It is important to understand that this keyword is not meant to rewrite
4543 errors returned by the server, but errors detected and returned by HAProxy.
4544 This is why the list of supported errors is limited to a small set.
4545
Willy Tarreauae94d4d2011-05-11 16:28:49 +02004546 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
4547
Willy Tarreau2769aa02007-12-27 18:26:09 +01004548 Note that both keyword return the HTTP 303 status code, which tells the
4549 client to fetch the designated URL using the same HTTP GET method. This
4550 solves the usual problems associated with "errorloc" and the 302 code. It is
4551 possible that some very old browsers designed before HTTP/1.1 do not support
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01004552 it, but no such problem has been reported till now.
Willy Tarreau2769aa02007-12-27 18:26:09 +01004553
Christopher Faulet3b967c12020-05-15 15:47:44 +02004554 See also : "http-error", "errorfile", "errorloc", "errorloc302"
Willy Tarreau2769aa02007-12-27 18:26:09 +01004555
4556
Simon Horman51a1cf62015-02-03 13:00:44 +09004557email-alert from <emailaddr>
4558 Declare the from email address to be used in both the envelope and header
Davor Ocelice9ed2812017-12-25 17:49:28 +01004559 of email alerts. This is the address that email alerts are sent from.
Simon Horman51a1cf62015-02-03 13:00:44 +09004560 May be used in sections: defaults | frontend | listen | backend
4561 yes | yes | yes | yes
4562
4563 Arguments :
4564
4565 <emailaddr> is the from email address to use when sending email alerts
4566
4567 Also requires "email-alert mailers" and "email-alert to" to be set
4568 and if so sending email alerts is enabled for the proxy.
4569
Simon Horman64e34162015-02-06 11:11:57 +09004570 See also : "email-alert level", "email-alert mailers",
Cyril Bonté307ee1e2015-09-28 23:16:06 +02004571 "email-alert myhostname", "email-alert to", section 3.6 about
4572 mailers.
Simon Horman64e34162015-02-06 11:11:57 +09004573
4574
4575email-alert level <level>
4576 Declare the maximum log level of messages for which email alerts will be
4577 sent. This acts as a filter on the sending of email alerts.
4578 May be used in sections: defaults | frontend | listen | backend
4579 yes | yes | yes | yes
4580
4581 Arguments :
4582
4583 <level> One of the 8 syslog levels:
4584 emerg alert crit err warning notice info debug
4585 The above syslog levels are ordered from lowest to highest.
4586
4587 By default level is alert
4588
4589 Also requires "email-alert from", "email-alert mailers" and
4590 "email-alert to" to be set and if so sending email alerts is enabled
4591 for the proxy.
4592
Simon Horman1421e212015-04-30 13:10:35 +09004593 Alerts are sent when :
4594
4595 * An un-paused server is marked as down and <level> is alert or lower
4596 * A paused server is marked as down and <level> is notice or lower
4597 * A server is marked as up or enters the drain state and <level>
4598 is notice or lower
4599 * "option log-health-checks" is enabled, <level> is info or lower,
4600 and a health check status update occurs
4601
Simon Horman64e34162015-02-06 11:11:57 +09004602 See also : "email-alert from", "email-alert mailers",
4603 "email-alert myhostname", "email-alert to",
Simon Horman51a1cf62015-02-03 13:00:44 +09004604 section 3.6 about mailers.
4605
4606
4607email-alert mailers <mailersect>
4608 Declare the mailers to be used when sending email alerts
4609 May be used in sections: defaults | frontend | listen | backend
4610 yes | yes | yes | yes
4611
4612 Arguments :
4613
4614 <mailersect> is the name of the mailers section to send email alerts.
4615
4616 Also requires "email-alert from" and "email-alert to" to be set
4617 and if so sending email alerts is enabled for the proxy.
4618
Simon Horman64e34162015-02-06 11:11:57 +09004619 See also : "email-alert from", "email-alert level", "email-alert myhostname",
4620 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09004621
4622
4623email-alert myhostname <hostname>
4624 Declare the to hostname address to be used when communicating with
4625 mailers.
4626 May be used in sections: defaults | frontend | listen | backend
4627 yes | yes | yes | yes
4628
4629 Arguments :
4630
Baptiste Assmann738bad92015-12-21 15:27:53 +01004631 <hostname> is the hostname to use when communicating with mailers
Simon Horman51a1cf62015-02-03 13:00:44 +09004632
4633 By default the systems hostname is used.
4634
4635 Also requires "email-alert from", "email-alert mailers" and
4636 "email-alert to" to be set and if so sending email alerts is enabled
4637 for the proxy.
4638
Simon Horman64e34162015-02-06 11:11:57 +09004639 See also : "email-alert from", "email-alert level", "email-alert mailers",
4640 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09004641
4642
4643email-alert to <emailaddr>
Davor Ocelice9ed2812017-12-25 17:49:28 +01004644 Declare both the recipient address in the envelope and to address in the
Simon Horman51a1cf62015-02-03 13:00:44 +09004645 header of email alerts. This is the address that email alerts are sent to.
4646 May be used in sections: defaults | frontend | listen | backend
4647 yes | yes | yes | yes
4648
4649 Arguments :
4650
4651 <emailaddr> is the to email address to use when sending email alerts
4652
4653 Also requires "email-alert mailers" and "email-alert to" to be set
4654 and if so sending email alerts is enabled for the proxy.
4655
Simon Horman64e34162015-02-06 11:11:57 +09004656 See also : "email-alert from", "email-alert level", "email-alert mailers",
Simon Horman51a1cf62015-02-03 13:00:44 +09004657 "email-alert myhostname", section 3.6 about mailers.
4658
4659
Willy Tarreau4de91492010-01-22 19:10:05 +01004660force-persist { if | unless } <condition>
4661 Declare a condition to force persistence on down servers
4662 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01004663 no | no | yes | yes
Willy Tarreau4de91492010-01-22 19:10:05 +01004664
4665 By default, requests are not dispatched to down servers. It is possible to
4666 force this using "option persist", but it is unconditional and redispatches
4667 to a valid server if "option redispatch" is set. That leaves with very little
4668 possibilities to force some requests to reach a server which is artificially
4669 marked down for maintenance operations.
4670
4671 The "force-persist" statement allows one to declare various ACL-based
4672 conditions which, when met, will cause a request to ignore the down status of
4673 a server and still try to connect to it. That makes it possible to start a
4674 server, still replying an error to the health checks, and run a specially
4675 configured browser to test the service. Among the handy methods, one could
4676 use a specific source IP address, or a specific cookie. The cookie also has
4677 the advantage that it can easily be added/removed on the browser from a test
4678 page. Once the service is validated, it is then possible to open the service
4679 to the world by returning a valid response to health checks.
4680
4681 The forced persistence is enabled when an "if" condition is met, or unless an
4682 "unless" condition is met. The final redispatch is always disabled when this
4683 is used.
4684
Cyril Bonté0d4bf012010-04-25 23:21:46 +02004685 See also : "option redispatch", "ignore-persist", "persist",
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +02004686 and section 7 about ACL usage.
Willy Tarreau4de91492010-01-22 19:10:05 +01004687
Christopher Fauletc3fe5332016-04-07 15:30:10 +02004688
4689filter <name> [param*]
4690 Add the filter <name> in the filter list attached to the proxy.
4691 May be used in sections : defaults | frontend | listen | backend
4692 no | yes | yes | yes
4693 Arguments :
4694 <name> is the name of the filter. Officially supported filters are
4695 referenced in section 9.
4696
Tim Düsterhus4896c442016-11-29 02:15:19 +01004697 <param*> is a list of parameters accepted by the filter <name>. The
Christopher Fauletc3fe5332016-04-07 15:30:10 +02004698 parsing of these parameters are the responsibility of the
Tim Düsterhus4896c442016-11-29 02:15:19 +01004699 filter. Please refer to the documentation of the corresponding
4700 filter (section 9) for all details on the supported parameters.
Christopher Fauletc3fe5332016-04-07 15:30:10 +02004701
4702 Multiple occurrences of the filter line can be used for the same proxy. The
4703 same filter can be referenced many times if needed.
4704
4705 Example:
4706 listen
4707 bind *:80
4708
4709 filter trace name BEFORE-HTTP-COMP
4710 filter compression
4711 filter trace name AFTER-HTTP-COMP
4712
4713 compression algo gzip
4714 compression offload
4715
4716 server srv1 192.168.0.1:80
4717
4718 See also : section 9.
4719
Willy Tarreau4de91492010-01-22 19:10:05 +01004720
Willy Tarreau2769aa02007-12-27 18:26:09 +01004721fullconn <conns>
4722 Specify at what backend load the servers will reach their maxconn
4723 May be used in sections : defaults | frontend | listen | backend
4724 yes | no | yes | yes
4725 Arguments :
4726 <conns> is the number of connections on the backend which will make the
4727 servers use the maximal number of connections.
4728
Willy Tarreau198a7442008-01-17 12:05:32 +01004729 When a server has a "maxconn" parameter specified, it means that its number
Willy Tarreau2769aa02007-12-27 18:26:09 +01004730 of concurrent connections will never go higher. Additionally, if it has a
Willy Tarreau198a7442008-01-17 12:05:32 +01004731 "minconn" parameter, it indicates a dynamic limit following the backend's
Willy Tarreau2769aa02007-12-27 18:26:09 +01004732 load. The server will then always accept at least <minconn> connections,
4733 never more than <maxconn>, and the limit will be on the ramp between both
4734 values when the backend has less than <conns> concurrent connections. This
4735 makes it possible to limit the load on the servers during normal loads, but
4736 push it further for important loads without overloading the servers during
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004737 exceptional loads.
Willy Tarreau2769aa02007-12-27 18:26:09 +01004738
Willy Tarreaufbb78422011-06-05 15:38:35 +02004739 Since it's hard to get this value right, haproxy automatically sets it to
4740 10% of the sum of the maxconns of all frontends that may branch to this
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01004741 backend (based on "use_backend" and "default_backend" rules). That way it's
4742 safe to leave it unset. However, "use_backend" involving dynamic names are
4743 not counted since there is no way to know if they could match or not.
Willy Tarreaufbb78422011-06-05 15:38:35 +02004744
Willy Tarreau2769aa02007-12-27 18:26:09 +01004745 Example :
4746 # The servers will accept between 100 and 1000 concurrent connections each
4747 # and the maximum of 1000 will be reached when the backend reaches 10000
4748 # connections.
4749 backend dynamic
4750 fullconn 10000
4751 server srv1 dyn1:80 minconn 100 maxconn 1000
4752 server srv2 dyn2:80 minconn 100 maxconn 1000
4753
4754 See also : "maxconn", "server"
4755
4756
Willy Tarreauab0a5192020-10-09 19:07:01 +02004757grace <time> (deprecated)
Willy Tarreau2769aa02007-12-27 18:26:09 +01004758 Maintain a proxy operational for some time after a soft stop
4759 May be used in sections : defaults | frontend | listen | backend
Cyril Bonté99ed3272010-01-24 23:29:44 +01004760 yes | yes | yes | yes
Willy Tarreau2769aa02007-12-27 18:26:09 +01004761 Arguments :
4762 <time> is the time (by default in milliseconds) for which the instance
4763 will remain operational with the frontend sockets still listening
4764 when a soft-stop is received via the SIGUSR1 signal.
4765
4766 This may be used to ensure that the services disappear in a certain order.
4767 This was designed so that frontends which are dedicated to monitoring by an
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004768 external equipment fail immediately while other ones remain up for the time
Willy Tarreau2769aa02007-12-27 18:26:09 +01004769 needed by the equipment to detect the failure.
4770
4771 Note that currently, there is very little benefit in using this parameter,
4772 and it may in fact complicate the soft-reconfiguration process more than
4773 simplify it.
4774
Willy Tarreau0ba27502007-12-24 16:55:16 +01004775
Andrew Rodland17be45e2016-10-25 17:04:12 -04004776hash-balance-factor <factor>
4777 Specify the balancing factor for bounded-load consistent hashing
4778 May be used in sections : defaults | frontend | listen | backend
4779 yes | no | no | yes
4780 Arguments :
4781 <factor> is the control for the maximum number of concurrent requests to
4782 send to a server, expressed as a percentage of the average number
Frédéric Lécaille93d33162019-03-06 09:35:59 +01004783 of concurrent requests across all of the active servers.
Andrew Rodland17be45e2016-10-25 17:04:12 -04004784
4785 Specifying a "hash-balance-factor" for a server with "hash-type consistent"
4786 enables an algorithm that prevents any one server from getting too many
4787 requests at once, even if some hash buckets receive many more requests than
4788 others. Setting <factor> to 0 (the default) disables the feature. Otherwise,
4789 <factor> is a percentage greater than 100. For example, if <factor> is 150,
4790 then no server will be allowed to have a load more than 1.5 times the average.
4791 If server weights are used, they will be respected.
4792
4793 If the first-choice server is disqualified, the algorithm will choose another
4794 server based on the request hash, until a server with additional capacity is
4795 found. A higher <factor> allows more imbalance between the servers, while a
4796 lower <factor> means that more servers will be checked on average, affecting
4797 performance. Reasonable values are from 125 to 200.
4798
Willy Tarreau760e81d2018-05-03 07:20:40 +02004799 This setting is also used by "balance random" which internally relies on the
4800 consistent hashing mechanism.
4801
Andrew Rodland17be45e2016-10-25 17:04:12 -04004802 See also : "balance" and "hash-type".
4803
4804
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004805hash-type <method> <function> <modifier>
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004806 Specify a method to use for mapping hashes to servers
4807 May be used in sections : defaults | frontend | listen | backend
4808 yes | no | yes | yes
4809 Arguments :
Bhaskar98634f02013-10-29 23:30:51 -04004810 <method> is the method used to select a server from the hash computed by
4811 the <function> :
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004812
Bhaskar98634f02013-10-29 23:30:51 -04004813 map-based the hash table is a static array containing all alive servers.
4814 The hashes will be very smooth, will consider weights, but
4815 will be static in that weight changes while a server is up
4816 will be ignored. This means that there will be no slow start.
4817 Also, since a server is selected by its position in the array,
4818 most mappings are changed when the server count changes. This
4819 means that when a server goes up or down, or when a server is
4820 added to a farm, most connections will be redistributed to
4821 different servers. This can be inconvenient with caches for
4822 instance.
Willy Tarreau798a39c2010-11-24 15:04:29 +01004823
Bhaskar98634f02013-10-29 23:30:51 -04004824 consistent the hash table is a tree filled with many occurrences of each
4825 server. The hash key is looked up in the tree and the closest
4826 server is chosen. This hash is dynamic, it supports changing
4827 weights while the servers are up, so it is compatible with the
4828 slow start feature. It has the advantage that when a server
4829 goes up or down, only its associations are moved. When a
4830 server is added to the farm, only a few part of the mappings
4831 are redistributed, making it an ideal method for caches.
4832 However, due to its principle, the distribution will never be
4833 very smooth and it may sometimes be necessary to adjust a
4834 server's weight or its ID to get a more balanced distribution.
4835 In order to get the same distribution on multiple load
4836 balancers, it is important that all servers have the exact
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004837 same IDs. Note: consistent hash uses sdbm and avalanche if no
4838 hash function is specified.
Bhaskar98634f02013-10-29 23:30:51 -04004839
4840 <function> is the hash function to be used :
4841
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03004842 sdbm this function was created initially for sdbm (a public-domain
Bhaskar98634f02013-10-29 23:30:51 -04004843 reimplementation of ndbm) database library. It was found to do
4844 well in scrambling bits, causing better distribution of the keys
4845 and fewer splits. It also happens to be a good general hashing
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004846 function with good distribution, unless the total server weight
4847 is a multiple of 64, in which case applying the avalanche
4848 modifier may help.
Bhaskar98634f02013-10-29 23:30:51 -04004849
4850 djb2 this function was first proposed by Dan Bernstein many years ago
4851 on comp.lang.c. Studies have shown that for certain workload this
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004852 function provides a better distribution than sdbm. It generally
4853 works well with text-based inputs though it can perform extremely
4854 poorly with numeric-only input or when the total server weight is
4855 a multiple of 33, unless the avalanche modifier is also used.
4856
Willy Tarreaua0f42712013-11-14 14:30:35 +01004857 wt6 this function was designed for haproxy while testing other
4858 functions in the past. It is not as smooth as the other ones, but
4859 is much less sensible to the input data set or to the number of
4860 servers. It can make sense as an alternative to sdbm+avalanche or
4861 djb2+avalanche for consistent hashing or when hashing on numeric
4862 data such as a source IP address or a visitor identifier in a URL
4863 parameter.
4864
Willy Tarreau324f07f2015-01-20 19:44:50 +01004865 crc32 this is the most common CRC32 implementation as used in Ethernet,
4866 gzip, PNG, etc. It is slower than the other ones but may provide
4867 a better distribution or less predictable results especially when
4868 used on strings.
4869
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004870 <modifier> indicates an optional method applied after hashing the key :
4871
4872 avalanche This directive indicates that the result from the hash
4873 function above should not be used in its raw form but that
4874 a 4-byte full avalanche hash must be applied first. The
4875 purpose of this step is to mix the resulting bits from the
4876 previous hash in order to avoid any undesired effect when
4877 the input contains some limited values or when the number of
4878 servers is a multiple of one of the hash's components (64
4879 for SDBM, 33 for DJB2). Enabling avalanche tends to make the
4880 result less predictable, but it's also not as smooth as when
4881 using the original function. Some testing might be needed
4882 with some workloads. This hash is one of the many proposed
4883 by Bob Jenkins.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004884
Bhaskar98634f02013-10-29 23:30:51 -04004885 The default hash type is "map-based" and is recommended for most usages. The
4886 default function is "sdbm", the selection of a function should be based on
4887 the range of the values being hashed.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004888
Andrew Rodland17be45e2016-10-25 17:04:12 -04004889 See also : "balance", "hash-balance-factor", "server"
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004890
4891
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01004892http-after-response <action> <options...> [ { if | unless } <condition> ]
4893 Access control for all Layer 7 responses (server, applet/service and internal
4894 ones).
4895
4896 May be used in sections: defaults | frontend | listen | backend
4897 no | yes | yes | yes
4898
4899 The http-after-response statement defines a set of rules which apply to layer
4900 7 processing. The rules are evaluated in their declaration order when they
4901 are met in a frontend, listen or backend section. Any rule may optionally be
4902 followed by an ACL-based condition, in which case it will only be evaluated
4903 if the condition is true. Since these rules apply on responses, the backend
4904 rules are applied first, followed by the frontend's rules.
4905
4906 Unlike http-response rules, these ones are applied on all responses, the
4907 server ones but also to all responses generated by HAProxy. These rules are
4908 evaluated at the end of the responses analysis, before the data forwarding.
4909
4910 The first keyword is the rule's action. The supported actions are described
4911 below.
4912
4913 There is no limit to the number of http-after-response statements per
4914 instance.
4915
4916 Example:
4917 http-after-response set-header Strict-Transport-Security "max-age=31536000"
4918 http-after-response set-header Cache-Control "no-store,no-cache,private"
4919 http-after-response set-header Pragma "no-cache"
4920
4921http-after-response add-header <name> <fmt> [ { if | unless } <condition> ]
4922
4923 This appends an HTTP header field whose name is specified in <name> and whose
4924 value is defined by <fmt> which follows the log-format rules (see Custom Log
4925 Format in section 8.2.4). This may be used to send a cookie to a client for
4926 example, or to pass some internal information.
4927 This rule is not final, so it is possible to add other similar rules.
4928 Note that header addition is performed immediately, so one rule might reuse
4929 the resulting header from a previous rule.
4930
4931http-after-response allow [ { if | unless } <condition> ]
4932
4933 This stops the evaluation of the rules and lets the response pass the check.
4934 No further "http-after-response" rules are evaluated.
4935
Maciej Zdeb36662462020-11-20 13:58:48 +00004936http-after-response del-header <name> [ -m <meth> ] [ { if | unless } <condition> ]
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01004937
Maciej Zdeb36662462020-11-20 13:58:48 +00004938 This removes all HTTP header fields whose name is specified in <name>. <meth>
4939 is the matching method, applied on the header name. Supported matching methods
4940 are "str" (exact match), "beg" (prefix match), "end" (suffix match), "sub"
4941 (substring match) and "reg" (regex match). If not specified, exact matching
4942 method is used.
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01004943
4944http-after-response replace-header <name> <regex-match> <replace-fmt>
4945 [ { if | unless } <condition> ]
4946
4947 This works like "http-response replace-header".
4948
4949 Example:
4950 http-after-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2
4951
4952 # applied to:
4953 Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT
4954
4955 # outputs:
4956 Set-Cookie: C=1;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT
4957
4958 # assuming the backend IP is 192.168.1.20.
4959
4960http-after-response replace-value <name> <regex-match> <replace-fmt>
4961 [ { if | unless } <condition> ]
4962
4963 This works like "http-response replace-value".
4964
4965 Example:
4966 http-after-response replace-value Cache-control ^public$ private
4967
4968 # applied to:
4969 Cache-Control: max-age=3600, public
4970
4971 # outputs:
4972 Cache-Control: max-age=3600, private
4973
4974http-after-response set-header <name> <fmt> [ { if | unless } <condition> ]
4975
4976 This does the same as "add-header" except that the header name is first
4977 removed if it existed. This is useful when passing security information to
4978 the server, where the header must not be manipulated by external users.
4979
4980http-after-response set-status <status> [reason <str>]
4981 [ { if | unless } <condition> ]
4982
4983 This replaces the response status code with <status> which must be an integer
4984 between 100 and 999. Optionally, a custom reason text can be provided defined
4985 by <str>, or the default reason for the specified code will be used as a
4986 fallback.
4987
4988 Example:
4989 # return "431 Request Header Fields Too Large"
4990 http-response set-status 431
4991 # return "503 Slow Down", custom reason
4992 http-response set-status 503 reason "Slow Down"
4993
4994http-after-response set-var(<var-name>) <expr> [ { if | unless } <condition> ]
4995
4996 This is used to set the contents of a variable. The variable is declared
4997 inline.
4998
4999 Arguments:
5000 <var-name> The name of the variable starts with an indication about its
5001 scope. The scopes allowed are:
5002 "proc" : the variable is shared with the whole process
5003 "sess" : the variable is shared with the whole session
5004 "txn" : the variable is shared with the transaction
5005 (request and response)
5006 "req" : the variable is shared only during request
5007 processing
5008 "res" : the variable is shared only during response
5009 processing
5010 This prefix is followed by a name. The separator is a '.'.
5011 The name may only contain characters 'a-z', 'A-Z', '0-9', '.'
5012 and '_'.
5013
5014 <expr> Is a standard HAProxy expression formed by a sample-fetch
5015 followed by some converters.
5016
5017 Example:
5018 http-after-response set-var(sess.last_redir) res.hdr(location)
5019
5020http-after-response strict-mode { on | off }
5021
5022 This enables or disables the strict rewriting mode for following rules. It
5023 does not affect rules declared before it and it is only applicable on rules
5024 performing a rewrite on the responses. When the strict mode is enabled, any
5025 rewrite failure triggers an internal error. Otherwise, such errors are
5026 silently ignored. The purpose of the strict rewriting mode is to make some
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05005027 rewrites optional while others must be performed to continue the response
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01005028 processing.
5029
5030 By default, the strict rewriting mode is enabled. Its value is also reset
5031 when a ruleset evaluation ends. So, for instance, if you change the mode on
Daniel Corbett67a82712020-07-06 23:01:19 -04005032 the backend, the default mode is restored when HAProxy starts the frontend
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01005033 rules evaluation.
5034
5035http-after-response unset-var(<var-name>) [ { if | unless } <condition> ]
5036
5037 This is used to unset a variable. See "http-after-response set-var" for
5038 details about <var-name>.
5039
5040 Example:
5041 http-after-response unset-var(sess.last_redir)
5042
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005043
5044http-check comment <string>
5045 Defines a comment for the following the http-check rule, reported in logs if
5046 it fails.
5047 May be used in sections : defaults | frontend | listen | backend
5048 yes | no | yes | yes
5049
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005050 Arguments :
5051 <string> is the comment message to add in logs if the following http-check
5052 rule fails.
5053
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005054 It only works for connect, send and expect rules. It is useful to make
5055 user-friendly error reporting.
5056
Daniel Corbett67a82712020-07-06 23:01:19 -04005057 See also : "option httpchk", "http-check connect", "http-check send" and
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005058 "http-check expect".
5059
5060
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005061http-check connect [default] [port <expr>] [addr <ip>] [send-proxy]
5062 [via-socks4] [ssl] [sni <sni>] [alpn <alpn>] [linger]
Christopher Fauletedc6ed92020-04-23 16:27:59 +02005063 [proto <name>] [comment <msg>]
Christopher Faulete5870d82020-04-15 11:32:03 +02005064 Opens a new connection to perform an HTTP health check
5065 May be used in sections : defaults | frontend | listen | backend
5066 yes | no | yes | yes
5067
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005068 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005069 comment <msg> defines a message to report if the rule evaluation fails.
5070
Christopher Faulete5870d82020-04-15 11:32:03 +02005071 default Use default options of the server line to do the health
Daniel Corbett67a82712020-07-06 23:01:19 -04005072 checks. The server options are used only if not redefined.
Christopher Faulete5870d82020-04-15 11:32:03 +02005073
5074 port <expr> if not set, check port or server port is used.
5075 It tells HAProxy where to open the connection to.
5076 <port> must be a valid TCP port source integer, from 1 to
5077 65535 or an sample-fetch expression.
5078
5079 addr <ip> defines the IP address to do the health check.
5080
5081 send-proxy send a PROXY protocol string
5082
5083 via-socks4 enables outgoing health checks using upstream socks4 proxy.
5084
5085 ssl opens a ciphered connection
5086
5087 sni <sni> specifies the SNI to use to do health checks over SSL.
5088
5089 alpn <alpn> defines which protocols to advertise with ALPN. The protocol
5090 list consists in a comma-delimited list of protocol names,
5091 for instance: "h2,http/1.1". If it is not set, the server ALPN
5092 is used.
5093
Christopher Fauletedc6ed92020-04-23 16:27:59 +02005094 proto <name> forces the multiplexer's protocol to use for this connection.
5095 It must be an HTTP mux protocol and it must be usable on the
5096 backend side. The list of available protocols is reported in
5097 haproxy -vv.
5098
Christopher Faulete5870d82020-04-15 11:32:03 +02005099 linger cleanly close the connection instead of using a single RST.
5100
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005101 Just like tcp-check health checks, it is possible to configure the connection
5102 to use to perform HTTP health check. This directive should also be used to
5103 describe a scenario involving several request/response exchanges, possibly on
5104 different ports or with different servers.
5105
5106 When there are no TCP port configured on the server line neither server port
5107 directive, then the first step of the http-check sequence must be to specify
5108 the port with a "http-check connect".
5109
5110 In an http-check ruleset a 'connect' is required, it is also mandatory to start
5111 the ruleset with a 'connect' rule. Purpose is to ensure admin know what they
5112 do.
5113
5114 When a connect must start the ruleset, if may still be preceded by set-var,
5115 unset-var or comment rules.
5116
5117 Examples :
Christopher Faulete5870d82020-04-15 11:32:03 +02005118 # check HTTP and HTTPs services on a server.
5119 # first open port 80 thanks to server line port directive, then
5120 # tcp-check opens port 443, ciphered and run a request on it:
5121 option httpchk
5122
5123 http-check connect
Christopher Fauleta5c14ef2020-04-29 14:19:13 +02005124 http-check send meth GET uri / ver HTTP/1.1 hdr host haproxy.1wt.eu
Christopher Faulet8021a5f2020-04-24 13:53:12 +02005125 http-check expect status 200-399
Christopher Faulete5870d82020-04-15 11:32:03 +02005126 http-check connect port 443 ssl sni haproxy.1wt.eu
Christopher Fauleta5c14ef2020-04-29 14:19:13 +02005127 http-check send meth GET uri / ver HTTP/1.1 hdr host haproxy.1wt.eu
Christopher Faulet8021a5f2020-04-24 13:53:12 +02005128 http-check expect status 200-399
Christopher Faulete5870d82020-04-15 11:32:03 +02005129
5130 server www 10.0.0.1 check port 80
5131
5132 See also : "option httpchk", "http-check send", "http-check expect"
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01005133
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005134
Willy Tarreau0ba27502007-12-24 16:55:16 +01005135http-check disable-on-404
5136 Enable a maintenance mode upon HTTP/404 response to health-checks
5137 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau2769aa02007-12-27 18:26:09 +01005138 yes | no | yes | yes
Willy Tarreau0ba27502007-12-24 16:55:16 +01005139 Arguments : none
5140
5141 When this option is set, a server which returns an HTTP code 404 will be
5142 excluded from further load-balancing, but will still receive persistent
5143 connections. This provides a very convenient method for Web administrators
5144 to perform a graceful shutdown of their servers. It is also important to note
5145 that a server which is detected as failed while it was in this mode will not
5146 generate an alert, just a notice. If the server responds 2xx or 3xx again, it
5147 will immediately be reinserted into the farm. The status on the stats page
5148 reports "NOLB" for a server in this mode. It is important to note that this
Willy Tarreaubd741542010-03-16 18:46:54 +01005149 option only works in conjunction with the "httpchk" option. If this option
5150 is used with "http-check expect", then it has precedence over it so that 404
5151 responses will still be considered as soft-stop.
5152
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005153 See also : "option httpchk" and "http-check expect".
Willy Tarreaubd741542010-03-16 18:46:54 +01005154
5155
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005156http-check expect [min-recv <int>] [comment <msg>]
Christopher Faulete5870d82020-04-15 11:32:03 +02005157 [ok-status <st>] [error-status <st>] [tout-status <st>]
5158 [on-success <fmt>] [on-error <fmt>] [status-code <expr>]
5159 [!] <match> <pattern>
Jamie Gloudonaaa21002012-08-25 00:18:33 -04005160 Make HTTP health checks consider response contents or specific status codes
Willy Tarreaubd741542010-03-16 18:46:54 +01005161 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau1ee51a62011-08-19 20:04:17 +02005162 yes | no | yes | yes
Christopher Faulete5870d82020-04-15 11:32:03 +02005163
Willy Tarreaubd741542010-03-16 18:46:54 +01005164 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005165 comment <msg> defines a message to report if the rule evaluation fails.
5166
Christopher Faulete5870d82020-04-15 11:32:03 +02005167 min-recv is optional and can define the minimum amount of data required to
5168 evaluate the current expect rule. If the number of received bytes
5169 is under this limit, the check will wait for more data. This
5170 option can be used to resolve some ambiguous matching rules or to
5171 avoid executing costly regex matches on content known to be still
5172 incomplete. If an exact string is used, the minimum between the
5173 string length and this parameter is used. This parameter is
5174 ignored if it is set to -1. If the expect rule does not match,
5175 the check will wait for more data. If set to 0, the evaluation
5176 result is always conclusive.
5177
5178 ok-status <st> is optional and can be used to set the check status if
5179 the expect rule is successfully evaluated and if it is
5180 the last rule in the tcp-check ruleset. "L7OK", "L7OKC",
Christopher Fauletd888f0f2020-05-07 07:40:17 +02005181 "L6OK" and "L4OK" are supported :
5182 - L7OK : check passed on layer 7
5183 - L7OKC : check conditionally passed on layer 7, for
5184 example 404 with disable-on-404
5185 - L6OK : check passed on layer 6
5186 - L4OK : check passed on layer 4
5187 By default "L7OK" is used.
Christopher Faulete5870d82020-04-15 11:32:03 +02005188
5189 error-status <st> is optional and can be used to set the check status if
5190 an error occurred during the expect rule evaluation.
Christopher Fauletd888f0f2020-05-07 07:40:17 +02005191 "L7RSP", "L7STS", "L6RSP" and "L4CON" are supported :
5192 - L7RSP : layer 7 invalid response - protocol error
5193 - L7STS : layer 7 response error, for example HTTP 5xx
5194 - L6RSP : layer 6 invalid response - protocol error
5195 - L4CON : layer 1-4 connection problem
5196 By default "L7RSP" is used.
Christopher Faulete5870d82020-04-15 11:32:03 +02005197
5198 tout-status <st> is optional and can be used to set the check status if
5199 a timeout occurred during the expect rule evaluation.
Christopher Fauletd888f0f2020-05-07 07:40:17 +02005200 "L7TOUT", "L6TOUT", and "L4TOUT" are supported :
5201 - L7TOUT : layer 7 (HTTP/SMTP) timeout
5202 - L6TOUT : layer 6 (SSL) timeout
5203 - L4TOUT : layer 1-4 timeout
Christopher Faulete5870d82020-04-15 11:32:03 +02005204 By default "L7TOUT" is used.
5205
5206 on-success <fmt> is optional and can be used to customize the
5207 informational message reported in logs if the expect
5208 rule is successfully evaluated and if it is the last rule
5209 in the tcp-check ruleset. <fmt> is a log-format string.
5210
5211 on-error <fmt> is optional and can be used to customize the
5212 informational message reported in logs if an error
5213 occurred during the expect rule evaluation. <fmt> is a
5214 log-format string.
5215
Willy Tarreaubd741542010-03-16 18:46:54 +01005216 <match> is a keyword indicating how to look for a specific pattern in the
Christopher Fauletb5594262020-05-05 20:23:13 +02005217 response. The keyword may be one of "status", "rstatus", "hdr",
5218 "fhdr", "string", or "rstring". The keyword may be preceded by an
Willy Tarreaubd741542010-03-16 18:46:54 +01005219 exclamation mark ("!") to negate the match. Spaces are allowed
5220 between the exclamation mark and the keyword. See below for more
5221 details on the supported keywords.
5222
Christopher Faulet39708192020-05-05 10:47:36 +02005223 <pattern> is the pattern to look for. It may be a string, a regular
5224 expression or a more complex pattern with several arguments. If
5225 the string pattern contains spaces, they must be escaped with the
5226 usual backslash ('\').
Willy Tarreaubd741542010-03-16 18:46:54 +01005227
5228 By default, "option httpchk" considers that response statuses 2xx and 3xx
5229 are valid, and that others are invalid. When "http-check expect" is used,
5230 it defines what is considered valid or invalid. Only one "http-check"
5231 statement is supported in a backend. If a server fails to respond or times
5232 out, the check obviously fails. The available matches are :
5233
Christopher Faulet8021a5f2020-04-24 13:53:12 +02005234 status <codes> : test the status codes found parsing <codes> string. it
5235 must be a comma-separated list of status codes or range
5236 codes. A health check response will be considered as
5237 valid if the response's status code matches any status
5238 code or is inside any range of the list. If the "status"
5239 keyword is prefixed with "!", then the response will be
5240 considered invalid if the status code matches.
Willy Tarreaubd741542010-03-16 18:46:54 +01005241
5242 rstatus <regex> : test a regular expression for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04005243 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01005244 response's status code matches the expression. If the
5245 "rstatus" keyword is prefixed with "!", then the response
5246 will be considered invalid if the status code matches.
5247 This is mostly used to check for multiple codes.
5248
Christopher Fauletb5594262020-05-05 20:23:13 +02005249 hdr { name | name-lf } [ -m <meth> ] <name>
5250 [ { value | value-lf } [ -m <meth> ] <value> :
Christopher Faulet39708192020-05-05 10:47:36 +02005251 test the specified header pattern on the HTTP response
5252 headers. The name pattern is mandatory but the value
5253 pattern is optional. If not specified, only the header
5254 presence is verified. <meth> is the matching method,
5255 applied on the header name or the header value. Supported
5256 matching methods are "str" (exact match), "beg" (prefix
5257 match), "end" (suffix match), "sub" (substring match) or
5258 "reg" (regex match). If not specified, exact matching
Christopher Fauletb5594262020-05-05 20:23:13 +02005259 method is used. If the "name-lf" parameter is used,
5260 <name> is evaluated as a log-format string. If "value-lf"
5261 parameter is used, <value> is evaluated as a log-format
5262 string. These parameters cannot be used with the regex
5263 matching method. Finally, the header value is considered
5264 as comma-separated list. Note that matchings are case
5265 insensitive on the header names.
5266
5267 fhdr { name | name-lf } [ -m <meth> ] <name>
5268 [ { value | value-lf } [ -m <meth> ] <value> :
5269 test the specified full header pattern on the HTTP
5270 response headers. It does exactly the same than "hdr"
5271 keyword, except the full header value is tested, commas
5272 are not considered as delimiters.
Christopher Faulet39708192020-05-05 10:47:36 +02005273
Willy Tarreaubd741542010-03-16 18:46:54 +01005274 string <string> : test the exact string match in the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04005275 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01005276 response's body contains this exact string. If the
5277 "string" keyword is prefixed with "!", then the response
5278 will be considered invalid if the body contains this
5279 string. This can be used to look for a mandatory word at
5280 the end of a dynamic page, or to detect a failure when a
Davor Ocelice9ed2812017-12-25 17:49:28 +01005281 specific error appears on the check page (e.g. a stack
Willy Tarreaubd741542010-03-16 18:46:54 +01005282 trace).
5283
5284 rstring <regex> : test a regular expression on the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04005285 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01005286 response's body matches this expression. If the "rstring"
5287 keyword is prefixed with "!", then the response will be
5288 considered invalid if the body matches the expression.
5289 This can be used to look for a mandatory word at the end
5290 of a dynamic page, or to detect a failure when a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01005291 error appears on the check page (e.g. a stack trace).
Willy Tarreaubd741542010-03-16 18:46:54 +01005292
Christopher Fauletaaab0832020-05-05 15:54:22 +02005293 string-lf <fmt> : test a log-format string match in the HTTP response body.
5294 A health check response will be considered valid if the
5295 response's body contains the string resulting of the
5296 evaluation of <fmt>, which follows the log-format rules.
5297 If prefixed with "!", then the response will be
5298 considered invalid if the body contains the string.
5299
Willy Tarreaubd741542010-03-16 18:46:54 +01005300 It is important to note that the responses will be limited to a certain size
Christopher Faulet7151a122020-11-25 17:20:57 +01005301 defined by the global "tune.bufsize" option, which defaults to 16384 bytes.
Willy Tarreaubd741542010-03-16 18:46:54 +01005302 Thus, too large responses may not contain the mandatory pattern when using
5303 "string" or "rstring". If a large response is absolutely required, it is
5304 possible to change the default max size by setting the global variable.
5305 However, it is worth keeping in mind that parsing very large responses can
5306 waste some CPU cycles, especially when regular expressions are used, and that
5307 it is always better to focus the checks on smaller resources.
5308
Christopher Faulete5870d82020-04-15 11:32:03 +02005309 In an http-check ruleset, the last expect rule may be implicit. If no expect
5310 rule is specified after the last "http-check send", an implicit expect rule
5311 is defined to match on 2xx or 3xx status codes. It means this rule is also
5312 defined if there is no "http-check" rule at all, when only "option httpchk"
5313 is set.
Cyril Bonté32602d22015-01-30 00:07:07 +01005314
Willy Tarreaubd741542010-03-16 18:46:54 +01005315 Last, if "http-check expect" is combined with "http-check disable-on-404",
5316 then this last one has precedence when the server responds with 404.
5317
5318 Examples :
5319 # only accept status 200 as valid
Christopher Faulet8021a5f2020-04-24 13:53:12 +02005320 http-check expect status 200,201,300-310
Willy Tarreaubd741542010-03-16 18:46:54 +01005321
Christopher Faulet39708192020-05-05 10:47:36 +02005322 # be sure a sessid coookie is set
5323 http-check expect header name "set-cookie" value -m beg "sessid="
5324
Willy Tarreaubd741542010-03-16 18:46:54 +01005325 # consider SQL errors as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01005326 http-check expect ! string SQL\ Error
Willy Tarreaubd741542010-03-16 18:46:54 +01005327
5328 # consider status 5xx only as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01005329 http-check expect ! rstatus ^5
Willy Tarreaubd741542010-03-16 18:46:54 +01005330
5331 # check that we have a correct hexadecimal tag before /html
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03005332 http-check expect rstring <!--tag:[0-9a-f]*--></html>
Willy Tarreau0ba27502007-12-24 16:55:16 +01005333
Christopher Faulete5870d82020-04-15 11:32:03 +02005334 See also : "option httpchk", "http-check connect", "http-check disable-on-404"
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005335 and "http-check send".
Willy Tarreau2769aa02007-12-27 18:26:09 +01005336
5337
Christopher Faulet7c95f5f2020-05-06 15:06:34 +02005338http-check send [meth <method>] [{ uri <uri> | uri-lf <fmt> }>] [ver <version>]
Christopher Faulet574e7bd2020-05-06 15:38:58 +02005339 [hdr <name> <fmt>]* [{ body <string> | body-lf <fmt> }]
5340 [comment <msg>]
Christopher Faulet8acb1282020-04-09 08:44:06 +02005341 Add a possible list of headers and/or a body to the request sent during HTTP
5342 health checks.
5343 May be used in sections : defaults | frontend | listen | backend
5344 yes | no | yes | yes
5345 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005346 comment <msg> defines a message to report if the rule evaluation fails.
5347
Christopher Faulete5870d82020-04-15 11:32:03 +02005348 meth <method> is the optional HTTP method used with the requests. When not
5349 set, the "OPTIONS" method is used, as it generally requires
5350 low server processing and is easy to filter out from the
5351 logs. Any method may be used, though it is not recommended
5352 to invent non-standard ones.
5353
Christopher Faulet7c95f5f2020-05-06 15:06:34 +02005354 uri <uri> is optional and set the URI referenced in the HTTP requests
5355 to the string <uri>. It defaults to "/" which is accessible
5356 by default on almost any server, but may be changed to any
5357 other URI. Query strings are permitted.
5358
5359 uri-lf <fmt> is optional and set the URI referenced in the HTTP requests
5360 using the log-format string <fmt>. It defaults to "/" which
5361 is accessible by default on almost any server, but may be
5362 changed to any other URI. Query strings are permitted.
Christopher Faulet8acb1282020-04-09 08:44:06 +02005363
Christopher Faulet907701b2020-04-28 09:37:00 +02005364 ver <version> is the optional HTTP version string. It defaults to
Christopher Faulete5870d82020-04-15 11:32:03 +02005365 "HTTP/1.0" but some servers might behave incorrectly in HTTP
Daniel Corbett67a82712020-07-06 23:01:19 -04005366 1.0, so turning it to HTTP/1.1 may sometimes help. Note that
Christopher Faulete5870d82020-04-15 11:32:03 +02005367 the Host field is mandatory in HTTP/1.1, use "hdr" argument
5368 to add it.
5369
5370 hdr <name> <fmt> adds the HTTP header field whose name is specified in
5371 <name> and whose value is defined by <fmt>, which follows
5372 to the log-format rules.
5373
5374 body <string> add the body defined by <string> to the request sent during
5375 HTTP health checks. If defined, the "Content-Length" header
5376 is thus automatically added to the request.
Christopher Faulet8acb1282020-04-09 08:44:06 +02005377
Christopher Faulet574e7bd2020-05-06 15:38:58 +02005378 body-lf <fmt> add the body defined by the log-format string <fmt> to the
5379 request sent during HTTP health checks. If defined, the
5380 "Content-Length" header is thus automatically added to the
5381 request.
5382
Christopher Faulet8acb1282020-04-09 08:44:06 +02005383 In addition to the request line defined by the "option httpchk" directive,
5384 this one is the valid way to add some headers and optionally a body to the
5385 request sent during HTTP health checks. If a body is defined, the associate
Christopher Faulet9df910c2020-04-29 14:20:47 +02005386 "Content-Length" header is automatically added. Thus, this header or
5387 "Transfer-encoding" header should not be present in the request provided by
5388 "http-check send". If so, it will be ignored. The old trick consisting to add
5389 headers after the version string on the "option httpchk" line is now
Amaury Denoyelleaea63022020-12-22 14:08:52 +01005390 deprecated.
Christopher Faulet8acb1282020-04-09 08:44:06 +02005391
Christopher Faulete5870d82020-04-15 11:32:03 +02005392 Also "http-check send" doesn't support HTTP keep-alive. Keep in mind that it
Amaury Denoyelleaea63022020-12-22 14:08:52 +01005393 will automatically append a "Connection: close" header, unless a Connection
5394 header has already already been configured via a hdr entry.
Christopher Faulet9df910c2020-04-29 14:20:47 +02005395
5396 Note that the Host header and the request authority, when both defined, are
5397 automatically synchronized. It means when the HTTP request is sent, when a
5398 Host is inserted in the request, the request authority is accordingly
5399 updated. Thus, don't be surprised if the Host header value overwrites the
5400 configured request authority.
5401
5402 Note also for now, no Host header is automatically added in HTTP/1.1 or above
5403 requests. You should add it explicitly.
Christopher Faulete5870d82020-04-15 11:32:03 +02005404
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005405 See also : "option httpchk", "http-check send-state" and "http-check expect".
Christopher Faulet8acb1282020-04-09 08:44:06 +02005406
5407
Willy Tarreauef781042010-01-27 11:53:01 +01005408http-check send-state
5409 Enable emission of a state header with HTTP health checks
5410 May be used in sections : defaults | frontend | listen | backend
5411 yes | no | yes | yes
5412 Arguments : none
5413
5414 When this option is set, haproxy will systematically send a special header
5415 "X-Haproxy-Server-State" with a list of parameters indicating to each server
5416 how they are seen by haproxy. This can be used for instance when a server is
5417 manipulated without access to haproxy and the operator needs to know whether
5418 haproxy still sees it up or not, or if the server is the last one in a farm.
5419
5420 The header is composed of fields delimited by semi-colons, the first of which
5421 is a word ("UP", "DOWN", "NOLB"), possibly followed by a number of valid
5422 checks on the total number before transition, just as appears in the stats
5423 interface. Next headers are in the form "<variable>=<value>", indicating in
5424 no specific order some values available in the stats interface :
Joseph Lynch514061c2015-01-15 17:52:59 -08005425 - a variable "address", containing the address of the backend server.
5426 This corresponds to the <address> field in the server declaration. For
5427 unix domain sockets, it will read "unix".
5428
5429 - a variable "port", containing the port of the backend server. This
5430 corresponds to the <port> field in the server declaration. For unix
5431 domain sockets, it will read "unix".
5432
Willy Tarreauef781042010-01-27 11:53:01 +01005433 - a variable "name", containing the name of the backend followed by a slash
5434 ("/") then the name of the server. This can be used when a server is
5435 checked in multiple backends.
5436
5437 - a variable "node" containing the name of the haproxy node, as set in the
5438 global "node" variable, otherwise the system's hostname if unspecified.
5439
5440 - a variable "weight" indicating the weight of the server, a slash ("/")
5441 and the total weight of the farm (just counting usable servers). This
5442 helps to know if other servers are available to handle the load when this
5443 one fails.
5444
5445 - a variable "scur" indicating the current number of concurrent connections
5446 on the server, followed by a slash ("/") then the total number of
5447 connections on all servers of the same backend.
5448
5449 - a variable "qcur" indicating the current number of requests in the
5450 server's queue.
5451
5452 Example of a header received by the application server :
5453 >>> X-Haproxy-Server-State: UP 2/3; name=bck/srv2; node=lb1; weight=1/2; \
5454 scur=13/22; qcur=0
5455
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005456 See also : "option httpchk", "http-check disable-on-404" and
5457 "http-check send".
Willy Tarreauef781042010-01-27 11:53:01 +01005458
Christopher Faulete5870d82020-04-15 11:32:03 +02005459
5460http-check set-var(<var-name>) <expr>
Christopher Faulete5870d82020-04-15 11:32:03 +02005461 This operation sets the content of a variable. The variable is declared inline.
Christopher Faulete5870d82020-04-15 11:32:03 +02005462 May be used in sections: defaults | frontend | listen | backend
5463 yes | no | yes | yes
5464
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005465 Arguments :
Christopher Faulete5870d82020-04-15 11:32:03 +02005466 <var-name> The name of the variable starts with an indication about its
5467 scope. The scopes allowed for http-check are:
5468 "proc" : the variable is shared with the whole process.
5469 "sess" : the variable is shared with the tcp-check session.
5470 "check": the variable is declared for the lifetime of the tcp-check.
5471 This prefix is followed by a name. The separator is a '.'.
5472 The name may only contain characters 'a-z', 'A-Z', '0-9', '.',
5473 and '-'.
5474
5475 <expr> Is a sample-fetch expression potentially followed by converters.
5476
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005477 Examples :
5478 http-check set-var(check.port) int(1234)
Christopher Faulete5870d82020-04-15 11:32:03 +02005479
5480
5481http-check unset-var(<var-name>)
Christopher Faulete5870d82020-04-15 11:32:03 +02005482 Free a reference to a variable within its scope.
Christopher Faulete5870d82020-04-15 11:32:03 +02005483 May be used in sections: defaults | frontend | listen | backend
5484 yes | no | yes | yes
5485
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005486 Arguments :
Christopher Faulete5870d82020-04-15 11:32:03 +02005487 <var-name> The name of the variable starts with an indication about its
5488 scope. The scopes allowed for http-check are:
5489 "proc" : the variable is shared with the whole process.
5490 "sess" : the variable is shared with the tcp-check session.
5491 "check": the variable is declared for the lifetime of the tcp-check.
5492 This prefix is followed by a name. The separator is a '.'.
5493 The name may only contain characters 'a-z', 'A-Z', '0-9', '.',
5494 and '-'.
5495
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005496 Examples :
5497 http-check unset-var(check.port)
Christopher Faulete5870d82020-04-15 11:32:03 +02005498
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005499
Christopher Faulet3b967c12020-05-15 15:47:44 +02005500http-error status <code> [content-type <type>]
5501 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
5502 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
5503 [ hdr <name> <fmt> ]*
5504 Defines a custom error message to use instead of errors generated by HAProxy.
5505 May be used in sections : defaults | frontend | listen | backend
5506 yes | yes | yes | yes
5507 Arguments :
Ilya Shipitsin11057a32020-06-21 21:18:27 +05005508 status <code> is the HTTP status code. It must be specified.
Christopher Faulet3b967c12020-05-15 15:47:44 +02005509 Currently, HAProxy is capable of generating codes
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02005510 200, 400, 401, 403, 404, 405, 407, 408, 410, 413, 425,
5511 429, 500, 502, 503, and 504.
Christopher Faulet3b967c12020-05-15 15:47:44 +02005512
5513 content-type <type> is the response content type, for instance
5514 "text/plain". This parameter is ignored and should be
5515 omitted when an errorfile is configured or when the
5516 payload is empty. Otherwise, it must be defined.
5517
5518 default-errorfiles Reset the previously defined error message for current
5519 proxy for the status <code>. If used on a backend, the
5520 frontend error message is used, if defined. If used on
5521 a frontend, the default error message is used.
5522
5523 errorfile <file> designates a file containing the full HTTP response.
5524 It is recommended to follow the common practice of
5525 appending ".http" to the filename so that people do
5526 not confuse the response with HTML error pages, and to
5527 use absolute paths, since files are read before any
5528 chroot is performed.
5529
5530 errorfiles <name> designates the http-errors section to use to import
5531 the error message with the status code <code>. If no
5532 such message is found, the proxy's error messages are
5533 considered.
5534
5535 file <file> specifies the file to use as response payload. If the
5536 file is not empty, its content-type must be set as
5537 argument to "content-type", otherwise, any
5538 "content-type" argument is ignored. <file> is
5539 considered as a raw string.
5540
5541 string <str> specifies the raw string to use as response payload.
5542 The content-type must always be set as argument to
5543 "content-type".
5544
5545 lf-file <file> specifies the file to use as response payload. If the
5546 file is not empty, its content-type must be set as
5547 argument to "content-type", otherwise, any
5548 "content-type" argument is ignored. <file> is
5549 evaluated as a log-format string.
5550
5551 lf-string <str> specifies the log-format string to use as response
5552 payload. The content-type must always be set as
5553 argument to "content-type".
5554
5555 hdr <name> <fmt> adds to the response the HTTP header field whose name
5556 is specified in <name> and whose value is defined by
5557 <fmt>, which follows to the log-format rules.
5558 This parameter is ignored if an errorfile is used.
5559
5560 This directive may be used instead of "errorfile", to define a custom error
5561 message. As "errorfile" directive, it is used for errors detected and
5562 returned by HAProxy. If an errorfile is defined, it is parsed when HAProxy
5563 starts and must be valid according to the HTTP standards. The generated
5564 response must not exceed the configured buffer size (BUFFSIZE), otherwise an
5565 internal error will be returned. Finally, if you consider to use some
5566 http-after-response rules to rewrite these errors, the reserved buffer space
5567 should be available (see "tune.maxrewrite").
5568
5569 The files are read at the same time as the configuration and kept in memory.
5570 For this reason, the errors continue to be returned even when the process is
5571 chrooted, and no file change is considered while the process is running.
5572
5573 See also : "errorfile", "errorfiles", "errorloc", "errorloc302",
5574 "errorloc303" and section 3.8 about http-errors.
5575
5576
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005577http-request <action> [options...] [ { if | unless } <condition> ]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01005578 Access control for Layer 7 requests
5579
5580 May be used in sections: defaults | frontend | listen | backend
5581 no | yes | yes | yes
5582
Willy Tarreau20b0de52012-12-24 15:45:22 +01005583 The http-request statement defines a set of rules which apply to layer 7
5584 processing. The rules are evaluated in their declaration order when they are
5585 met in a frontend, listen or backend section. Any rule may optionally be
5586 followed by an ACL-based condition, in which case it will only be evaluated
5587 if the condition is true.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01005588
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005589 The first keyword is the rule's action. The supported actions are described
5590 below.
Willy Tarreau20b0de52012-12-24 15:45:22 +01005591
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005592 There is no limit to the number of http-request statements per instance.
Willy Tarreau20b0de52012-12-24 15:45:22 +01005593
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005594 Example:
5595 acl nagios src 192.168.129.3
5596 acl local_net src 192.168.0.0/16
5597 acl auth_ok http_auth(L1)
Willy Tarreau20b0de52012-12-24 15:45:22 +01005598
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005599 http-request allow if nagios
5600 http-request allow if local_net auth_ok
5601 http-request auth realm Gimme if local_net auth_ok
5602 http-request deny
Willy Tarreau81499eb2012-12-27 12:19:02 +01005603
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005604 Example:
5605 acl key req.hdr(X-Add-Acl-Key) -m found
5606 acl add path /addacl
5607 acl del path /delacl
Willy Tarreau20b0de52012-12-24 15:45:22 +01005608
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005609 acl myhost hdr(Host) -f myhost.lst
Willy Tarreau20b0de52012-12-24 15:45:22 +01005610
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005611 http-request add-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key add
5612 http-request del-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key del
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02005613
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005614 Example:
5615 acl value req.hdr(X-Value) -m found
5616 acl setmap path /setmap
5617 acl delmap path /delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06005618
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005619 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06005620
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005621 http-request set-map(map.lst) %[src] %[req.hdr(X-Value)] if setmap value
5622 http-request del-map(map.lst) %[src] if delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06005623
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005624 See also : "stats http-request", section 3.4 about userlists and section 7
5625 about ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06005626
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005627http-request add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005628
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005629 This is used to add a new entry into an ACL. The ACL must be loaded from a
5630 file (even a dummy empty file). The file name of the ACL to be updated is
5631 passed between parentheses. It takes one argument: <key fmt>, which follows
5632 log-format rules, to collect content of the new entry. It performs a lookup
5633 in the ACL before insertion, to avoid duplicated (or more) values. This
5634 lookup is done by a linear search and can be expensive with large lists!
5635 It is the equivalent of the "add acl" command from the stats socket, but can
5636 be triggered by an HTTP request.
Sasha Pachev218f0642014-06-16 12:05:59 -06005637
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005638http-request add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005639
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005640 This appends an HTTP header field whose name is specified in <name> and
5641 whose value is defined by <fmt> which follows the log-format rules (see
5642 Custom Log Format in section 8.2.4). This is particularly useful to pass
5643 connection-specific information to the server (e.g. the client's SSL
5644 certificate), or to combine several headers into one. This rule is not
5645 final, so it is possible to add other similar rules. Note that header
5646 addition is performed immediately, so one rule might reuse the resulting
5647 header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06005648
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005649http-request allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005650
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005651 This stops the evaluation of the rules and lets the request pass the check.
5652 No further "http-request" rules are evaluated.
Sasha Pachev218f0642014-06-16 12:05:59 -06005653
Sasha Pachev218f0642014-06-16 12:05:59 -06005654
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005655http-request auth [realm <realm>] [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005656
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005657 This stops the evaluation of the rules and immediately responds with an
5658 HTTP 401 or 407 error code to invite the user to present a valid user name
5659 and password. No further "http-request" rules are evaluated. An optional
5660 "realm" parameter is supported, it sets the authentication realm that is
5661 returned with the response (typically the application's name).
Sasha Pachev218f0642014-06-16 12:05:59 -06005662
Christopher Faulet612f2ea2020-05-27 09:57:28 +02005663 The corresponding proxy's error message is used. It may be customized using
5664 an "errorfile" or an "http-error" directive. For 401 responses, all
5665 occurrences of the WWW-Authenticate header are removed and replaced by a new
5666 one with a basic authentication challenge for realm "<realm>". For 407
5667 responses, the same is done on the Proxy-Authenticate header. If the error
5668 message must not be altered, consider to use "http-request return" rule
5669 instead.
5670
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005671 Example:
5672 acl auth_ok http_auth_group(L1) G1
5673 http-request auth unless auth_ok
Sasha Pachev218f0642014-06-16 12:05:59 -06005674
Jarno Huuskonen251a6b72019-01-04 14:05:02 +02005675http-request cache-use <name> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005676
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02005677 See section 6.2 about cache setup.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005678
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005679http-request capture <sample> [ len <length> | id <id> ]
5680 [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005681
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005682 This captures sample expression <sample> from the request buffer, and
5683 converts it to a string of at most <len> characters. The resulting string is
5684 stored into the next request "capture" slot, so it will possibly appear next
5685 to some captured HTTP headers. It will then automatically appear in the logs,
5686 and it will be possible to extract it using sample fetch rules to feed it
5687 into headers or anything. The length should be limited given that this size
5688 will be allocated for each capture during the whole session life.
5689 Please check section 7.3 (Fetching samples) and "capture request header" for
5690 more information.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005691
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005692 If the keyword "id" is used instead of "len", the action tries to store the
5693 captured string in a previously declared capture slot. This is useful to run
5694 captures in backends. The slot id can be declared by a previous directive
Baptiste Assmann19a69b32020-01-16 14:34:22 +01005695 "http-request capture" or with the "declare capture" keyword.
5696
5697 When using this action in a backend, double check that the relevant
5698 frontend(s) have the required capture slots otherwise, this rule will be
5699 ignored at run time. This can't be detected at configuration parsing time
5700 due to HAProxy's ability to dynamically resolve backend name at runtime.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005701
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005702http-request del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005703
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005704 This is used to delete an entry from an ACL. The ACL must be loaded from a
5705 file (even a dummy empty file). The file name of the ACL to be updated is
5706 passed between parentheses. It takes one argument: <key fmt>, which follows
5707 log-format rules, to collect content of the entry to delete.
5708 It is the equivalent of the "del acl" command from the stats socket, but can
5709 be triggered by an HTTP request.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005710
Maciej Zdeb36662462020-11-20 13:58:48 +00005711http-request del-header <name> [ -m <meth> ] [ { if | unless } <condition> ]
Willy Tarreauf4c43c12013-06-11 17:01:13 +02005712
Maciej Zdeb36662462020-11-20 13:58:48 +00005713 This removes all HTTP header fields whose name is specified in <name>. <meth>
5714 is the matching method, applied on the header name. Supported matching methods
5715 are "str" (exact match), "beg" (prefix match), "end" (suffix match), "sub"
5716 (substring match) and "reg" (regex match). If not specified, exact matching
5717 method is used.
Willy Tarreau9a355ec2013-06-11 17:45:46 +02005718
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005719http-request del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau42cf39e2013-06-11 18:51:32 +02005720
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005721 This is used to delete an entry from a MAP. The MAP must be loaded from a
5722 file (even a dummy empty file). The file name of the MAP to be updated is
5723 passed between parentheses. It takes one argument: <key fmt>, which follows
5724 log-format rules, to collect content of the entry to delete.
5725 It takes one argument: "file name" It is the equivalent of the "del map"
5726 command from the stats socket, but can be triggered by an HTTP request.
Willy Tarreau51347ed2013-06-11 19:34:13 +02005727
Christopher Faulet5cb513a2020-05-13 17:56:56 +02005728http-request deny [deny_status <status>] [ { if | unless } <condition> ]
5729http-request deny [ { status | deny_status } <code>] [content-type <type>]
5730 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
5731 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
5732 [ hdr <name> <fmt> ]*
5733 [ { if | unless } <condition> ]
Patrick Hemmer268a7072018-05-11 12:52:31 -04005734
Christopher Faulet5cb513a2020-05-13 17:56:56 +02005735 This stops the evaluation of the rules and immediately rejects the request.
5736 By default an HTTP 403 error is returned. But the response may be customized
5737 using same syntax than "http-request return" rules. Thus, see "http-request
Ilya Shipitsin11057a32020-06-21 21:18:27 +05005738 return" for details. For compatibility purpose, when no argument is defined,
Christopher Faulet5cb513a2020-05-13 17:56:56 +02005739 or only "deny_status", the argument "default-errorfiles" is implied. It means
5740 "http-request deny [deny_status <status>]" is an alias of
5741 "http-request deny [status <status>] default-errorfiles".
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005742 No further "http-request" rules are evaluated.
Christopher Faulet5cb513a2020-05-13 17:56:56 +02005743 See also "http-request return".
Patrick Hemmer268a7072018-05-11 12:52:31 -04005744
Olivier Houchard602bf7d2019-05-10 13:59:15 +02005745http-request disable-l7-retry [ { if | unless } <condition> ]
5746 This disables any attempt to retry the request if it fails for any other
5747 reason than a connection failure. This can be useful for example to make
5748 sure POST requests aren't retried on failure.
5749
Baptiste Assmann333939c2019-01-21 08:34:50 +01005750http-request do-resolve(<var>,<resolvers>,[ipv4,ipv6]) <expr> :
5751
5752 This action performs a DNS resolution of the output of <expr> and stores
5753 the result in the variable <var>. It uses the DNS resolvers section
5754 pointed by <resolvers>.
5755 It is possible to choose a resolution preference using the optional
5756 arguments 'ipv4' or 'ipv6'.
5757 When performing the DNS resolution, the client side connection is on
5758 pause waiting till the end of the resolution.
5759 If an IP address can be found, it is stored into <var>. If any kind of
5760 error occurs, then <var> is not set.
5761 One can use this action to discover a server IP address at run time and
5762 based on information found in the request (IE a Host header).
5763 If this action is used to find the server's IP address (using the
5764 "set-dst" action), then the server IP address in the backend must be set
5765 to 0.0.0.0.
5766
5767 Example:
5768 resolvers mydns
5769 nameserver local 127.0.0.53:53
5770 nameserver google 8.8.8.8:53
5771 timeout retry 1s
5772 hold valid 10s
5773 hold nx 3s
5774 hold other 3s
5775 hold obsolete 0s
5776 accepted_payload_size 8192
5777
5778 frontend fe
5779 bind 10.42.0.1:80
5780 http-request do-resolve(txn.myip,mydns,ipv4) hdr(Host),lower
5781 http-request capture var(txn.myip) len 40
5782
5783 # return 503 when the variable is not set,
5784 # which mean DNS resolution error
5785 use_backend b_503 unless { var(txn.myip) -m found }
5786
5787 default_backend be
5788
5789 backend b_503
5790 # dummy backend used to return 503.
5791 # one can use the errorfile directive to send a nice
5792 # 503 error page to end users
5793
5794 backend be
5795 # rule to prevent HAProxy from reconnecting to services
5796 # on the local network (forged DNS name used to scan the network)
5797 http-request deny if { var(txn.myip) -m ip 127.0.0.0/8 10.0.0.0/8 }
5798 http-request set-dst var(txn.myip)
5799 server clear 0.0.0.0:0
5800
5801 NOTE: Don't forget to set the "protection" rules to ensure HAProxy won't
5802 be used to scan the network or worst won't loop over itself...
5803
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01005804http-request early-hint <name> <fmt> [ { if | unless } <condition> ]
5805
5806 This is used to build an HTTP 103 Early Hints response prior to any other one.
5807 This appends an HTTP header field to this response whose name is specified in
5808 <name> and whose value is defined by <fmt> which follows the log-format rules
5809 (see Custom Log Format in section 8.2.4). This is particularly useful to pass
Frédéric Lécaille3aac1062018-11-13 09:42:13 +01005810 to the client some Link headers to preload resources required to render the
5811 HTML documents.
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01005812
5813 See RFC 8297 for more information.
5814
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005815http-request redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005816
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005817 This performs an HTTP redirection based on a redirect rule. This is exactly
5818 the same as the "redirect" statement except that it inserts a redirect rule
5819 which can be processed in the middle of other "http-request" rules and that
5820 these rules use the "log-format" strings. See the "redirect" keyword for the
5821 rule's syntax.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005822
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005823http-request reject [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005824
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005825 This stops the evaluation of the rules and immediately closes the connection
5826 without sending any response. It acts similarly to the
5827 "tcp-request content reject" rules. It can be useful to force an immediate
5828 connection closure on HTTP/2 connections.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005829
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005830http-request replace-header <name> <match-regex> <replace-fmt>
5831 [ { if | unless } <condition> ]
Willy Tarreaua9083d02015-05-08 15:27:59 +02005832
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05005833 This matches the value of all occurrences of header field <name> against
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005834 <match-regex>. Matching is performed case-sensitively. Matching values are
5835 completely replaced by <replace-fmt>. Format characters are allowed in
5836 <replace-fmt> and work like <fmt> arguments in "http-request add-header".
5837 Standard back-references using the backslash ('\') followed by a number are
5838 supported.
Thierry FOURNIER82bf70d2015-05-26 17:58:29 +02005839
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005840 This action acts on whole header lines, regardless of the number of values
5841 they may contain. Thus it is well-suited to process headers naturally
5842 containing commas in their value, such as If-Modified-Since. Headers that
5843 contain a comma-separated list of values, such as Accept, should be processed
5844 using "http-request replace-value".
William Lallemand86d0df02017-11-24 21:36:45 +01005845
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005846 Example:
5847 http-request replace-header Cookie foo=([^;]*);(.*) foo=\1;ip=%bi;\2
5848
5849 # applied to:
5850 Cookie: foo=foobar; expires=Tue, 14-Jun-2016 01:40:45 GMT;
5851
5852 # outputs:
5853 Cookie: foo=foobar;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT;
5854
5855 # assuming the backend IP is 192.168.1.20
Willy Tarreau09448f72014-06-25 18:12:15 +02005856
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005857 http-request replace-header User-Agent curl foo
5858
5859 # applied to:
5860 User-Agent: curl/7.47.0
Willy Tarreau09448f72014-06-25 18:12:15 +02005861
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005862 # outputs:
5863 User-Agent: foo
Willy Tarreau09448f72014-06-25 18:12:15 +02005864
Willy Tarreau262c3f12019-12-17 06:52:51 +01005865http-request replace-path <match-regex> <replace-fmt>
5866 [ { if | unless } <condition> ]
5867
5868 This works like "replace-header" except that it works on the request's path
5869 component instead of a header. The path component starts at the first '/'
Christopher Faulet82c83322020-09-02 14:16:59 +02005870 after an optional scheme+authority and ends before the question mark. Thus,
5871 the replacement does not modify the scheme, the authority and the
5872 query-string.
Willy Tarreau262c3f12019-12-17 06:52:51 +01005873
5874 It is worth noting that regular expressions may be more expensive to evaluate
5875 than certain ACLs, so rare replacements may benefit from a condition to avoid
5876 performing the evaluation at all if it does not match.
5877
5878 Example:
5879 # prefix /foo : turn /bar?q=1 into /foo/bar?q=1 :
5880 http-request replace-path (.*) /foo\1
5881
Willy Tarreau262c3f12019-12-17 06:52:51 +01005882 # strip /foo : turn /foo/bar?q=1 into /bar?q=1
5883 http-request replace-path /foo/(.*) /\1
5884 # or more efficient if only some requests match :
5885 http-request replace-path /foo/(.*) /\1 if { url_beg /foo/ }
5886
Christopher Faulet312294f2020-09-02 17:17:44 +02005887http-request replace-pathq <match-regex> <replace-fmt>
5888 [ { if | unless } <condition> ]
5889
5890 This does the same as "http-request replace-path" except that the path
5891 contains the query-string if any is present. Thus, the path and the
5892 query-string are replaced.
5893
5894 Example:
5895 # suffix /foo : turn /bar?q=1 into /bar/foo?q=1 :
5896 http-request replace-pathq ([^?]*)(\?(.*))? \1/foo\2
5897
Willy Tarreau33810222019-06-12 17:44:02 +02005898http-request replace-uri <match-regex> <replace-fmt>
5899 [ { if | unless } <condition> ]
5900
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005901 This works like "replace-header" except that it works on the request's URI part
5902 instead of a header. The URI part may contain an optional scheme, authority or
5903 query string. These are considered to be part of the value that is matched
5904 against.
5905
5906 It is worth noting that regular expressions may be more expensive to evaluate
5907 than certain ACLs, so rare replacements may benefit from a condition to avoid
5908 performing the evaluation at all if it does not match.
Willy Tarreau33810222019-06-12 17:44:02 +02005909
Willy Tarreau62b59132019-12-17 06:51:20 +01005910 IMPORTANT NOTE: historically in HTTP/1.x, the vast majority of requests sent
5911 by browsers use the "origin form", which differs from the "absolute form" in
5912 that they do not contain a scheme nor authority in the URI portion. Mostly
5913 only requests sent to proxies, those forged by hand and some emitted by
5914 certain applications use the absolute form. As such, "replace-uri" usually
5915 works fine most of the time in HTTP/1.x with rules starting with a "/". But
5916 with HTTP/2, clients are encouraged to send absolute URIs only, which look
5917 like the ones HTTP/1 clients use to talk to proxies. Such partial replace-uri
5918 rules may then fail in HTTP/2 when they work in HTTP/1. Either the rules need
Willy Tarreau262c3f12019-12-17 06:52:51 +01005919 to be adapted to optionally match a scheme and authority, or replace-path
5920 should be used.
Willy Tarreau33810222019-06-12 17:44:02 +02005921
Willy Tarreau62b59132019-12-17 06:51:20 +01005922 Example:
5923 # rewrite all "http" absolute requests to "https":
5924 http-request replace-uri ^http://(.*) https://\1
Willy Tarreau33810222019-06-12 17:44:02 +02005925
Willy Tarreau62b59132019-12-17 06:51:20 +01005926 # prefix /foo : turn /bar?q=1 into /foo/bar?q=1 :
5927 http-request replace-uri ([^/:]*://[^/]*)?(.*) \1/foo\2
Willy Tarreau33810222019-06-12 17:44:02 +02005928
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005929http-request replace-value <name> <match-regex> <replace-fmt>
5930 [ { if | unless } <condition> ]
Willy Tarreau09448f72014-06-25 18:12:15 +02005931
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005932 This works like "replace-header" except that it matches the regex against
5933 every comma-delimited value of the header field <name> instead of the
5934 entire header. This is suited for all headers which are allowed to carry
5935 more than one value. An example could be the Accept header.
Willy Tarreau09448f72014-06-25 18:12:15 +02005936
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005937 Example:
5938 http-request replace-value X-Forwarded-For ^192\.168\.(.*)$ 172.16.\1
Thierry FOURNIER236657b2015-08-19 08:25:14 +02005939
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005940 # applied to:
5941 X-Forwarded-For: 192.168.10.1, 192.168.13.24, 10.0.0.37
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02005942
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005943 # outputs:
5944 X-Forwarded-For: 172.16.10.1, 172.16.13.24, 10.0.0.37
Frédéric Lécaille6778b272018-01-29 15:22:53 +01005945
Christopher Faulet24231ab2020-01-24 17:44:23 +01005946http-request return [status <code>] [content-type <type>]
5947 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
5948 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
Christopher Faulet4a2c1422020-01-31 17:36:01 +01005949 [ hdr <name> <fmt> ]*
Christopher Faulet24231ab2020-01-24 17:44:23 +01005950 [ { if | unless } <condition> ]
5951
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05005952 This stops the evaluation of the rules and immediately returns a response. The
Christopher Faulet24231ab2020-01-24 17:44:23 +01005953 default status code used for the response is 200. It can be optionally
5954 specified as an arguments to "status". The response content-type may also be
Daniel Corbett67a82712020-07-06 23:01:19 -04005955 specified as an argument to "content-type". Finally the response itself may
Sébastien Grossab877122020-10-08 10:06:03 +02005956 be defined. It can be a full HTTP response specifying the errorfile to use,
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05005957 or the response payload specifying the file or the string to use. These rules
Christopher Faulet24231ab2020-01-24 17:44:23 +01005958 are followed to create the response :
5959
5960 * If neither the errorfile nor the payload to use is defined, a dummy
5961 response is returned. Only the "status" argument is considered. It can be
5962 any code in the range [200, 599]. The "content-type" argument, if any, is
5963 ignored.
5964
5965 * If "default-errorfiles" argument is set, the proxy's errorfiles are
5966 considered. If the "status" argument is defined, it must be one of the
Daniel Corbett67a82712020-07-06 23:01:19 -04005967 status code handled by haproxy (200, 400, 403, 404, 405, 408, 410, 413,
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02005968 425, 429, 500, 502, 503, and 504). The "content-type" argument, if any,
5969 is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01005970
5971 * If a specific errorfile is defined, with an "errorfile" argument, the
5972 corresponding file, containing a full HTTP response, is returned. Only the
5973 "status" argument is considered. It must be one of the status code handled
Daniel Corbett67a82712020-07-06 23:01:19 -04005974 by haproxy (200, 400, 403, 404, 405, 408, 410, 413, 425, 429, 500, 502, 503,
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02005975 and 504). The "content-type" argument, if any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01005976
5977 * If an http-errors section is defined, with an "errorfiles" argument, the
5978 corresponding file in the specified http-errors section, containing a full
5979 HTTP response, is returned. Only the "status" argument is considered. It
Daniel Corbett67a82712020-07-06 23:01:19 -04005980 must be one of the status code handled by haproxy (200, 400, 403, 404, 405,
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02005981 408, 410, 413, 425, 429, 500, 502, 503, and 504). The "content-type"
5982 argument, if any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01005983
5984 * If a "file" or a "lf-file" argument is specified, the file's content is
5985 used as the response payload. If the file is not empty, its content-type
5986 must be set as argument to "content-type". Otherwise, any "content-type"
5987 argument is ignored. With a "lf-file" argument, the file's content is
5988 evaluated as a log-format string. With a "file" argument, it is considered
5989 as a raw content.
5990
5991 * If a "string" or "lf-string" argument is specified, the defined string is
5992 used as the response payload. The content-type must always be set as
5993 argument to "content-type". With a "lf-string" argument, the string is
5994 evaluated as a log-format string. With a "string" argument, it is
5995 considered as a raw string.
5996
Sébastien Grossab877122020-10-08 10:06:03 +02005997 When the response is not based on an errorfile, it is possible to append HTTP
Christopher Faulet4a2c1422020-01-31 17:36:01 +01005998 header fields to the response using "hdr" arguments. Otherwise, all "hdr"
5999 arguments are ignored. For each one, the header name is specified in <name>
6000 and its value is defined by <fmt> which follows the log-format rules.
6001
Christopher Faulet24231ab2020-01-24 17:44:23 +01006002 Note that the generated response must be smaller than a buffer. And to avoid
6003 any warning, when an errorfile or a raw file is loaded, the buffer space
Sébastien Grossab877122020-10-08 10:06:03 +02006004 reserved for the headers rewriting should also be free.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006005
6006 No further "http-request" rules are evaluated.
6007
6008 Example:
Daniel Corbett67a82712020-07-06 23:01:19 -04006009 http-request return errorfile /etc/haproxy/errorfiles/200.http \
Christopher Faulet24231ab2020-01-24 17:44:23 +01006010 if { path /ping }
6011
6012 http-request return content-type image/x-icon file /var/www/favicon.ico \
6013 if { path /favicon.ico }
6014
6015 http-request return status 403 content-type text/plain \
6016 lf-string "Access denied. IP %[src] is blacklisted." \
6017 if { src -f /etc/haproxy/blacklist.lst }
6018
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006019http-request sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
6020http-request sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006021
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006022 This actions increments the GPC0 or GPC1 counter according with the sticky
6023 counter designated by <sc-id>. If an error occurs, this action silently fails
6024 and the actions evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006025
Cédric Dufour0d7712d2019-11-06 18:38:53 +01006026http-request sc-set-gpt0(<sc-id>) { <int> | <expr> }
6027 [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006028
Cédric Dufour0d7712d2019-11-06 18:38:53 +01006029 This action sets the 32-bit unsigned GPT0 tag according to the sticky counter
6030 designated by <sc-id> and the value of <int>/<expr>. The expected result is a
6031 boolean. If an error occurs, this action silently fails and the actions
6032 evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006033
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006034http-request set-dst <expr> [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006035
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006036 This is used to set the destination IP address to the value of specified
6037 expression. Useful when a proxy in front of HAProxy rewrites destination IP,
6038 but provides the correct IP in a HTTP header; or you want to mask the IP for
6039 privacy. If you want to connect to the new address/port, use '0.0.0.0:0' as a
6040 server address in the backend.
Christopher Faulet85d79c92016-11-09 16:54:56 +01006041
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006042 Arguments:
6043 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
6044 by some converters.
Christopher Faulet85d79c92016-11-09 16:54:56 +01006045
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006046 Example:
6047 http-request set-dst hdr(x-dst)
6048 http-request set-dst dst,ipmask(24)
Christopher Faulet85d79c92016-11-09 16:54:56 +01006049
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006050 When possible, set-dst preserves the original destination port as long as the
6051 address family allows it, otherwise the destination port is set to 0.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02006052
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006053http-request set-dst-port <expr> [ { if | unless } <condition> ]
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02006054
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006055 This is used to set the destination port address to the value of specified
6056 expression. If you want to connect to the new address/port, use '0.0.0.0:0'
6057 as a server address in the backend.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02006058
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006059 Arguments:
6060 <expr> Is a standard HAProxy expression formed by a sample-fetch
6061 followed by some converters.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02006062
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006063 Example:
6064 http-request set-dst-port hdr(x-port)
6065 http-request set-dst-port int(4000)
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02006066
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006067 When possible, set-dst-port preserves the original destination address as
6068 long as the address family supports a port, otherwise it forces the
6069 destination address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +02006070
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006071http-request set-header <name> <fmt> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02006072
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006073 This does the same as "http-request add-header" except that the header name
6074 is first removed if it existed. This is useful when passing security
6075 information to the server, where the header must not be manipulated by
6076 external users. Note that the new value is computed before the removal so it
6077 is possible to concatenate a value to an existing header.
William Lallemand44be6402016-05-25 01:51:35 +02006078
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006079 Example:
6080 http-request set-header X-Haproxy-Current-Date %T
6081 http-request set-header X-SSL %[ssl_fc]
6082 http-request set-header X-SSL-Session_ID %[ssl_fc_session_id,hex]
6083 http-request set-header X-SSL-Client-Verify %[ssl_c_verify]
6084 http-request set-header X-SSL-Client-DN %{+Q}[ssl_c_s_dn]
6085 http-request set-header X-SSL-Client-CN %{+Q}[ssl_c_s_dn(cn)]
6086 http-request set-header X-SSL-Issuer %{+Q}[ssl_c_i_dn]
6087 http-request set-header X-SSL-Client-NotBefore %{+Q}[ssl_c_notbefore]
6088 http-request set-header X-SSL-Client-NotAfter %{+Q}[ssl_c_notafter]
William Lallemand44be6402016-05-25 01:51:35 +02006089
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006090http-request set-log-level <level> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02006091
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006092 This is used to change the log level of the current request when a certain
6093 condition is met. Valid levels are the 8 syslog levels (see the "log"
6094 keyword) plus the special level "silent" which disables logging for this
6095 request. This rule is not final so the last matching rule wins. This rule
6096 can be useful to disable health checks coming from another equipment.
William Lallemand13e9b0c2016-05-25 02:34:07 +02006097
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006098http-request set-map(<file-name>) <key fmt> <value fmt>
6099 [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02006100
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006101 This is used to add a new entry into a MAP. The MAP must be loaded from a
6102 file (even a dummy empty file). The file name of the MAP to be updated is
6103 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
6104 log-format rules, used to collect MAP key, and <value fmt>, which follows
6105 log-format rules, used to collect content for the new entry.
6106 It performs a lookup in the MAP before insertion, to avoid duplicated (or
6107 more) values. This lookup is done by a linear search and can be expensive
6108 with large lists! It is the equivalent of the "set map" command from the
6109 stats socket, but can be triggered by an HTTP request.
William Lallemand13e9b0c2016-05-25 02:34:07 +02006110
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006111http-request set-mark <mark> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02006112
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006113 This is used to set the Netfilter MARK on all packets sent to the client to
6114 the value passed in <mark> on platforms which support it. This value is an
6115 unsigned 32 bit value which can be matched by netfilter and by the routing
6116 table. It can be expressed both in decimal or hexadecimal format (prefixed by
6117 "0x"). This can be useful to force certain packets to take a different route
6118 (for example a cheaper network path for bulk downloads). This works on Linux
6119 kernels 2.6.32 and above and requires admin privileges.
Willy Tarreau00005ce2016-10-21 15:07:45 +02006120
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006121http-request set-method <fmt> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02006122
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006123 This rewrites the request method with the result of the evaluation of format
6124 string <fmt>. There should be very few valid reasons for having to do so as
6125 this is more likely to break something than to fix it.
William Lallemand13e9b0c2016-05-25 02:34:07 +02006126
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006127http-request set-nice <nice> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02006128
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006129 This sets the "nice" factor of the current request being processed. It only
6130 has effect against the other requests being processed at the same time.
6131 The default value is 0, unless altered by the "nice" setting on the "bind"
6132 line. The accepted range is -1024..1024. The higher the value, the nicest
6133 the request will be. Lower values will make the request more important than
6134 other ones. This can be useful to improve the speed of some requests, or
6135 lower the priority of non-important requests. Using this setting without
6136 prior experimentation can cause some major slowdown.
William Lallemand13e9b0c2016-05-25 02:34:07 +02006137
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006138http-request set-path <fmt> [ { if | unless } <condition> ]
Willy Tarreau00005ce2016-10-21 15:07:45 +02006139
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006140 This rewrites the request path with the result of the evaluation of format
6141 string <fmt>. The query string, if any, is left intact. If a scheme and
6142 authority is found before the path, they are left intact as well. If the
6143 request doesn't have a path ("*"), this one is replaced with the format.
6144 This can be used to prepend a directory component in front of a path for
6145 example. See also "http-request set-query" and "http-request set-uri".
Willy Tarreau2d392c22015-08-24 01:43:45 +02006146
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006147 Example :
6148 # prepend the host name before the path
6149 http-request set-path /%[hdr(host)]%[path]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006150
Christopher Faulet312294f2020-09-02 17:17:44 +02006151http-request set-pathq <fmt> [ { if | unless } <condition> ]
6152
6153 This does the same as "http-request set-path" except that the query-string is
6154 also rewritten. It may be used to remove the query-string, including the
6155 question mark (it is not possible using "http-request set-query").
6156
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006157http-request set-priority-class <expr> [ { if | unless } <condition> ]
Olivier Houchardccaa7de2017-10-02 11:51:03 +02006158
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006159 This is used to set the queue priority class of the current request.
6160 The value must be a sample expression which converts to an integer in the
6161 range -2047..2047. Results outside this range will be truncated.
6162 The priority class determines the order in which queued requests are
6163 processed. Lower values have higher priority.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006164
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006165http-request set-priority-offset <expr> [ { if | unless } <condition> ]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006166
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006167 This is used to set the queue priority timestamp offset of the current
6168 request. The value must be a sample expression which converts to an integer
6169 in the range -524287..524287. Results outside this range will be truncated.
6170 When a request is queued, it is ordered first by the priority class, then by
6171 the current timestamp adjusted by the given offset in milliseconds. Lower
6172 values have higher priority.
6173 Note that the resulting timestamp is is only tracked with enough precision
6174 for 524,287ms (8m44s287ms). If the request is queued long enough to where the
6175 adjusted timestamp exceeds this value, it will be misidentified as highest
6176 priority. Thus it is important to set "timeout queue" to a value, where when
6177 combined with the offset, does not exceed this limit.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006178
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006179http-request set-query <fmt> [ { if | unless } <condition> ]
Willy Tarreau20b0de52012-12-24 15:45:22 +01006180
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006181 This rewrites the request's query string which appears after the first
6182 question mark ("?") with the result of the evaluation of format string <fmt>.
6183 The part prior to the question mark is left intact. If the request doesn't
6184 contain a question mark and the new value is not empty, then one is added at
6185 the end of the URI, followed by the new value. If a question mark was
6186 present, it will never be removed even if the value is empty. This can be
6187 used to add or remove parameters from the query string.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08006188
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006189 See also "http-request set-query" and "http-request set-uri".
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01006190
6191 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006192 # replace "%3D" with "=" in the query string
6193 http-request set-query %[query,regsub(%3D,=,g)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01006194
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006195http-request set-src <expr> [ { if | unless } <condition> ]
6196 This is used to set the source IP address to the value of specified
6197 expression. Useful when a proxy in front of HAProxy rewrites source IP, but
6198 provides the correct IP in a HTTP header; or you want to mask source IP for
Olivier Doucet56e31202020-04-21 09:32:56 +02006199 privacy. All subsequent calls to "src" fetch will return this value
6200 (see example).
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006201
6202 Arguments :
6203 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
6204 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01006205
Olivier Doucet56e31202020-04-21 09:32:56 +02006206 See also "option forwardfor".
6207
Cyril Bonté78caf842010-03-10 22:41:43 +01006208 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006209 http-request set-src hdr(x-forwarded-for)
6210 http-request set-src src,ipmask(24)
6211
Olivier Doucet56e31202020-04-21 09:32:56 +02006212 # After the masking this will track connections
6213 # based on the IP address with the last byte zeroed out.
6214 http-request track-sc0 src
6215
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006216 When possible, set-src preserves the original source port as long as the
6217 address family allows it, otherwise the source port is set to 0.
6218
6219http-request set-src-port <expr> [ { if | unless } <condition> ]
6220
6221 This is used to set the source port address to the value of specified
6222 expression.
6223
6224 Arguments:
6225 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
6226 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01006227
Willy Tarreau20b0de52012-12-24 15:45:22 +01006228 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006229 http-request set-src-port hdr(x-port)
6230 http-request set-src-port int(4000)
6231
6232 When possible, set-src-port preserves the original source address as long as
6233 the address family supports a port, otherwise it forces the source address to
6234 IPv4 "0.0.0.0" before rewriting the port.
6235
6236http-request set-tos <tos> [ { if | unless } <condition> ]
6237
6238 This is used to set the TOS or DSCP field value of packets sent to the client
6239 to the value passed in <tos> on platforms which support this. This value
6240 represents the whole 8 bits of the IP TOS field, and can be expressed both in
6241 decimal or hexadecimal format (prefixed by "0x"). Note that only the 6 higher
6242 bits are used in DSCP or TOS, and the two lower bits are always 0. This can
6243 be used to adjust some routing behavior on border routers based on some
6244 information from the request.
6245
6246 See RFC 2474, 2597, 3260 and 4594 for more information.
6247
6248http-request set-uri <fmt> [ { if | unless } <condition> ]
6249
6250 This rewrites the request URI with the result of the evaluation of format
6251 string <fmt>. The scheme, authority, path and query string are all replaced
6252 at once. This can be used to rewrite hosts in front of proxies, or to
6253 perform complex modifications to the URI such as moving parts between the
6254 path and the query string.
6255 See also "http-request set-path" and "http-request set-query".
6256
6257http-request set-var(<var-name>) <expr> [ { if | unless } <condition> ]
6258
6259 This is used to set the contents of a variable. The variable is declared
6260 inline.
6261
6262 Arguments:
6263 <var-name> The name of the variable starts with an indication about its
6264 scope. The scopes allowed are:
6265 "proc" : the variable is shared with the whole process
6266 "sess" : the variable is shared with the whole session
6267 "txn" : the variable is shared with the transaction
6268 (request and response)
6269 "req" : the variable is shared only during request
6270 processing
6271 "res" : the variable is shared only during response
6272 processing
6273 This prefix is followed by a name. The separator is a '.'.
6274 The name may only contain characters 'a-z', 'A-Z', '0-9'
6275 and '_'.
6276
6277 <expr> Is a standard HAProxy expression formed by a sample-fetch
6278 followed by some converters.
Willy Tarreau20b0de52012-12-24 15:45:22 +01006279
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006280 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006281 http-request set-var(req.my_var) req.fhdr(user-agent),lower
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006282
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006283http-request send-spoe-group <engine-name> <group-name>
6284 [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006285
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006286 This action is used to trigger sending of a group of SPOE messages. To do so,
6287 the SPOE engine used to send messages must be defined, as well as the SPOE
6288 group to send. Of course, the SPOE engine must refer to an existing SPOE
6289 filter. If not engine name is provided on the SPOE filter line, the SPOE
6290 agent name must be used.
6291
6292 Arguments:
6293 <engine-name> The SPOE engine name.
6294
6295 <group-name> The SPOE group name as specified in the engine
6296 configuration.
6297
6298http-request silent-drop [ { if | unless } <condition> ]
6299
6300 This stops the evaluation of the rules and makes the client-facing connection
6301 suddenly disappear using a system-dependent way that tries to prevent the
6302 client from being notified. The effect it then that the client still sees an
6303 established connection while there's none on HAProxy. The purpose is to
6304 achieve a comparable effect to "tarpit" except that it doesn't use any local
6305 resource at all on the machine running HAProxy. It can resist much higher
6306 loads than "tarpit", and slow down stronger attackers. It is important to
6307 understand the impact of using this mechanism. All stateful equipment placed
6308 between the client and HAProxy (firewalls, proxies, load balancers) will also
6309 keep the established connection for a long time and may suffer from this
6310 action.
6311 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
6312 option is used to block the emission of a TCP reset. On other systems, the
6313 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
6314 router, though it's still delivered to local networks. Do not use it unless
6315 you fully understand how it works.
6316
Christopher Faulet46f95542019-12-20 10:07:22 +01006317http-request strict-mode { on | off }
6318
6319 This enables or disables the strict rewriting mode for following rules. It
6320 does not affect rules declared before it and it is only applicable on rules
6321 performing a rewrite on the requests. When the strict mode is enabled, any
6322 rewrite failure triggers an internal error. Otherwise, such errors are
6323 silently ignored. The purpose of the strict rewriting mode is to make some
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05006324 rewrites optional while others must be performed to continue the request
Christopher Faulet46f95542019-12-20 10:07:22 +01006325 processing.
6326
Christopher Faulet1aea50e2020-01-17 16:03:53 +01006327 By default, the strict rewriting mode is enabled. Its value is also reset
Christopher Faulet46f95542019-12-20 10:07:22 +01006328 when a ruleset evaluation ends. So, for instance, if you change the mode on
6329 the frontend, the default mode is restored when HAProxy starts the backend
6330 rules evaluation.
6331
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006332http-request tarpit [deny_status <status>] [ { if | unless } <condition> ]
6333http-request tarpit [ { status | deny_status } <code>] [content-type <type>]
6334 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
6335 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
6336 [ hdr <name> <fmt> ]*
6337 [ { if | unless } <condition> ]
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006338
6339 This stops the evaluation of the rules and immediately blocks the request
6340 without responding for a delay specified by "timeout tarpit" or
6341 "timeout connect" if the former is not set. After that delay, if the client
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006342 is still connected, a response is returned so that the client does not
6343 suspect it has been tarpitted. Logs will report the flags "PT". The goal of
6344 the tarpit rule is to slow down robots during an attack when they're limited
6345 on the number of concurrent requests. It can be very efficient against very
6346 dumb robots, and will significantly reduce the load on firewalls compared to
6347 a "deny" rule. But when facing "correctly" developed robots, it can make
6348 things worse by forcing haproxy and the front firewall to support insane
6349 number of concurrent connections. By default an HTTP error 500 is returned.
6350 But the response may be customized using same syntax than
6351 "http-request return" rules. Thus, see "http-request return" for details.
Ilya Shipitsin11057a32020-06-21 21:18:27 +05006352 For compatibility purpose, when no argument is defined, or only "deny_status",
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006353 the argument "default-errorfiles" is implied. It means
6354 "http-request tarpit [deny_status <status>]" is an alias of
6355 "http-request tarpit [status <status>] default-errorfiles".
6356 No further "http-request" rules are evaluated.
6357 See also "http-request return" and "http-request silent-drop".
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006358
6359http-request track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
6360http-request track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
6361http-request track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
6362
6363 This enables tracking of sticky counters from current request. These rules do
6364 not stop evaluation and do not change default action. The number of counters
6365 that may be simultaneously tracked by the same connection is set in
6366 MAX_SESS_STKCTR at build time (reported in haproxy -vv) which defaults to 3,
Matteo Contrini1857b8c2020-10-16 17:35:54 +02006367 so the track-sc number is between 0 and (MAX_SESS_STKCTR-1). The first
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006368 "track-sc0" rule executed enables tracking of the counters of the specified
6369 table as the first set. The first "track-sc1" rule executed enables tracking
6370 of the counters of the specified table as the second set. The first
6371 "track-sc2" rule executed enables tracking of the counters of the specified
6372 table as the third set. It is a recommended practice to use the first set of
6373 counters for the per-frontend counters and the second set for the per-backend
6374 ones. But this is just a guideline, all may be used everywhere.
6375
6376 Arguments :
6377 <key> is mandatory, and is a sample expression rule as described in
6378 section 7.3. It describes what elements of the incoming request or
6379 connection will be analyzed, extracted, combined, and used to
6380 select which table entry to update the counters.
6381
6382 <table> is an optional table to be used instead of the default one, which
6383 is the stick-table declared in the current proxy. All the counters
6384 for the matches and updates for the key will then be performed in
6385 that table until the session ends.
6386
6387 Once a "track-sc*" rule is executed, the key is looked up in the table and if
6388 it is not found, an entry is allocated for it. Then a pointer to that entry
6389 is kept during all the session's life, and this entry's counters are updated
6390 as often as possible, every time the session's counters are updated, and also
6391 systematically when the session ends. Counters are only updated for events
6392 that happen after the tracking has been started. As an exception, connection
6393 counters and request counters are systematically updated so that they reflect
6394 useful information.
6395
6396 If the entry tracks concurrent connection counters, one connection is counted
6397 for as long as the entry is tracked, and the entry will not expire during
6398 that time. Tracking counters also provides a performance advantage over just
6399 checking the keys, because only one table lookup is performed for all ACL
6400 checks that make use of it.
6401
6402http-request unset-var(<var-name>) [ { if | unless } <condition> ]
6403
6404 This is used to unset a variable. See above for details about <var-name>.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006405
6406 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006407 http-request unset-var(req.my_var)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006408
Christopher Faulet579d83b2019-11-22 15:34:17 +01006409http-request use-service <service-name> [ { if | unless } <condition> ]
6410
6411 This directive executes the configured HTTP service to reply to the request
6412 and stops the evaluation of the rules. An HTTP service may choose to reply by
6413 sending any valid HTTP response or it may immediately close the connection
6414 without sending any response. Outside natives services, for instance the
6415 Prometheus exporter, it is possible to write your own services in Lua. No
6416 further "http-request" rules are evaluated.
6417
6418 Arguments :
6419 <service-name> is mandatory. It is the service to call
6420
6421 Example:
6422 http-request use-service prometheus-exporter if { path /metrics }
6423
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006424http-request wait-for-handshake [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006425
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006426 This will delay the processing of the request until the SSL handshake
6427 happened. This is mostly useful to delay processing early data until we're
6428 sure they are valid.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006429
Willy Tarreauef781042010-01-27 11:53:01 +01006430
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006431http-response <action> <options...> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02006432 Access control for Layer 7 responses
6433
6434 May be used in sections: defaults | frontend | listen | backend
6435 no | yes | yes | yes
6436
6437 The http-response statement defines a set of rules which apply to layer 7
6438 processing. The rules are evaluated in their declaration order when they are
6439 met in a frontend, listen or backend section. Any rule may optionally be
6440 followed by an ACL-based condition, in which case it will only be evaluated
6441 if the condition is true. Since these rules apply on responses, the backend
6442 rules are applied first, followed by the frontend's rules.
6443
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006444 The first keyword is the rule's action. The supported actions are described
6445 below.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02006446
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006447 There is no limit to the number of http-response statements per instance.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02006448
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006449 Example:
6450 acl key_acl res.hdr(X-Acl-Key) -m found
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02006451
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006452 acl myhost hdr(Host) -f myhost.lst
Sasha Pachev218f0642014-06-16 12:05:59 -06006453
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006454 http-response add-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
6455 http-response del-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
Sasha Pachev218f0642014-06-16 12:05:59 -06006456
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006457 Example:
6458 acl value res.hdr(X-Value) -m found
Sasha Pachev218f0642014-06-16 12:05:59 -06006459
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006460 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06006461
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006462 http-response set-map(map.lst) %[src] %[res.hdr(X-Value)] if value
6463 http-response del-map(map.lst) %[src] if ! value
Sasha Pachev218f0642014-06-16 12:05:59 -06006464
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006465 See also : "http-request", section 3.4 about userlists and section 7 about
6466 ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06006467
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006468http-response add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006469
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006470 This is used to add a new entry into an ACL. The ACL must be loaded from a
6471 file (even a dummy empty file). The file name of the ACL to be updated is
6472 passed between parentheses. It takes one argument: <key fmt>, which follows
6473 log-format rules, to collect content of the new entry. It performs a lookup
6474 in the ACL before insertion, to avoid duplicated (or more) values.
6475 This lookup is done by a linear search and can be expensive with large lists!
6476 It is the equivalent of the "add acl" command from the stats socket, but can
6477 be triggered by an HTTP response.
Sasha Pachev218f0642014-06-16 12:05:59 -06006478
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006479http-response add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006480
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006481 This appends an HTTP header field whose name is specified in <name> and whose
6482 value is defined by <fmt> which follows the log-format rules (see Custom Log
6483 Format in section 8.2.4). This may be used to send a cookie to a client for
6484 example, or to pass some internal information.
6485 This rule is not final, so it is possible to add other similar rules.
6486 Note that header addition is performed immediately, so one rule might reuse
6487 the resulting header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06006488
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006489http-response allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006490
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006491 This stops the evaluation of the rules and lets the response pass the check.
6492 No further "http-response" rules are evaluated for the current section.
Sasha Pachev218f0642014-06-16 12:05:59 -06006493
Jarno Huuskonen251a6b72019-01-04 14:05:02 +02006494http-response cache-store <name> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006495
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02006496 See section 6.2 about cache setup.
Sasha Pachev218f0642014-06-16 12:05:59 -06006497
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006498http-response capture <sample> id <id> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006499
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006500 This captures sample expression <sample> from the response buffer, and
6501 converts it to a string. The resulting string is stored into the next request
6502 "capture" slot, so it will possibly appear next to some captured HTTP
6503 headers. It will then automatically appear in the logs, and it will be
6504 possible to extract it using sample fetch rules to feed it into headers or
6505 anything. Please check section 7.3 (Fetching samples) and
6506 "capture response header" for more information.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02006507
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006508 The keyword "id" is the id of the capture slot which is used for storing the
6509 string. The capture slot must be defined in an associated frontend.
6510 This is useful to run captures in backends. The slot id can be declared by a
6511 previous directive "http-response capture" or with the "declare capture"
6512 keyword.
Baptiste Assmann19a69b32020-01-16 14:34:22 +01006513
6514 When using this action in a backend, double check that the relevant
6515 frontend(s) have the required capture slots otherwise, this rule will be
6516 ignored at run time. This can't be detected at configuration parsing time
6517 due to HAProxy's ability to dynamically resolve backend name at runtime.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02006518
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006519http-response del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02006520
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006521 This is used to delete an entry from an ACL. The ACL must be loaded from a
6522 file (even a dummy empty file). The file name of the ACL to be updated is
6523 passed between parentheses. It takes one argument: <key fmt>, which follows
6524 log-format rules, to collect content of the entry to delete.
6525 It is the equivalent of the "del acl" command from the stats socket, but can
6526 be triggered by an HTTP response.
Willy Tarreauf4c43c12013-06-11 17:01:13 +02006527
Maciej Zdeb36662462020-11-20 13:58:48 +00006528http-response del-header <name> [ -m <meth> ] [ { if | unless } <condition> ]
Willy Tarreau9a355ec2013-06-11 17:45:46 +02006529
Maciej Zdeb36662462020-11-20 13:58:48 +00006530 This removes all HTTP header fields whose name is specified in <name>. <meth>
6531 is the matching method, applied on the header name. Supported matching methods
6532 are "str" (exact match), "beg" (prefix match), "end" (suffix match), "sub"
6533 (substring match) and "reg" (regex match). If not specified, exact matching
6534 method is used.
Willy Tarreau42cf39e2013-06-11 18:51:32 +02006535
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006536http-response del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau51347ed2013-06-11 19:34:13 +02006537
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006538 This is used to delete an entry from a MAP. The MAP must be loaded from a
6539 file (even a dummy empty file). The file name of the MAP to be updated is
6540 passed between parentheses. It takes one argument: <key fmt>, which follows
6541 log-format rules, to collect content of the entry to delete.
6542 It takes one argument: "file name" It is the equivalent of the "del map"
6543 command from the stats socket, but can be triggered by an HTTP response.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006544
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006545http-response deny [deny_status <status>] [ { if | unless } <condition> ]
6546http-response deny [ { status | deny_status } <code>] [content-type <type>]
6547 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
6548 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
6549 [ hdr <name> <fmt> ]*
6550 [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006551
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006552 This stops the evaluation of the rules and immediately rejects the response.
6553 By default an HTTP 502 error is returned. But the response may be customized
6554 using same syntax than "http-response return" rules. Thus, see
Ilya Shipitsin11057a32020-06-21 21:18:27 +05006555 "http-response return" for details. For compatibility purpose, when no
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006556 argument is defined, or only "deny_status", the argument "default-errorfiles"
6557 is implied. It means "http-response deny [deny_status <status>]" is an alias
6558 of "http-response deny [status <status>] default-errorfiles".
Christopher Faulet040c8cd2020-01-13 16:43:45 +01006559 No further "http-response" rules are evaluated.
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006560 See also "http-response return".
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006561
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006562http-response redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006563
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006564 This performs an HTTP redirection based on a redirect rule.
6565 This supports a format string similarly to "http-request redirect" rules,
6566 with the exception that only the "location" type of redirect is possible on
6567 the response. See the "redirect" keyword for the rule's syntax. When a
6568 redirect rule is applied during a response, connections to the server are
6569 closed so that no data can be forwarded from the server to the client.
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02006570
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006571http-response replace-header <name> <regex-match> <replace-fmt>
6572 [ { if | unless } <condition> ]
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02006573
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006574 This works like "http-request replace-header" except that it works on the
6575 server's response instead of the client's request.
William Lallemand86d0df02017-11-24 21:36:45 +01006576
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006577 Example:
6578 http-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2
Willy Tarreau51d861a2015-05-22 17:30:48 +02006579
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006580 # applied to:
6581 Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006582
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006583 # outputs:
6584 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 +02006585
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006586 # assuming the backend IP is 192.168.1.20.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006587
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006588http-response replace-value <name> <regex-match> <replace-fmt>
6589 [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006590
Tim Duesterhus6bd909b2020-01-17 15:53:18 +01006591 This works like "http-request replace-value" except that it works on the
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006592 server's response instead of the client's request.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006593
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006594 Example:
6595 http-response replace-value Cache-control ^public$ private
Christopher Faulet85d79c92016-11-09 16:54:56 +01006596
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006597 # applied to:
6598 Cache-Control: max-age=3600, public
Christopher Faulet85d79c92016-11-09 16:54:56 +01006599
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006600 # outputs:
6601 Cache-Control: max-age=3600, private
Christopher Faulet85d79c92016-11-09 16:54:56 +01006602
Christopher Faulet24231ab2020-01-24 17:44:23 +01006603http-response return [status <code>] [content-type <type>]
6604 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
6605 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
Christopher Faulet4a2c1422020-01-31 17:36:01 +01006606 [ hdr <name> <value> ]*
Christopher Faulet24231ab2020-01-24 17:44:23 +01006607 [ { if | unless } <condition> ]
6608
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05006609 This stops the evaluation of the rules and immediately returns a response. The
Christopher Faulet24231ab2020-01-24 17:44:23 +01006610 default status code used for the response is 200. It can be optionally
6611 specified as an arguments to "status". The response content-type may also be
Daniel Corbett67a82712020-07-06 23:01:19 -04006612 specified as an argument to "content-type". Finally the response itself may
Christopher Faulet24231ab2020-01-24 17:44:23 +01006613 be defined. If can be a full HTTP response specifying the errorfile to use,
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05006614 or the response payload specifying the file or the string to use. These rules
Christopher Faulet24231ab2020-01-24 17:44:23 +01006615 are followed to create the response :
6616
6617 * If neither the errorfile nor the payload to use is defined, a dummy
6618 response is returned. Only the "status" argument is considered. It can be
6619 any code in the range [200, 599]. The "content-type" argument, if any, is
6620 ignored.
6621
6622 * If "default-errorfiles" argument is set, the proxy's errorfiles are
6623 considered. If the "status" argument is defined, it must be one of the
Daniel Corbett67a82712020-07-06 23:01:19 -04006624 status code handled by haproxy (200, 400, 403, 404, 405, 408, 410, 413,
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02006625 425, 429, 500, 502, 503, and 504). The "content-type" argument, if any,
6626 is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006627
6628 * If a specific errorfile is defined, with an "errorfile" argument, the
6629 corresponding file, containing a full HTTP response, is returned. Only the
6630 "status" argument is considered. It must be one of the status code handled
Daniel Corbett67a82712020-07-06 23:01:19 -04006631 by haproxy (200, 400, 403, 404, 405, 408, 410, 413, 425, 429, 500, 502, 503,
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02006632 and 504). The "content-type" argument, if any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006633
6634 * If an http-errors section is defined, with an "errorfiles" argument, the
6635 corresponding file in the specified http-errors section, containing a full
6636 HTTP response, is returned. Only the "status" argument is considered. It
Daniel Corbett67a82712020-07-06 23:01:19 -04006637 must be one of the status code handled by haproxy (200, 400, 403, 404, 405,
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02006638 408, 410, 413, 425, 429, 500, 502, 503, and 504). The "content-type"
6639 argument, if any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006640
6641 * If a "file" or a "lf-file" argument is specified, the file's content is
6642 used as the response payload. If the file is not empty, its content-type
6643 must be set as argument to "content-type". Otherwise, any "content-type"
6644 argument is ignored. With a "lf-file" argument, the file's content is
6645 evaluated as a log-format string. With a "file" argument, it is considered
6646 as a raw content.
6647
6648 * If a "string" or "lf-string" argument is specified, the defined string is
6649 used as the response payload. The content-type must always be set as
6650 argument to "content-type". With a "lf-string" argument, the string is
6651 evaluated as a log-format string. With a "string" argument, it is
6652 considered as a raw string.
6653
Christopher Faulet4a2c1422020-01-31 17:36:01 +01006654 When the response is not based an errorfile, it is possible to appends HTTP
6655 header fields to the response using "hdr" arguments. Otherwise, all "hdr"
6656 arguments are ignored. For each one, the header name is specified in <name>
6657 and its value is defined by <fmt> which follows the log-format rules.
6658
Christopher Faulet24231ab2020-01-24 17:44:23 +01006659 Note that the generated response must be smaller than a buffer. And to avoid
6660 any warning, when an errorfile or a raw file is loaded, the buffer space
Ilya Shipitsin11057a32020-06-21 21:18:27 +05006661 reserved to the headers rewriting should also be free.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006662
6663 No further "http-response" rules are evaluated.
6664
6665 Example:
Daniel Corbett67a82712020-07-06 23:01:19 -04006666 http-response return errorfile /etc/haproxy/errorfiles/200.http \
Christopher Faulet24231ab2020-01-24 17:44:23 +01006667 if { status eq 404 }
6668
6669 http-response return content-type text/plain \
6670 string "This is the end !" \
6671 if { status eq 500 }
6672
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006673http-response sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
6674http-response sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Ruoshan Huange4edc6b2016-07-14 15:07:45 +08006675
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006676 This action increments the GPC0 or GPC1 counter according with the sticky
6677 counter designated by <sc-id>. If an error occurs, this action silently fails
6678 and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +02006679
Cédric Dufour0d7712d2019-11-06 18:38:53 +01006680http-response sc-set-gpt0(<sc-id>) { <int> | <expr> }
6681 [ { if | unless } <condition> ]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02006682
Cédric Dufour0d7712d2019-11-06 18:38:53 +01006683 This action sets the 32-bit unsigned GPT0 tag according to the sticky counter
6684 designated by <sc-id> and the value of <int>/<expr>. The expected result is a
6685 boolean. If an error occurs, this action silently fails and the actions
6686 evaluation continues.
Frédéric Lécaille6778b272018-01-29 15:22:53 +01006687
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006688http-response send-spoe-group [ { if | unless } <condition> ]
Willy Tarreau2d392c22015-08-24 01:43:45 +02006689
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006690 This action is used to trigger sending of a group of SPOE messages. To do so,
6691 the SPOE engine used to send messages must be defined, as well as the SPOE
6692 group to send. Of course, the SPOE engine must refer to an existing SPOE
6693 filter. If not engine name is provided on the SPOE filter line, the SPOE
6694 agent name must be used.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006695
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006696 Arguments:
6697 <engine-name> The SPOE engine name.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006698
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006699 <group-name> The SPOE group name as specified in the engine
6700 configuration.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006701
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006702http-response set-header <name> <fmt> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02006703
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006704 This does the same as "add-header" except that the header name is first
6705 removed if it existed. This is useful when passing security information to
6706 the server, where the header must not be manipulated by external users.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02006707
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006708http-response set-log-level <level> [ { if | unless } <condition> ]
6709
6710 This is used to change the log level of the current request when a certain
6711 condition is met. Valid levels are the 8 syslog levels (see the "log"
6712 keyword) plus the special level "silent" which disables logging for this
6713 request. This rule is not final so the last matching rule wins. This rule can
6714 be useful to disable health checks coming from another equipment.
6715
6716http-response set-map(<file-name>) <key fmt> <value fmt>
6717
6718 This is used to add a new entry into a MAP. The MAP must be loaded from a
6719 file (even a dummy empty file). The file name of the MAP to be updated is
6720 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
6721 log-format rules, used to collect MAP key, and <value fmt>, which follows
6722 log-format rules, used to collect content for the new entry. It performs a
6723 lookup in the MAP before insertion, to avoid duplicated (or more) values.
6724 This lookup is done by a linear search and can be expensive with large lists!
6725 It is the equivalent of the "set map" command from the stats socket, but can
6726 be triggered by an HTTP response.
6727
6728http-response set-mark <mark> [ { if | unless } <condition> ]
6729
6730 This is used to set the Netfilter MARK on all packets sent to the client to
6731 the value passed in <mark> on platforms which support it. This value is an
6732 unsigned 32 bit value which can be matched by netfilter and by the routing
6733 table. It can be expressed both in decimal or hexadecimal format (prefixed
6734 by "0x"). This can be useful to force certain packets to take a different
6735 route (for example a cheaper network path for bulk downloads). This works on
6736 Linux kernels 2.6.32 and above and requires admin privileges.
6737
6738http-response set-nice <nice> [ { if | unless } <condition> ]
6739
6740 This sets the "nice" factor of the current request being processed.
6741 It only has effect against the other requests being processed at the same
6742 time. The default value is 0, unless altered by the "nice" setting on the
6743 "bind" line. The accepted range is -1024..1024. The higher the value, the
6744 nicest the request will be. Lower values will make the request more important
6745 than other ones. This can be useful to improve the speed of some requests, or
6746 lower the priority of non-important requests. Using this setting without
6747 prior experimentation can cause some major slowdown.
6748
6749http-response set-status <status> [reason <str>]
6750 [ { if | unless } <condition> ]
6751
6752 This replaces the response status code with <status> which must be an integer
6753 between 100 and 999. Optionally, a custom reason text can be provided defined
6754 by <str>, or the default reason for the specified code will be used as a
6755 fallback.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08006756
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006757 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006758 # return "431 Request Header Fields Too Large"
6759 http-response set-status 431
6760 # return "503 Slow Down", custom reason
6761 http-response set-status 503 reason "Slow Down".
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006762
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006763http-response set-tos <tos> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006764
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006765 This is used to set the TOS or DSCP field value of packets sent to the client
6766 to the value passed in <tos> on platforms which support this.
6767 This value represents the whole 8 bits of the IP TOS field, and can be
6768 expressed both in decimal or hexadecimal format (prefixed by "0x"). Note that
6769 only the 6 higher bits are used in DSCP or TOS, and the two lower bits are
6770 always 0. This can be used to adjust some routing behavior on border routers
6771 based on some information from the request.
6772
6773 See RFC 2474, 2597, 3260 and 4594 for more information.
6774
6775http-response set-var(<var-name>) <expr> [ { if | unless } <condition> ]
6776
6777 This is used to set the contents of a variable. The variable is declared
6778 inline.
6779
6780 Arguments:
6781 <var-name> The name of the variable starts with an indication about its
6782 scope. The scopes allowed are:
6783 "proc" : the variable is shared with the whole process
6784 "sess" : the variable is shared with the whole session
6785 "txn" : the variable is shared with the transaction
6786 (request and response)
6787 "req" : the variable is shared only during request
6788 processing
6789 "res" : the variable is shared only during response
6790 processing
6791 This prefix is followed by a name. The separator is a '.'.
6792 The name may only contain characters 'a-z', 'A-Z', '0-9', '.'
6793 and '_'.
6794
6795 <expr> Is a standard HAProxy expression formed by a sample-fetch
6796 followed by some converters.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006797
6798 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006799 http-response set-var(sess.last_redir) res.hdr(location)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006800
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006801http-response silent-drop [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006802
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006803 This stops the evaluation of the rules and makes the client-facing connection
6804 suddenly disappear using a system-dependent way that tries to prevent the
6805 client from being notified. The effect it then that the client still sees an
6806 established connection while there's none on HAProxy. The purpose is to
6807 achieve a comparable effect to "tarpit" except that it doesn't use any local
6808 resource at all on the machine running HAProxy. It can resist much higher
6809 loads than "tarpit", and slow down stronger attackers. It is important to
6810 understand the impact of using this mechanism. All stateful equipment placed
6811 between the client and HAProxy (firewalls, proxies, load balancers) will also
6812 keep the established connection for a long time and may suffer from this
6813 action.
6814 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
6815 option is used to block the emission of a TCP reset. On other systems, the
6816 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
6817 router, though it's still delivered to local networks. Do not use it unless
6818 you fully understand how it works.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006819
Christopher Faulet46f95542019-12-20 10:07:22 +01006820http-response strict-mode { on | off }
6821
6822 This enables or disables the strict rewriting mode for following rules. It
6823 does not affect rules declared before it and it is only applicable on rules
6824 performing a rewrite on the responses. When the strict mode is enabled, any
6825 rewrite failure triggers an internal error. Otherwise, such errors are
6826 silently ignored. The purpose of the strict rewriting mode is to make some
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05006827 rewrites optional while others must be performed to continue the response
Christopher Faulet46f95542019-12-20 10:07:22 +01006828 processing.
6829
Christopher Faulet1aea50e2020-01-17 16:03:53 +01006830 By default, the strict rewriting mode is enabled. Its value is also reset
Christopher Faulet46f95542019-12-20 10:07:22 +01006831 when a ruleset evaluation ends. So, for instance, if you change the mode on
Daniel Corbett67a82712020-07-06 23:01:19 -04006832 the backend, the default mode is restored when HAProxy starts the frontend
Christopher Faulet46f95542019-12-20 10:07:22 +01006833 rules evaluation.
6834
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006835http-response track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
6836http-response track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
6837http-response track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02006838
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006839 This enables tracking of sticky counters from current response. Please refer
6840 to "http-request track-sc" for a complete description. The only difference
6841 from "http-request track-sc" is the <key> sample expression can only make use
6842 of samples in response (e.g. res.*, status etc.) and samples below Layer 6
6843 (e.g. SSL-related samples, see section 7.3.4). If the sample is not
6844 supported, haproxy will fail and warn while parsing the config.
6845
6846http-response unset-var(<var-name>) [ { if | unless } <condition> ]
6847
6848 This is used to unset a variable. See "http-response set-var" for details
6849 about <var-name>.
6850
6851 Example:
6852 http-response unset-var(sess.last_redir)
6853
Baptiste Assmann5ecb77f2013-10-06 23:24:13 +02006854
Willy Tarreau30631952015-08-06 15:05:24 +02006855http-reuse { never | safe | aggressive | always }
6856 Declare how idle HTTP connections may be shared between requests
6857
6858 May be used in sections: defaults | frontend | listen | backend
6859 yes | no | yes | yes
6860
6861 By default, a connection established between haproxy and the backend server
Olivier Houchard86006a52018-12-14 19:37:49 +01006862 which is considered safe for reuse is moved back to the server's idle
6863 connections pool so that any other request can make use of it. This is the
6864 "safe" strategy below.
Willy Tarreau30631952015-08-06 15:05:24 +02006865
6866 The argument indicates the desired connection reuse strategy :
6867
Olivier Houchard86006a52018-12-14 19:37:49 +01006868 - "never" : idle connections are never shared between sessions. This mode
6869 may be enforced to cancel a different strategy inherited from
6870 a defaults section or for troubleshooting. For example, if an
6871 old bogus application considers that multiple requests over
6872 the same connection come from the same client and it is not
6873 possible to fix the application, it may be desirable to
6874 disable connection sharing in a single backend. An example of
6875 such an application could be an old haproxy using cookie
6876 insertion in tunnel mode and not checking any request past the
6877 first one.
Willy Tarreau30631952015-08-06 15:05:24 +02006878
Olivier Houchard86006a52018-12-14 19:37:49 +01006879 - "safe" : this is the default and the recommended strategy. The first
6880 request of a session is always sent over its own connection,
6881 and only subsequent requests may be dispatched over other
6882 existing connections. This ensures that in case the server
6883 closes the connection when the request is being sent, the
6884 browser can decide to silently retry it. Since it is exactly
6885 equivalent to regular keep-alive, there should be no side
Amaury Denoyelle27179652020-10-14 18:17:12 +02006886 effects. There is also a special handling for the connections
6887 using protocols subject to Head-of-line blocking (backend with
6888 h2 or fcgi). In this case, when at least one stream is
6889 processed, the used connection is reserved to handle streams
6890 of the same session. When no more streams are processed, the
6891 connection is released and can be reused.
Willy Tarreau30631952015-08-06 15:05:24 +02006892
6893 - "aggressive" : this mode may be useful in webservices environments where
6894 all servers are not necessarily known and where it would be
6895 appreciable to deliver most first requests over existing
6896 connections. In this case, first requests are only delivered
6897 over existing connections that have been reused at least once,
6898 proving that the server correctly supports connection reuse.
6899 It should only be used when it's sure that the client can
6900 retry a failed request once in a while and where the benefit
Michael Prokop4438c602019-05-24 10:25:45 +02006901 of aggressive connection reuse significantly outweighs the
Willy Tarreau30631952015-08-06 15:05:24 +02006902 downsides of rare connection failures.
6903
6904 - "always" : this mode is only recommended when the path to the server is
6905 known for never breaking existing connections quickly after
6906 releasing them. It allows the first request of a session to be
6907 sent to an existing connection. This can provide a significant
6908 performance increase over the "safe" strategy when the backend
6909 is a cache farm, since such components tend to show a
Davor Ocelice9ed2812017-12-25 17:49:28 +01006910 consistent behavior and will benefit from the connection
Willy Tarreau30631952015-08-06 15:05:24 +02006911 sharing. It is recommended that the "http-keep-alive" timeout
6912 remains low in this mode so that no dead connections remain
6913 usable. In most cases, this will lead to the same performance
6914 gains as "aggressive" but with more risks. It should only be
6915 used when it improves the situation over "aggressive".
6916
6917 When http connection sharing is enabled, a great care is taken to respect the
Davor Ocelice9ed2812017-12-25 17:49:28 +01006918 connection properties and compatibility. Specifically :
6919 - connections made with "usesrc" followed by a client-dependent value
6920 ("client", "clientip", "hdr_ip") are marked private and never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02006921
Amaury Denoyelle7239c242020-10-15 16:41:09 +02006922 - connections sent to a server with a variable value as TLS SNI extension
6923 are marked private and are never shared. This is not the case if the SNI
6924 is guaranteed to be a constant, as for example using a literal string;
Willy Tarreau30631952015-08-06 15:05:24 +02006925
Lukas Tribusfd9b68c2018-10-27 20:06:59 +02006926 - connections with certain bogus authentication schemes (relying on the
6927 connection) like NTLM are detected, marked private and are never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02006928
Lukas Tribuse8adfeb2019-11-06 11:50:25 +01006929 A connection pool is involved and configurable with "pool-max-conn".
Willy Tarreau30631952015-08-06 15:05:24 +02006930
6931 Note: connection reuse improves the accuracy of the "server maxconn" setting,
6932 because almost no new connection will be established while idle connections
6933 remain available. This is particularly true with the "always" strategy.
6934
6935 See also : "option http-keep-alive", "server maxconn"
6936
6937
Mark Lamourinec2247f02012-01-04 13:02:01 -05006938http-send-name-header [<header>]
6939 Add the server name to a request. Use the header string given by <header>
Mark Lamourinec2247f02012-01-04 13:02:01 -05006940 May be used in sections: defaults | frontend | listen | backend
6941 yes | no | yes | yes
Mark Lamourinec2247f02012-01-04 13:02:01 -05006942 Arguments :
Mark Lamourinec2247f02012-01-04 13:02:01 -05006943 <header> The header string to use to send the server name
6944
Willy Tarreau81bef7e2019-10-07 14:58:02 +02006945 The "http-send-name-header" statement causes the header field named <header>
6946 to be set to the name of the target server at the moment the request is about
6947 to be sent on the wire. Any existing occurrences of this header are removed.
6948 Upon retries and redispatches, the header field is updated to always reflect
6949 the server being attempted to connect to. Given that this header is modified
6950 very late in the connection setup, it may have unexpected effects on already
6951 modified headers. For example using it with transport-level header such as
6952 connection, content-length, transfer-encoding and so on will likely result in
6953 invalid requests being sent to the server. Additionally it has been reported
6954 that this directive is currently being used as a way to overwrite the Host
6955 header field in outgoing requests; while this trick has been known to work
6956 as a side effect of the feature for some time, it is not officially supported
6957 and might possibly not work anymore in a future version depending on the
6958 technical difficulties this feature induces. A long-term solution instead
6959 consists in fixing the application which required this trick so that it binds
6960 to the correct host name.
Mark Lamourinec2247f02012-01-04 13:02:01 -05006961
6962 See also : "server"
6963
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01006964id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +02006965 Set a persistent ID to a proxy.
6966 May be used in sections : defaults | frontend | listen | backend
6967 no | yes | yes | yes
6968 Arguments : none
6969
6970 Set a persistent ID for the proxy. This ID must be unique and positive.
6971 An unused ID will automatically be assigned if unset. The first assigned
6972 value will be 1. This ID is currently only returned in statistics.
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01006973
6974
Cyril Bonté0d4bf012010-04-25 23:21:46 +02006975ignore-persist { if | unless } <condition>
6976 Declare a condition to ignore persistence
6977 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01006978 no | no | yes | yes
Cyril Bonté0d4bf012010-04-25 23:21:46 +02006979
6980 By default, when cookie persistence is enabled, every requests containing
6981 the cookie are unconditionally persistent (assuming the target server is up
6982 and running).
6983
6984 The "ignore-persist" statement allows one to declare various ACL-based
6985 conditions which, when met, will cause a request to ignore persistence.
6986 This is sometimes useful to load balance requests for static files, which
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03006987 often don't require persistence. This can also be used to fully disable
Cyril Bonté0d4bf012010-04-25 23:21:46 +02006988 persistence for a specific User-Agent (for example, some web crawler bots).
6989
Cyril Bonté0d4bf012010-04-25 23:21:46 +02006990 The persistence is ignored when an "if" condition is met, or unless an
6991 "unless" condition is met.
6992
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03006993 Example:
6994 acl url_static path_beg /static /images /img /css
6995 acl url_static path_end .gif .png .jpg .css .js
6996 ignore-persist if url_static
6997
Cyril Bonté0d4bf012010-04-25 23:21:46 +02006998 See also : "force-persist", "cookie", and section 7 about ACL usage.
6999
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007000load-server-state-from-file { global | local | none }
7001 Allow seamless reload of HAProxy
7002 May be used in sections: defaults | frontend | listen | backend
7003 yes | no | yes | yes
7004
7005 This directive points HAProxy to a file where server state from previous
7006 running process has been saved. That way, when starting up, before handling
7007 traffic, the new process can apply old states to servers exactly has if no
Davor Ocelice9ed2812017-12-25 17:49:28 +01007008 reload occurred. The purpose of the "load-server-state-from-file" directive is
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007009 to tell haproxy which file to use. For now, only 2 arguments to either prevent
7010 loading state or load states from a file containing all backends and servers.
7011 The state file can be generated by running the command "show servers state"
7012 over the stats socket and redirect output.
7013
Davor Ocelice9ed2812017-12-25 17:49:28 +01007014 The format of the file is versioned and is very specific. To understand it,
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007015 please read the documentation of the "show servers state" command (chapter
Willy Tarreau1af20c72017-06-23 16:01:14 +02007016 9.3 of Management Guide).
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007017
7018 Arguments:
7019 global load the content of the file pointed by the global directive
7020 named "server-state-file".
7021
7022 local load the content of the file pointed by the directive
7023 "server-state-file-name" if set. If not set, then the backend
7024 name is used as a file name.
7025
7026 none don't load any stat for this backend
7027
7028 Notes:
Willy Tarreaue5a60682016-11-09 14:54:53 +01007029 - server's IP address is preserved across reloads by default, but the
7030 order can be changed thanks to the server's "init-addr" setting. This
7031 means that an IP address change performed on the CLI at run time will
Davor Ocelice9ed2812017-12-25 17:49:28 +01007032 be preserved, and that any change to the local resolver (e.g. /etc/hosts)
Willy Tarreaue5a60682016-11-09 14:54:53 +01007033 will possibly not have any effect if the state file is in use.
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007034
7035 - server's weight is applied from previous running process unless it has
7036 has changed between previous and new configuration files.
7037
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007038 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007039
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007040 global
7041 stats socket /tmp/socket
7042 server-state-file /tmp/server_state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007043
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007044 defaults
7045 load-server-state-from-file global
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007046
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007047 backend bk
7048 server s1 127.0.0.1:22 check weight 11
7049 server s2 127.0.0.1:22 check weight 12
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007050
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007051
7052 Then one can run :
7053
7054 socat /tmp/socket - <<< "show servers state" > /tmp/server_state
7055
7056 Content of the file /tmp/server_state would be like this:
7057
7058 1
7059 # <field names skipped for the doc example>
7060 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
7061 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
7062
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007063 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007064
7065 global
7066 stats socket /tmp/socket
7067 server-state-base /etc/haproxy/states
7068
7069 defaults
7070 load-server-state-from-file local
7071
7072 backend bk
7073 server s1 127.0.0.1:22 check weight 11
7074 server s2 127.0.0.1:22 check weight 12
7075
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007076
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007077 Then one can run :
7078
7079 socat /tmp/socket - <<< "show servers state bk" > /etc/haproxy/states/bk
7080
7081 Content of the file /etc/haproxy/states/bk would be like this:
7082
7083 1
7084 # <field names skipped for the doc example>
7085 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
7086 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
7087
7088 See also: "server-state-file", "server-state-file-name", and
7089 "show servers state"
7090
Cyril Bonté0d4bf012010-04-25 23:21:46 +02007091
Willy Tarreau2769aa02007-12-27 18:26:09 +01007092log global
Jan Wagnerf2f5c4e2020-12-17 22:22:32 +01007093log <address> [len <length>] [format <format>] [sample <ranges>:<sample_size>]
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02007094 <facility> [<level> [<minlevel>]]
William Lallemand0f99e342011-10-12 17:50:54 +02007095no log
Willy Tarreau2769aa02007-12-27 18:26:09 +01007096 Enable per-instance logging of events and traffic.
7097 May be used in sections : defaults | frontend | listen | backend
7098 yes | yes | yes | yes
William Lallemand0f99e342011-10-12 17:50:54 +02007099
7100 Prefix :
7101 no should be used when the logger list must be flushed. For example,
7102 if you don't want to inherit from the default logger list. This
7103 prefix does not allow arguments.
7104
Willy Tarreau2769aa02007-12-27 18:26:09 +01007105 Arguments :
7106 global should be used when the instance's logging parameters are the
7107 same as the global ones. This is the most common usage. "global"
7108 replaces <address>, <facility> and <level> with those of the log
7109 entries found in the "global" section. Only one "log global"
7110 statement may be used per instance, and this form takes no other
7111 parameter.
7112
7113 <address> indicates where to send the logs. It takes the same format as
7114 for the "global" section's logs, and can be one of :
7115
7116 - An IPv4 address optionally followed by a colon (':') and a UDP
7117 port. If no port is specified, 514 is used by default (the
7118 standard syslog port).
7119
David du Colombier24bb5f52011-03-17 10:40:23 +01007120 - An IPv6 address followed by a colon (':') and optionally a UDP
7121 port. If no port is specified, 514 is used by default (the
7122 standard syslog port).
7123
Willy Tarreau2769aa02007-12-27 18:26:09 +01007124 - A filesystem path to a UNIX domain socket, keeping in mind
7125 considerations for chroot (be sure the path is accessible
7126 inside the chroot) and uid/gid (be sure the path is
Davor Ocelice9ed2812017-12-25 17:49:28 +01007127 appropriately writable).
Willy Tarreau2769aa02007-12-27 18:26:09 +01007128
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01007129 - A file descriptor number in the form "fd@<number>", which may
7130 point to a pipe, terminal, or socket. In this case unbuffered
7131 logs are used and one writev() call per log is performed. This
7132 is a bit expensive but acceptable for most workloads. Messages
7133 sent this way will not be truncated but may be dropped, in
7134 which case the DroppedLogs counter will be incremented. The
7135 writev() call is atomic even on pipes for messages up to
7136 PIPE_BUF size, which POSIX recommends to be at least 512 and
7137 which is 4096 bytes on most modern operating systems. Any
7138 larger message may be interleaved with messages from other
7139 processes. Exceptionally for debugging purposes the file
7140 descriptor may also be directed to a file, but doing so will
7141 significantly slow haproxy down as non-blocking calls will be
7142 ignored. Also there will be no way to purge nor rotate this
7143 file without restarting the process. Note that the configured
7144 syslog format is preserved, so the output is suitable for use
Willy Tarreauc1b06452018-11-12 11:57:56 +01007145 with a TCP syslog server. See also the "short" and "raw"
7146 formats below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01007147
7148 - "stdout" / "stderr", which are respectively aliases for "fd@1"
7149 and "fd@2", see above.
7150
Willy Tarreauc046d162019-08-30 15:24:59 +02007151 - A ring buffer in the form "ring@<name>", which will correspond
7152 to an in-memory ring buffer accessible over the CLI using the
7153 "show events" command, which will also list existing rings and
7154 their sizes. Such buffers are lost on reload or restart but
7155 when used as a complement this can help troubleshooting by
7156 having the logs instantly available.
7157
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01007158 You may want to reference some environment variables in the
7159 address parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01007160
Willy Tarreau18324f52014-06-27 18:10:07 +02007161 <length> is an optional maximum line length. Log lines larger than this
7162 value will be truncated before being sent. The reason is that
7163 syslog servers act differently on log line length. All servers
7164 support the default value of 1024, but some servers simply drop
7165 larger lines while others do log them. If a server supports long
7166 lines, it may make sense to set this value here in order to avoid
7167 truncating long lines. Similarly, if a server drops long lines,
7168 it is preferable to truncate them before sending them. Accepted
7169 values are 80 to 65535 inclusive. The default value of 1024 is
7170 generally fine for all standard usages. Some specific cases of
Davor Ocelice9ed2812017-12-25 17:49:28 +01007171 long captures or JSON-formatted logs may require larger values.
Willy Tarreau18324f52014-06-27 18:10:07 +02007172
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02007173 <ranges> A list of comma-separated ranges to identify the logs to sample.
7174 This is used to balance the load of the logs to send to the log
7175 server. The limits of the ranges cannot be null. They are numbered
7176 from 1. The size or period (in number of logs) of the sample must
7177 be set with <sample_size> parameter.
7178
7179 <sample_size>
7180 The size of the sample in number of logs to consider when balancing
7181 their logging loads. It is used to balance the load of the logs to
7182 send to the syslog server. This size must be greater or equal to the
7183 maximum of the high limits of the ranges.
7184 (see also <ranges> parameter).
7185
Willy Tarreauadb345d2018-11-12 07:56:13 +01007186 <format> is the log format used when generating syslog messages. It may be
7187 one of the following :
7188
Emeric Bruna0338b92020-11-27 16:24:34 +01007189 local Analog to rfc3164 syslog message format except that hostname
7190 field is stripped. This is the default.
7191 Note: option "log-send-hostname" switches the default to
7192 rfc3164.
7193
7194 rfc3164 The RFC3164 syslog message format.
Willy Tarreauadb345d2018-11-12 07:56:13 +01007195 (https://tools.ietf.org/html/rfc3164)
7196
7197 rfc5424 The RFC5424 syslog message format.
7198 (https://tools.ietf.org/html/rfc5424)
7199
Emeric Brun54648852020-07-06 15:54:06 +02007200 priority A message containing only a level plus syslog facility between
7201 angle brackets such as '<63>', followed by the text. The PID,
7202 date, time, process name and system name are omitted. This is
7203 designed to be used with a local log server.
7204
Willy Tarreaue8746a02018-11-12 08:45:00 +01007205 short A message containing only a level between angle brackets such as
7206 '<3>', followed by the text. The PID, date, time, process name
7207 and system name are omitted. This is designed to be used with a
7208 local log server. This format is compatible with what the
7209 systemd logger consumes.
7210
Emeric Brun54648852020-07-06 15:54:06 +02007211 timed A message containing only a level between angle brackets such as
7212 '<3>', followed by ISO date and by the text. The PID, process
7213 name and system name are omitted. This is designed to be
7214 used with a local log server.
7215
7216 iso A message containing only the ISO date, followed by the text.
7217 The PID, process name and system name are omitted. This is
7218 designed to be used with a local log server.
7219
Willy Tarreauc1b06452018-11-12 11:57:56 +01007220 raw A message containing only the text. The level, PID, date, time,
7221 process name and system name are omitted. This is designed to
7222 be used in containers or during development, where the severity
7223 only depends on the file descriptor used (stdout/stderr).
7224
Willy Tarreau2769aa02007-12-27 18:26:09 +01007225 <facility> must be one of the 24 standard syslog facilities :
7226
Willy Tarreaue8746a02018-11-12 08:45:00 +01007227 kern user mail daemon auth syslog lpr news
7228 uucp cron auth2 ftp ntp audit alert cron2
7229 local0 local1 local2 local3 local4 local5 local6 local7
7230
Willy Tarreauc1b06452018-11-12 11:57:56 +01007231 Note that the facility is ignored for the "short" and "raw"
7232 formats, but still required as a positional field. It is
7233 recommended to use "daemon" in this case to make it clear that
7234 it's only supposed to be used locally.
Willy Tarreau2769aa02007-12-27 18:26:09 +01007235
7236 <level> is optional and can be specified to filter outgoing messages. By
7237 default, all messages are sent. If a level is specified, only
7238 messages with a severity at least as important as this level
Willy Tarreauf7edefa2009-05-10 17:20:05 +02007239 will be sent. An optional minimum level can be specified. If it
7240 is set, logs emitted with a more severe level than this one will
7241 be capped to this level. This is used to avoid sending "emerg"
7242 messages on all terminals on some default syslog configurations.
7243 Eight levels are known :
Willy Tarreau2769aa02007-12-27 18:26:09 +01007244
7245 emerg alert crit err warning notice info debug
7246
William Lallemand0f99e342011-10-12 17:50:54 +02007247 It is important to keep in mind that it is the frontend which decides what to
7248 log from a connection, and that in case of content switching, the log entries
7249 from the backend will be ignored. Connections are logged at level "info".
Willy Tarreaucc6c8912009-02-22 10:53:55 +01007250
7251 However, backend log declaration define how and where servers status changes
7252 will be logged. Level "notice" will be used to indicate a server going up,
7253 "warning" will be used for termination signals and definitive service
7254 termination, and "alert" will be used for when a server goes down.
7255
7256 Note : According to RFC3164, messages are truncated to 1024 bytes before
7257 being emitted.
Willy Tarreau2769aa02007-12-27 18:26:09 +01007258
7259 Example :
7260 log global
Willy Tarreauc1b06452018-11-12 11:57:56 +01007261 log stdout format short daemon # send log to systemd
7262 log stdout format raw daemon # send everything to stdout
7263 log stderr format raw daemon notice # send important events to stderr
Willy Tarreauf7edefa2009-05-10 17:20:05 +02007264 log 127.0.0.1:514 local0 notice # only send important events
7265 log 127.0.0.1:514 local0 notice notice # same but limit output level
William Lallemandb2f07452015-05-12 14:27:13 +02007266 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
Willy Tarreaudad36a32013-03-11 01:20:04 +01007267
Willy Tarreau2769aa02007-12-27 18:26:09 +01007268
William Lallemand48940402012-01-30 16:47:22 +01007269log-format <string>
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01007270 Specifies the log format string to use for traffic logs
7271 May be used in sections: defaults | frontend | listen | backend
7272 yes | yes | yes | no
William Lallemand48940402012-01-30 16:47:22 +01007273
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01007274 This directive specifies the log format string that will be used for all logs
7275 resulting from traffic passing through the frontend using this line. If the
7276 directive is used in a defaults section, all subsequent frontends will use
7277 the same log format. Please see section 8.2.4 which covers the log format
7278 string in depth.
William Lallemand48940402012-01-30 16:47:22 +01007279
Guillaume de Lafond29f45602017-03-31 19:52:15 +02007280 "log-format" directive overrides previous "option tcplog", "log-format" and
7281 "option httplog" directives.
7282
Dragan Dosen7ad31542015-09-28 17:16:47 +02007283log-format-sd <string>
7284 Specifies the RFC5424 structured-data log format string
7285 May be used in sections: defaults | frontend | listen | backend
7286 yes | yes | yes | no
7287
7288 This directive specifies the RFC5424 structured-data log format string that
7289 will be used for all logs resulting from traffic passing through the frontend
7290 using this line. If the directive is used in a defaults section, all
7291 subsequent frontends will use the same log format. Please see section 8.2.4
7292 which covers the log format string in depth.
7293
7294 See https://tools.ietf.org/html/rfc5424#section-6.3 for more information
7295 about the RFC5424 structured-data part.
7296
7297 Note : This log format string will be used only for loggers that have set
7298 log format to "rfc5424".
7299
7300 Example :
7301 log-format-sd [exampleSDID@1234\ bytes=\"%B\"\ status=\"%ST\"]
7302
7303
Willy Tarreau094af4e2015-01-07 15:03:42 +01007304log-tag <string>
7305 Specifies the log tag to use for all outgoing logs
7306 May be used in sections: defaults | frontend | listen | backend
7307 yes | yes | yes | yes
7308
7309 Sets the tag field in the syslog header to this string. It defaults to the
7310 log-tag set in the global section, otherwise the program name as launched
7311 from the command line, which usually is "haproxy". Sometimes it can be useful
7312 to differentiate between multiple processes running on the same host, or to
7313 differentiate customer instances running in the same process. In the backend,
7314 logs about servers up/down will use this tag. As a hint, it can be convenient
7315 to set a log-tag related to a hosted customer in a defaults section then put
7316 all the frontends and backends for that customer, then start another customer
7317 in a new defaults section. See also the global "log-tag" directive.
Willy Tarreau2769aa02007-12-27 18:26:09 +01007318
Willy Tarreauc35362a2014-04-25 13:58:37 +02007319max-keep-alive-queue <value>
7320 Set the maximum server queue size for maintaining keep-alive connections
7321 May be used in sections: defaults | frontend | listen | backend
7322 yes | no | yes | yes
7323
7324 HTTP keep-alive tries to reuse the same server connection whenever possible,
7325 but sometimes it can be counter-productive, for example if a server has a lot
7326 of connections while other ones are idle. This is especially true for static
7327 servers.
7328
7329 The purpose of this setting is to set a threshold on the number of queued
7330 connections at which haproxy stops trying to reuse the same server and prefers
7331 to find another one. The default value, -1, means there is no limit. A value
7332 of zero means that keep-alive requests will never be queued. For very close
7333 servers which can be reached with a low latency and which are not sensible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01007334 breaking keep-alive, a low value is recommended (e.g. local static server can
Willy Tarreauc35362a2014-04-25 13:58:37 +02007335 use a value of 10 or less). For remote servers suffering from a high latency,
7336 higher values might be needed to cover for the latency and/or the cost of
7337 picking a different server.
7338
7339 Note that this has no impact on responses which are maintained to the same
7340 server consecutively to a 401 response. They will still go to the same server
7341 even if they have to be queued.
7342
7343 See also : "option http-server-close", "option prefer-last-server", server
7344 "maxconn" and cookie persistence.
7345
Olivier Houcharda4d4fdf2018-12-14 19:27:06 +01007346max-session-srv-conns <nb>
7347 Set the maximum number of outgoing connections we can keep idling for a given
7348 client session. The default is 5 (it precisely equals MAX_SRV_LIST which is
7349 defined at build time).
Willy Tarreauc35362a2014-04-25 13:58:37 +02007350
Willy Tarreau2769aa02007-12-27 18:26:09 +01007351maxconn <conns>
7352 Fix the maximum number of concurrent connections on a frontend
7353 May be used in sections : defaults | frontend | listen | backend
7354 yes | yes | yes | no
7355 Arguments :
7356 <conns> is the maximum number of concurrent connections the frontend will
7357 accept to serve. Excess connections will be queued by the system
7358 in the socket's listen queue and will be served once a connection
7359 closes.
7360
7361 If the system supports it, it can be useful on big sites to raise this limit
7362 very high so that haproxy manages connection queues, instead of leaving the
7363 clients with unanswered connection attempts. This value should not exceed the
7364 global maxconn. Also, keep in mind that a connection contains two buffers
Baptiste Assmann79fb45d2016-03-06 23:34:31 +01007365 of tune.bufsize (16kB by default) each, as well as some other data resulting
7366 in about 33 kB of RAM being consumed per established connection. That means
7367 that a medium system equipped with 1GB of RAM can withstand around
7368 20000-25000 concurrent connections if properly tuned.
Willy Tarreau2769aa02007-12-27 18:26:09 +01007369
7370 Also, when <conns> is set to large values, it is possible that the servers
7371 are not sized to accept such loads, and for this reason it is generally wise
7372 to assign them some reasonable connection limits.
7373
Willy Tarreauc8d5b952019-02-27 17:25:52 +01007374 When this value is set to zero, which is the default, the global "maxconn"
7375 value is used.
Vincent Bernat6341be52012-06-27 17:18:30 +02007376
Willy Tarreau2769aa02007-12-27 18:26:09 +01007377 See also : "server", global section's "maxconn", "fullconn"
7378
7379
Willy Tarreau77e0dae2020-10-14 15:44:27 +02007380mode { tcp|http }
Willy Tarreau2769aa02007-12-27 18:26:09 +01007381 Set the running mode or protocol of the instance
7382 May be used in sections : defaults | frontend | listen | backend
7383 yes | yes | yes | yes
7384 Arguments :
7385 tcp The instance will work in pure TCP mode. A full-duplex connection
7386 will be established between clients and servers, and no layer 7
7387 examination will be performed. This is the default mode. It
7388 should be used for SSL, SSH, SMTP, ...
7389
7390 http The instance will work in HTTP mode. The client request will be
7391 analyzed in depth before connecting to any server. Any request
7392 which is not RFC-compliant will be rejected. Layer 7 filtering,
7393 processing and switching will be possible. This is the mode which
7394 brings HAProxy most of its value.
7395
Cyril Bonté108cf6e2012-04-21 23:30:29 +02007396 When doing content switching, it is mandatory that the frontend and the
7397 backend are in the same mode (generally HTTP), otherwise the configuration
7398 will be refused.
Willy Tarreau2769aa02007-12-27 18:26:09 +01007399
Cyril Bonté108cf6e2012-04-21 23:30:29 +02007400 Example :
Willy Tarreau2769aa02007-12-27 18:26:09 +01007401 defaults http_instances
7402 mode http
7403
Willy Tarreau0ba27502007-12-24 16:55:16 +01007404
Cyril Bontéf0c60612010-02-06 14:44:47 +01007405monitor fail { if | unless } <condition>
Willy Tarreau2769aa02007-12-27 18:26:09 +01007406 Add a condition to report a failure to a monitor HTTP request.
Willy Tarreau0ba27502007-12-24 16:55:16 +01007407 May be used in sections : defaults | frontend | listen | backend
7408 no | yes | yes | no
Willy Tarreau0ba27502007-12-24 16:55:16 +01007409 Arguments :
7410 if <cond> the monitor request will fail if the condition is satisfied,
7411 and will succeed otherwise. The condition should describe a
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007412 combined test which must induce a failure if all conditions
Willy Tarreau0ba27502007-12-24 16:55:16 +01007413 are met, for instance a low number of servers both in a
7414 backend and its backup.
7415
7416 unless <cond> the monitor request will succeed only if the condition is
7417 satisfied, and will fail otherwise. Such a condition may be
7418 based on a test on the presence of a minimum number of active
7419 servers in a list of backends.
7420
7421 This statement adds a condition which can force the response to a monitor
7422 request to report a failure. By default, when an external component queries
7423 the URI dedicated to monitoring, a 200 response is returned. When one of the
7424 conditions above is met, haproxy will return 503 instead of 200. This is
7425 very useful to report a site failure to an external component which may base
7426 routing advertisements between multiple sites on the availability reported by
7427 haproxy. In this case, one would rely on an ACL involving the "nbsrv"
Willy Tarreauae94d4d2011-05-11 16:28:49 +02007428 criterion. Note that "monitor fail" only works in HTTP mode. Both status
7429 messages may be tweaked using "errorfile" or "errorloc" if needed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01007430
7431 Example:
7432 frontend www
Willy Tarreau2769aa02007-12-27 18:26:09 +01007433 mode http
Willy Tarreau0ba27502007-12-24 16:55:16 +01007434 acl site_dead nbsrv(dynamic) lt 2
7435 acl site_dead nbsrv(static) lt 2
7436 monitor-uri /site_alive
7437 monitor fail if site_dead
7438
Willy Tarreau9e9919d2020-10-14 15:55:23 +02007439 See also : "monitor-uri", "errorfile", "errorloc"
Willy Tarreau2769aa02007-12-27 18:26:09 +01007440
7441
Willy Tarreau2769aa02007-12-27 18:26:09 +01007442monitor-uri <uri>
7443 Intercept a URI used by external components' monitor requests
7444 May be used in sections : defaults | frontend | listen | backend
7445 yes | yes | yes | no
7446 Arguments :
7447 <uri> is the exact URI which we want to intercept to return HAProxy's
7448 health status instead of forwarding the request.
7449
7450 When an HTTP request referencing <uri> will be received on a frontend,
7451 HAProxy will not forward it nor log it, but instead will return either
7452 "HTTP/1.0 200 OK" or "HTTP/1.0 503 Service unavailable", depending on failure
7453 conditions defined with "monitor fail". This is normally enough for any
7454 front-end HTTP probe to detect that the service is UP and running without
7455 forwarding the request to a backend server. Note that the HTTP method, the
7456 version and all headers are ignored, but the request must at least be valid
7457 at the HTTP level. This keyword may only be used with an HTTP-mode frontend.
7458
Willy Tarreau721d8e02017-12-01 18:25:08 +01007459 Monitor requests are processed very early, just after the request is parsed
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02007460 and even before any "http-request". The only rulesets applied before are the
7461 tcp-request ones. They cannot be logged either, and it is the intended
7462 purpose. They are only used to report HAProxy's health to an upper component,
7463 nothing more. However, it is possible to add any number of conditions using
7464 "monitor fail" and ACLs so that the result can be adjusted to whatever check
7465 can be imagined (most often the number of available servers in a backend).
Willy Tarreau2769aa02007-12-27 18:26:09 +01007466
Christopher Faulet6072beb2020-02-18 15:34:58 +01007467 Note: if <uri> starts by a slash ('/'), the matching is performed against the
7468 request's path instead of the request's uri. It is a workaround to let
7469 the HTTP/2 requests match the monitor-uri. Indeed, in HTTP/2, clients
7470 are encouraged to send absolute URIs only.
7471
Willy Tarreau2769aa02007-12-27 18:26:09 +01007472 Example :
7473 # Use /haproxy_test to report haproxy's status
7474 frontend www
7475 mode http
7476 monitor-uri /haproxy_test
7477
Willy Tarreau9e9919d2020-10-14 15:55:23 +02007478 See also : "monitor fail"
Willy Tarreau2769aa02007-12-27 18:26:09 +01007479
Willy Tarreau0ba27502007-12-24 16:55:16 +01007480
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007481option abortonclose
7482no option abortonclose
7483 Enable or disable early dropping of aborted requests pending in queues.
7484 May be used in sections : defaults | frontend | listen | backend
7485 yes | no | yes | yes
7486 Arguments : none
7487
7488 In presence of very high loads, the servers will take some time to respond.
7489 The per-instance connection queue will inflate, and the response time will
7490 increase respective to the size of the queue times the average per-session
7491 response time. When clients will wait for more than a few seconds, they will
Willy Tarreau198a7442008-01-17 12:05:32 +01007492 often hit the "STOP" button on their browser, leaving a useless request in
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007493 the queue, and slowing down other users, and the servers as well, because the
7494 request will eventually be served, then aborted at the first error
7495 encountered while delivering the response.
7496
7497 As there is no way to distinguish between a full STOP and a simple output
7498 close on the client side, HTTP agents should be conservative and consider
7499 that the client might only have closed its output channel while waiting for
7500 the response. However, this introduces risks of congestion when lots of users
7501 do the same, and is completely useless nowadays because probably no client at
7502 all will close the session while waiting for the response. Some HTTP agents
Davor Ocelice9ed2812017-12-25 17:49:28 +01007503 support this behavior (Squid, Apache, HAProxy), and others do not (TUX, most
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007504 hardware-based load balancers). So the probability for a closed input channel
Willy Tarreau198a7442008-01-17 12:05:32 +01007505 to represent a user hitting the "STOP" button is close to 100%, and the risk
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007506 of being the single component to break rare but valid traffic is extremely
7507 low, which adds to the temptation to be able to abort a session early while
7508 still not served and not pollute the servers.
7509
Davor Ocelice9ed2812017-12-25 17:49:28 +01007510 In HAProxy, the user can choose the desired behavior using the option
7511 "abortonclose". By default (without the option) the behavior is HTTP
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007512 compliant and aborted requests will be served. But when the option is
7513 specified, a session with an incoming channel closed will be aborted while
7514 it is still possible, either pending in the queue for a connection slot, or
7515 during the connection establishment if the server has not yet acknowledged
7516 the connection request. This considerably reduces the queue size and the load
7517 on saturated servers when users are tempted to click on STOP, which in turn
Willy Tarreaud72758d2010-01-12 10:42:19 +01007518 reduces the response time for other users.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007519
7520 If this option has been enabled in a "defaults" section, it can be disabled
7521 in a specific instance by prepending the "no" keyword before it.
7522
7523 See also : "timeout queue" and server's "maxconn" and "maxqueue" parameters
7524
7525
Willy Tarreau4076a152009-04-02 15:18:36 +02007526option accept-invalid-http-request
7527no option accept-invalid-http-request
7528 Enable or disable relaxing of HTTP request parsing
7529 May be used in sections : defaults | frontend | listen | backend
7530 yes | yes | yes | no
7531 Arguments : none
7532
Willy Tarreau91852eb2015-05-01 13:26:00 +02007533 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02007534 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01007535 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02007536 forbidden characters are essentially used to build attacks exploiting server
7537 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
7538 server will emit invalid header names for whatever reason (configuration,
7539 implementation) and the issue will not be immediately fixed. In such a case,
7540 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau422246e2012-01-07 23:54:13 +01007541 even if that does not make sense, by specifying this option. Similarly, the
7542 list of characters allowed to appear in a URI is well defined by RFC3986, and
7543 chars 0-31, 32 (space), 34 ('"'), 60 ('<'), 62 ('>'), 92 ('\'), 94 ('^'), 96
7544 ('`'), 123 ('{'), 124 ('|'), 125 ('}'), 127 (delete) and anything above are
Davor Ocelice9ed2812017-12-25 17:49:28 +01007545 not allowed at all. HAProxy always blocks a number of them (0..32, 127). The
Willy Tarreau91852eb2015-05-01 13:26:00 +02007546 remaining ones are blocked by default unless this option is enabled. This
Willy Tarreau13317662015-05-01 13:47:08 +02007547 option also relaxes the test on the HTTP version, it allows HTTP/0.9 requests
7548 to pass through (no version specified) and multiple digits for both the major
7549 and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02007550
7551 This option should never be enabled by default as it hides application bugs
7552 and open security breaches. It should only be deployed after a problem has
7553 been confirmed.
7554
7555 When this option is enabled, erroneous header names will still be accepted in
7556 requests, but the complete request will be captured in order to permit later
Willy Tarreau422246e2012-01-07 23:54:13 +01007557 analysis using the "show errors" request on the UNIX stats socket. Similarly,
7558 requests containing invalid chars in the URI part will be logged. Doing this
Willy Tarreau4076a152009-04-02 15:18:36 +02007559 also helps confirming that the issue has been solved.
7560
7561 If this option has been enabled in a "defaults" section, it can be disabled
7562 in a specific instance by prepending the "no" keyword before it.
7563
7564 See also : "option accept-invalid-http-response" and "show errors" on the
7565 stats socket.
7566
7567
7568option accept-invalid-http-response
7569no option accept-invalid-http-response
7570 Enable or disable relaxing of HTTP response parsing
7571 May be used in sections : defaults | frontend | listen | backend
7572 yes | no | yes | yes
7573 Arguments : none
7574
Willy Tarreau91852eb2015-05-01 13:26:00 +02007575 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02007576 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01007577 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02007578 forbidden characters are essentially used to build attacks exploiting server
7579 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
7580 server will emit invalid header names for whatever reason (configuration,
7581 implementation) and the issue will not be immediately fixed. In such a case,
7582 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau91852eb2015-05-01 13:26:00 +02007583 even if that does not make sense, by specifying this option. This option also
7584 relaxes the test on the HTTP version format, it allows multiple digits for
7585 both the major and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02007586
7587 This option should never be enabled by default as it hides application bugs
7588 and open security breaches. It should only be deployed after a problem has
7589 been confirmed.
7590
7591 When this option is enabled, erroneous header names will still be accepted in
7592 responses, but the complete response will be captured in order to permit
7593 later analysis using the "show errors" request on the UNIX stats socket.
7594 Doing this also helps confirming that the issue has been solved.
7595
7596 If this option has been enabled in a "defaults" section, it can be disabled
7597 in a specific instance by prepending the "no" keyword before it.
7598
7599 See also : "option accept-invalid-http-request" and "show errors" on the
7600 stats socket.
7601
7602
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007603option allbackups
7604no option allbackups
7605 Use either all backup servers at a time or only the first one
7606 May be used in sections : defaults | frontend | listen | backend
7607 yes | no | yes | yes
7608 Arguments : none
7609
7610 By default, the first operational backup server gets all traffic when normal
7611 servers are all down. Sometimes, it may be preferred to use multiple backups
7612 at once, because one will not be enough. When "option allbackups" is enabled,
7613 the load balancing will be performed among all backup servers when all normal
7614 ones are unavailable. The same load balancing algorithm will be used and the
7615 servers' weights will be respected. Thus, there will not be any priority
7616 order between the backup servers anymore.
7617
7618 This option is mostly used with static server farms dedicated to return a
7619 "sorry" page when an application is completely offline.
7620
7621 If this option has been enabled in a "defaults" section, it can be disabled
7622 in a specific instance by prepending the "no" keyword before it.
7623
7624
7625option checkcache
7626no option checkcache
Godbach7056a352013-12-11 20:01:07 +08007627 Analyze all server responses and block responses with cacheable cookies
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007628 May be used in sections : defaults | frontend | listen | backend
7629 yes | no | yes | yes
7630 Arguments : none
7631
7632 Some high-level frameworks set application cookies everywhere and do not
7633 always let enough control to the developer to manage how the responses should
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007634 be cached. When a session cookie is returned on a cacheable object, there is a
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007635 high risk of session crossing or stealing between users traversing the same
7636 caches. In some situations, it is better to block the response than to let
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02007637 some sensitive session information go in the wild.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007638
7639 The option "checkcache" enables deep inspection of all server responses for
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007640 strict compliance with HTTP specification in terms of cacheability. It
Willy Tarreau198a7442008-01-17 12:05:32 +01007641 carefully checks "Cache-control", "Pragma" and "Set-cookie" headers in server
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007642 response to check if there's a risk of caching a cookie on a client-side
7643 proxy. When this option is enabled, the only responses which can be delivered
Willy Tarreau198a7442008-01-17 12:05:32 +01007644 to the client are :
Davor Ocelice9ed2812017-12-25 17:49:28 +01007645 - all those without "Set-Cookie" header;
Willy Tarreauc55ddce2017-12-21 11:41:38 +01007646 - all those with a return code other than 200, 203, 204, 206, 300, 301,
7647 404, 405, 410, 414, 501, provided that the server has not set a
Davor Ocelice9ed2812017-12-25 17:49:28 +01007648 "Cache-control: public" header field;
Willy Tarreau24ea0bc2017-12-21 11:32:55 +01007649 - all those that result from a request using a method other than GET, HEAD,
7650 OPTIONS, TRACE, provided that the server has not set a 'Cache-Control:
Davor Ocelice9ed2812017-12-25 17:49:28 +01007651 public' header field;
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007652 - those with a 'Pragma: no-cache' header
7653 - those with a 'Cache-control: private' header
7654 - those with a 'Cache-control: no-store' header
7655 - those with a 'Cache-control: max-age=0' header
7656 - those with a 'Cache-control: s-maxage=0' header
7657 - those with a 'Cache-control: no-cache' header
7658 - those with a 'Cache-control: no-cache="set-cookie"' header
7659 - those with a 'Cache-control: no-cache="set-cookie,' header
7660 (allowing other fields after set-cookie)
7661
7662 If a response doesn't respect these requirements, then it will be blocked
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02007663 just as if it was from an "http-response deny" rule, with an "HTTP 502 bad
7664 gateway". The session state shows "PH--" meaning that the proxy blocked the
7665 response during headers processing. Additionally, an alert will be sent in
7666 the logs so that admins are informed that there's something to be fixed.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007667
7668 Due to the high impact on the application, the application should be tested
7669 in depth with the option enabled before going to production. It is also a
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01007670 good practice to always activate it during tests, even if it is not used in
Davor Ocelice9ed2812017-12-25 17:49:28 +01007671 production, as it will report potentially dangerous application behaviors.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007672
7673 If this option has been enabled in a "defaults" section, it can be disabled
7674 in a specific instance by prepending the "no" keyword before it.
7675
7676
7677option clitcpka
7678no option clitcpka
7679 Enable or disable the sending of TCP keepalive packets on the client side
7680 May be used in sections : defaults | frontend | listen | backend
7681 yes | yes | yes | no
7682 Arguments : none
7683
7684 When there is a firewall or any session-aware component between a client and
7685 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01007686 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007687 components decides to expire a session which has remained idle for too long.
7688
7689 Enabling socket-level TCP keep-alives makes the system regularly send packets
7690 to the other end of the connection, leaving it active. The delay between
7691 keep-alive probes is controlled by the system only and depends both on the
7692 operating system and its tuning parameters.
7693
7694 It is important to understand that keep-alive packets are neither emitted nor
7695 received at the application level. It is only the network stacks which sees
7696 them. For this reason, even if one side of the proxy already uses keep-alives
7697 to maintain its connection alive, those keep-alive packets will not be
7698 forwarded to the other side of the proxy.
7699
7700 Please note that this has nothing to do with HTTP keep-alive.
7701
7702 Using option "clitcpka" enables the emission of TCP keep-alive probes on the
7703 client side of a connection, which should help when session expirations are
7704 noticed between HAProxy and a client.
7705
7706 If this option has been enabled in a "defaults" section, it can be disabled
7707 in a specific instance by prepending the "no" keyword before it.
7708
7709 See also : "option srvtcpka", "option tcpka"
7710
7711
Willy Tarreau0ba27502007-12-24 16:55:16 +01007712option contstats
7713 Enable continuous traffic statistics updates
7714 May be used in sections : defaults | frontend | listen | backend
7715 yes | yes | yes | no
7716 Arguments : none
7717
7718 By default, counters used for statistics calculation are incremented
7719 only when a session finishes. It works quite well when serving small
7720 objects, but with big ones (for example large images or archives) or
7721 with A/V streaming, a graph generated from haproxy counters looks like
Willy Tarreaudef0d222016-11-08 22:03:00 +01007722 a hedgehog. With this option enabled counters get incremented frequently
7723 along the session, typically every 5 seconds, which is often enough to
7724 produce clean graphs. Recounting touches a hotpath directly so it is not
7725 not enabled by default, as it can cause a lot of wakeups for very large
7726 session counts and cause a small performance drop.
Willy Tarreau0ba27502007-12-24 16:55:16 +01007727
Christopher Faulet89aed322020-06-02 17:33:56 +02007728option disable-h2-upgrade
7729no option disable-h2-upgrade
7730 Enable or disable the implicit HTTP/2 upgrade from an HTTP/1.x client
7731 connection.
7732 May be used in sections : defaults | frontend | listen | backend
7733 yes | yes | yes | no
7734 Arguments : none
7735
7736 By default, HAProxy is able to implicitly upgrade an HTTP/1.x client
7737 connection to an HTTP/2 connection if the first request it receives from a
7738 given HTTP connection matches the HTTP/2 connection preface (i.e. the string
7739 "PRI * HTTP/2.0\r\n\r\nSM\r\n\r\n"). This way, it is possible to support
7740 HTTP/1.x and HTTP/2 clients on a non-SSL connections. This option must be used to
7741 disable the implicit upgrade. Note this implicit upgrade is only supported
7742 for HTTP proxies, thus this option too. Note also it is possible to force the
7743 HTTP/2 on clear connections by specifying "proto h2" on the bind line.
7744
7745 If this option has been enabled in a "defaults" section, it can be disabled
7746 in a specific instance by prepending the "no" keyword before it.
Willy Tarreau0ba27502007-12-24 16:55:16 +01007747
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02007748option dontlog-normal
7749no option dontlog-normal
7750 Enable or disable logging of normal, successful connections
7751 May be used in sections : defaults | frontend | listen | backend
7752 yes | yes | yes | no
7753 Arguments : none
7754
7755 There are large sites dealing with several thousand connections per second
7756 and for which logging is a major pain. Some of them are even forced to turn
7757 logs off and cannot debug production issues. Setting this option ensures that
7758 normal connections, those which experience no error, no timeout, no retry nor
7759 redispatch, will not be logged. This leaves disk space for anomalies. In HTTP
7760 mode, the response status code is checked and return codes 5xx will still be
7761 logged.
7762
7763 It is strongly discouraged to use this option as most of the time, the key to
7764 complex issues is in the normal logs which will not be logged here. If you
7765 need to separate logs, see the "log-separate-errors" option instead.
7766
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007767 See also : "log", "dontlognull", "log-separate-errors" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02007768 logging.
7769
7770
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007771option dontlognull
7772no option dontlognull
7773 Enable or disable logging of null connections
7774 May be used in sections : defaults | frontend | listen | backend
7775 yes | yes | yes | no
7776 Arguments : none
7777
7778 In certain environments, there are components which will regularly connect to
7779 various systems to ensure that they are still alive. It can be the case from
7780 another load balancer as well as from monitoring systems. By default, even a
7781 simple port probe or scan will produce a log. If those connections pollute
7782 the logs too much, it is possible to enable option "dontlognull" to indicate
7783 that a connection on which no data has been transferred will not be logged,
Willy Tarreau0f228a02015-05-01 15:37:53 +02007784 which typically corresponds to those probes. Note that errors will still be
7785 returned to the client and accounted for in the stats. If this is not what is
7786 desired, option http-ignore-probes can be used instead.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007787
7788 It is generally recommended not to use this option in uncontrolled
Davor Ocelice9ed2812017-12-25 17:49:28 +01007789 environments (e.g. internet), otherwise scans and other malicious activities
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007790 would not be logged.
7791
7792 If this option has been enabled in a "defaults" section, it can be disabled
7793 in a specific instance by prepending the "no" keyword before it.
7794
Willy Tarreau9e9919d2020-10-14 15:55:23 +02007795 See also : "log", "http-ignore-probes", "monitor-uri", and
Willy Tarreau0f228a02015-05-01 15:37:53 +02007796 section 8 about logging.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007797
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007798
Willy Tarreau87cf5142011-08-19 22:57:24 +02007799option forwardfor [ except <network> ] [ header <name> ] [ if-none ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01007800 Enable insertion of the X-Forwarded-For header to requests sent to servers
7801 May be used in sections : defaults | frontend | listen | backend
7802 yes | yes | yes | yes
7803 Arguments :
7804 <network> is an optional argument used to disable this option for sources
7805 matching <network>
Ross Westaf72a1d2008-08-03 10:51:45 +02007806 <name> an optional argument to specify a different "X-Forwarded-For"
Willy Tarreaud72758d2010-01-12 10:42:19 +01007807 header name.
Willy Tarreauc27debf2008-01-06 08:57:02 +01007808
7809 Since HAProxy works in reverse-proxy mode, the servers see its IP address as
7810 their client address. This is sometimes annoying when the client's IP address
7811 is expected in server logs. To solve this problem, the well-known HTTP header
7812 "X-Forwarded-For" may be added by HAProxy to all requests sent to the server.
7813 This header contains a value representing the client's IP address. Since this
7814 header is always appended at the end of the existing header list, the server
7815 must be configured to always use the last occurrence of this header only. See
Ross Westaf72a1d2008-08-03 10:51:45 +02007816 the server's manual to find how to enable use of this standard header. Note
7817 that only the last occurrence of the header must be used, since it is really
7818 possible that the client has already brought one.
7819
Willy Tarreaud72758d2010-01-12 10:42:19 +01007820 The keyword "header" may be used to supply a different header name to replace
Ross Westaf72a1d2008-08-03 10:51:45 +02007821 the default "X-Forwarded-For". This can be useful where you might already
Davor Ocelice9ed2812017-12-25 17:49:28 +01007822 have a "X-Forwarded-For" header from a different application (e.g. stunnel),
Willy Tarreaud72758d2010-01-12 10:42:19 +01007823 and you need preserve it. Also if your backend server doesn't use the
Davor Ocelice9ed2812017-12-25 17:49:28 +01007824 "X-Forwarded-For" header and requires different one (e.g. Zeus Web Servers
Ross Westaf72a1d2008-08-03 10:51:45 +02007825 require "X-Cluster-Client-IP").
Willy Tarreauc27debf2008-01-06 08:57:02 +01007826
7827 Sometimes, a same HAProxy instance may be shared between a direct client
7828 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
7829 used to decrypt HTTPS traffic). It is possible to disable the addition of the
7830 header for a known source address or network by adding the "except" keyword
7831 followed by the network address. In this case, any source IP matching the
7832 network will not cause an addition of this header. Most common uses are with
7833 private networks or 127.0.0.1.
7834
Willy Tarreau87cf5142011-08-19 22:57:24 +02007835 Alternatively, the keyword "if-none" states that the header will only be
7836 added if it is not present. This should only be used in perfectly trusted
7837 environment, as this might cause a security issue if headers reaching haproxy
7838 are under the control of the end-user.
7839
Christopher Fauleteac279f2021-04-06 09:01:09 +02007840 Only IPv4 addresses are supported. "http-request add-header" or "http-request
7841 set-header" rules may be used to work around this limitation.
7842
Willy Tarreauc27debf2008-01-06 08:57:02 +01007843 This option may be specified either in the frontend or in the backend. If at
Ross Westaf72a1d2008-08-03 10:51:45 +02007844 least one of them uses it, the header will be added. Note that the backend's
7845 setting of the header subargument takes precedence over the frontend's if
Willy Tarreau87cf5142011-08-19 22:57:24 +02007846 both are defined. In the case of the "if-none" argument, if at least one of
7847 the frontend or the backend does not specify it, it wants the addition to be
7848 mandatory, so it wins.
Willy Tarreauc27debf2008-01-06 08:57:02 +01007849
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007850 Example :
Willy Tarreauc27debf2008-01-06 08:57:02 +01007851 # Public HTTP address also used by stunnel on the same machine
7852 frontend www
7853 mode http
7854 option forwardfor except 127.0.0.1 # stunnel already adds the header
7855
Ross Westaf72a1d2008-08-03 10:51:45 +02007856 # Those servers want the IP Address in X-Client
7857 backend www
7858 mode http
7859 option forwardfor header X-Client
7860
Willy Tarreau87cf5142011-08-19 22:57:24 +02007861 See also : "option httpclose", "option http-server-close",
Christopher Faulet315b39c2018-09-21 16:26:19 +02007862 "option http-keep-alive"
Willy Tarreauc27debf2008-01-06 08:57:02 +01007863
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02007864
Christopher Faulet98fbe952019-07-22 16:18:24 +02007865option h1-case-adjust-bogus-client
7866no option h1-case-adjust-bogus-client
7867 Enable or disable the case adjustment of HTTP/1 headers sent to bogus clients
7868 May be used in sections : defaults | frontend | listen | backend
7869 yes | yes | yes | no
7870 Arguments : none
7871
7872 There is no standard case for header names because, as stated in RFC7230,
7873 they are case-insensitive. So applications must handle them in a case-
7874 insensitive manner. But some bogus applications violate the standards and
7875 erroneously rely on the cases most commonly used by browsers. This problem
7876 becomes critical with HTTP/2 because all header names must be exchanged in
7877 lower case, and HAProxy follows the same convention. All header names are
7878 sent in lower case to clients and servers, regardless of the HTTP version.
7879
7880 When HAProxy receives an HTTP/1 response, its header names are converted to
7881 lower case and manipulated and sent this way to the clients. If a client is
7882 known to violate the HTTP standards and to fail to process a response coming
7883 from HAProxy, it is possible to transform the lower case header names to a
7884 different format when the response is formatted and sent to the client, by
7885 enabling this option and specifying the list of headers to be reformatted
7886 using the global directives "h1-case-adjust" or "h1-case-adjust-file". This
7887 must only be a temporary workaround for the time it takes the client to be
7888 fixed, because clients which require such workarounds might be vulnerable to
7889 content smuggling attacks and must absolutely be fixed.
7890
7891 Please note that this option will not affect standards-compliant clients.
7892
7893 If this option has been enabled in a "defaults" section, it can be disabled
7894 in a specific instance by prepending the "no" keyword before it.
7895
7896 See also: "option h1-case-adjust-bogus-server", "h1-case-adjust",
7897 "h1-case-adjust-file".
7898
7899
7900option h1-case-adjust-bogus-server
7901no option h1-case-adjust-bogus-server
7902 Enable or disable the case adjustment of HTTP/1 headers sent to bogus servers
7903 May be used in sections : defaults | frontend | listen | backend
7904 yes | no | yes | yes
7905 Arguments : none
7906
7907 There is no standard case for header names because, as stated in RFC7230,
7908 they are case-insensitive. So applications must handle them in a case-
7909 insensitive manner. But some bogus applications violate the standards and
7910 erroneously rely on the cases most commonly used by browsers. This problem
7911 becomes critical with HTTP/2 because all header names must be exchanged in
7912 lower case, and HAProxy follows the same convention. All header names are
7913 sent in lower case to clients and servers, regardless of the HTTP version.
7914
7915 When HAProxy receives an HTTP/1 request, its header names are converted to
7916 lower case and manipulated and sent this way to the servers. If a server is
7917 known to violate the HTTP standards and to fail to process a request coming
7918 from HAProxy, it is possible to transform the lower case header names to a
7919 different format when the request is formatted and sent to the server, by
7920 enabling this option and specifying the list of headers to be reformatted
7921 using the global directives "h1-case-adjust" or "h1-case-adjust-file". This
7922 must only be a temporary workaround for the time it takes the server to be
7923 fixed, because servers which require such workarounds might be vulnerable to
7924 content smuggling attacks and must absolutely be fixed.
7925
7926 Please note that this option will not affect standards-compliant servers.
7927
7928 If this option has been enabled in a "defaults" section, it can be disabled
7929 in a specific instance by prepending the "no" keyword before it.
7930
7931 See also: "option h1-case-adjust-bogus-client", "h1-case-adjust",
7932 "h1-case-adjust-file".
7933
7934
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02007935option http-buffer-request
7936no option http-buffer-request
7937 Enable or disable waiting for whole HTTP request body before proceeding
7938 May be used in sections : defaults | frontend | listen | backend
7939 yes | yes | yes | yes
7940 Arguments : none
7941
7942 It is sometimes desirable to wait for the body of an HTTP request before
7943 taking a decision. This is what is being done by "balance url_param" for
7944 example. The first use case is to buffer requests from slow clients before
7945 connecting to the server. Another use case consists in taking the routing
7946 decision based on the request body's contents. This option placed in a
7947 frontend or backend forces the HTTP processing to wait until either the whole
Christopher Faulet6db8a2e2019-11-19 16:27:25 +01007948 body is received or the request buffer is full. It can have undesired side
7949 effects with some applications abusing HTTP by expecting unbuffered
7950 transmissions between the frontend and the backend, so this should definitely
7951 not be used by default.
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02007952
Baptiste Assmanneccdf432015-10-28 13:49:01 +01007953 See also : "option http-no-delay", "timeout http-request"
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02007954
7955
Willy Tarreau0f228a02015-05-01 15:37:53 +02007956option http-ignore-probes
7957no option http-ignore-probes
7958 Enable or disable logging of null connections and request timeouts
7959 May be used in sections : defaults | frontend | listen | backend
7960 yes | yes | yes | no
7961 Arguments : none
7962
7963 Recently some browsers started to implement a "pre-connect" feature
7964 consisting in speculatively connecting to some recently visited web sites
7965 just in case the user would like to visit them. This results in many
7966 connections being established to web sites, which end up in 408 Request
7967 Timeout if the timeout strikes first, or 400 Bad Request when the browser
7968 decides to close them first. These ones pollute the log and feed the error
7969 counters. There was already "option dontlognull" but it's insufficient in
7970 this case. Instead, this option does the following things :
7971 - prevent any 400/408 message from being sent to the client if nothing
Davor Ocelice9ed2812017-12-25 17:49:28 +01007972 was received over a connection before it was closed;
7973 - prevent any log from being emitted in this situation;
Willy Tarreau0f228a02015-05-01 15:37:53 +02007974 - prevent any error counter from being incremented
7975
7976 That way the empty connection is silently ignored. Note that it is better
7977 not to use this unless it is clear that it is needed, because it will hide
7978 real problems. The most common reason for not receiving a request and seeing
7979 a 408 is due to an MTU inconsistency between the client and an intermediary
7980 element such as a VPN, which blocks too large packets. These issues are
7981 generally seen with POST requests as well as GET with large cookies. The logs
7982 are often the only way to detect them.
7983
7984 If this option has been enabled in a "defaults" section, it can be disabled
7985 in a specific instance by prepending the "no" keyword before it.
7986
7987 See also : "log", "dontlognull", "errorfile", and section 8 about logging.
7988
7989
Willy Tarreau16bfb022010-01-16 19:48:41 +01007990option http-keep-alive
7991no option http-keep-alive
7992 Enable or disable HTTP keep-alive from client to server
7993 May be used in sections : defaults | frontend | listen | backend
7994 yes | yes | yes | yes
7995 Arguments : none
7996
Willy Tarreau70dffda2014-01-30 03:07:23 +01007997 By default HAProxy operates in keep-alive mode with regards to persistent
7998 connections: for each connection it processes each request and response, and
Christopher Faulet315b39c2018-09-21 16:26:19 +02007999 leaves the connection idle on both sides between the end of a response and
8000 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02008001 as "option http-server-close" or "option httpclose". This option allows to
8002 set back the keep-alive mode, which can be useful when another mode was used
8003 in a defaults section.
Willy Tarreau70dffda2014-01-30 03:07:23 +01008004
8005 Setting "option http-keep-alive" enables HTTP keep-alive mode on the client-
8006 and server- sides. This provides the lowest latency on the client side (slow
Willy Tarreau16bfb022010-01-16 19:48:41 +01008007 network) and the fastest session reuse on the server side at the expense
8008 of maintaining idle connections to the servers. In general, it is possible
8009 with this option to achieve approximately twice the request rate that the
8010 "http-server-close" option achieves on small objects. There are mainly two
8011 situations where this option may be useful :
8012
8013 - when the server is non-HTTP compliant and authenticates the connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01008014 instead of requests (e.g. NTLM authentication)
Willy Tarreau16bfb022010-01-16 19:48:41 +01008015
8016 - when the cost of establishing the connection to the server is significant
8017 compared to the cost of retrieving the associated object from the server.
8018
8019 This last case can happen when the server is a fast static server of cache.
8020 In this case, the server will need to be properly tuned to support high enough
8021 connection counts because connections will last until the client sends another
8022 request.
8023
8024 If the client request has to go to another backend or another server due to
8025 content switching or the load balancing algorithm, the idle connection will
Willy Tarreau9420b122013-12-15 18:58:25 +01008026 immediately be closed and a new one re-opened. Option "prefer-last-server" is
8027 available to try optimize server selection so that if the server currently
8028 attached to an idle connection is usable, it will be used.
Willy Tarreau16bfb022010-01-16 19:48:41 +01008029
Willy Tarreau16bfb022010-01-16 19:48:41 +01008030 At the moment, logs will not indicate whether requests came from the same
8031 session or not. The accept date reported in the logs corresponds to the end
8032 of the previous request, and the request time corresponds to the time spent
8033 waiting for a new request. The keep-alive request time is still bound to the
8034 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
8035 not set.
8036
Christopher Faulet159e6672019-07-16 15:09:52 +02008037 This option disables and replaces any previous "option httpclose" or "option
8038 http-server-close". When backend and frontend options differ, all of these 4
8039 options have precedence over "option http-keep-alive".
Willy Tarreau16bfb022010-01-16 19:48:41 +01008040
Christopher Faulet315b39c2018-09-21 16:26:19 +02008041 See also : "option httpclose",, "option http-server-close",
Willy Tarreau9420b122013-12-15 18:58:25 +01008042 "option prefer-last-server", "option http-pretend-keepalive",
Frédéric Lécaille93d33162019-03-06 09:35:59 +01008043 and "1.1. The HTTP transaction model".
Willy Tarreau16bfb022010-01-16 19:48:41 +01008044
8045
Willy Tarreau96e31212011-05-30 18:10:30 +02008046option http-no-delay
8047no option http-no-delay
8048 Instruct the system to favor low interactive delays over performance in HTTP
8049 May be used in sections : defaults | frontend | listen | backend
8050 yes | yes | yes | yes
8051 Arguments : none
8052
8053 In HTTP, each payload is unidirectional and has no notion of interactivity.
8054 Any agent is expected to queue data somewhat for a reasonably low delay.
8055 There are some very rare server-to-server applications that abuse the HTTP
8056 protocol and expect the payload phase to be highly interactive, with many
8057 interleaved data chunks in both directions within a single request. This is
8058 absolutely not supported by the HTTP specification and will not work across
8059 most proxies or servers. When such applications attempt to do this through
8060 haproxy, it works but they will experience high delays due to the network
8061 optimizations which favor performance by instructing the system to wait for
8062 enough data to be available in order to only send full packets. Typical
8063 delays are around 200 ms per round trip. Note that this only happens with
8064 abnormal uses. Normal uses such as CONNECT requests nor WebSockets are not
8065 affected.
8066
8067 When "option http-no-delay" is present in either the frontend or the backend
8068 used by a connection, all such optimizations will be disabled in order to
8069 make the exchanges as fast as possible. Of course this offers no guarantee on
8070 the functionality, as it may break at any other place. But if it works via
8071 HAProxy, it will work as fast as possible. This option should never be used
8072 by default, and should never be used at all unless such a buggy application
8073 is discovered. The impact of using this option is an increase of bandwidth
8074 usage and CPU usage, which may significantly lower performance in high
8075 latency environments.
8076
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02008077 See also : "option http-buffer-request"
8078
Willy Tarreau96e31212011-05-30 18:10:30 +02008079
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008080option http-pretend-keepalive
8081no option http-pretend-keepalive
8082 Define whether haproxy will announce keepalive to the server or not
8083 May be used in sections : defaults | frontend | listen | backend
Christopher Faulet98db9762018-09-21 10:25:19 +02008084 yes | no | yes | yes
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008085 Arguments : none
8086
Christopher Faulet315b39c2018-09-21 16:26:19 +02008087 When running with "option http-server-close" or "option httpclose", haproxy
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008088 adds a "Connection: close" header to the request forwarded to the server.
8089 Unfortunately, when some servers see this header, they automatically refrain
8090 from using the chunked encoding for responses of unknown length, while this
8091 is totally unrelated. The immediate effect is that this prevents haproxy from
8092 maintaining the client connection alive. A second effect is that a client or
8093 a cache could receive an incomplete response without being aware of it, and
8094 consider the response complete.
8095
8096 By setting "option http-pretend-keepalive", haproxy will make the server
8097 believe it will keep the connection alive. The server will then not fall back
8098 to the abnormal undesired above. When haproxy gets the whole response, it
8099 will close the connection with the server just as it would do with the
Christopher Faulet315b39c2018-09-21 16:26:19 +02008100 "option httpclose". That way the client gets a normal response and the
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008101 connection is correctly closed on the server side.
8102
8103 It is recommended not to enable this option by default, because most servers
8104 will more efficiently close the connection themselves after the last packet,
8105 and release its buffers slightly earlier. Also, the added packet on the
8106 network could slightly reduce the overall peak performance. However it is
8107 worth noting that when this option is enabled, haproxy will have slightly
8108 less work to do. So if haproxy is the bottleneck on the whole architecture,
8109 enabling this option might save a few CPU cycles.
8110
Christopher Faulet98db9762018-09-21 10:25:19 +02008111 This option may be set in backend and listen sections. Using it in a frontend
8112 section will be ignored and a warning will be reported during startup. It is
8113 a backend related option, so there is no real reason to set it on a
8114 frontend. This option may be combined with "option httpclose", which will
8115 cause keepalive to be announced to the server and close to be announced to
8116 the client. This practice is discouraged though.
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008117
8118 If this option has been enabled in a "defaults" section, it can be disabled
8119 in a specific instance by prepending the "no" keyword before it.
8120
Christopher Faulet315b39c2018-09-21 16:26:19 +02008121 See also : "option httpclose", "option http-server-close", and
Willy Tarreau16bfb022010-01-16 19:48:41 +01008122 "option http-keep-alive"
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008123
Willy Tarreauc27debf2008-01-06 08:57:02 +01008124
Willy Tarreaub608feb2010-01-02 22:47:18 +01008125option http-server-close
8126no option http-server-close
8127 Enable or disable HTTP connection closing on the server side
8128 May be used in sections : defaults | frontend | listen | backend
8129 yes | yes | yes | yes
8130 Arguments : none
8131
Willy Tarreau70dffda2014-01-30 03:07:23 +01008132 By default HAProxy operates in keep-alive mode with regards to persistent
8133 connections: for each connection it processes each request and response, and
8134 leaves the connection idle on both sides between the end of a response and
8135 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02008136 as "option http-server-close" or "option httpclose". Setting "option
8137 http-server-close" enables HTTP connection-close mode on the server side
8138 while keeping the ability to support HTTP keep-alive and pipelining on the
8139 client side. This provides the lowest latency on the client side (slow
8140 network) and the fastest session reuse on the server side to save server
8141 resources, similarly to "option httpclose". It also permits non-keepalive
8142 capable servers to be served in keep-alive mode to the clients if they
8143 conform to the requirements of RFC7230. Please note that some servers do not
8144 always conform to those requirements when they see "Connection: close" in the
8145 request. The effect will be that keep-alive will never be used. A workaround
8146 consists in enabling "option http-pretend-keepalive".
Willy Tarreaub608feb2010-01-02 22:47:18 +01008147
8148 At the moment, logs will not indicate whether requests came from the same
8149 session or not. The accept date reported in the logs corresponds to the end
8150 of the previous request, and the request time corresponds to the time spent
8151 waiting for a new request. The keep-alive request time is still bound to the
Willy Tarreaub16a5742010-01-10 14:46:16 +01008152 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
8153 not set.
Willy Tarreaub608feb2010-01-02 22:47:18 +01008154
8155 This option may be set both in a frontend and in a backend. It is enabled if
8156 at least one of the frontend or backend holding a connection has it enabled.
Christopher Faulet159e6672019-07-16 15:09:52 +02008157 It disables and replaces any previous "option httpclose" or "option
8158 http-keep-alive". Please check section 4 ("Proxies") to see how this option
8159 combines with others when frontend and backend options differ.
Willy Tarreaub608feb2010-01-02 22:47:18 +01008160
8161 If this option has been enabled in a "defaults" section, it can be disabled
8162 in a specific instance by prepending the "no" keyword before it.
8163
Christopher Faulet315b39c2018-09-21 16:26:19 +02008164 See also : "option httpclose", "option http-pretend-keepalive",
8165 "option http-keep-alive", and "1.1. The HTTP transaction model".
Willy Tarreaub608feb2010-01-02 22:47:18 +01008166
Willy Tarreau88d349d2010-01-25 12:15:43 +01008167option http-use-proxy-header
Cyril Bontéf0c60612010-02-06 14:44:47 +01008168no option http-use-proxy-header
Willy Tarreau88d349d2010-01-25 12:15:43 +01008169 Make use of non-standard Proxy-Connection header instead of Connection
8170 May be used in sections : defaults | frontend | listen | backend
8171 yes | yes | yes | no
8172 Arguments : none
8173
Lukas Tribus23953682017-04-28 13:24:30 +00008174 While RFC7230 explicitly states that HTTP/1.1 agents must use the
Willy Tarreau88d349d2010-01-25 12:15:43 +01008175 Connection header to indicate their wish of persistent or non-persistent
8176 connections, both browsers and proxies ignore this header for proxied
8177 connections and make use of the undocumented, non-standard Proxy-Connection
8178 header instead. The issue begins when trying to put a load balancer between
8179 browsers and such proxies, because there will be a difference between what
8180 haproxy understands and what the client and the proxy agree on.
8181
8182 By setting this option in a frontend, haproxy can automatically switch to use
8183 that non-standard header if it sees proxied requests. A proxied request is
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01008184 defined here as one where the URI begins with neither a '/' nor a '*'. This
8185 is incompatible with the HTTP tunnel mode. Note that this option can only be
8186 specified in a frontend and will affect the request along its whole life.
Willy Tarreau88d349d2010-01-25 12:15:43 +01008187
Willy Tarreau844a7e72010-01-31 21:46:18 +01008188 Also, when this option is set, a request which requires authentication will
8189 automatically switch to use proxy authentication headers if it is itself a
8190 proxied request. That makes it possible to check or enforce authentication in
8191 front of an existing proxy.
8192
Willy Tarreau88d349d2010-01-25 12:15:43 +01008193 This option should normally never be used, except in front of a proxy.
8194
Christopher Faulet315b39c2018-09-21 16:26:19 +02008195 See also : "option httpclose", and "option http-server-close".
Willy Tarreau88d349d2010-01-25 12:15:43 +01008196
Willy Tarreaud63335a2010-02-26 12:56:52 +01008197option httpchk
8198option httpchk <uri>
8199option httpchk <method> <uri>
8200option httpchk <method> <uri> <version>
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02008201 Enables HTTP protocol to check on the servers health
Willy Tarreaud63335a2010-02-26 12:56:52 +01008202 May be used in sections : defaults | frontend | listen | backend
8203 yes | no | yes | yes
8204 Arguments :
8205 <method> is the optional HTTP method used with the requests. When not set,
8206 the "OPTIONS" method is used, as it generally requires low server
8207 processing and is easy to filter out from the logs. Any method
8208 may be used, though it is not recommended to invent non-standard
8209 ones.
8210
8211 <uri> is the URI referenced in the HTTP requests. It defaults to " / "
8212 which is accessible by default on almost any server, but may be
8213 changed to any other URI. Query strings are permitted.
8214
8215 <version> is the optional HTTP version string. It defaults to "HTTP/1.0"
8216 but some servers might behave incorrectly in HTTP 1.0, so turning
8217 it to HTTP/1.1 may sometimes help. Note that the Host field is
Christopher Faulet8acb1282020-04-09 08:44:06 +02008218 mandatory in HTTP/1.1, use "http-check send" directive to add it.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008219
8220 By default, server health checks only consist in trying to establish a TCP
8221 connection. When "option httpchk" is specified, a complete HTTP request is
8222 sent once the TCP connection is established, and responses 2xx and 3xx are
8223 considered valid, while all other ones indicate a server failure, including
8224 the lack of any response.
8225
Christopher Faulete5870d82020-04-15 11:32:03 +02008226 Combined with "http-check" directives, it is possible to customize the
8227 request sent during the HTTP health checks or the matching rules on the
8228 response. It is also possible to configure a send/expect sequence, just like
8229 with the directive "tcp-check" for TCP health checks.
8230
8231 The server configuration is used by default to open connections to perform
8232 HTTP health checks. By it is also possible to overwrite server parameters
8233 using "http-check connect" rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008234
Christopher Faulete5870d82020-04-15 11:32:03 +02008235 "httpchk" option does not necessarily require an HTTP backend, it also works
8236 with plain TCP backends. This is particularly useful to check simple scripts
Christopher Faulet14cd3162020-04-16 14:50:06 +02008237 bound to some dedicated ports using the inetd daemon. However, it will always
Daniel Corbett67a82712020-07-06 23:01:19 -04008238 internally relies on an HTX multiplexer. Thus, it means the request
Christopher Faulet14cd3162020-04-16 14:50:06 +02008239 formatting and the response parsing will be strict.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008240
Christopher Faulet8acb1282020-04-09 08:44:06 +02008241 Note : For a while, there was no way to add headers or body in the request
8242 used for HTTP health checks. So a workaround was to hide it at the end
8243 of the version string with a "\r\n" after the version. It is now
8244 deprecated. The directive "http-check send" must be used instead.
8245
Willy Tarreaud63335a2010-02-26 12:56:52 +01008246 Examples :
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02008247 # Relay HTTPS traffic to Apache instance and check service availability
8248 # using HTTP request "OPTIONS * HTTP/1.1" on port 80.
8249 backend https_relay
8250 mode tcp
8251 option httpchk OPTIONS * HTTP/1.1
8252 http-check send hdr Host www
8253 server apache1 192.168.1.1:443 check port 80
Willy Tarreaud63335a2010-02-26 12:56:52 +01008254
Simon Hormanafc47ee2013-11-25 10:46:35 +09008255 See also : "option ssl-hello-chk", "option smtpchk", "option mysql-check",
8256 "option pgsql-check", "http-check" and the "check", "port" and
8257 "inter" server options.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008258
8259
Willy Tarreauc27debf2008-01-06 08:57:02 +01008260option httpclose
8261no option httpclose
Christopher Faulet315b39c2018-09-21 16:26:19 +02008262 Enable or disable HTTP connection closing
Willy Tarreauc27debf2008-01-06 08:57:02 +01008263 May be used in sections : defaults | frontend | listen | backend
8264 yes | yes | yes | yes
8265 Arguments : none
8266
Willy Tarreau70dffda2014-01-30 03:07:23 +01008267 By default HAProxy operates in keep-alive mode with regards to persistent
8268 connections: for each connection it processes each request and response, and
8269 leaves the connection idle on both sides between the end of a response and
8270 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02008271 as "option http-server-close" or "option httpclose".
Willy Tarreau70dffda2014-01-30 03:07:23 +01008272
Christopher Faulet315b39c2018-09-21 16:26:19 +02008273 If "option httpclose" is set, HAProxy will close connections with the server
8274 and the client as soon as the request and the response are received. It will
John Roeslerfb2fce12019-07-10 15:45:51 -05008275 also check if a "Connection: close" header is already set in each direction,
Christopher Faulet315b39c2018-09-21 16:26:19 +02008276 and will add one if missing. Any "Connection" header different from "close"
8277 will also be removed.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008278
Christopher Faulet315b39c2018-09-21 16:26:19 +02008279 This option may also be combined with "option http-pretend-keepalive", which
8280 will disable sending of the "Connection: close" header, but will still cause
8281 the connection to be closed once the whole response is received.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008282
8283 This option may be set both in a frontend and in a backend. It is enabled if
8284 at least one of the frontend or backend holding a connection has it enabled.
Christopher Faulet159e6672019-07-16 15:09:52 +02008285 It disables and replaces any previous "option http-server-close" or "option
8286 http-keep-alive". Please check section 4 ("Proxies") to see how this option
8287 combines with others when frontend and backend options differ.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008288
8289 If this option has been enabled in a "defaults" section, it can be disabled
8290 in a specific instance by prepending the "no" keyword before it.
8291
Christopher Faulet315b39c2018-09-21 16:26:19 +02008292 See also : "option http-server-close" and "1.1. The HTTP transaction model".
Willy Tarreauc27debf2008-01-06 08:57:02 +01008293
8294
Emeric Brun3a058f32009-06-30 18:26:00 +02008295option httplog [ clf ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01008296 Enable logging of HTTP request, session state and timers
8297 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01008298 yes | yes | yes | no
Emeric Brun3a058f32009-06-30 18:26:00 +02008299 Arguments :
8300 clf if the "clf" argument is added, then the output format will be
8301 the CLF format instead of HAProxy's default HTTP format. You can
8302 use this when you need to feed HAProxy's logs through a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01008303 log analyzer which only support the CLF format and which is not
Emeric Brun3a058f32009-06-30 18:26:00 +02008304 extensible.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008305
8306 By default, the log output format is very poor, as it only contains the
8307 source and destination addresses, and the instance name. By specifying
8308 "option httplog", each log line turns into a much richer format including,
8309 but not limited to, the HTTP request, the connection timers, the session
8310 status, the connections numbers, the captured headers and cookies, the
8311 frontend, backend and server name, and of course the source address and
8312 ports.
8313
PiBa-NLbd556bf2014-12-11 21:31:54 +01008314 Specifying only "option httplog" will automatically clear the 'clf' mode
8315 if it was set by default.
Emeric Brun3a058f32009-06-30 18:26:00 +02008316
Guillaume de Lafond29f45602017-03-31 19:52:15 +02008317 "option httplog" overrides any previous "log-format" directive.
8318
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008319 See also : section 8 about logging.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008320
Willy Tarreau55165fe2009-05-10 12:02:55 +02008321
8322option http_proxy
8323no option http_proxy
8324 Enable or disable plain HTTP proxy mode
8325 May be used in sections : defaults | frontend | listen | backend
8326 yes | yes | yes | yes
8327 Arguments : none
8328
8329 It sometimes happens that people need a pure HTTP proxy which understands
8330 basic proxy requests without caching nor any fancy feature. In this case,
8331 it may be worth setting up an HAProxy instance with the "option http_proxy"
8332 set. In this mode, no server is declared, and the connection is forwarded to
8333 the IP address and port found in the URL after the "http://" scheme.
8334
8335 No host address resolution is performed, so this only works when pure IP
8336 addresses are passed. Since this option's usage perimeter is rather limited,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01008337 it will probably be used only by experts who know they need exactly it. This
8338 is incompatible with the HTTP tunnel mode.
Willy Tarreau55165fe2009-05-10 12:02:55 +02008339
8340 If this option has been enabled in a "defaults" section, it can be disabled
8341 in a specific instance by prepending the "no" keyword before it.
8342
8343 Example :
8344 # this backend understands HTTP proxy requests and forwards them directly.
8345 backend direct_forward
8346 option httpclose
8347 option http_proxy
8348
8349 See also : "option httpclose"
8350
Willy Tarreau211ad242009-10-03 21:45:07 +02008351
Jamie Gloudon801a0a32012-08-25 00:18:33 -04008352option independent-streams
8353no option independent-streams
8354 Enable or disable independent timeout processing for both directions
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02008355 May be used in sections : defaults | frontend | listen | backend
8356 yes | yes | yes | yes
8357 Arguments : none
8358
8359 By default, when data is sent over a socket, both the write timeout and the
8360 read timeout for that socket are refreshed, because we consider that there is
8361 activity on that socket, and we have no other means of guessing if we should
8362 receive data or not.
8363
Davor Ocelice9ed2812017-12-25 17:49:28 +01008364 While this default behavior is desirable for almost all applications, there
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02008365 exists a situation where it is desirable to disable it, and only refresh the
8366 read timeout if there are incoming data. This happens on sessions with large
8367 timeouts and low amounts of exchanged data such as telnet session. If the
8368 server suddenly disappears, the output data accumulates in the system's
8369 socket buffers, both timeouts are correctly refreshed, and there is no way
8370 to know the server does not receive them, so we don't timeout. However, when
8371 the underlying protocol always echoes sent data, it would be enough by itself
8372 to detect the issue using the read timeout. Note that this problem does not
8373 happen with more verbose protocols because data won't accumulate long in the
8374 socket buffers.
8375
8376 When this option is set on the frontend, it will disable read timeout updates
8377 on data sent to the client. There probably is little use of this case. When
8378 the option is set on the backend, it will disable read timeout updates on
8379 data sent to the server. Doing so will typically break large HTTP posts from
8380 slow lines, so use it with caution.
8381
Willy Tarreauce887fd2012-05-12 12:50:00 +02008382 See also : "timeout client", "timeout server" and "timeout tunnel"
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02008383
8384
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02008385option ldap-check
8386 Use LDAPv3 health checks for server testing
8387 May be used in sections : defaults | frontend | listen | backend
8388 yes | no | yes | yes
8389 Arguments : none
8390
8391 It is possible to test that the server correctly talks LDAPv3 instead of just
8392 testing that it accepts the TCP connection. When this option is set, an
8393 LDAPv3 anonymous simple bind message is sent to the server, and the response
8394 is analyzed to find an LDAPv3 bind response message.
8395
8396 The server is considered valid only when the LDAP response contains success
8397 resultCode (http://tools.ietf.org/html/rfc4511#section-4.1.9).
8398
8399 Logging of bind requests is server dependent see your documentation how to
8400 configure it.
8401
8402 Example :
8403 option ldap-check
8404
8405 See also : "option httpchk"
8406
8407
Simon Horman98637e52014-06-20 12:30:16 +09008408option external-check
8409 Use external processes for server health checks
8410 May be used in sections : defaults | frontend | listen | backend
8411 yes | no | yes | yes
8412
8413 It is possible to test the health of a server using an external command.
8414 This is achieved by running the executable set using "external-check
8415 command".
8416
8417 Requires the "external-check" global to be set.
8418
8419 See also : "external-check", "external-check command", "external-check path"
8420
8421
Willy Tarreau211ad242009-10-03 21:45:07 +02008422option log-health-checks
8423no option log-health-checks
Willy Tarreaubef1b322014-05-13 21:01:39 +02008424 Enable or disable logging of health checks status updates
Willy Tarreau211ad242009-10-03 21:45:07 +02008425 May be used in sections : defaults | frontend | listen | backend
8426 yes | no | yes | yes
8427 Arguments : none
8428
Willy Tarreaubef1b322014-05-13 21:01:39 +02008429 By default, failed health check are logged if server is UP and successful
8430 health checks are logged if server is DOWN, so the amount of additional
8431 information is limited.
Willy Tarreau211ad242009-10-03 21:45:07 +02008432
Willy Tarreaubef1b322014-05-13 21:01:39 +02008433 When this option is enabled, any change of the health check status or to
8434 the server's health will be logged, so that it becomes possible to know
8435 that a server was failing occasional checks before crashing, or exactly when
8436 it failed to respond a valid HTTP status, then when the port started to
8437 reject connections, then when the server stopped responding at all.
8438
Davor Ocelice9ed2812017-12-25 17:49:28 +01008439 Note that status changes not caused by health checks (e.g. enable/disable on
Willy Tarreaubef1b322014-05-13 21:01:39 +02008440 the CLI) are intentionally not logged by this option.
Willy Tarreau211ad242009-10-03 21:45:07 +02008441
Willy Tarreaubef1b322014-05-13 21:01:39 +02008442 See also: "option httpchk", "option ldap-check", "option mysql-check",
8443 "option pgsql-check", "option redis-check", "option smtpchk",
8444 "option tcp-check", "log" and section 8 about logging.
Willy Tarreau211ad242009-10-03 21:45:07 +02008445
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02008446
8447option log-separate-errors
8448no option log-separate-errors
8449 Change log level for non-completely successful connections
8450 May be used in sections : defaults | frontend | listen | backend
8451 yes | yes | yes | no
8452 Arguments : none
8453
8454 Sometimes looking for errors in logs is not easy. This option makes haproxy
8455 raise the level of logs containing potentially interesting information such
8456 as errors, timeouts, retries, redispatches, or HTTP status codes 5xx. The
8457 level changes from "info" to "err". This makes it possible to log them
8458 separately to a different file with most syslog daemons. Be careful not to
8459 remove them from the original file, otherwise you would lose ordering which
8460 provides very important information.
8461
8462 Using this option, large sites dealing with several thousand connections per
8463 second may log normal traffic to a rotating buffer and only archive smaller
8464 error logs.
8465
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008466 See also : "log", "dontlognull", "dontlog-normal" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02008467 logging.
8468
Willy Tarreauc27debf2008-01-06 08:57:02 +01008469
8470option logasap
8471no option logasap
Jerome Magnin95fb57b2020-04-23 19:01:17 +02008472 Enable or disable early logging.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008473 May be used in sections : defaults | frontend | listen | backend
8474 yes | yes | yes | no
8475 Arguments : none
8476
Jerome Magnin95fb57b2020-04-23 19:01:17 +02008477 By default, logs are emitted when all the log format variables and sample
8478 fetches used in the definition of the log-format string return a value, or
8479 when the session is terminated. This allows the built in log-format strings
8480 to account for the transfer time, or the number of bytes in log messages.
8481
8482 When handling long lived connections such as large file transfers or RDP,
8483 it may take a while for the request or connection to appear in the logs.
8484 Using "option logasap", the log message is created as soon as the server
8485 connection is established in mode tcp, or as soon as the server sends the
8486 complete headers in mode http. Missing information in the logs will be the
Ilya Shipitsin4329a9a2020-05-05 21:17:10 +05008487 total number of bytes which will only indicate the amount of data transferred
Jerome Magnin95fb57b2020-04-23 19:01:17 +02008488 before the message was created and the total time which will not take the
8489 remainder of the connection life or transfer time into account. For the case
8490 of HTTP, it is good practice to capture the Content-Length response header
8491 so that the logs at least indicate how many bytes are expected to be
Ilya Shipitsin4329a9a2020-05-05 21:17:10 +05008492 transferred.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008493
Willy Tarreaucc6c8912009-02-22 10:53:55 +01008494 Examples :
8495 listen http_proxy 0.0.0.0:80
8496 mode http
8497 option httplog
8498 option logasap
8499 log 192.168.2.200 local3
8500
8501 >>> Feb 6 12:14:14 localhost \
8502 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
8503 static/srv1 9/10/7/14/+30 200 +243 - - ---- 3/1/1/1/0 1/0 \
8504 "GET /image.iso HTTP/1.0"
8505
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008506 See also : "option httplog", "capture response header", and section 8 about
Willy Tarreauc27debf2008-01-06 08:57:02 +01008507 logging.
8508
8509
Christopher Faulet62f79fe2020-05-18 18:13:03 +02008510option mysql-check [ user <username> [ { post-41 | pre-41 } ] ]
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02008511 Use MySQL health checks for server testing
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01008512 May be used in sections : defaults | frontend | listen | backend
8513 yes | no | yes | yes
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02008514 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008515 <username> This is the username which will be used when connecting to MySQL
8516 server.
Christopher Faulet62f79fe2020-05-18 18:13:03 +02008517 post-41 Send post v4.1 client compatible checks (the default)
8518 pre-41 Send pre v4.1 client compatible checks
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02008519
8520 If you specify a username, the check consists of sending two MySQL packet,
8521 one Client Authentication packet, and one QUIT packet, to correctly close
Davor Ocelice9ed2812017-12-25 17:49:28 +01008522 MySQL session. We then parse the MySQL Handshake Initialization packet and/or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02008523 Error packet. It is a basic but useful test which does not produce error nor
Daniel Blacka1d2bbc2021-07-01 12:09:32 +10008524 aborted connect on the server. However, it requires an unlocked authorised
8525 user without a password. To create a basic limited user in MySQL with
8526 optional resource limits:
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02008527
Daniel Blacka1d2bbc2021-07-01 12:09:32 +10008528 CREATE USER '<username>'@'<ip_of_haproxy|network_of_haproxy/netmask>'
8529 /*!50701 WITH MAX_QUERIES_PER_HOUR 1 MAX_UPDATES_PER_HOUR 0 */
8530 /*M!100201 MAX_STATEMENT_TIME 0.0001 */;
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02008531
8532 If you don't specify a username (it is deprecated and not recommended), the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008533 check only consists in parsing the Mysql Handshake Initialization packet or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02008534 Error packet, we don't send anything in this mode. It was reported that it
8535 can generate lockout if check is too frequent and/or if there is not enough
8536 traffic. In fact, you need in this case to check MySQL "max_connect_errors"
8537 value as if a connection is established successfully within fewer than MySQL
8538 "max_connect_errors" attempts after a previous connection was interrupted,
8539 the error count for the host is cleared to zero. If HAProxy's server get
8540 blocked, the "FLUSH HOSTS" statement is the only way to unblock it.
8541
8542 Remember that this does not check database presence nor database consistency.
8543 To do this, you can use an external check with xinetd for example.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01008544
Hervé COMMOWICK212f7782011-06-10 14:05:59 +02008545 The check requires MySQL >=3.22, for older version, please use TCP check.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01008546
8547 Most often, an incoming MySQL server needs to see the client's IP address for
8548 various purposes, including IP privilege matching and connection logging.
8549 When possible, it is often wise to masquerade the client's IP address when
8550 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02008551 which requires the transparent proxy feature to be compiled in, and the MySQL
8552 server to route the client via the machine hosting haproxy.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01008553
8554 See also: "option httpchk"
8555
8556
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008557option nolinger
8558no option nolinger
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01008559 Enable or disable immediate session resource cleaning after close
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008560 May be used in sections: defaults | frontend | listen | backend
8561 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008562 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008563
Davor Ocelice9ed2812017-12-25 17:49:28 +01008564 When clients or servers abort connections in a dirty way (e.g. they are
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008565 physically disconnected), the session timeouts triggers and the session is
8566 closed. But it will remain in FIN_WAIT1 state for some time in the system,
8567 using some resources and possibly limiting the ability to establish newer
8568 connections.
8569
8570 When this happens, it is possible to activate "option nolinger" which forces
8571 the system to immediately remove any socket's pending data on close. Thus,
Willy Tarreau4a321032020-10-16 04:55:19 +02008572 a TCP RST is emitted, any pending data are truncated, and the session is
8573 instantly purged from the system's tables. The generally visible effect for
8574 a client is that responses are truncated if the close happens with a last
8575 block of data (e.g. on a redirect or error response). On the server side,
8576 it may help release the source ports immediately when forwarding a client
8577 aborts in tunnels. In both cases, TCP resets are emitted and given that
8578 the session is instantly destroyed, there will be no retransmit. On a lossy
8579 network this can increase problems, especially when there is a firewall on
8580 the lossy side, because the firewall might see and process the reset (hence
8581 purge its session) and block any further traffic for this session,, including
8582 retransmits from the other side. So if the other side doesn't receive it,
8583 it will never receive any RST again, and the firewall might log many blocked
8584 packets.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008585
Willy Tarreau4a321032020-10-16 04:55:19 +02008586 For all these reasons, it is strongly recommended NOT to use this option,
8587 unless absolutely needed as a last resort. In most situations, using the
8588 "client-fin" or "server-fin" timeouts achieves similar results with a more
8589 reliable behavior. On Linux it's also possible to use the "tcp-ut" bind or
8590 server setting.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008591
8592 This option may be used both on frontends and backends, depending on the side
8593 where it is required. Use it on the frontend for clients, and on the backend
Willy Tarreau4a321032020-10-16 04:55:19 +02008594 for servers. While this option is technically supported in "defaults"
8595 sections, it must really not be used there as it risks to accidently
8596 propagate to sections that must no use it and to cause problems there.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008597
8598 If this option has been enabled in a "defaults" section, it can be disabled
8599 in a specific instance by prepending the "no" keyword before it.
8600
Willy Tarreau4a321032020-10-16 04:55:19 +02008601 See also: "timeout client-fin", "timeout server-fin", "tcp-ut" bind or server
8602 keywords.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008603
Willy Tarreau55165fe2009-05-10 12:02:55 +02008604option originalto [ except <network> ] [ header <name> ]
8605 Enable insertion of the X-Original-To header to requests sent to servers
8606 May be used in sections : defaults | frontend | listen | backend
8607 yes | yes | yes | yes
8608 Arguments :
8609 <network> is an optional argument used to disable this option for sources
8610 matching <network>
8611 <name> an optional argument to specify a different "X-Original-To"
8612 header name.
8613
8614 Since HAProxy can work in transparent mode, every request from a client can
8615 be redirected to the proxy and HAProxy itself can proxy every request to a
8616 complex SQUID environment and the destination host from SO_ORIGINAL_DST will
8617 be lost. This is annoying when you want access rules based on destination ip
8618 addresses. To solve this problem, a new HTTP header "X-Original-To" may be
8619 added by HAProxy to all requests sent to the server. This header contains a
8620 value representing the original destination IP address. Since this must be
8621 configured to always use the last occurrence of this header only. Note that
8622 only the last occurrence of the header must be used, since it is really
8623 possible that the client has already brought one.
8624
8625 The keyword "header" may be used to supply a different header name to replace
8626 the default "X-Original-To". This can be useful where you might already
8627 have a "X-Original-To" header from a different application, and you need
8628 preserve it. Also if your backend server doesn't use the "X-Original-To"
8629 header and requires different one.
8630
8631 Sometimes, a same HAProxy instance may be shared between a direct client
8632 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
8633 used to decrypt HTTPS traffic). It is possible to disable the addition of the
8634 header for a known source address or network by adding the "except" keyword
8635 followed by the network address. In this case, any source IP matching the
8636 network will not cause an addition of this header. Most common uses are with
8637 private networks or 127.0.0.1.
8638
Christopher Fauleteac279f2021-04-06 09:01:09 +02008639 Only IPv4 addresses are supported. "http-request add-header" or "http-request
8640 set-header" rules may be used to work around this limitation.
8641
Willy Tarreau55165fe2009-05-10 12:02:55 +02008642 This option may be specified either in the frontend or in the backend. If at
8643 least one of them uses it, the header will be added. Note that the backend's
8644 setting of the header subargument takes precedence over the frontend's if
8645 both are defined.
8646
Willy Tarreau55165fe2009-05-10 12:02:55 +02008647 Examples :
8648 # Original Destination address
8649 frontend www
8650 mode http
8651 option originalto except 127.0.0.1
8652
8653 # Those servers want the IP Address in X-Client-Dst
8654 backend www
8655 mode http
8656 option originalto header X-Client-Dst
8657
Christopher Faulet315b39c2018-09-21 16:26:19 +02008658 See also : "option httpclose", "option http-server-close".
Willy Tarreau55165fe2009-05-10 12:02:55 +02008659
8660
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008661option persist
8662no option persist
8663 Enable or disable forced persistence on down servers
8664 May be used in sections: defaults | frontend | listen | backend
8665 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008666 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008667
8668 When an HTTP request reaches a backend with a cookie which references a dead
8669 server, by default it is redispatched to another server. It is possible to
8670 force the request to be sent to the dead server first using "option persist"
8671 if absolutely needed. A common use case is when servers are under extreme
8672 load and spend their time flapping. In this case, the users would still be
8673 directed to the server they opened the session on, in the hope they would be
8674 correctly served. It is recommended to use "option redispatch" in conjunction
8675 with this option so that in the event it would not be possible to connect to
8676 the server at all (server definitely dead), the client would finally be
8677 redirected to another valid server.
8678
8679 If this option has been enabled in a "defaults" section, it can be disabled
8680 in a specific instance by prepending the "no" keyword before it.
8681
Willy Tarreau4de91492010-01-22 19:10:05 +01008682 See also : "option redispatch", "retries", "force-persist"
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008683
8684
Willy Tarreau0c122822013-12-15 18:49:01 +01008685option pgsql-check [ user <username> ]
8686 Use PostgreSQL health checks for server testing
8687 May be used in sections : defaults | frontend | listen | backend
8688 yes | no | yes | yes
8689 Arguments :
8690 <username> This is the username which will be used when connecting to
8691 PostgreSQL server.
8692
8693 The check sends a PostgreSQL StartupMessage and waits for either
8694 Authentication request or ErrorResponse message. It is a basic but useful
8695 test which does not produce error nor aborted connect on the server.
8696 This check is identical with the "mysql-check".
8697
8698 See also: "option httpchk"
8699
8700
Willy Tarreau9420b122013-12-15 18:58:25 +01008701option prefer-last-server
8702no option prefer-last-server
8703 Allow multiple load balanced requests to remain on the same server
8704 May be used in sections: defaults | frontend | listen | backend
8705 yes | no | yes | yes
8706 Arguments : none
8707
8708 When the load balancing algorithm in use is not deterministic, and a previous
8709 request was sent to a server to which haproxy still holds a connection, it is
8710 sometimes desirable that subsequent requests on a same session go to the same
8711 server as much as possible. Note that this is different from persistence, as
8712 we only indicate a preference which haproxy tries to apply without any form
8713 of warranty. The real use is for keep-alive connections sent to servers. When
8714 this option is used, haproxy will try to reuse the same connection that is
8715 attached to the server instead of rebalancing to another server, causing a
8716 close of the connection. This can make sense for static file servers. It does
Willy Tarreau068621e2013-12-23 15:11:25 +01008717 not make much sense to use this in combination with hashing algorithms. Note,
8718 haproxy already automatically tries to stick to a server which sends a 401 or
Lukas Tribus80512b12018-10-27 20:07:40 +02008719 to a proxy which sends a 407 (authentication required), when the load
8720 balancing algorithm is not deterministic. This is mandatory for use with the
8721 broken NTLM authentication challenge, and significantly helps in
Willy Tarreau068621e2013-12-23 15:11:25 +01008722 troubleshooting some faulty applications. Option prefer-last-server might be
8723 desirable in these environments as well, to avoid redistributing the traffic
8724 after every other response.
Willy Tarreau9420b122013-12-15 18:58:25 +01008725
8726 If this option has been enabled in a "defaults" section, it can be disabled
8727 in a specific instance by prepending the "no" keyword before it.
8728
8729 See also: "option http-keep-alive"
8730
8731
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008732option redispatch
Joseph Lynch726ab712015-05-11 23:25:34 -07008733option redispatch <interval>
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008734no option redispatch
8735 Enable or disable session redistribution in case of connection failure
8736 May be used in sections: defaults | frontend | listen | backend
8737 yes | no | yes | yes
Joseph Lynch726ab712015-05-11 23:25:34 -07008738 Arguments :
8739 <interval> The optional integer value that controls how often redispatches
8740 occur when retrying connections. Positive value P indicates a
8741 redispatch is desired on every Pth retry, and negative value
Davor Ocelice9ed2812017-12-25 17:49:28 +01008742 N indicate a redispatch is desired on the Nth retry prior to the
Joseph Lynch726ab712015-05-11 23:25:34 -07008743 last retry. For example, the default of -1 preserves the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008744 historical behavior of redispatching on the last retry, a
Joseph Lynch726ab712015-05-11 23:25:34 -07008745 positive value of 1 would indicate a redispatch on every retry,
8746 and a positive value of 3 would indicate a redispatch on every
8747 third retry. You can disable redispatches with a value of 0.
8748
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008749
8750 In HTTP mode, if a server designated by a cookie is down, clients may
8751 definitely stick to it because they cannot flush the cookie, so they will not
8752 be able to access the service anymore.
8753
Willy Tarreau59884a62019-01-02 14:48:31 +01008754 Specifying "option redispatch" will allow the proxy to break cookie or
8755 consistent hash based persistence and redistribute them to a working server.
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008756
Olivier Carrère6e6f59b2020-04-15 11:30:18 +02008757 Active servers are selected from a subset of the list of available
8758 servers. Active servers that are not down or in maintenance (i.e., whose
8759 health is not checked or that have been checked as "up"), are selected in the
8760 following order:
8761
8762 1. Any active, non-backup server, if any, or,
8763
8764 2. If the "allbackups" option is not set, the first backup server in the
8765 list, or
8766
8767 3. If the "allbackups" option is set, any backup server.
8768
8769 When a retry occurs, HAProxy tries to select another server than the last
8770 one. The new server is selected from the current list of servers.
8771
8772 Sometimes, if the list is updated between retries (e.g., if numerous retries
8773 occur and last longer than the time needed to check that a server is down,
8774 remove it from the list and fall back on the list of backup servers),
8775 connections may be redirected to a backup server, though.
8776
Joseph Lynch726ab712015-05-11 23:25:34 -07008777 It also allows to retry connections to another server in case of multiple
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008778 connection failures. Of course, it requires having "retries" set to a nonzero
8779 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01008780
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008781 If this option has been enabled in a "defaults" section, it can be disabled
8782 in a specific instance by prepending the "no" keyword before it.
8783
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02008784 See also : "retries", "force-persist"
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008785
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008786
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02008787option redis-check
8788 Use redis health checks for server testing
8789 May be used in sections : defaults | frontend | listen | backend
8790 yes | no | yes | yes
8791 Arguments : none
8792
8793 It is possible to test that the server correctly talks REDIS protocol instead
8794 of just testing that it accepts the TCP connection. When this option is set,
8795 a PING redis command is sent to the server, and the response is analyzed to
8796 find the "+PONG" response message.
8797
8798 Example :
8799 option redis-check
8800
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03008801 See also : "option httpchk", "option tcp-check", "tcp-check expect"
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02008802
8803
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008804option smtpchk
8805option smtpchk <hello> <domain>
8806 Use SMTP health checks for server testing
8807 May be used in sections : defaults | frontend | listen | backend
8808 yes | no | yes | yes
Willy Tarreaud72758d2010-01-12 10:42:19 +01008809 Arguments :
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008810 <hello> is an optional argument. It is the "hello" command to use. It can
Lukas Tribus27935782018-10-01 02:00:16 +02008811 be either "HELO" (for SMTP) or "EHLO" (for ESMTP). All other
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008812 values will be turned into the default command ("HELO").
8813
8814 <domain> is the domain name to present to the server. It may only be
8815 specified (and is mandatory) if the hello command has been
8816 specified. By default, "localhost" is used.
8817
8818 When "option smtpchk" is set, the health checks will consist in TCP
8819 connections followed by an SMTP command. By default, this command is
8820 "HELO localhost". The server's return code is analyzed and only return codes
8821 starting with a "2" will be considered as valid. All other responses,
8822 including a lack of response will constitute an error and will indicate a
8823 dead server.
8824
8825 This test is meant to be used with SMTP servers or relays. Depending on the
8826 request, it is possible that some servers do not log each connection attempt,
Davor Ocelice9ed2812017-12-25 17:49:28 +01008827 so you may want to experiment to improve the behavior. Using telnet on port
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008828 25 is often easier than adjusting the configuration.
8829
8830 Most often, an incoming SMTP server needs to see the client's IP address for
8831 various purposes, including spam filtering, anti-spoofing and logging. When
8832 possible, it is often wise to masquerade the client's IP address when
8833 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02008834 which requires the transparent proxy feature to be compiled in.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008835
8836 Example :
8837 option smtpchk HELO mydomain.org
8838
8839 See also : "option httpchk", "source"
8840
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008841
Krzysztof Piotr Oledzkiaeebf9b2009-10-04 15:43:17 +02008842option socket-stats
8843no option socket-stats
8844
8845 Enable or disable collecting & providing separate statistics for each socket.
8846 May be used in sections : defaults | frontend | listen | backend
8847 yes | yes | yes | no
8848
8849 Arguments : none
8850
8851
Willy Tarreauff4f82d2009-02-06 11:28:13 +01008852option splice-auto
8853no option splice-auto
8854 Enable or disable automatic kernel acceleration on sockets in both directions
8855 May be used in sections : defaults | frontend | listen | backend
8856 yes | yes | yes | yes
8857 Arguments : none
8858
8859 When this option is enabled either on a frontend or on a backend, haproxy
8860 will automatically evaluate the opportunity to use kernel tcp splicing to
Davor Ocelice9ed2812017-12-25 17:49:28 +01008861 forward data between the client and the server, in either direction. HAProxy
Willy Tarreauff4f82d2009-02-06 11:28:13 +01008862 uses heuristics to estimate if kernel splicing might improve performance or
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01008863 not. Both directions are handled independently. Note that the heuristics used
Willy Tarreauff4f82d2009-02-06 11:28:13 +01008864 are not much aggressive in order to limit excessive use of splicing. This
8865 option requires splicing to be enabled at compile time, and may be globally
8866 disabled with the global option "nosplice". Since splice uses pipes, using it
8867 requires that there are enough spare pipes.
8868
8869 Important note: kernel-based TCP splicing is a Linux-specific feature which
8870 first appeared in kernel 2.6.25. It offers kernel-based acceleration to
8871 transfer data between sockets without copying these data to user-space, thus
8872 providing noticeable performance gains and CPU cycles savings. Since many
8873 early implementations are buggy, corrupt data and/or are inefficient, this
8874 feature is not enabled by default, and it should be used with extreme care.
8875 While it is not possible to detect the correctness of an implementation,
8876 2.6.29 is the first version offering a properly working implementation. In
8877 case of doubt, splicing may be globally disabled using the global "nosplice"
8878 keyword.
8879
8880 Example :
8881 option splice-auto
8882
8883 If this option has been enabled in a "defaults" section, it can be disabled
8884 in a specific instance by prepending the "no" keyword before it.
8885
8886 See also : "option splice-request", "option splice-response", and global
8887 options "nosplice" and "maxpipes"
8888
8889
8890option splice-request
8891no option splice-request
8892 Enable or disable automatic kernel acceleration on sockets for requests
8893 May be used in sections : defaults | frontend | listen | backend
8894 yes | yes | yes | yes
8895 Arguments : none
8896
8897 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04008898 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01008899 the client to the server. It might still use the recv/send scheme if there
8900 are no spare pipes left. This option requires splicing to be enabled at
8901 compile time, and may be globally disabled with the global option "nosplice".
8902 Since splice uses pipes, using it requires that there are enough spare pipes.
8903
8904 Important note: see "option splice-auto" for usage limitations.
8905
8906 Example :
8907 option splice-request
8908
8909 If this option has been enabled in a "defaults" section, it can be disabled
8910 in a specific instance by prepending the "no" keyword before it.
8911
8912 See also : "option splice-auto", "option splice-response", and global options
8913 "nosplice" and "maxpipes"
8914
8915
8916option splice-response
8917no option splice-response
8918 Enable or disable automatic kernel acceleration on sockets for responses
8919 May be used in sections : defaults | frontend | listen | backend
8920 yes | yes | yes | yes
8921 Arguments : none
8922
8923 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04008924 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01008925 the server to the client. It might still use the recv/send scheme if there
8926 are no spare pipes left. This option requires splicing to be enabled at
8927 compile time, and may be globally disabled with the global option "nosplice".
8928 Since splice uses pipes, using it requires that there are enough spare pipes.
8929
8930 Important note: see "option splice-auto" for usage limitations.
8931
8932 Example :
8933 option splice-response
8934
8935 If this option has been enabled in a "defaults" section, it can be disabled
8936 in a specific instance by prepending the "no" keyword before it.
8937
8938 See also : "option splice-auto", "option splice-request", and global options
8939 "nosplice" and "maxpipes"
8940
8941
Christopher Fauletba7bc162016-11-07 21:07:38 +01008942option spop-check
8943 Use SPOP health checks for server testing
8944 May be used in sections : defaults | frontend | listen | backend
8945 no | no | no | yes
8946 Arguments : none
8947
8948 It is possible to test that the server correctly talks SPOP protocol instead
8949 of just testing that it accepts the TCP connection. When this option is set,
8950 a HELLO handshake is performed between HAProxy and the server, and the
8951 response is analyzed to check no error is reported.
8952
8953 Example :
8954 option spop-check
8955
8956 See also : "option httpchk"
8957
8958
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008959option srvtcpka
8960no option srvtcpka
8961 Enable or disable the sending of TCP keepalive packets on the server side
8962 May be used in sections : defaults | frontend | listen | backend
8963 yes | no | yes | yes
8964 Arguments : none
8965
8966 When there is a firewall or any session-aware component between a client and
8967 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01008968 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008969 components decides to expire a session which has remained idle for too long.
8970
8971 Enabling socket-level TCP keep-alives makes the system regularly send packets
8972 to the other end of the connection, leaving it active. The delay between
8973 keep-alive probes is controlled by the system only and depends both on the
8974 operating system and its tuning parameters.
8975
8976 It is important to understand that keep-alive packets are neither emitted nor
8977 received at the application level. It is only the network stacks which sees
8978 them. For this reason, even if one side of the proxy already uses keep-alives
8979 to maintain its connection alive, those keep-alive packets will not be
8980 forwarded to the other side of the proxy.
8981
8982 Please note that this has nothing to do with HTTP keep-alive.
8983
8984 Using option "srvtcpka" enables the emission of TCP keep-alive probes on the
8985 server side of a connection, which should help when session expirations are
8986 noticed between HAProxy and a server.
8987
8988 If this option has been enabled in a "defaults" section, it can be disabled
8989 in a specific instance by prepending the "no" keyword before it.
8990
8991 See also : "option clitcpka", "option tcpka"
8992
8993
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008994option ssl-hello-chk
8995 Use SSLv3 client hello health checks for server testing
8996 May be used in sections : defaults | frontend | listen | backend
8997 yes | no | yes | yes
8998 Arguments : none
8999
9000 When some SSL-based protocols are relayed in TCP mode through HAProxy, it is
9001 possible to test that the server correctly talks SSL instead of just testing
9002 that it accepts the TCP connection. When "option ssl-hello-chk" is set, pure
9003 SSLv3 client hello messages are sent once the connection is established to
9004 the server, and the response is analyzed to find an SSL server hello message.
9005 The server is considered valid only when the response contains this server
9006 hello message.
9007
9008 All servers tested till there correctly reply to SSLv3 client hello messages,
9009 and most servers tested do not even log the requests containing only hello
9010 messages, which is appreciable.
9011
Willy Tarreau763a95b2012-10-04 23:15:39 +02009012 Note that this check works even when SSL support was not built into haproxy
9013 because it forges the SSL message. When SSL support is available, it is best
9014 to use native SSL health checks instead of this one.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009015
Willy Tarreau763a95b2012-10-04 23:15:39 +02009016 See also: "option httpchk", "check-ssl"
9017
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009018
Willy Tarreaued179852013-12-16 01:07:00 +01009019option tcp-check
9020 Perform health checks using tcp-check send/expect sequences
9021 May be used in sections: defaults | frontend | listen | backend
9022 yes | no | yes | yes
9023
9024 This health check method is intended to be combined with "tcp-check" command
9025 lists in order to support send/expect types of health check sequences.
9026
9027 TCP checks currently support 4 modes of operations :
9028 - no "tcp-check" directive : the health check only consists in a connection
9029 attempt, which remains the default mode.
9030
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009031 - "tcp-check send" or "tcp-check send-binary" only is mentioned : this is
Willy Tarreaued179852013-12-16 01:07:00 +01009032 used to send a string along with a connection opening. With some
9033 protocols, it helps sending a "QUIT" message for example that prevents
9034 the server from logging a connection error for each health check. The
9035 check result will still be based on the ability to open the connection
9036 only.
9037
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009038 - "tcp-check expect" only is mentioned : this is used to test a banner.
Willy Tarreaued179852013-12-16 01:07:00 +01009039 The connection is opened and haproxy waits for the server to present some
9040 contents which must validate some rules. The check result will be based
9041 on the matching between the contents and the rules. This is suited for
9042 POP, IMAP, SMTP, FTP, SSH, TELNET.
9043
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009044 - both "tcp-check send" and "tcp-check expect" are mentioned : this is
Davor Ocelice9ed2812017-12-25 17:49:28 +01009045 used to test a hello-type protocol. HAProxy sends a message, the server
9046 responds and its response is analyzed. the check result will be based on
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009047 the matching between the response contents and the rules. This is often
Willy Tarreaued179852013-12-16 01:07:00 +01009048 suited for protocols which require a binding or a request/response model.
9049 LDAP, MySQL, Redis and SSL are example of such protocols, though they
9050 already all have their dedicated checks with a deeper understanding of
9051 the respective protocols.
9052 In this mode, many questions may be sent and many answers may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01009053 analyzed.
Willy Tarreaued179852013-12-16 01:07:00 +01009054
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02009055 A fifth mode can be used to insert comments in different steps of the script.
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009056
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02009057 For each tcp-check rule you create, you can add a "comment" directive,
9058 followed by a string. This string will be reported in the log and stderr in
9059 debug mode. It is useful to make user-friendly error reporting. The
9060 "comment" is of course optional.
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009061
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02009062 During the execution of a health check, a variable scope is made available to
9063 store data samples, using the "tcp-check set-var" operation. Freeing those
9064 variable is possible using "tcp-check unset-var".
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +01009065
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009066
Willy Tarreaued179852013-12-16 01:07:00 +01009067 Examples :
Davor Ocelice9ed2812017-12-25 17:49:28 +01009068 # perform a POP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01009069 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009070 tcp-check expect string +OK\ POP3\ ready comment POP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01009071
Davor Ocelice9ed2812017-12-25 17:49:28 +01009072 # perform an IMAP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01009073 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009074 tcp-check expect string *\ OK\ IMAP4\ ready comment IMAP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01009075
9076 # look for the redis master server after ensuring it speaks well
9077 # redis protocol, then it exits properly.
Davor Ocelice9ed2812017-12-25 17:49:28 +01009078 # (send a command then analyze the response 3 times)
Willy Tarreaued179852013-12-16 01:07:00 +01009079 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009080 tcp-check comment PING\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01009081 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02009082 tcp-check expect string +PONG
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009083 tcp-check comment role\ check
Willy Tarreaued179852013-12-16 01:07:00 +01009084 tcp-check send info\ replication\r\n
9085 tcp-check expect string role:master
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009086 tcp-check comment QUIT\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01009087 tcp-check send QUIT\r\n
9088 tcp-check expect string +OK
9089
Davor Ocelice9ed2812017-12-25 17:49:28 +01009090 forge a HTTP request, then analyze the response
Willy Tarreaued179852013-12-16 01:07:00 +01009091 (send many headers before analyzing)
9092 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009093 tcp-check comment forge\ and\ send\ HTTP\ request
Willy Tarreaued179852013-12-16 01:07:00 +01009094 tcp-check send HEAD\ /\ HTTP/1.1\r\n
9095 tcp-check send Host:\ www.mydomain.com\r\n
9096 tcp-check send User-Agent:\ HAProxy\ tcpcheck\r\n
9097 tcp-check send \r\n
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009098 tcp-check expect rstring HTTP/1\..\ (2..|3..) comment check\ HTTP\ response
Willy Tarreaued179852013-12-16 01:07:00 +01009099
9100
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02009101 See also : "tcp-check connect", "tcp-check expect" and "tcp-check send".
Willy Tarreaued179852013-12-16 01:07:00 +01009102
9103
Willy Tarreau9ea05a72009-06-14 12:07:01 +02009104option tcp-smart-accept
9105no option tcp-smart-accept
9106 Enable or disable the saving of one ACK packet during the accept sequence
9107 May be used in sections : defaults | frontend | listen | backend
9108 yes | yes | yes | no
9109 Arguments : none
9110
9111 When an HTTP connection request comes in, the system acknowledges it on
9112 behalf of HAProxy, then the client immediately sends its request, and the
9113 system acknowledges it too while it is notifying HAProxy about the new
9114 connection. HAProxy then reads the request and responds. This means that we
9115 have one TCP ACK sent by the system for nothing, because the request could
9116 very well be acknowledged by HAProxy when it sends its response.
9117
9118 For this reason, in HTTP mode, HAProxy automatically asks the system to avoid
9119 sending this useless ACK on platforms which support it (currently at least
9120 Linux). It must not cause any problem, because the system will send it anyway
9121 after 40 ms if the response takes more time than expected to come.
9122
9123 During complex network debugging sessions, it may be desirable to disable
9124 this optimization because delayed ACKs can make troubleshooting more complex
9125 when trying to identify where packets are delayed. It is then possible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01009126 fall back to normal behavior by specifying "no option tcp-smart-accept".
Willy Tarreau9ea05a72009-06-14 12:07:01 +02009127
9128 It is also possible to force it for non-HTTP proxies by simply specifying
9129 "option tcp-smart-accept". For instance, it can make sense with some services
9130 such as SMTP where the server speaks first.
9131
9132 It is recommended to avoid forcing this option in a defaults section. In case
9133 of doubt, consider setting it back to automatic values by prepending the
9134 "default" keyword before it, or disabling it using the "no" keyword.
9135
Willy Tarreaud88edf22009-06-14 15:48:17 +02009136 See also : "option tcp-smart-connect"
9137
9138
9139option tcp-smart-connect
9140no option tcp-smart-connect
9141 Enable or disable the saving of one ACK packet during the connect sequence
9142 May be used in sections : defaults | frontend | listen | backend
9143 yes | no | yes | yes
9144 Arguments : none
9145
9146 On certain systems (at least Linux), HAProxy can ask the kernel not to
9147 immediately send an empty ACK upon a connection request, but to directly
9148 send the buffer request instead. This saves one packet on the network and
9149 thus boosts performance. It can also be useful for some servers, because they
9150 immediately get the request along with the incoming connection.
9151
9152 This feature is enabled when "option tcp-smart-connect" is set in a backend.
9153 It is not enabled by default because it makes network troubleshooting more
9154 complex.
9155
9156 It only makes sense to enable it with protocols where the client speaks first
9157 such as HTTP. In other situations, if there is no data to send in place of
9158 the ACK, a normal ACK is sent.
9159
9160 If this option has been enabled in a "defaults" section, it can be disabled
9161 in a specific instance by prepending the "no" keyword before it.
9162
9163 See also : "option tcp-smart-accept"
9164
Willy Tarreau9ea05a72009-06-14 12:07:01 +02009165
Willy Tarreaubf1f8162007-12-28 17:42:56 +01009166option tcpka
9167 Enable or disable the sending of TCP keepalive packets on both sides
9168 May be used in sections : defaults | frontend | listen | backend
9169 yes | yes | yes | yes
9170 Arguments : none
9171
9172 When there is a firewall or any session-aware component between a client and
9173 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01009174 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01009175 components decides to expire a session which has remained idle for too long.
9176
9177 Enabling socket-level TCP keep-alives makes the system regularly send packets
9178 to the other end of the connection, leaving it active. The delay between
9179 keep-alive probes is controlled by the system only and depends both on the
9180 operating system and its tuning parameters.
9181
9182 It is important to understand that keep-alive packets are neither emitted nor
9183 received at the application level. It is only the network stacks which sees
9184 them. For this reason, even if one side of the proxy already uses keep-alives
9185 to maintain its connection alive, those keep-alive packets will not be
9186 forwarded to the other side of the proxy.
9187
9188 Please note that this has nothing to do with HTTP keep-alive.
9189
9190 Using option "tcpka" enables the emission of TCP keep-alive probes on both
9191 the client and server sides of a connection. Note that this is meaningful
9192 only in "defaults" or "listen" sections. If this option is used in a
9193 frontend, only the client side will get keep-alives, and if this option is
9194 used in a backend, only the server side will get keep-alives. For this
9195 reason, it is strongly recommended to explicitly use "option clitcpka" and
9196 "option srvtcpka" when the configuration is split between frontends and
9197 backends.
9198
9199 See also : "option clitcpka", "option srvtcpka"
9200
Willy Tarreau844e3c52008-01-11 16:28:18 +01009201
9202option tcplog
9203 Enable advanced logging of TCP connections with session state and timers
9204 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01009205 yes | yes | yes | no
Willy Tarreau844e3c52008-01-11 16:28:18 +01009206 Arguments : none
9207
9208 By default, the log output format is very poor, as it only contains the
9209 source and destination addresses, and the instance name. By specifying
9210 "option tcplog", each log line turns into a much richer format including, but
9211 not limited to, the connection timers, the session status, the connections
9212 numbers, the frontend, backend and server name, and of course the source
9213 address and ports. This option is useful for pure TCP proxies in order to
9214 find which of the client or server disconnects or times out. For normal HTTP
9215 proxies, it's better to use "option httplog" which is even more complete.
9216
Guillaume de Lafond29f45602017-03-31 19:52:15 +02009217 "option tcplog" overrides any previous "log-format" directive.
9218
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009219 See also : "option httplog", and section 8 about logging.
Willy Tarreau844e3c52008-01-11 16:28:18 +01009220
9221
Willy Tarreau844e3c52008-01-11 16:28:18 +01009222option transparent
9223no option transparent
9224 Enable client-side transparent proxying
9225 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +01009226 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +01009227 Arguments : none
9228
9229 This option was introduced in order to provide layer 7 persistence to layer 3
9230 load balancers. The idea is to use the OS's ability to redirect an incoming
9231 connection for a remote address to a local process (here HAProxy), and let
9232 this process know what address was initially requested. When this option is
9233 used, sessions without cookies will be forwarded to the original destination
9234 IP address of the incoming request (which should match that of another
9235 equipment), while requests with cookies will still be forwarded to the
9236 appropriate server.
9237
9238 Note that contrary to a common belief, this option does NOT make HAProxy
9239 present the client's IP to the server when establishing the connection.
9240
Willy Tarreaua1146052011-03-01 09:51:54 +01009241 See also: the "usesrc" argument of the "source" keyword, and the
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009242 "transparent" option of the "bind" keyword.
Willy Tarreau844e3c52008-01-11 16:28:18 +01009243
Willy Tarreaubf1f8162007-12-28 17:42:56 +01009244
Simon Horman98637e52014-06-20 12:30:16 +09009245external-check command <command>
9246 Executable to run when performing an external-check
9247 May be used in sections : defaults | frontend | listen | backend
9248 yes | no | yes | yes
9249
9250 Arguments :
9251 <command> is the external command to run
9252
Simon Horman98637e52014-06-20 12:30:16 +09009253 The arguments passed to the to the command are:
9254
Cyril Bonté777be862014-12-02 21:21:35 +01009255 <proxy_address> <proxy_port> <server_address> <server_port>
Simon Horman98637e52014-06-20 12:30:16 +09009256
Cyril Bonté777be862014-12-02 21:21:35 +01009257 The <proxy_address> and <proxy_port> are derived from the first listener
9258 that is either IPv4, IPv6 or a UNIX socket. In the case of a UNIX socket
9259 listener the proxy_address will be the path of the socket and the
9260 <proxy_port> will be the string "NOT_USED". In a backend section, it's not
9261 possible to determine a listener, and both <proxy_address> and <proxy_port>
9262 will have the string value "NOT_USED".
Simon Horman98637e52014-06-20 12:30:16 +09009263
Cyril Bonté72cda2a2014-12-27 22:28:39 +01009264 Some values are also provided through environment variables.
9265
9266 Environment variables :
9267 HAPROXY_PROXY_ADDR The first bind address if available (or empty if not
9268 applicable, for example in a "backend" section).
9269
9270 HAPROXY_PROXY_ID The backend id.
9271
9272 HAPROXY_PROXY_NAME The backend name.
9273
9274 HAPROXY_PROXY_PORT The first bind port if available (or empty if not
9275 applicable, for example in a "backend" section or
9276 for a UNIX socket).
9277
9278 HAPROXY_SERVER_ADDR The server address.
9279
9280 HAPROXY_SERVER_CURCONN The current number of connections on the server.
9281
9282 HAPROXY_SERVER_ID The server id.
9283
9284 HAPROXY_SERVER_MAXCONN The server max connections.
9285
9286 HAPROXY_SERVER_NAME The server name.
9287
9288 HAPROXY_SERVER_PORT The server port if available (or empty for a UNIX
9289 socket).
9290
9291 PATH The PATH environment variable used when executing
9292 the command may be set using "external-check path".
9293
William Lallemand4d03e432019-06-14 15:35:37 +02009294 See also "2.3. Environment variables" for other variables.
9295
Simon Horman98637e52014-06-20 12:30:16 +09009296 If the command executed and exits with a zero status then the check is
9297 considered to have passed, otherwise the check is considered to have
9298 failed.
9299
9300 Example :
9301 external-check command /bin/true
9302
9303 See also : "external-check", "option external-check", "external-check path"
9304
9305
9306external-check path <path>
9307 The value of the PATH environment variable used when running an external-check
9308 May be used in sections : defaults | frontend | listen | backend
9309 yes | no | yes | yes
9310
9311 Arguments :
9312 <path> is the path used when executing external command to run
9313
9314 The default path is "".
9315
9316 Example :
9317 external-check path "/usr/bin:/bin"
9318
9319 See also : "external-check", "option external-check",
9320 "external-check command"
9321
9322
Emeric Brun647caf12009-06-30 17:57:00 +02009323persist rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02009324persist rdp-cookie(<name>)
Emeric Brun647caf12009-06-30 17:57:00 +02009325 Enable RDP cookie-based persistence
9326 May be used in sections : defaults | frontend | listen | backend
9327 yes | no | yes | yes
9328 Arguments :
9329 <name> is the optional name of the RDP cookie to check. If omitted, the
Willy Tarreau61e28f22010-05-16 22:31:05 +02009330 default cookie name "msts" will be used. There currently is no
9331 valid reason to change this name.
Emeric Brun647caf12009-06-30 17:57:00 +02009332
9333 This statement enables persistence based on an RDP cookie. The RDP cookie
9334 contains all information required to find the server in the list of known
Davor Ocelice9ed2812017-12-25 17:49:28 +01009335 servers. So when this option is set in the backend, the request is analyzed
Emeric Brun647caf12009-06-30 17:57:00 +02009336 and if an RDP cookie is found, it is decoded. If it matches a known server
9337 which is still UP (or if "option persist" is set), then the connection is
9338 forwarded to this server.
9339
9340 Note that this only makes sense in a TCP backend, but for this to work, the
9341 frontend must have waited long enough to ensure that an RDP cookie is present
9342 in the request buffer. This is the same requirement as with the "rdp-cookie"
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009343 load-balancing method. Thus it is highly recommended to put all statements in
Emeric Brun647caf12009-06-30 17:57:00 +02009344 a single "listen" section.
9345
Willy Tarreau61e28f22010-05-16 22:31:05 +02009346 Also, it is important to understand that the terminal server will emit this
9347 RDP cookie only if it is configured for "token redirection mode", which means
9348 that the "IP address redirection" option is disabled.
9349
Emeric Brun647caf12009-06-30 17:57:00 +02009350 Example :
9351 listen tse-farm
9352 bind :3389
9353 # wait up to 5s for an RDP cookie in the request
9354 tcp-request inspect-delay 5s
9355 tcp-request content accept if RDP_COOKIE
9356 # apply RDP cookie persistence
9357 persist rdp-cookie
9358 # if server is unknown, let's balance on the same cookie.
Cyril Bontédc4d9032012-04-08 21:57:39 +02009359 # alternatively, "balance leastconn" may be useful too.
Emeric Brun647caf12009-06-30 17:57:00 +02009360 balance rdp-cookie
9361 server srv1 1.1.1.1:3389
9362 server srv2 1.1.1.2:3389
9363
Simon Hormanab814e02011-06-24 14:50:20 +09009364 See also : "balance rdp-cookie", "tcp-request", the "req_rdp_cookie" ACL and
9365 the rdp_cookie pattern fetch function.
Emeric Brun647caf12009-06-30 17:57:00 +02009366
9367
Willy Tarreau3a7d2072009-03-05 23:48:25 +01009368rate-limit sessions <rate>
9369 Set a limit on the number of new sessions accepted per second on a frontend
9370 May be used in sections : defaults | frontend | listen | backend
9371 yes | yes | yes | no
9372 Arguments :
9373 <rate> The <rate> parameter is an integer designating the maximum number
9374 of new sessions per second to accept on the frontend.
9375
9376 When the frontend reaches the specified number of new sessions per second, it
9377 stops accepting new connections until the rate drops below the limit again.
9378 During this time, the pending sessions will be kept in the socket's backlog
9379 (in system buffers) and haproxy will not even be aware that sessions are
9380 pending. When applying very low limit on a highly loaded service, it may make
9381 sense to increase the socket's backlog using the "backlog" keyword.
9382
9383 This feature is particularly efficient at blocking connection-based attacks
9384 or service abuse on fragile servers. Since the session rate is measured every
9385 millisecond, it is extremely accurate. Also, the limit applies immediately,
9386 no delay is needed at all to detect the threshold.
9387
9388 Example : limit the connection rate on SMTP to 10 per second max
9389 listen smtp
9390 mode tcp
9391 bind :25
9392 rate-limit sessions 10
Panagiotis Panagiotopoulos7282d8e2016-02-11 16:37:15 +02009393 server smtp1 127.0.0.1:1025
Willy Tarreau3a7d2072009-03-05 23:48:25 +01009394
Willy Tarreaua17c2d92011-07-25 08:16:20 +02009395 Note : when the maximum rate is reached, the frontend's status is not changed
9396 but its sockets appear as "WAITING" in the statistics if the
9397 "socket-stats" option is enabled.
Willy Tarreau3a7d2072009-03-05 23:48:25 +01009398
9399 See also : the "backlog" keyword and the "fe_sess_rate" ACL criterion.
9400
9401
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009402redirect location <loc> [code <code>] <option> [{if | unless} <condition>]
9403redirect prefix <pfx> [code <code>] <option> [{if | unless} <condition>]
9404redirect scheme <sch> [code <code>] <option> [{if | unless} <condition>]
Willy Tarreaub463dfb2008-06-07 23:08:56 +02009405 Return an HTTP redirection if/unless a condition is matched
9406 May be used in sections : defaults | frontend | listen | backend
9407 no | yes | yes | yes
9408
9409 If/unless the condition is matched, the HTTP request will lead to a redirect
Willy Tarreauf285f542010-01-03 20:03:03 +01009410 response. If no condition is specified, the redirect applies unconditionally.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02009411
Willy Tarreau0140f252008-11-19 21:07:09 +01009412 Arguments :
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009413 <loc> With "redirect location", the exact value in <loc> is placed into
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01009414 the HTTP "Location" header. When used in an "http-request" rule,
9415 <loc> value follows the log-format rules and can include some
9416 dynamic values (see Custom Log Format in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009417
9418 <pfx> With "redirect prefix", the "Location" header is built from the
9419 concatenation of <pfx> and the complete URI path, including the
9420 query string, unless the "drop-query" option is specified (see
9421 below). As a special case, if <pfx> equals exactly "/", then
9422 nothing is inserted before the original URI. It allows one to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01009423 redirect to the same URL (for instance, to insert a cookie). When
9424 used in an "http-request" rule, <pfx> value follows the log-format
9425 rules and can include some dynamic values (see Custom Log Format
9426 in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009427
9428 <sch> With "redirect scheme", then the "Location" header is built by
9429 concatenating <sch> with "://" then the first occurrence of the
9430 "Host" header, and then the URI path, including the query string
9431 unless the "drop-query" option is specified (see below). If no
9432 path is found or if the path is "*", then "/" is used instead. If
9433 no "Host" header is found, then an empty host component will be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009434 returned, which most recent browsers interpret as redirecting to
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009435 the same host. This directive is mostly used to redirect HTTP to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01009436 HTTPS. When used in an "http-request" rule, <sch> value follows
9437 the log-format rules and can include some dynamic values (see
9438 Custom Log Format in section 8.2.4).
Willy Tarreau0140f252008-11-19 21:07:09 +01009439
9440 <code> The code is optional. It indicates which type of HTTP redirection
Willy Tarreaub67fdc42013-03-29 19:28:11 +01009441 is desired. Only codes 301, 302, 303, 307 and 308 are supported,
9442 with 302 used by default if no code is specified. 301 means
9443 "Moved permanently", and a browser may cache the Location. 302
Baptiste Assmannea849c02015-08-03 11:42:50 +02009444 means "Moved temporarily" and means that the browser should not
Willy Tarreaub67fdc42013-03-29 19:28:11 +01009445 cache the redirection. 303 is equivalent to 302 except that the
9446 browser will fetch the location with a GET method. 307 is just
9447 like 302 but makes it clear that the same method must be reused.
9448 Likewise, 308 replaces 301 if the same method must be used.
Willy Tarreau0140f252008-11-19 21:07:09 +01009449
9450 <option> There are several options which can be specified to adjust the
Davor Ocelice9ed2812017-12-25 17:49:28 +01009451 expected behavior of a redirection :
Willy Tarreau0140f252008-11-19 21:07:09 +01009452
9453 - "drop-query"
9454 When this keyword is used in a prefix-based redirection, then the
9455 location will be set without any possible query-string, which is useful
9456 for directing users to a non-secure page for instance. It has no effect
9457 with a location-type redirect.
9458
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01009459 - "append-slash"
9460 This keyword may be used in conjunction with "drop-query" to redirect
9461 users who use a URL not ending with a '/' to the same one with the '/'.
9462 It can be useful to ensure that search engines will only see one URL.
9463 For this, a return code 301 is preferred.
9464
Willy Tarreau0140f252008-11-19 21:07:09 +01009465 - "set-cookie NAME[=value]"
9466 A "Set-Cookie" header will be added with NAME (and optionally "=value")
9467 to the response. This is sometimes used to indicate that a user has
9468 been seen, for instance to protect against some types of DoS. No other
9469 cookie option is added, so the cookie will be a session cookie. Note
9470 that for a browser, a sole cookie name without an equal sign is
9471 different from a cookie with an equal sign.
9472
9473 - "clear-cookie NAME[=]"
9474 A "Set-Cookie" header will be added with NAME (and optionally "="), but
9475 with the "Max-Age" attribute set to zero. This will tell the browser to
9476 delete this cookie. It is useful for instance on logout pages. It is
9477 important to note that clearing the cookie "NAME" will not remove a
9478 cookie set with "NAME=value". You have to clear the cookie "NAME=" for
9479 that, because the browser makes the difference.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02009480
9481 Example: move the login URL only to HTTPS.
9482 acl clear dst_port 80
9483 acl secure dst_port 8080
9484 acl login_page url_beg /login
Willy Tarreau0140f252008-11-19 21:07:09 +01009485 acl logout url_beg /logout
Willy Tarreau79da4692008-11-19 20:03:04 +01009486 acl uid_given url_reg /login?userid=[^&]+
Willy Tarreau0140f252008-11-19 21:07:09 +01009487 acl cookie_set hdr_sub(cookie) SEEN=1
9488
9489 redirect prefix https://mysite.com set-cookie SEEN=1 if !cookie_set
Willy Tarreau79da4692008-11-19 20:03:04 +01009490 redirect prefix https://mysite.com if login_page !secure
9491 redirect prefix http://mysite.com drop-query if login_page !uid_given
9492 redirect location http://mysite.com/ if !login_page secure
Willy Tarreau0140f252008-11-19 21:07:09 +01009493 redirect location / clear-cookie USERID= if logout
Willy Tarreaub463dfb2008-06-07 23:08:56 +02009494
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01009495 Example: send redirects for request for articles without a '/'.
9496 acl missing_slash path_reg ^/article/[^/]*$
9497 redirect code 301 prefix / drop-query append-slash if missing_slash
9498
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009499 Example: redirect all HTTP traffic to HTTPS when SSL is handled by haproxy.
David BERARDe7153042012-11-03 00:11:31 +01009500 redirect scheme https if !{ ssl_fc }
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009501
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01009502 Example: append 'www.' prefix in front of all hosts not having it
Coen Rosdorff596659b2016-04-11 11:33:49 +02009503 http-request redirect code 301 location \
9504 http://www.%[hdr(host)]%[capture.req.uri] \
9505 unless { hdr_beg(host) -i www }
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01009506
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009507 See section 7 about ACL usage.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02009508
Willy Tarreau303c0352008-01-17 19:01:39 +01009509
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02009510retries <value>
9511 Set the number of retries to perform on a server after a connection failure
9512 May be used in sections: defaults | frontend | listen | backend
9513 yes | no | yes | yes
9514 Arguments :
9515 <value> is the number of times a connection attempt should be retried on
9516 a server when a connection either is refused or times out. The
9517 default value is 3.
9518
9519 It is important to understand that this value applies to the number of
9520 connection attempts, not full requests. When a connection has effectively
9521 been established to a server, there will be no more retry.
9522
9523 In order to avoid immediate reconnections to a server which is restarting,
Joseph Lynch726ab712015-05-11 23:25:34 -07009524 a turn-around timer of min("timeout connect", one second) is applied before
9525 a retry occurs.
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02009526
9527 When "option redispatch" is set, the last retry may be performed on another
9528 server even if a cookie references a different server.
9529
9530 See also : "option redispatch"
9531
9532
Olivier Houcharda254a372019-04-05 15:30:12 +02009533retry-on [list of keywords]
Jerome Magnin5ce3c142020-05-13 20:09:57 +02009534 Specify when to attempt to automatically retry a failed request.
9535 This setting is only valid when "mode" is set to http and is silently ignored
9536 otherwise.
Olivier Houcharda254a372019-04-05 15:30:12 +02009537 May be used in sections: defaults | frontend | listen | backend
9538 yes | no | yes | yes
9539 Arguments :
9540 <keywords> is a list of keywords or HTTP status codes, each representing a
9541 type of failure event on which an attempt to retry the request
9542 is desired. Please read the notes at the bottom before changing
9543 this setting. The following keywords are supported :
9544
9545 none never retry
9546
9547 conn-failure retry when the connection or the SSL handshake failed
9548 and the request could not be sent. This is the default.
9549
9550 empty-response retry when the server connection was closed after part
9551 of the request was sent, and nothing was received from
9552 the server. This type of failure may be caused by the
9553 request timeout on the server side, poor network
9554 condition, or a server crash or restart while
9555 processing the request.
9556
Olivier Houcharde3249a92019-05-03 23:01:47 +02009557 junk-response retry when the server returned something not looking
9558 like a complete HTTP response. This includes partial
9559 responses headers as well as non-HTTP contents. It
9560 usually is a bad idea to retry on such events, which
9561 may be caused a configuration issue (wrong server port)
9562 or by the request being harmful to the server (buffer
9563 overflow attack for example).
9564
Olivier Houcharda254a372019-04-05 15:30:12 +02009565 response-timeout the server timeout stroke while waiting for the server
9566 to respond to the request. This may be caused by poor
9567 network condition, the reuse of an idle connection
9568 which has expired on the path, or by the request being
9569 extremely expensive to process. It generally is a bad
9570 idea to retry on such events on servers dealing with
9571 heavy database processing (full scans, etc) as it may
9572 amplify denial of service attacks.
9573
Olivier Houchard865d8392019-05-03 22:46:27 +02009574 0rtt-rejected retry requests which were sent over early data and were
9575 rejected by the server. These requests are generally
9576 considered to be safe to retry.
9577
Olivier Houcharda254a372019-04-05 15:30:12 +02009578 <status> any HTTP status code among "404" (Not Found), "408"
9579 (Request Timeout), "425" (Too Early), "500" (Server
9580 Error), "501" (Not Implemented), "502" (Bad Gateway),
9581 "503" (Service Unavailable), "504" (Gateway Timeout).
9582
Olivier Houchardddf0e032019-05-10 18:05:40 +02009583 all-retryable-errors
9584 retry request for any error that are considered
9585 retryable. This currently activates "conn-failure",
9586 "empty-response", "junk-response", "response-timeout",
9587 "0rtt-rejected", "500", "502", "503", and "504".
9588
Olivier Houcharda254a372019-04-05 15:30:12 +02009589 Using this directive replaces any previous settings with the new ones; it is
9590 not cumulative.
9591
9592 Please note that using anything other than "none" and "conn-failure" requires
9593 to allocate a buffer and copy the whole request into it, so it has memory and
9594 performance impacts. Requests not fitting in a single buffer will never be
9595 retried (see the global tune.bufsize setting).
9596
9597 You have to make sure the application has a replay protection mechanism built
9598 in such as a unique transaction IDs passed in requests, or that replaying the
9599 same request has no consequence, or it is very dangerous to use any retry-on
9600 value beside "conn-failure" and "none". Static file servers and caches are
9601 generally considered safe against any type of retry. Using a status code can
9602 be useful to quickly leave a server showing an abnormal behavior (out of
9603 memory, file system issues, etc), but in this case it may be a good idea to
9604 immediately redispatch the connection to another server (please see "option
9605 redispatch" for this). Last, it is important to understand that most causes
9606 of failures are the requests themselves and that retrying a request causing a
9607 server to misbehave will often make the situation even worse for this server,
9608 or for the whole service in case of redispatch.
9609
9610 Unless you know exactly how the application deals with replayed requests, you
9611 should not use this directive.
9612
9613 The default is "conn-failure".
9614
9615 See also: "retries", "option redispatch", "tune.bufsize"
9616
David du Colombier486df472011-03-17 10:40:26 +01009617server <name> <address>[:[port]] [param*]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009618 Declare a server in a backend
9619 May be used in sections : defaults | frontend | listen | backend
9620 no | no | yes | yes
9621 Arguments :
9622 <name> is the internal name assigned to this server. This name will
Davor Ocelice9ed2812017-12-25 17:49:28 +01009623 appear in logs and alerts. If "http-send-name-header" is
Mark Lamourinec2247f02012-01-04 13:02:01 -05009624 set, it will be added to the request header sent to the server.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009625
David du Colombier486df472011-03-17 10:40:26 +01009626 <address> is the IPv4 or IPv6 address of the server. Alternatively, a
9627 resolvable hostname is supported, but this name will be resolved
9628 during start-up. Address "0.0.0.0" or "*" has a special meaning.
9629 It indicates that the connection will be forwarded to the same IP
Willy Tarreaud669a4f2010-07-13 14:49:50 +02009630 address as the one from the client connection. This is useful in
9631 transparent proxy architectures where the client's connection is
9632 intercepted and haproxy must forward to the original destination
9633 address. This is more or less what the "transparent" keyword does
9634 except that with a server it's possible to limit concurrency and
Willy Tarreau24709282013-03-10 21:32:12 +01009635 to report statistics. Optionally, an address family prefix may be
9636 used before the address to force the family regardless of the
9637 address format, which can be useful to specify a path to a unix
9638 socket with no slash ('/'). Currently supported prefixes are :
9639 - 'ipv4@' -> address is always IPv4
9640 - 'ipv6@' -> address is always IPv6
9641 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02009642 - 'abns@' -> address is in abstract namespace (Linux only)
William Lallemand2fe7dd02018-09-11 16:51:29 +02009643 - 'sockpair@' -> address is the FD of a connected unix
9644 socket or of a socketpair. During a connection, the
9645 backend creates a pair of connected sockets, and passes
9646 one of them over the FD. The bind part will use the
9647 received socket as the client FD. Should be used
9648 carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02009649 You may want to reference some environment variables in the
9650 address parameter, see section 2.3 about environment
Willy Tarreau6a031d12016-11-07 19:42:35 +01009651 variables. The "init-addr" setting can be used to modify the way
9652 IP addresses should be resolved upon startup.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009653
Willy Tarreaub6205fd2012-09-24 12:27:33 +02009654 <port> is an optional port specification. If set, all connections will
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009655 be sent to this port. If unset, the same port the client
9656 connected to will be used. The port may also be prefixed by a "+"
9657 or a "-". In this case, the server's port will be determined by
9658 adding this value to the client's port.
9659
9660 <param*> is a list of parameters for this server. The "server" keywords
9661 accepts an important number of options and has a complete section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009662 dedicated to it. Please refer to section 5 for more details.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009663
9664 Examples :
9665 server first 10.1.1.1:1080 cookie first check inter 1000
9666 server second 10.1.1.2:1080 cookie second check inter 1000
Willy Tarreau24709282013-03-10 21:32:12 +01009667 server transp ipv4@
William Lallemandb2f07452015-05-12 14:27:13 +02009668 server backup "${SRV_BACKUP}:1080" backup
9669 server www1_dc1 "${LAN_DC1}.101:80"
9670 server www1_dc2 "${LAN_DC2}.101:80"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009671
Willy Tarreau55dcaf62015-09-27 15:03:15 +02009672 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
9673 sun_path length is used for the address length. Some other programs
9674 such as socat use the string length only by default. Pass the option
9675 ",unix-tightsocklen=0" to any abstract socket definition in socat to
9676 make it compatible with HAProxy's.
9677
Mark Lamourinec2247f02012-01-04 13:02:01 -05009678 See also: "default-server", "http-send-name-header" and section 5 about
9679 server options
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009680
Christopher Faulet05df94b2021-02-12 09:27:10 +01009681server-state-file-name [ { use-backend-name | <file> } ]
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02009682 Set the server state file to read, load and apply to servers available in
Christopher Faulet05df94b2021-02-12 09:27:10 +01009683 this backend.
9684 May be used in sections: defaults | frontend | listen | backend
9685 no | no | yes | yes
9686
9687 It only applies when the directive "load-server-state-from-file" is set to
9688 "local". When <file> is not provided, if "use-backend-name" is used or if
9689 this directive is not set, then backend name is used. If <file> starts with a
9690 slash '/', then it is considered as an absolute path. Otherwise, <file> is
9691 concatenated to the global directive "server-state-base".
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02009692
9693 Example: the minimal configuration below would make HAProxy look for the
9694 state server file '/etc/haproxy/states/bk':
9695
9696 global
9697 server-state-file-base /etc/haproxy/states
9698
Willy Tarreauc9c6cdb2020-03-05 16:03:58 +01009699 backend bk
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02009700 load-server-state-from-file
9701
Christopher Faulet05df94b2021-02-12 09:27:10 +01009702 See also: "server-state-base", "load-server-state-from-file", and
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02009703 "show servers state"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009704
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02009705server-template <prefix> <num | range> <fqdn>[:<port>] [params*]
9706 Set a template to initialize servers with shared parameters.
9707 The names of these servers are built from <prefix> and <num | range> parameters.
9708 May be used in sections : defaults | frontend | listen | backend
9709 no | no | yes | yes
9710
9711 Arguments:
9712 <prefix> A prefix for the server names to be built.
9713
9714 <num | range>
9715 If <num> is provided, this template initializes <num> servers
9716 with 1 up to <num> as server name suffixes. A range of numbers
9717 <num_low>-<num_high> may also be used to use <num_low> up to
9718 <num_high> as server name suffixes.
9719
9720 <fqdn> A FQDN for all the servers this template initializes.
9721
9722 <port> Same meaning as "server" <port> argument (see "server" keyword).
9723
9724 <params*>
9725 Remaining server parameters among all those supported by "server"
9726 keyword.
9727
9728 Examples:
9729 # Initializes 3 servers with srv1, srv2 and srv3 as names,
9730 # google.com as FQDN, and health-check enabled.
9731 server-template srv 1-3 google.com:80 check
9732
9733 # or
9734 server-template srv 3 google.com:80 check
9735
9736 # would be equivalent to:
9737 server srv1 google.com:80 check
9738 server srv2 google.com:80 check
9739 server srv3 google.com:80 check
9740
9741
9742
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009743source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +02009744source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreaud53f96b2009-02-04 18:46:54 +01009745source <addr>[:<port>] [interface <name>]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009746 Set the source address for outgoing connections
9747 May be used in sections : defaults | frontend | listen | backend
9748 yes | no | yes | yes
9749 Arguments :
9750 <addr> is the IPv4 address HAProxy will bind to before connecting to a
9751 server. This address is also used as a source for health checks.
Willy Tarreau24709282013-03-10 21:32:12 +01009752
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009753 The default value of 0.0.0.0 means that the system will select
Willy Tarreau24709282013-03-10 21:32:12 +01009754 the most appropriate address to reach its destination. Optionally
9755 an address family prefix may be used before the address to force
9756 the family regardless of the address format, which can be useful
9757 to specify a path to a unix socket with no slash ('/'). Currently
9758 supported prefixes are :
9759 - 'ipv4@' -> address is always IPv4
9760 - 'ipv6@' -> address is always IPv6
9761 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02009762 - 'abns@' -> address is in abstract namespace (Linux only)
Cyril Bonté307ee1e2015-09-28 23:16:06 +02009763 You may want to reference some environment variables in the
9764 address parameter, see section 2.3 about environment variables.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009765
9766 <port> is an optional port. It is normally not needed but may be useful
9767 in some very specific contexts. The default value of zero means
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02009768 the system will select a free port. Note that port ranges are not
9769 supported in the backend. If you want to force port ranges, you
9770 have to specify them on each "server" line.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009771
9772 <addr2> is the IP address to present to the server when connections are
9773 forwarded in full transparent proxy mode. This is currently only
9774 supported on some patched Linux kernels. When this address is
9775 specified, clients connecting to the server will be presented
9776 with this address, while health checks will still use the address
9777 <addr>.
9778
9779 <port2> is the optional port to present to the server when connections
9780 are forwarded in full transparent proxy mode (see <addr2> above).
9781 The default value of zero means the system will select a free
9782 port.
9783
Willy Tarreaubce70882009-09-07 11:51:47 +02009784 <hdr> is the name of a HTTP header in which to fetch the IP to bind to.
9785 This is the name of a comma-separated header list which can
9786 contain multiple IP addresses. By default, the last occurrence is
9787 used. This is designed to work with the X-Forwarded-For header
Baptiste Assmannea3e73b2013-02-02 23:47:49 +01009788 and to automatically bind to the client's IP address as seen
Willy Tarreaubce70882009-09-07 11:51:47 +02009789 by previous proxy, typically Stunnel. In order to use another
9790 occurrence from the last one, please see the <occ> parameter
9791 below. When the header (or occurrence) is not found, no binding
9792 is performed so that the proxy's default IP address is used. Also
9793 keep in mind that the header name is case insensitive, as for any
9794 HTTP header.
9795
9796 <occ> is the occurrence number of a value to be used in a multi-value
9797 header. This is to be used in conjunction with "hdr_ip(<hdr>)",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04009798 in order to specify which occurrence to use for the source IP
Willy Tarreaubce70882009-09-07 11:51:47 +02009799 address. Positive values indicate a position from the first
9800 occurrence, 1 being the first one. Negative values indicate
9801 positions relative to the last one, -1 being the last one. This
9802 is helpful for situations where an X-Forwarded-For header is set
9803 at the entry point of an infrastructure and must be used several
9804 proxy layers away. When this value is not specified, -1 is
9805 assumed. Passing a zero here disables the feature.
9806
Willy Tarreaud53f96b2009-02-04 18:46:54 +01009807 <name> is an optional interface name to which to bind to for outgoing
9808 traffic. On systems supporting this features (currently, only
9809 Linux), this allows one to bind all traffic to the server to
9810 this interface even if it is not the one the system would select
9811 based on routing tables. This should be used with extreme care.
9812 Note that using this option requires root privileges.
9813
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009814 The "source" keyword is useful in complex environments where a specific
9815 address only is allowed to connect to the servers. It may be needed when a
9816 private address must be used through a public gateway for instance, and it is
9817 known that the system cannot determine the adequate source address by itself.
9818
9819 An extension which is available on certain patched Linux kernels may be used
9820 through the "usesrc" optional keyword. It makes it possible to connect to the
9821 servers with an IP address which does not belong to the system itself. This
9822 is called "full transparent proxy mode". For this to work, the destination
9823 servers have to route their traffic back to this address through the machine
9824 running HAProxy, and IP forwarding must generally be enabled on this machine.
9825
9826 In this "full transparent proxy" mode, it is possible to force a specific IP
9827 address to be presented to the servers. This is not much used in fact. A more
9828 common use is to tell HAProxy to present the client's IP address. For this,
9829 there are two methods :
9830
9831 - present the client's IP and port addresses. This is the most transparent
9832 mode, but it can cause problems when IP connection tracking is enabled on
9833 the machine, because a same connection may be seen twice with different
9834 states. However, this solution presents the huge advantage of not
9835 limiting the system to the 64k outgoing address+port couples, because all
9836 of the client ranges may be used.
9837
9838 - present only the client's IP address and select a spare port. This
9839 solution is still quite elegant but slightly less transparent (downstream
9840 firewalls logs will not match upstream's). It also presents the downside
9841 of limiting the number of concurrent connections to the usual 64k ports.
9842 However, since the upstream and downstream ports are different, local IP
9843 connection tracking on the machine will not be upset by the reuse of the
9844 same session.
9845
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009846 This option sets the default source for all servers in the backend. It may
9847 also be specified in a "defaults" section. Finer source address specification
9848 is possible at the server level using the "source" server option. Refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009849 section 5 for more information.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009850
Baptiste Assmann91bd3372015-07-17 21:59:42 +02009851 In order to work, "usesrc" requires root privileges.
9852
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009853 Examples :
9854 backend private
9855 # Connect to the servers using our 192.168.1.200 source address
9856 source 192.168.1.200
9857
9858 backend transparent_ssl1
9859 # Connect to the SSL farm from the client's source address
9860 source 192.168.1.200 usesrc clientip
9861
9862 backend transparent_ssl2
9863 # Connect to the SSL farm from the client's source address and port
9864 # not recommended if IP conntrack is present on the local machine.
9865 source 192.168.1.200 usesrc client
9866
9867 backend transparent_ssl3
9868 # Connect to the SSL farm from the client's source address. It
9869 # is more conntrack-friendly.
9870 source 192.168.1.200 usesrc clientip
9871
9872 backend transparent_smtp
9873 # Connect to the SMTP farm from the client's source address/port
9874 # with Tproxy version 4.
9875 source 0.0.0.0 usesrc clientip
9876
Willy Tarreaubce70882009-09-07 11:51:47 +02009877 backend transparent_http
9878 # Connect to the servers using the client's IP as seen by previous
9879 # proxy.
9880 source 0.0.0.0 usesrc hdr_ip(x-forwarded-for,-1)
9881
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009882 See also : the "source" server option in section 5, the Tproxy patches for
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009883 the Linux kernel on www.balabit.com, the "bind" keyword.
9884
Willy Tarreau844e3c52008-01-11 16:28:18 +01009885
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09009886srvtcpka-cnt <count>
9887 Sets the maximum number of keepalive probes TCP should send before dropping
9888 the connection on the server side.
9889 May be used in sections : defaults | frontend | listen | backend
9890 yes | no | yes | yes
9891 Arguments :
9892 <count> is the maximum number of keepalive probes.
9893
9894 This keyword corresponds to the socket option TCP_KEEPCNT. If this keyword
9895 is not specified, system-wide TCP parameter (tcp_keepalive_probes) is used.
Willy Tarreau52543212020-07-09 05:58:51 +02009896 The availability of this setting depends on the operating system. It is
9897 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09009898
9899 See also : "option srvtcpka", "srvtcpka-idle", "srvtcpka-intvl".
9900
9901
9902srvtcpka-idle <timeout>
9903 Sets the time the connection needs to remain idle before TCP starts sending
9904 keepalive probes, if enabled the sending of TCP keepalive packets on the
9905 server side.
9906 May be used in sections : defaults | frontend | listen | backend
9907 yes | no | yes | yes
9908 Arguments :
9909 <timeout> is the time the connection needs to remain idle before TCP starts
9910 sending keepalive probes. It is specified in seconds by default,
9911 but can be in any other unit if the number is suffixed by the
9912 unit, as explained at the top of this document.
9913
9914 This keyword corresponds to the socket option TCP_KEEPIDLE. If this keyword
9915 is not specified, system-wide TCP parameter (tcp_keepalive_time) is used.
Willy Tarreau52543212020-07-09 05:58:51 +02009916 The availability of this setting depends on the operating system. It is
9917 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09009918
9919 See also : "option srvtcpka", "srvtcpka-cnt", "srvtcpka-intvl".
9920
9921
9922srvtcpka-intvl <timeout>
9923 Sets the time between individual keepalive probes on the server side.
9924 May be used in sections : defaults | frontend | listen | backend
9925 yes | no | yes | yes
9926 Arguments :
9927 <timeout> is the time between individual keepalive probes. It is specified
9928 in seconds by default, but can be in any other unit if the number
9929 is suffixed by the unit, as explained at the top of this
9930 document.
9931
9932 This keyword corresponds to the socket option TCP_KEEPINTVL. If this keyword
9933 is not specified, system-wide TCP parameter (tcp_keepalive_intvl) is used.
Willy Tarreau52543212020-07-09 05:58:51 +02009934 The availability of this setting depends on the operating system. It is
9935 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09009936
9937 See also : "option srvtcpka", "srvtcpka-cnt", "srvtcpka-idle".
9938
9939
Cyril Bonté66c327d2010-10-12 00:14:37 +02009940stats admin { if | unless } <cond>
9941 Enable statistics admin level if/unless a condition is matched
9942 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02009943 no | yes | yes | yes
Cyril Bonté66c327d2010-10-12 00:14:37 +02009944
9945 This statement enables the statistics admin level if/unless a condition is
9946 matched.
9947
9948 The admin level allows to enable/disable servers from the web interface. By
9949 default, statistics page is read-only for security reasons.
9950
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009951 Note : Consider not using this feature in multi-process mode (nbproc > 1)
9952 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01009953 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009954
Cyril Bonté23b39d92011-02-10 22:54:44 +01009955 Currently, the POST request is limited to the buffer size minus the reserved
9956 buffer space, which means that if the list of servers is too long, the
9957 request won't be processed. It is recommended to alter few servers at a
9958 time.
Cyril Bonté66c327d2010-10-12 00:14:37 +02009959
9960 Example :
9961 # statistics admin level only for localhost
9962 backend stats_localhost
9963 stats enable
9964 stats admin if LOCALHOST
9965
9966 Example :
9967 # statistics admin level always enabled because of the authentication
9968 backend stats_auth
9969 stats enable
9970 stats auth admin:AdMiN123
9971 stats admin if TRUE
9972
9973 Example :
9974 # statistics admin level depends on the authenticated user
9975 userlist stats-auth
9976 group admin users admin
9977 user admin insecure-password AdMiN123
9978 group readonly users haproxy
9979 user haproxy insecure-password haproxy
9980
9981 backend stats_auth
9982 stats enable
9983 acl AUTH http_auth(stats-auth)
9984 acl AUTH_ADMIN http_auth_group(stats-auth) admin
9985 stats http-request auth unless AUTH
9986 stats admin if AUTH_ADMIN
9987
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009988 See also : "stats enable", "stats auth", "stats http-request", "nbproc",
9989 "bind-process", section 3.4 about userlists and section 7 about
9990 ACL usage.
Cyril Bonté66c327d2010-10-12 00:14:37 +02009991
9992
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009993stats auth <user>:<passwd>
9994 Enable statistics with authentication and grant access to an account
9995 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02009996 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009997 Arguments :
9998 <user> is a user name to grant access to
9999
10000 <passwd> is the cleartext password associated to this user
10001
10002 This statement enables statistics with default settings, and restricts access
10003 to declared users only. It may be repeated as many times as necessary to
10004 allow as many users as desired. When a user tries to access the statistics
10005 without a valid account, a "401 Forbidden" response will be returned so that
10006 the browser asks the user to provide a valid user and password. The real
10007 which will be returned to the browser is configurable using "stats realm".
10008
10009 Since the authentication method is HTTP Basic Authentication, the passwords
10010 circulate in cleartext on the network. Thus, it was decided that the
10011 configuration file would also use cleartext passwords to remind the users
Willy Tarreau3c92c5f2011-08-28 09:45:47 +020010012 that those ones should not be sensitive and not shared with any other account.
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010013
10014 It is also possible to reduce the scope of the proxies which appear in the
10015 report using "stats scope".
10016
10017 Though this statement alone is enough to enable statistics reporting, it is
10018 recommended to set all other settings in order to avoid relying on default
10019 unobvious parameters.
10020
10021 Example :
10022 # public access (limited to this backend only)
10023 backend public_www
10024 server srv1 192.168.0.1:80
10025 stats enable
10026 stats hide-version
10027 stats scope .
10028 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010029 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010030 stats auth admin1:AdMiN123
10031 stats auth admin2:AdMiN321
10032
10033 # internal monitoring access (unlimited)
10034 backend private_monitoring
10035 stats enable
10036 stats uri /admin?stats
10037 stats refresh 5s
10038
10039 See also : "stats enable", "stats realm", "stats scope", "stats uri"
10040
10041
10042stats enable
10043 Enable statistics reporting with default settings
10044 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010045 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010046 Arguments : none
10047
10048 This statement enables statistics reporting with default settings defined
10049 at build time. Unless stated otherwise, these settings are used :
10050 - stats uri : /haproxy?stats
10051 - stats realm : "HAProxy Statistics"
10052 - stats auth : no authentication
10053 - stats scope : no restriction
10054
10055 Though this statement alone is enough to enable statistics reporting, it is
10056 recommended to set all other settings in order to avoid relying on default
10057 unobvious parameters.
10058
10059 Example :
10060 # public access (limited to this backend only)
10061 backend public_www
10062 server srv1 192.168.0.1:80
10063 stats enable
10064 stats hide-version
10065 stats scope .
10066 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010067 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010068 stats auth admin1:AdMiN123
10069 stats auth admin2:AdMiN321
10070
10071 # internal monitoring access (unlimited)
10072 backend private_monitoring
10073 stats enable
10074 stats uri /admin?stats
10075 stats refresh 5s
10076
10077 See also : "stats auth", "stats realm", "stats uri"
10078
10079
Willy Tarreaud63335a2010-02-26 12:56:52 +010010080stats hide-version
10081 Enable statistics and hide HAProxy version reporting
Willy Tarreau1d45b7c2009-08-16 10:29:18 +020010082 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010083 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +010010084 Arguments : none
Willy Tarreau1d45b7c2009-08-16 10:29:18 +020010085
Willy Tarreaud63335a2010-02-26 12:56:52 +010010086 By default, the stats page reports some useful status information along with
10087 the statistics. Among them is HAProxy's version. However, it is generally
10088 considered dangerous to report precise version to anyone, as it can help them
10089 target known weaknesses with specific attacks. The "stats hide-version"
10090 statement removes the version from the statistics report. This is recommended
10091 for public sites or any site with a weak login/password.
Willy Tarreau1d45b7c2009-08-16 10:29:18 +020010092
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +020010093 Though this statement alone is enough to enable statistics reporting, it is
10094 recommended to set all other settings in order to avoid relying on default
10095 unobvious parameters.
10096
Willy Tarreaud63335a2010-02-26 12:56:52 +010010097 Example :
10098 # public access (limited to this backend only)
10099 backend public_www
10100 server srv1 192.168.0.1:80
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +020010101 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +010010102 stats hide-version
10103 stats scope .
10104 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010105 stats realm HAProxy\ Statistics
Willy Tarreaud63335a2010-02-26 12:56:52 +010010106 stats auth admin1:AdMiN123
10107 stats auth admin2:AdMiN321
Willy Tarreau1d45b7c2009-08-16 10:29:18 +020010108
Willy Tarreau1d45b7c2009-08-16 10:29:18 +020010109 # internal monitoring access (unlimited)
10110 backend private_monitoring
10111 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +010010112 stats uri /admin?stats
10113 stats refresh 5s
Krzysztof Piotr Oledzki15514c22010-01-04 16:03:09 +010010114
Willy Tarreaud63335a2010-02-26 12:56:52 +010010115 See also : "stats auth", "stats enable", "stats realm", "stats uri"
Willy Tarreau1d45b7c2009-08-16 10:29:18 +020010116
Willy Tarreau983e01e2010-01-11 18:42:06 +010010117
Cyril Bonté2be1b3f2010-09-30 23:46:30 +020010118stats http-request { allow | deny | auth [realm <realm>] }
10119 [ { if | unless } <condition> ]
10120 Access control for statistics
10121
10122 May be used in sections: defaults | frontend | listen | backend
10123 no | no | yes | yes
10124
10125 As "http-request", these set of options allow to fine control access to
10126 statistics. Each option may be followed by if/unless and acl.
10127 First option with matched condition (or option without condition) is final.
10128 For "deny" a 403 error will be returned, for "allow" normal processing is
10129 performed, for "auth" a 401/407 error code is returned so the client
10130 should be asked to enter a username and password.
10131
10132 There is no fixed limit to the number of http-request statements per
10133 instance.
10134
10135 See also : "http-request", section 3.4 about userlists and section 7
10136 about ACL usage.
10137
10138
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010139stats realm <realm>
10140 Enable statistics and set authentication realm
10141 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010142 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010143 Arguments :
10144 <realm> is the name of the HTTP Basic Authentication realm reported to
10145 the browser. The browser uses it to display it in the pop-up
10146 inviting the user to enter a valid username and password.
10147
10148 The realm is read as a single word, so any spaces in it should be escaped
10149 using a backslash ('\').
10150
10151 This statement is useful only in conjunction with "stats auth" since it is
10152 only related to authentication.
10153
10154 Though this statement alone is enough to enable statistics reporting, it is
10155 recommended to set all other settings in order to avoid relying on default
10156 unobvious parameters.
10157
10158 Example :
10159 # public access (limited to this backend only)
10160 backend public_www
10161 server srv1 192.168.0.1:80
10162 stats enable
10163 stats hide-version
10164 stats scope .
10165 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010166 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010167 stats auth admin1:AdMiN123
10168 stats auth admin2:AdMiN321
10169
10170 # internal monitoring access (unlimited)
10171 backend private_monitoring
10172 stats enable
10173 stats uri /admin?stats
10174 stats refresh 5s
10175
10176 See also : "stats auth", "stats enable", "stats uri"
10177
10178
10179stats refresh <delay>
10180 Enable statistics with automatic refresh
10181 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010182 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010183 Arguments :
10184 <delay> is the suggested refresh delay, specified in seconds, which will
10185 be returned to the browser consulting the report page. While the
10186 browser is free to apply any delay, it will generally respect it
10187 and refresh the page this every seconds. The refresh interval may
10188 be specified in any other non-default time unit, by suffixing the
10189 unit after the value, as explained at the top of this document.
10190
10191 This statement is useful on monitoring displays with a permanent page
10192 reporting the load balancer's activity. When set, the HTML report page will
10193 include a link "refresh"/"stop refresh" so that the user can select whether
Jackie Tapia749f74c2020-07-22 18:59:40 -050010194 they want automatic refresh of the page or not.
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010195
10196 Though this statement alone is enough to enable statistics reporting, it is
10197 recommended to set all other settings in order to avoid relying on default
10198 unobvious parameters.
10199
10200 Example :
10201 # public access (limited to this backend only)
10202 backend public_www
10203 server srv1 192.168.0.1:80
10204 stats enable
10205 stats hide-version
10206 stats scope .
10207 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010208 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010209 stats auth admin1:AdMiN123
10210 stats auth admin2:AdMiN321
10211
10212 # internal monitoring access (unlimited)
10213 backend private_monitoring
10214 stats enable
10215 stats uri /admin?stats
10216 stats refresh 5s
10217
10218 See also : "stats auth", "stats enable", "stats realm", "stats uri"
10219
10220
10221stats scope { <name> | "." }
10222 Enable statistics and limit access scope
10223 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010224 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010225 Arguments :
10226 <name> is the name of a listen, frontend or backend section to be
10227 reported. The special name "." (a single dot) designates the
10228 section in which the statement appears.
10229
10230 When this statement is specified, only the sections enumerated with this
10231 statement will appear in the report. All other ones will be hidden. This
10232 statement may appear as many times as needed if multiple sections need to be
10233 reported. Please note that the name checking is performed as simple string
10234 comparisons, and that it is never checked that a give section name really
10235 exists.
10236
10237 Though this statement alone is enough to enable statistics reporting, it is
10238 recommended to set all other settings in order to avoid relying on default
10239 unobvious parameters.
10240
10241 Example :
10242 # public access (limited to this backend only)
10243 backend public_www
10244 server srv1 192.168.0.1:80
10245 stats enable
10246 stats hide-version
10247 stats scope .
10248 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010249 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010250 stats auth admin1:AdMiN123
10251 stats auth admin2:AdMiN321
10252
10253 # internal monitoring access (unlimited)
10254 backend private_monitoring
10255 stats enable
10256 stats uri /admin?stats
10257 stats refresh 5s
10258
10259 See also : "stats auth", "stats enable", "stats realm", "stats uri"
10260
Willy Tarreaud63335a2010-02-26 12:56:52 +010010261
Willy Tarreauc9705a12010-07-27 20:05:50 +020010262stats show-desc [ <desc> ]
Willy Tarreaud63335a2010-02-26 12:56:52 +010010263 Enable reporting of a description on the statistics page.
10264 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010265 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +010010266
Willy Tarreauc9705a12010-07-27 20:05:50 +020010267 <desc> is an optional description to be reported. If unspecified, the
Willy Tarreaud63335a2010-02-26 12:56:52 +010010268 description from global section is automatically used instead.
10269
10270 This statement is useful for users that offer shared services to their
10271 customers, where node or description should be different for each customer.
10272
10273 Though this statement alone is enough to enable statistics reporting, it is
10274 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +010010275 unobvious parameters. By default description is not shown.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010276
10277 Example :
10278 # internal monitoring access (unlimited)
10279 backend private_monitoring
10280 stats enable
10281 stats show-desc Master node for Europe, Asia, Africa
10282 stats uri /admin?stats
10283 stats refresh 5s
10284
10285 See also: "show-node", "stats enable", "stats uri" and "description" in
10286 global section.
10287
10288
10289stats show-legends
Willy Tarreaued2119c2014-04-24 22:10:39 +020010290 Enable reporting additional information on the statistics page
10291 May be used in sections : defaults | frontend | listen | backend
10292 yes | yes | yes | yes
10293 Arguments : none
10294
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010295 Enable reporting additional information on the statistics page :
Willy Tarreaud63335a2010-02-26 12:56:52 +010010296 - cap: capabilities (proxy)
10297 - mode: one of tcp, http or health (proxy)
10298 - id: SNMP ID (proxy, socket, server)
10299 - IP (socket, server)
10300 - cookie (backend, server)
10301
10302 Though this statement alone is enough to enable statistics reporting, it is
10303 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +010010304 unobvious parameters. Default behavior is not to show this information.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010305
10306 See also: "stats enable", "stats uri".
10307
10308
Amaury Denoyelle0b70a8a2020-10-05 11:49:45 +020010309stats show-modules
10310 Enable display of extra statistics module on the statistics page
10311 May be used in sections : defaults | frontend | listen | backend
10312 yes | yes | yes | yes
10313 Arguments : none
10314
10315 New columns are added at the end of the line containing the extra statistics
10316 values as a tooltip.
10317
10318 Though this statement alone is enough to enable statistics reporting, it is
10319 recommended to set all other settings in order to avoid relying on default
10320 unobvious parameters. Default behavior is not to show this information.
10321
10322 See also: "stats enable", "stats uri".
10323
10324
Willy Tarreaud63335a2010-02-26 12:56:52 +010010325stats show-node [ <name> ]
10326 Enable reporting of a host name on the statistics page.
10327 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010328 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +010010329 Arguments:
10330 <name> is an optional name to be reported. If unspecified, the
10331 node name from global section is automatically used instead.
10332
10333 This statement is useful for users that offer shared services to their
10334 customers, where node or description might be different on a stats page
Davor Ocelice9ed2812017-12-25 17:49:28 +010010335 provided for each customer. Default behavior is not to show host name.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010336
10337 Though this statement alone is enough to enable statistics reporting, it is
10338 recommended to set all other settings in order to avoid relying on default
10339 unobvious parameters.
10340
10341 Example:
10342 # internal monitoring access (unlimited)
10343 backend private_monitoring
10344 stats enable
10345 stats show-node Europe-1
10346 stats uri /admin?stats
10347 stats refresh 5s
10348
10349 See also: "show-desc", "stats enable", "stats uri", and "node" in global
10350 section.
10351
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010352
10353stats uri <prefix>
10354 Enable statistics and define the URI prefix to access them
10355 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010356 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010357 Arguments :
10358 <prefix> is the prefix of any URI which will be redirected to stats. This
10359 prefix may contain a question mark ('?') to indicate part of a
10360 query string.
10361
10362 The statistics URI is intercepted on the relayed traffic, so it appears as a
10363 page within the normal application. It is strongly advised to ensure that the
10364 selected URI will never appear in the application, otherwise it will never be
10365 possible to reach it in the application.
10366
10367 The default URI compiled in haproxy is "/haproxy?stats", but this may be
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010010368 changed at build time, so it's better to always explicitly specify it here.
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010369 It is generally a good idea to include a question mark in the URI so that
10370 intermediate proxies refrain from caching the results. Also, since any string
10371 beginning with the prefix will be accepted as a stats request, the question
10372 mark helps ensuring that no valid URI will begin with the same words.
10373
10374 It is sometimes very convenient to use "/" as the URI prefix, and put that
10375 statement in a "listen" instance of its own. That makes it easy to dedicate
10376 an address or a port to statistics only.
10377
10378 Though this statement alone is enough to enable statistics reporting, it is
10379 recommended to set all other settings in order to avoid relying on default
10380 unobvious parameters.
10381
10382 Example :
10383 # public access (limited to this backend only)
10384 backend public_www
10385 server srv1 192.168.0.1:80
10386 stats enable
10387 stats hide-version
10388 stats scope .
10389 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010390 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010391 stats auth admin1:AdMiN123
10392 stats auth admin2:AdMiN321
10393
10394 # internal monitoring access (unlimited)
10395 backend private_monitoring
10396 stats enable
10397 stats uri /admin?stats
10398 stats refresh 5s
10399
10400 See also : "stats auth", "stats enable", "stats realm"
10401
10402
Willy Tarreaud63335a2010-02-26 12:56:52 +010010403stick match <pattern> [table <table>] [{if | unless} <cond>]
10404 Define a request pattern matching condition to stick a user to a server
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010405 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaud63335a2010-02-26 12:56:52 +010010406 no | no | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010407
10408 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +020010409 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010410 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +010010411 will be analyzed in the hope to find a matching entry in a
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010412 stickiness table. This rule is mandatory.
10413
10414 <table> is an optional stickiness table name. If unspecified, the same
10415 backend's table is used. A stickiness table is declared using
10416 the "stick-table" statement.
10417
10418 <cond> is an optional matching condition. It makes it possible to match
10419 on a certain criterion only when other conditions are met (or
10420 not met). For instance, it could be used to match on a source IP
10421 address except when a request passes through a known proxy, in
10422 which case we'd match on a header containing that IP address.
10423
10424 Some protocols or applications require complex stickiness rules and cannot
10425 always simply rely on cookies nor hashing. The "stick match" statement
10426 describes a rule to extract the stickiness criterion from an incoming request
10427 or connection. See section 7 for a complete list of possible patterns and
10428 transformation rules.
10429
10430 The table has to be declared using the "stick-table" statement. It must be of
10431 a type compatible with the pattern. By default it is the one which is present
10432 in the same backend. It is possible to share a table with other backends by
10433 referencing it using the "table" keyword. If another table is referenced,
10434 the server's ID inside the backends are used. By default, all server IDs
10435 start at 1 in each backend, so the server ordering is enough. But in case of
10436 doubt, it is highly recommended to force server IDs using their "id" setting.
10437
10438 It is possible to restrict the conditions where a "stick match" statement
10439 will apply, using "if" or "unless" followed by a condition. See section 7 for
10440 ACL based conditions.
10441
10442 There is no limit on the number of "stick match" statements. The first that
10443 applies and matches will cause the request to be directed to the same server
10444 as was used for the request which created the entry. That way, multiple
10445 matches can be used as fallbacks.
10446
10447 The stick rules are checked after the persistence cookies, so they will not
10448 affect stickiness if a cookie has already been used to select a server. That
10449 way, it becomes very easy to insert cookies and match on IP addresses in
10450 order to maintain stickiness between HTTP and HTTPS.
10451
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010452 Note : Consider not using this feature in multi-process mode (nbproc > 1)
10453 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +010010454 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010455
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010456 Example :
10457 # forward SMTP users to the same server they just used for POP in the
10458 # last 30 minutes
10459 backend pop
10460 mode tcp
10461 balance roundrobin
10462 stick store-request src
10463 stick-table type ip size 200k expire 30m
10464 server s1 192.168.1.1:110
10465 server s2 192.168.1.1:110
10466
10467 backend smtp
10468 mode tcp
10469 balance roundrobin
10470 stick match src table pop
10471 server s1 192.168.1.1:25
10472 server s2 192.168.1.1:25
10473
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010474 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +020010475 about ACLs and samples fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010476
10477
10478stick on <pattern> [table <table>] [{if | unless} <condition>]
10479 Define a request pattern to associate a user to a server
10480 May be used in sections : defaults | frontend | listen | backend
10481 no | no | yes | yes
10482
10483 Note : This form is exactly equivalent to "stick match" followed by
10484 "stick store-request", all with the same arguments. Please refer
10485 to both keywords for details. It is only provided as a convenience
10486 for writing more maintainable configurations.
10487
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010488 Note : Consider not using this feature in multi-process mode (nbproc > 1)
10489 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +010010490 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010491
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010492 Examples :
10493 # The following form ...
Willy Tarreauec579d82010-02-26 19:15:04 +010010494 stick on src table pop if !localhost
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010495
10496 # ...is strictly equivalent to this one :
10497 stick match src table pop if !localhost
10498 stick store-request src table pop if !localhost
10499
10500
10501 # Use cookie persistence for HTTP, and stick on source address for HTTPS as
10502 # well as HTTP without cookie. Share the same table between both accesses.
10503 backend http
10504 mode http
10505 balance roundrobin
10506 stick on src table https
10507 cookie SRV insert indirect nocache
10508 server s1 192.168.1.1:80 cookie s1
10509 server s2 192.168.1.1:80 cookie s2
10510
10511 backend https
10512 mode tcp
10513 balance roundrobin
10514 stick-table type ip size 200k expire 30m
10515 stick on src
10516 server s1 192.168.1.1:443
10517 server s2 192.168.1.1:443
10518
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010519 See also : "stick match", "stick store-request", "nbproc" and "bind-process".
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010520
10521
10522stick store-request <pattern> [table <table>] [{if | unless} <condition>]
10523 Define a request pattern used to create an entry in a stickiness table
10524 May be used in sections : defaults | frontend | listen | backend
10525 no | no | yes | yes
10526
10527 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +020010528 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010529 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +010010530 will be analyzed, extracted and stored in the table once a
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010531 server is selected.
10532
10533 <table> is an optional stickiness table name. If unspecified, the same
10534 backend's table is used. A stickiness table is declared using
10535 the "stick-table" statement.
10536
10537 <cond> is an optional storage condition. It makes it possible to store
10538 certain criteria only when some conditions are met (or not met).
10539 For instance, it could be used to store the source IP address
10540 except when the request passes through a known proxy, in which
10541 case we'd store a converted form of a header containing that IP
10542 address.
10543
10544 Some protocols or applications require complex stickiness rules and cannot
10545 always simply rely on cookies nor hashing. The "stick store-request" statement
10546 describes a rule to decide what to extract from the request and when to do
10547 it, in order to store it into a stickiness table for further requests to
10548 match it using the "stick match" statement. Obviously the extracted part must
10549 make sense and have a chance to be matched in a further request. Storing a
10550 client's IP address for instance often makes sense. Storing an ID found in a
10551 URL parameter also makes sense. Storing a source port will almost never make
10552 any sense because it will be randomly matched. See section 7 for a complete
10553 list of possible patterns and transformation rules.
10554
10555 The table has to be declared using the "stick-table" statement. It must be of
10556 a type compatible with the pattern. By default it is the one which is present
10557 in the same backend. It is possible to share a table with other backends by
10558 referencing it using the "table" keyword. If another table is referenced,
10559 the server's ID inside the backends are used. By default, all server IDs
10560 start at 1 in each backend, so the server ordering is enough. But in case of
10561 doubt, it is highly recommended to force server IDs using their "id" setting.
10562
10563 It is possible to restrict the conditions where a "stick store-request"
10564 statement will apply, using "if" or "unless" followed by a condition. This
10565 condition will be evaluated while parsing the request, so any criteria can be
10566 used. See section 7 for ACL based conditions.
10567
10568 There is no limit on the number of "stick store-request" statements, but
10569 there is a limit of 8 simultaneous stores per request or response. This
10570 makes it possible to store up to 8 criteria, all extracted from either the
10571 request or the response, regardless of the number of rules. Only the 8 first
10572 ones which match will be kept. Using this, it is possible to feed multiple
10573 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +010010574 another protocol or access method. Using multiple store-request rules with
10575 the same table is possible and may be used to find the best criterion to rely
10576 on, by arranging the rules by decreasing preference order. Only the first
10577 extracted criterion for a given table will be stored. All subsequent store-
10578 request rules referencing the same table will be skipped and their ACLs will
10579 not be evaluated.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010580
10581 The "store-request" rules are evaluated once the server connection has been
10582 established, so that the table will contain the real server that processed
10583 the request.
10584
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010585 Note : Consider not using this feature in multi-process mode (nbproc > 1)
10586 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +010010587 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010588
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010589 Example :
10590 # forward SMTP users to the same server they just used for POP in the
10591 # last 30 minutes
10592 backend pop
10593 mode tcp
10594 balance roundrobin
10595 stick store-request src
10596 stick-table type ip size 200k expire 30m
10597 server s1 192.168.1.1:110
10598 server s2 192.168.1.1:110
10599
10600 backend smtp
10601 mode tcp
10602 balance roundrobin
10603 stick match src table pop
10604 server s1 192.168.1.1:25
10605 server s2 192.168.1.1:25
10606
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010607 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +020010608 about ACLs and sample fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010609
10610
Emeric Brun7c6b82e2010-09-24 16:34:28 +020010611stick-table type {ip | integer | string [len <length>] | binary [len <length>]}
Emeric Brunf099e792010-09-27 12:05:28 +020010612 size <size> [expire <expire>] [nopurge] [peers <peersect>]
10613 [store <data_type>]*
Godbach64cef792013-12-04 16:08:22 +080010614 Configure the stickiness table for the current section
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010615 May be used in sections : defaults | frontend | listen | backend
Willy Tarreauc00cdc22010-06-06 16:48:26 +020010616 no | yes | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010617
10618 Arguments :
10619 ip a table declared with "type ip" will only store IPv4 addresses.
10620 This form is very compact (about 50 bytes per entry) and allows
10621 very fast entry lookup and stores with almost no overhead. This
10622 is mainly used to store client source IP addresses.
10623
David du Colombier9a6d3c92011-03-17 10:40:24 +010010624 ipv6 a table declared with "type ipv6" will only store IPv6 addresses.
10625 This form is very compact (about 60 bytes per entry) and allows
10626 very fast entry lookup and stores with almost no overhead. This
10627 is mainly used to store client source IP addresses.
10628
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010629 integer a table declared with "type integer" will store 32bit integers
10630 which can represent a client identifier found in a request for
10631 instance.
10632
10633 string a table declared with "type string" will store substrings of up
10634 to <len> characters. If the string provided by the pattern
10635 extractor is larger than <len>, it will be truncated before
10636 being stored. During matching, at most <len> characters will be
10637 compared between the string in the table and the extracted
10638 pattern. When not specified, the string is automatically limited
Emeric Brun7c6b82e2010-09-24 16:34:28 +020010639 to 32 characters.
10640
10641 binary a table declared with "type binary" will store binary blocks
10642 of <len> bytes. If the block provided by the pattern
10643 extractor is larger than <len>, it will be truncated before
Willy Tarreaube722a22014-06-13 16:31:59 +020010644 being stored. If the block provided by the sample expression
Emeric Brun7c6b82e2010-09-24 16:34:28 +020010645 is shorter than <len>, it will be padded by 0. When not
10646 specified, the block is automatically limited to 32 bytes.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010647
10648 <length> is the maximum number of characters that will be stored in a
Emeric Brun7c6b82e2010-09-24 16:34:28 +020010649 "string" type table (See type "string" above). Or the number
10650 of bytes of the block in "binary" type table. Be careful when
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010651 changing this parameter as memory usage will proportionally
10652 increase.
10653
10654 <size> is the maximum number of entries that can fit in the table. This
Cyril Bonté78caf842010-03-10 22:41:43 +010010655 value directly impacts memory usage. Count approximately
10656 50 bytes per entry, plus the size of a string if any. The size
10657 supports suffixes "k", "m", "g" for 2^10, 2^20 and 2^30 factors.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010658
10659 [nopurge] indicates that we refuse to purge older entries when the table
10660 is full. When not specified and the table is full when haproxy
10661 wants to store an entry in it, it will flush a few of the oldest
10662 entries in order to release some space for the new ones. This is
Davor Ocelice9ed2812017-12-25 17:49:28 +010010663 most often the desired behavior. In some specific cases, it
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010664 be desirable to refuse new entries instead of purging the older
10665 ones. That may be the case when the amount of data to store is
10666 far above the hardware limits and we prefer not to offer access
10667 to new clients than to reject the ones already connected. When
10668 using this parameter, be sure to properly set the "expire"
10669 parameter (see below).
10670
Emeric Brunf099e792010-09-27 12:05:28 +020010671 <peersect> is the name of the peers section to use for replication. Entries
10672 which associate keys to server IDs are kept synchronized with
10673 the remote peers declared in this section. All entries are also
10674 automatically learned from the local peer (old process) during a
10675 soft restart.
10676
Willy Tarreau1abc6732015-05-01 19:21:02 +020010677 NOTE : each peers section may be referenced only by tables
10678 belonging to the same unique process.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010679
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010680 <expire> defines the maximum duration of an entry in the table since it
10681 was last created, refreshed or matched. The expiration delay is
10682 defined using the standard time format, similarly as the various
10683 timeouts. The maximum duration is slightly above 24 days. See
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +030010684 section 2.4 for more information. If this delay is not specified,
Cyril Bontédc4d9032012-04-08 21:57:39 +020010685 the session won't automatically expire, but older entries will
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010686 be removed once full. Be sure not to use the "nopurge" parameter
10687 if not expiration delay is specified.
10688
Willy Tarreau08d5f982010-06-06 13:34:54 +020010689 <data_type> is used to store additional information in the stick-table. This
10690 may be used by ACLs in order to control various criteria related
10691 to the activity of the client matching the stick-table. For each
10692 item specified here, the size of each entry will be inflated so
Willy Tarreauc9705a12010-07-27 20:05:50 +020010693 that the additional data can fit. Several data types may be
10694 stored with an entry. Multiple data types may be specified after
10695 the "store" keyword, as a comma-separated list. Alternatively,
10696 it is possible to repeat the "store" keyword followed by one or
10697 several data types. Except for the "server_id" type which is
10698 automatically detected and enabled, all data types must be
10699 explicitly declared to be stored. If an ACL references a data
10700 type which is not stored, the ACL will simply not match. Some
10701 data types require an argument which must be passed just after
10702 the type between parenthesis. See below for the supported data
10703 types and their arguments.
10704
10705 The data types that can be stored with an entry are the following :
10706 - server_id : this is an integer which holds the numeric ID of the server a
10707 request was assigned to. It is used by the "stick match", "stick store",
10708 and "stick on" rules. It is automatically enabled when referenced.
10709
10710 - gpc0 : first General Purpose Counter. It is a positive 32-bit integer
10711 integer which may be used for anything. Most of the time it will be used
10712 to put a special tag on some entries, for instance to note that a
Davor Ocelice9ed2812017-12-25 17:49:28 +010010713 specific behavior was detected and must be known for future matches.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010714
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010715 - gpc0_rate(<period>) : increment rate of the first General Purpose Counter
10716 over a period. It is a positive 32-bit integer integer which may be used
10717 for anything. Just like <gpc0>, it counts events, but instead of keeping
Davor Ocelice9ed2812017-12-25 17:49:28 +010010718 a cumulative number, it maintains the rate at which the counter is
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010719 incremented. Most of the time it will be used to measure the frequency of
Davor Ocelice9ed2812017-12-25 17:49:28 +010010720 occurrence of certain events (e.g. requests to a specific URL).
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010721
Frédéric Lécaille6778b272018-01-29 15:22:53 +010010722 - gpc1 : second General Purpose Counter. It is a positive 32-bit integer
10723 integer which may be used for anything. Most of the time it will be used
10724 to put a special tag on some entries, for instance to note that a
10725 specific behavior was detected and must be known for future matches.
10726
10727 - gpc1_rate(<period>) : increment rate of the second General Purpose Counter
10728 over a period. It is a positive 32-bit integer integer which may be used
10729 for anything. Just like <gpc1>, it counts events, but instead of keeping
10730 a cumulative number, it maintains the rate at which the counter is
10731 incremented. Most of the time it will be used to measure the frequency of
10732 occurrence of certain events (e.g. requests to a specific URL).
10733
Emeric Brunee3f0b12021-07-01 18:34:48 +020010734 - gpt0 : first General Purpose Tag. It is a positive 32-bit integer
10735 integer which may be used for anything. Most of the time it will be used
10736 to put a special tag on some entries, for instance to note that a
10737 specific behavior was detected and must be known for future matches
10738
Willy Tarreauc9705a12010-07-27 20:05:50 +020010739 - conn_cnt : Connection Count. It is a positive 32-bit integer which counts
10740 the absolute number of connections received from clients which matched
10741 this entry. It does not mean the connections were accepted, just that
10742 they were received.
10743
10744 - conn_cur : Current Connections. It is a positive 32-bit integer which
10745 stores the concurrent connection counts for the entry. It is incremented
10746 once an incoming connection matches the entry, and decremented once the
10747 connection leaves. That way it is possible to know at any time the exact
10748 number of concurrent connections for an entry.
10749
10750 - conn_rate(<period>) : frequency counter (takes 12 bytes). It takes an
10751 integer parameter <period> which indicates in milliseconds the length
10752 of the period over which the average is measured. It reports the average
10753 incoming connection rate over that period, in connections per period. The
10754 result is an integer which can be matched using ACLs.
10755
10756 - sess_cnt : Session Count. It is a positive 32-bit integer which counts
10757 the absolute number of sessions received from clients which matched this
10758 entry. A session is a connection that was accepted by the layer 4 rules.
10759
10760 - sess_rate(<period>) : frequency counter (takes 12 bytes). It takes an
10761 integer parameter <period> which indicates in milliseconds the length
10762 of the period over which the average is measured. It reports the average
10763 incoming session rate over that period, in sessions per period. The
10764 result is an integer which can be matched using ACLs.
10765
10766 - http_req_cnt : HTTP request Count. It is a positive 32-bit integer which
10767 counts the absolute number of HTTP requests received from clients which
10768 matched this entry. It does not matter whether they are valid requests or
10769 not. Note that this is different from sessions when keep-alive is used on
10770 the client side.
10771
10772 - http_req_rate(<period>) : frequency counter (takes 12 bytes). It takes an
10773 integer parameter <period> which indicates in milliseconds the length
10774 of the period over which the average is measured. It reports the average
10775 HTTP request rate over that period, in requests per period. The result is
10776 an integer which can be matched using ACLs. It does not matter whether
10777 they are valid requests or not. Note that this is different from sessions
10778 when keep-alive is used on the client side.
10779
10780 - http_err_cnt : HTTP Error Count. It is a positive 32-bit integer which
10781 counts the absolute number of HTTP requests errors induced by clients
10782 which matched this entry. Errors are counted on invalid and truncated
10783 requests, as well as on denied or tarpitted requests, and on failed
10784 authentications. If the server responds with 4xx, then the request is
10785 also counted as an error since it's an error triggered by the client
Davor Ocelice9ed2812017-12-25 17:49:28 +010010786 (e.g. vulnerability scan).
Willy Tarreauc9705a12010-07-27 20:05:50 +020010787
10788 - http_err_rate(<period>) : frequency counter (takes 12 bytes). It takes an
10789 integer parameter <period> which indicates in milliseconds the length
10790 of the period over which the average is measured. It reports the average
10791 HTTP request error rate over that period, in requests per period (see
10792 http_err_cnt above for what is accounted as an error). The result is an
10793 integer which can be matched using ACLs.
10794
10795 - bytes_in_cnt : client to server byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +010010796 integer which counts the cumulative number of bytes received from clients
Willy Tarreauc9705a12010-07-27 20:05:50 +020010797 which matched this entry. Headers are included in the count. This may be
10798 used to limit abuse of upload features on photo or video servers.
10799
10800 - bytes_in_rate(<period>) : frequency counter (takes 12 bytes). It takes an
10801 integer parameter <period> which indicates in milliseconds the length
10802 of the period over which the average is measured. It reports the average
10803 incoming bytes rate over that period, in bytes per period. It may be used
10804 to detect users which upload too much and too fast. Warning: with large
10805 uploads, it is possible that the amount of uploaded data will be counted
10806 once upon termination, thus causing spikes in the average transfer speed
10807 instead of having a smooth one. This may partially be smoothed with
10808 "option contstats" though this is not perfect yet. Use of byte_in_cnt is
10809 recommended for better fairness.
10810
10811 - bytes_out_cnt : server to client byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +010010812 integer which counts the cumulative number of bytes sent to clients which
Willy Tarreauc9705a12010-07-27 20:05:50 +020010813 matched this entry. Headers are included in the count. This may be used
10814 to limit abuse of bots sucking the whole site.
10815
10816 - bytes_out_rate(<period>) : frequency counter (takes 12 bytes). It takes
10817 an integer parameter <period> which indicates in milliseconds the length
10818 of the period over which the average is measured. It reports the average
10819 outgoing bytes rate over that period, in bytes per period. It may be used
10820 to detect users which download too much and too fast. Warning: with large
10821 transfers, it is possible that the amount of transferred data will be
10822 counted once upon termination, thus causing spikes in the average
10823 transfer speed instead of having a smooth one. This may partially be
10824 smoothed with "option contstats" though this is not perfect yet. Use of
10825 byte_out_cnt is recommended for better fairness.
Willy Tarreau08d5f982010-06-06 13:34:54 +020010826
Willy Tarreauc00cdc22010-06-06 16:48:26 +020010827 There is only one stick-table per proxy. At the moment of writing this doc,
10828 it does not seem useful to have multiple tables per proxy. If this happens
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010829 to be required, simply create a dummy backend with a stick-table in it and
10830 reference it.
10831
10832 It is important to understand that stickiness based on learning information
10833 has some limitations, including the fact that all learned associations are
Baptiste Assmann123ff042016-03-06 23:29:28 +010010834 lost upon restart unless peers are properly configured to transfer such
10835 information upon restart (recommended). In general it can be good as a
10836 complement but not always as an exclusive stickiness.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010837
Willy Tarreauc9705a12010-07-27 20:05:50 +020010838 Last, memory requirements may be important when storing many data types.
10839 Indeed, storing all indicators above at once in each entry requires 116 bytes
10840 per entry, or 116 MB for a 1-million entries table. This is definitely not
10841 something that can be ignored.
10842
10843 Example:
10844 # Keep track of counters of up to 1 million IP addresses over 5 minutes
10845 # and store a general purpose counter and the average connection rate
10846 # computed over a sliding window of 30 seconds.
10847 stick-table type ip size 1m expire 5m store gpc0,conn_rate(30s)
10848
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +030010849 See also : "stick match", "stick on", "stick store-request", section 2.4
David du Colombiera13d1b92011-03-17 10:40:22 +010010850 about time format and section 7 about ACLs.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010851
10852
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010853stick store-response <pattern> [table <table>] [{if | unless} <condition>]
Baptiste Assmann2f2d2ec2016-03-06 23:27:24 +010010854 Define a response pattern used to create an entry in a stickiness table
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010855 May be used in sections : defaults | frontend | listen | backend
10856 no | no | yes | yes
10857
10858 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +020010859 <pattern> is a sample expression rule as described in section 7.3. It
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010860 describes what elements of the response or connection will
Davor Ocelice9ed2812017-12-25 17:49:28 +010010861 be analyzed, extracted and stored in the table once a
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010862 server is selected.
10863
10864 <table> is an optional stickiness table name. If unspecified, the same
10865 backend's table is used. A stickiness table is declared using
10866 the "stick-table" statement.
10867
10868 <cond> is an optional storage condition. It makes it possible to store
10869 certain criteria only when some conditions are met (or not met).
10870 For instance, it could be used to store the SSL session ID only
10871 when the response is a SSL server hello.
10872
10873 Some protocols or applications require complex stickiness rules and cannot
10874 always simply rely on cookies nor hashing. The "stick store-response"
10875 statement describes a rule to decide what to extract from the response and
10876 when to do it, in order to store it into a stickiness table for further
10877 requests to match it using the "stick match" statement. Obviously the
10878 extracted part must make sense and have a chance to be matched in a further
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010879 request. Storing an ID found in a header of a response makes sense.
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010880 See section 7 for a complete list of possible patterns and transformation
10881 rules.
10882
10883 The table has to be declared using the "stick-table" statement. It must be of
10884 a type compatible with the pattern. By default it is the one which is present
10885 in the same backend. It is possible to share a table with other backends by
10886 referencing it using the "table" keyword. If another table is referenced,
10887 the server's ID inside the backends are used. By default, all server IDs
10888 start at 1 in each backend, so the server ordering is enough. But in case of
10889 doubt, it is highly recommended to force server IDs using their "id" setting.
10890
10891 It is possible to restrict the conditions where a "stick store-response"
10892 statement will apply, using "if" or "unless" followed by a condition. This
10893 condition will be evaluated while parsing the response, so any criteria can
10894 be used. See section 7 for ACL based conditions.
10895
10896 There is no limit on the number of "stick store-response" statements, but
10897 there is a limit of 8 simultaneous stores per request or response. This
10898 makes it possible to store up to 8 criteria, all extracted from either the
10899 request or the response, regardless of the number of rules. Only the 8 first
10900 ones which match will be kept. Using this, it is possible to feed multiple
10901 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +010010902 another protocol or access method. Using multiple store-response rules with
10903 the same table is possible and may be used to find the best criterion to rely
10904 on, by arranging the rules by decreasing preference order. Only the first
10905 extracted criterion for a given table will be stored. All subsequent store-
10906 response rules referencing the same table will be skipped and their ACLs will
10907 not be evaluated. However, even if a store-request rule references a table, a
10908 store-response rule may also use the same table. This means that each table
10909 may learn exactly one element from the request and one element from the
10910 response at once.
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010911
10912 The table will contain the real server that processed the request.
10913
10914 Example :
10915 # Learn SSL session ID from both request and response and create affinity.
10916 backend https
10917 mode tcp
10918 balance roundrobin
Cyril Bontédc4d9032012-04-08 21:57:39 +020010919 # maximum SSL session ID length is 32 bytes.
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010920 stick-table type binary len 32 size 30k expire 30m
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010921
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010922 acl clienthello req_ssl_hello_type 1
10923 acl serverhello rep_ssl_hello_type 2
10924
10925 # use tcp content accepts to detects ssl client and server hello.
10926 tcp-request inspect-delay 5s
10927 tcp-request content accept if clienthello
10928
10929 # no timeout on response inspect delay by default.
10930 tcp-response content accept if serverhello
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010931
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010932 # SSL session ID (SSLID) may be present on a client or server hello.
10933 # Its length is coded on 1 byte at offset 43 and its value starts
10934 # at offset 44.
10935
10936 # Match and learn on request if client hello.
10937 stick on payload_lv(43,1) if clienthello
10938
10939 # Learn on response if server hello.
10940 stick store-response payload_lv(43,1) if serverhello
Cyril Bontédc4d9032012-04-08 21:57:39 +020010941
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010942 server s1 192.168.1.1:443
10943 server s2 192.168.1.1:443
10944
10945 See also : "stick-table", "stick on", and section 7 about ACLs and pattern
10946 extraction.
10947
10948
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020010949tcp-check comment <string>
10950 Defines a comment for the following the tcp-check rule, reported in logs if
10951 it fails.
10952 May be used in sections : defaults | frontend | listen | backend
10953 yes | no | yes | yes
10954
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020010955 Arguments :
10956 <string> is the comment message to add in logs if the following tcp-check
10957 rule fails.
10958
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020010959 It only works for connect, send and expect rules. It is useful to make
10960 user-friendly error reporting.
10961
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020010962 See also : "option tcp-check", "tcp-check connect", "tcp-check send" and
10963 "tcp-check expect".
10964
10965
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020010966tcp-check connect [default] [port <expr>] [addr <ip>] [send-proxy] [via-socks4]
10967 [ssl] [sni <sni>] [alpn <alpn>] [linger]
Christopher Fauletedc6ed92020-04-23 16:27:59 +020010968 [proto <name>] [comment <msg>]
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010969 Opens a new connection
10970 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020010971 yes | no | yes | yes
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010972
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020010973 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020010974 comment <msg> defines a message to report if the rule evaluation fails.
10975
Christopher Faulet4dce5922020-03-30 13:54:42 +020010976 default Use default options of the server line to do the health
Daniel Corbett67a82712020-07-06 23:01:19 -040010977 checks. The server options are used only if not redefined.
Christopher Faulet4dce5922020-03-30 13:54:42 +020010978
Christopher Fauletb7d30092020-03-30 15:19:03 +020010979 port <expr> if not set, check port or server port is used.
Christopher Faulet5c288742020-03-31 08:15:58 +020010980 It tells HAProxy where to open the connection to.
10981 <port> must be a valid TCP port source integer, from 1 to
Christopher Fauletb7d30092020-03-30 15:19:03 +020010982 65535 or an sample-fetch expression.
Christopher Faulet5c288742020-03-31 08:15:58 +020010983
10984 addr <ip> defines the IP address to do the health check.
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010985
10986 send-proxy send a PROXY protocol string
10987
Christopher Faulet085426a2020-03-30 13:07:02 +020010988 via-socks4 enables outgoing health checks using upstream socks4 proxy.
10989
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010990 ssl opens a ciphered connection
10991
Christopher Faulet79b31d42020-03-30 13:00:05 +020010992 sni <sni> specifies the SNI to use to do health checks over SSL.
10993
Christopher Faulet98572322020-03-30 13:16:44 +020010994 alpn <alpn> defines which protocols to advertise with ALPN. The protocol
10995 list consists in a comma-delimited list of protocol names,
10996 for instance: "http/1.1,http/1.0" (without quotes).
10997 If it is not set, the server ALPN is used.
10998
Christopher Fauletedc6ed92020-04-23 16:27:59 +020010999 proto <name> forces the multiplexer's protocol to use for this connection.
11000 It must be a TCP mux protocol and it must be usable on the
11001 backend side. The list of available protocols is reported in
11002 haproxy -vv.
11003
Christopher Faulet5c288742020-03-31 08:15:58 +020011004 linger cleanly close the connection instead of using a single RST.
Gaetan Rivetf8ba6772020-02-07 15:37:17 +010011005
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011006 When an application lies on more than a single TCP port or when HAProxy
11007 load-balance many services in a single backend, it makes sense to probe all
11008 the services individually before considering a server as operational.
11009
11010 When there are no TCP port configured on the server line neither server port
11011 directive, then the 'tcp-check connect port <port>' must be the first step
11012 of the sequence.
11013
11014 In a tcp-check ruleset a 'connect' is required, it is also mandatory to start
11015 the ruleset with a 'connect' rule. Purpose is to ensure admin know what they
11016 do.
11017
11018 When a connect must start the ruleset, if may still be preceded by set-var,
11019 unset-var or comment rules.
11020
11021 Examples :
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011022 # check HTTP and HTTPs services on a server.
11023 # first open port 80 thanks to server line port directive, then
11024 # tcp-check opens port 443, ciphered and run a request on it:
11025 option tcp-check
11026 tcp-check connect
11027 tcp-check send GET\ /\ HTTP/1.0\r\n
11028 tcp-check send Host:\ haproxy.1wt.eu\r\n
11029 tcp-check send \r\n
11030 tcp-check expect rstring (2..|3..)
11031 tcp-check connect port 443 ssl
11032 tcp-check send GET\ /\ HTTP/1.0\r\n
11033 tcp-check send Host:\ haproxy.1wt.eu\r\n
11034 tcp-check send \r\n
11035 tcp-check expect rstring (2..|3..)
11036 server www 10.0.0.1 check port 80
11037
11038 # check both POP and IMAP from a single server:
11039 option tcp-check
Gaetan Rivetf8ba6772020-02-07 15:37:17 +010011040 tcp-check connect port 110 linger
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011041 tcp-check expect string +OK\ POP3\ ready
11042 tcp-check connect port 143
11043 tcp-check expect string *\ OK\ IMAP4\ ready
11044 server mail 10.0.0.1 check
11045
11046 See also : "option tcp-check", "tcp-check send", "tcp-check expect"
11047
11048
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011049tcp-check expect [min-recv <int>] [comment <msg>]
Christopher Fauletec07e382020-04-07 14:56:26 +020011050 [ok-status <st>] [error-status <st>] [tout-status <st>]
Christopher Faulet98cc57c2020-04-01 20:52:31 +020011051 [on-success <fmt>] [on-error <fmt>] [status-code <expr>]
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020011052 [!] <match> <pattern>
Davor Ocelice9ed2812017-12-25 17:49:28 +010011053 Specify data to be collected and analyzed during a generic health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011054 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020011055 yes | no | yes | yes
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011056
11057 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011058 comment <msg> defines a message to report if the rule evaluation fails.
11059
Gaetan Rivet1afd8262020-02-07 15:37:17 +010011060 min-recv is optional and can define the minimum amount of data required to
11061 evaluate the current expect rule. If the number of received bytes
11062 is under this limit, the check will wait for more data. This
11063 option can be used to resolve some ambiguous matching rules or to
11064 avoid executing costly regex matches on content known to be still
11065 incomplete. If an exact string (string or binary) is used, the
11066 minimum between the string length and this parameter is used.
11067 This parameter is ignored if it is set to -1. If the expect rule
11068 does not match, the check will wait for more data. If set to 0,
11069 the evaluation result is always conclusive.
11070
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011071 <match> is a keyword indicating how to look for a specific pattern in the
Gaetan Rivetefab6c62020-02-07 15:37:17 +010011072 response. The keyword may be one of "string", "rstring", "binary" or
11073 "rbinary".
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011074 The keyword may be preceded by an exclamation mark ("!") to negate
11075 the match. Spaces are allowed between the exclamation mark and the
11076 keyword. See below for more details on the supported keywords.
11077
Christopher Fauletec07e382020-04-07 14:56:26 +020011078 ok-status <st> is optional and can be used to set the check status if
11079 the expect rule is successfully evaluated and if it is
11080 the last rule in the tcp-check ruleset. "L7OK", "L7OKC",
Christopher Fauletd888f0f2020-05-07 07:40:17 +020011081 "L6OK" and "L4OK" are supported :
11082 - L7OK : check passed on layer 7
11083 - L7OKC : check conditionally passed on layer 7, for
11084 example 404 with disable-on-404
11085 - L6OK : check passed on layer 6
11086 - L4OK : check passed on layer 4
Christopher Fauletec07e382020-04-07 14:56:26 +020011087 By default "L7OK" is used.
11088
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020011089 error-status <st> is optional and can be used to set the check status if
11090 an error occurred during the expect rule evaluation.
Christopher Fauletd888f0f2020-05-07 07:40:17 +020011091 "L7RSP", "L7STS", "L6RSP" and "L4CON" are supported :
11092 - L7RSP : layer 7 invalid response - protocol error
11093 - L7STS : layer 7 response error, for example HTTP 5xx
11094 - L6RSP : layer 6 invalid response - protocol error
11095 - L4CON : layer 1-4 connection problem
11096 By default "L7RSP" is used.
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020011097
Christopher Fauletec07e382020-04-07 14:56:26 +020011098 tout-status <st> is optional and can be used to set the check status if
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020011099 a timeout occurred during the expect rule evaluation.
Christopher Fauletd888f0f2020-05-07 07:40:17 +020011100 "L7TOUT", "L6TOUT", and "L4TOUT" are supported :
11101 - L7TOUT : layer 7 (HTTP/SMTP) timeout
11102 - L6TOUT : layer 6 (SSL) timeout
11103 - L4TOUT : layer 1-4 timeout
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020011104 By default "L7TOUT" is used.
11105
Christopher Fauletbe52b4d2020-04-01 16:30:22 +020011106 on-success <fmt> is optional and can be used to customize the
11107 informational message reported in logs if the expect
11108 rule is successfully evaluated and if it is the last rule
11109 in the tcp-check ruleset. <fmt> is a log-format string.
11110
11111 on-error <fmt> is optional and can be used to customize the
11112 informational message reported in logs if an error
11113 occurred during the expect rule evaluation. <fmt> is a
11114 log-format string.
11115
Christopher Faulet98cc57c2020-04-01 20:52:31 +020011116 status-code <expr> is optional and can be used to set the check status code
11117 reported in logs, on success or on error. <expr> is a
11118 standard HAProxy expression formed by a sample-fetch
11119 followed by some converters.
11120
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011121 <pattern> is the pattern to look for. It may be a string or a regular
11122 expression. If the pattern contains spaces, they must be escaped
11123 with the usual backslash ('\').
11124 If the match is set to binary, then the pattern must be passed as
Davor Ocelice9ed2812017-12-25 17:49:28 +010011125 a series of hexadecimal digits in an even number. Each sequence of
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011126 two digits will represent a byte. The hexadecimal digits may be
11127 used upper or lower case.
11128
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011129 The available matches are intentionally similar to their http-check cousins :
11130
11131 string <string> : test the exact string matches in the response buffer.
11132 A health check response will be considered valid if the
11133 response's buffer contains this exact string. If the
11134 "string" keyword is prefixed with "!", then the response
11135 will be considered invalid if the body contains this
11136 string. This can be used to look for a mandatory pattern
11137 in a protocol response, or to detect a failure when a
11138 specific error appears in a protocol banner.
11139
11140 rstring <regex> : test a regular expression on the response buffer.
11141 A health check response will be considered valid if the
11142 response's buffer matches this expression. If the
11143 "rstring" keyword is prefixed with "!", then the response
11144 will be considered invalid if the body matches the
11145 expression.
11146
Christopher Fauletaaab0832020-05-05 15:54:22 +020011147 string-lf <fmt> : test a log-format string match in the response's buffer.
11148 A health check response will be considered valid if the
11149 response's buffer contains the string resulting of the
11150 evaluation of <fmt>, which follows the log-format rules.
11151 If prefixed with "!", then the response will be
11152 considered invalid if the buffer contains the string.
11153
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011154 binary <hexstring> : test the exact string in its hexadecimal form matches
11155 in the response buffer. A health check response will
11156 be considered valid if the response's buffer contains
11157 this exact hexadecimal string.
11158 Purpose is to match data on binary protocols.
11159
Gaetan Rivetefab6c62020-02-07 15:37:17 +010011160 rbinary <regex> : test a regular expression on the response buffer, like
11161 "rstring". However, the response buffer is transformed
11162 into its hexadecimal form, including NUL-bytes. This
11163 allows using all regex engines to match any binary
11164 content. The hexadecimal transformation takes twice the
11165 size of the original response. As such, the expected
11166 pattern should work on at-most half the response buffer
11167 size.
11168
Christopher Fauletaaab0832020-05-05 15:54:22 +020011169 binary-lf <hexfmt> : test a log-format string in its hexadecimal form
11170 match in the response's buffer. A health check response
11171 will be considered valid if the response's buffer
11172 contains the hexadecimal string resulting of the
11173 evaluation of <fmt>, which follows the log-format
11174 rules. If prefixed with "!", then the response will be
11175 considered invalid if the buffer contains the
11176 hexadecimal string. The hexadecimal string is converted
11177 in a binary string before matching the response's
11178 buffer.
11179
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011180 It is important to note that the responses will be limited to a certain size
Christopher Faulet7151a122020-11-25 17:20:57 +010011181 defined by the global "tune.bufsize" option, which defaults to 16384 bytes.
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011182 Thus, too large responses may not contain the mandatory pattern when using
11183 "string", "rstring" or binary. If a large response is absolutely required, it
11184 is possible to change the default max size by setting the global variable.
11185 However, it is worth keeping in mind that parsing very large responses can
11186 waste some CPU cycles, especially when regular expressions are used, and that
11187 it is always better to focus the checks on smaller resources. Also, in its
11188 current state, the check will not find any string nor regex past a null
11189 character in the response. Similarly it is not possible to request matching
11190 the null character.
11191
11192 Examples :
11193 # perform a POP check
11194 option tcp-check
11195 tcp-check expect string +OK\ POP3\ ready
11196
11197 # perform an IMAP check
11198 option tcp-check
11199 tcp-check expect string *\ OK\ IMAP4\ ready
11200
11201 # look for the redis master server
11202 option tcp-check
11203 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +020011204 tcp-check expect string +PONG
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011205 tcp-check send info\ replication\r\n
11206 tcp-check expect string role:master
11207 tcp-check send QUIT\r\n
11208 tcp-check expect string +OK
11209
11210
11211 See also : "option tcp-check", "tcp-check connect", "tcp-check send",
Christopher Faulet7151a122020-11-25 17:20:57 +010011212 "tcp-check send-binary", "http-check expect", tune.bufsize
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011213
11214
Christopher Fauletb50b3e62020-05-05 18:43:43 +020011215tcp-check send <data> [comment <msg>]
11216tcp-check send-lf <fmt> [comment <msg>]
11217 Specify a string or a log-format string to be sent as a question during a
11218 generic health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011219 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020011220 yes | no | yes | yes
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011221
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011222 Arguments :
11223 comment <msg> defines a message to report if the rule evaluation fails.
11224
Christopher Fauletb50b3e62020-05-05 18:43:43 +020011225 <data> is the string that will be sent during a generic health
11226 check session.
Christopher Faulet16fff672020-04-30 07:50:54 +020011227
Christopher Fauletb50b3e62020-05-05 18:43:43 +020011228 <fmt> is the log-format string that will be sent, once evaluated,
11229 during a generic health check session.
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011230
11231 Examples :
11232 # look for the redis master server
11233 option tcp-check
11234 tcp-check send info\ replication\r\n
11235 tcp-check expect string role:master
11236
11237 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
Christopher Faulet7151a122020-11-25 17:20:57 +010011238 "tcp-check send-binary", tune.bufsize
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011239
11240
Christopher Fauletb50b3e62020-05-05 18:43:43 +020011241tcp-check send-binary <hexstring> [comment <msg>]
11242tcp-check send-binary-lf <hexfmt> [comment <msg>]
11243 Specify an hex digits string or an hex digits log-format string to be sent as
11244 a binary question during a raw tcp health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011245 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020011246 yes | no | yes | yes
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011247
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011248 Arguments :
11249 comment <msg> defines a message to report if the rule evaluation fails.
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011250
Christopher Fauletb50b3e62020-05-05 18:43:43 +020011251 <hexstring> is the hexadecimal string that will be send, once converted
11252 to binary, during a generic health check session.
Christopher Faulet16fff672020-04-30 07:50:54 +020011253
Christopher Fauletb50b3e62020-05-05 18:43:43 +020011254 <hexfmt> is the hexadecimal log-format string that will be send, once
11255 evaluated and converted to binary, during a generic health
11256 check session.
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011257
11258 Examples :
11259 # redis check in binary
11260 option tcp-check
11261 tcp-check send-binary 50494e470d0a # PING\r\n
11262 tcp-check expect binary 2b504F4e47 # +PONG
11263
11264
11265 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
Christopher Faulet7151a122020-11-25 17:20:57 +010011266 "tcp-check send", tune.bufsize
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011267
11268
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011269tcp-check set-var(<var-name>) <expr>
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011270 This operation sets the content of a variable. The variable is declared inline.
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011271 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020011272 yes | no | yes | yes
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011273
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011274 Arguments :
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011275 <var-name> The name of the variable starts with an indication about its
11276 scope. The scopes allowed for tcp-check are:
11277 "proc" : the variable is shared with the whole process.
11278 "sess" : the variable is shared with the tcp-check session.
11279 "check": the variable is declared for the lifetime of the tcp-check.
11280 This prefix is followed by a name. The separator is a '.'.
11281 The name may only contain characters 'a-z', 'A-Z', '0-9', '.',
11282 and '-'.
11283
11284 <expr> Is a sample-fetch expression potentially followed by converters.
11285
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011286 Examples :
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011287 tcp-check set-var(check.port) int(1234)
11288
11289
11290tcp-check unset-var(<var-name>)
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011291 Free a reference to a variable within its scope.
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011292 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020011293 yes | no | yes | yes
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011294
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011295 Arguments :
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011296 <var-name> The name of the variable starts with an indication about its
11297 scope. The scopes allowed for tcp-check are:
11298 "proc" : the variable is shared with the whole process.
11299 "sess" : the variable is shared with the tcp-check session.
11300 "check": the variable is declared for the lifetime of the tcp-check.
11301 This prefix is followed by a name. The separator is a '.'.
11302 The name may only contain characters 'a-z', 'A-Z', '0-9', '.',
11303 and '-'.
11304
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011305 Examples :
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011306 tcp-check unset-var(check.port)
11307
11308
Willy Tarreaue9656522010-08-17 15:40:09 +020011309tcp-request connection <action> [{if | unless} <condition>]
11310 Perform an action on an incoming connection depending on a layer 4 condition
Willy Tarreau1a687942010-05-23 22:40:30 +020011311 May be used in sections : defaults | frontend | listen | backend
11312 no | yes | yes | no
Willy Tarreaue9656522010-08-17 15:40:09 +020011313 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +020011314 <action> defines the action to perform if the condition applies. See
11315 below.
Willy Tarreau1a687942010-05-23 22:40:30 +020011316
Willy Tarreaue9656522010-08-17 15:40:09 +020011317 <condition> is a standard layer4-only ACL-based condition (see section 7).
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011318
11319 Immediately after acceptance of a new incoming connection, it is possible to
11320 evaluate some conditions to decide whether this connection must be accepted
Willy Tarreaue9656522010-08-17 15:40:09 +020011321 or dropped or have its counters tracked. Those conditions cannot make use of
11322 any data contents because the connection has not been read from yet, and the
11323 buffers are not yet allocated. This is used to selectively and very quickly
11324 accept or drop connections from various sources with a very low overhead. If
11325 some contents need to be inspected in order to take the decision, the
11326 "tcp-request content" statements must be used instead.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011327
Willy Tarreaue9656522010-08-17 15:40:09 +020011328 The "tcp-request connection" rules are evaluated in their exact declaration
11329 order. If no rule matches or if there is no rule, the default action is to
11330 accept the incoming connection. There is no specific limit to the number of
11331 rules which may be inserted.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011332
Willy Tarreaua9083d02015-05-08 15:27:59 +020011333 Four types of actions are supported :
Willy Tarreaue9656522010-08-17 15:40:09 +020011334 - accept :
11335 accepts the connection if the condition is true (when used with "if")
11336 or false (when used with "unless"). The first such rule executed ends
11337 the rules evaluation.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011338
Willy Tarreaue9656522010-08-17 15:40:09 +020011339 - reject :
11340 rejects the connection if the condition is true (when used with "if")
11341 or false (when used with "unless"). The first such rule executed ends
11342 the rules evaluation. Rejected connections do not even become a
11343 session, which is why they are accounted separately for in the stats,
11344 as "denied connections". They are not considered for the session
11345 rate-limit and are not logged either. The reason is that these rules
11346 should only be used to filter extremely high connection rates such as
11347 the ones encountered during a massive DDoS attack. Under these extreme
11348 conditions, the simple action of logging each event would make the
11349 system collapse and would considerably lower the filtering capacity. If
11350 logging is absolutely desired, then "tcp-request content" rules should
Willy Tarreau4f614292016-10-21 17:49:36 +020011351 be used instead, as "tcp-request session" rules will not log either.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011352
Willy Tarreau4f0d9192013-06-11 20:40:55 +020011353 - expect-proxy layer4 :
11354 configures the client-facing connection to receive a PROXY protocol
11355 header before any byte is read from the socket. This is equivalent to
11356 having the "accept-proxy" keyword on the "bind" line, except that using
11357 the TCP rule allows the PROXY protocol to be accepted only for certain
11358 IP address ranges using an ACL. This is convenient when multiple layers
11359 of load balancers are passed through by traffic coming from public
11360 hosts.
11361
Bertrand Jacquin90759682016-06-06 15:35:39 +010011362 - expect-netscaler-cip layer4 :
11363 configures the client-facing connection to receive a NetScaler Client
11364 IP insertion protocol header before any byte is read from the socket.
11365 This is equivalent to having the "accept-netscaler-cip" keyword on the
11366 "bind" line, except that using the TCP rule allows the PROXY protocol
11367 to be accepted only for certain IP address ranges using an ACL. This
11368 is convenient when multiple layers of load balancers are passed
11369 through by traffic coming from public hosts.
11370
Willy Tarreau18bf01e2014-06-13 16:18:52 +020011371 - capture <sample> len <length> :
11372 This only applies to "tcp-request content" rules. It captures sample
11373 expression <sample> from the request buffer, and converts it to a
11374 string of at most <len> characters. The resulting string is stored into
11375 the next request "capture" slot, so it will possibly appear next to
11376 some captured HTTP headers. It will then automatically appear in the
11377 logs, and it will be possible to extract it using sample fetch rules to
11378 feed it into headers or anything. The length should be limited given
11379 that this size will be allocated for each capture during the whole
Willy Tarreaua9083d02015-05-08 15:27:59 +020011380 session life. Please check section 7.3 (Fetching samples) and "capture
11381 request header" for more information.
Willy Tarreau18bf01e2014-06-13 16:18:52 +020011382
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011383 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] :
Willy Tarreaue9656522010-08-17 15:40:09 +020011384 enables tracking of sticky counters from current connection. These
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020011385 rules do not stop evaluation and do not change default action. The
11386 number of counters that may be simultaneously tracked by the same
11387 connection is set in MAX_SESS_STKCTR at build time (reported in
John Roeslerfb2fce12019-07-10 15:45:51 -050011388 haproxy -vv) which defaults to 3, so the track-sc number is between 0
Matteo Contrini1857b8c2020-10-16 17:35:54 +020011389 and (MAX_SESS_STKCTR-1). The first "track-sc0" rule executed enables
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020011390 tracking of the counters of the specified table as the first set. The
11391 first "track-sc1" rule executed enables tracking of the counters of the
11392 specified table as the second set. The first "track-sc2" rule executed
11393 enables tracking of the counters of the specified table as the third
11394 set. It is a recommended practice to use the first set of counters for
11395 the per-frontend counters and the second set for the per-backend ones.
11396 But this is just a guideline, all may be used everywhere.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011397
Willy Tarreaue9656522010-08-17 15:40:09 +020011398 These actions take one or two arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +020011399 <key> is mandatory, and is a sample expression rule as described
Willy Tarreau74ca5042013-06-11 23:12:07 +020011400 in section 7.3. It describes what elements of the incoming
Davor Ocelice9ed2812017-12-25 17:49:28 +010011401 request or connection will be analyzed, extracted, combined,
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011402 and used to select which table entry to update the counters.
11403 Note that "tcp-request connection" cannot use content-based
11404 fetches.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011405
Willy Tarreaue9656522010-08-17 15:40:09 +020011406 <table> is an optional table to be used instead of the default one,
11407 which is the stick-table declared in the current proxy. All
11408 the counters for the matches and updates for the key will
11409 then be performed in that table until the session ends.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011410
Willy Tarreaue9656522010-08-17 15:40:09 +020011411 Once a "track-sc*" rule is executed, the key is looked up in the table
11412 and if it is not found, an entry is allocated for it. Then a pointer to
11413 that entry is kept during all the session's life, and this entry's
11414 counters are updated as often as possible, every time the session's
11415 counters are updated, and also systematically when the session ends.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011416 Counters are only updated for events that happen after the tracking has
11417 been started. For example, connection counters will not be updated when
11418 tracking layer 7 information, since the connection event happens before
11419 layer7 information is extracted.
11420
Willy Tarreaue9656522010-08-17 15:40:09 +020011421 If the entry tracks concurrent connection counters, one connection is
11422 counted for as long as the entry is tracked, and the entry will not
11423 expire during that time. Tracking counters also provides a performance
11424 advantage over just checking the keys, because only one table lookup is
11425 performed for all ACL checks that make use of it.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011426
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +020011427 - sc-inc-gpc0(<sc-id>):
11428 The "sc-inc-gpc0" increments the GPC0 counter according to the sticky
11429 counter designated by <sc-id>. If an error occurs, this action silently
11430 fails and the actions evaluation continues.
11431
Frédéric Lécaille6778b272018-01-29 15:22:53 +010011432 - sc-inc-gpc1(<sc-id>):
11433 The "sc-inc-gpc1" increments the GPC1 counter according to the sticky
11434 counter designated by <sc-id>. If an error occurs, this action silently
11435 fails and the actions evaluation continues.
11436
Cédric Dufour0d7712d2019-11-06 18:38:53 +010011437 - sc-set-gpt0(<sc-id>) { <int> | <expr> }:
11438 This action sets the 32-bit unsigned GPT0 tag according to the sticky
11439 counter designated by <sc-id> and the value of <int>/<expr>. The
11440 expected result is a boolean. If an error occurs, this action silently
11441 fails and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +020011442
William Lallemand2e785f22016-05-25 01:48:42 +020011443 - set-src <expr> :
11444 Is used to set the source IP address to the value of specified
11445 expression. Useful if you want to mask source IP for privacy.
11446 If you want to provide an IP from a HTTP header use "http-request
Cyril Bonté6c81d5f2018-10-17 00:14:51 +020011447 set-src".
William Lallemand2e785f22016-05-25 01:48:42 +020011448
Cyril Bonté6c81d5f2018-10-17 00:14:51 +020011449 Arguments:
11450 <expr> Is a standard HAProxy expression formed by a sample-fetch
11451 followed by some converters.
William Lallemand2e785f22016-05-25 01:48:42 +020011452
11453 Example:
William Lallemand2e785f22016-05-25 01:48:42 +020011454 tcp-request connection set-src src,ipmask(24)
11455
Willy Tarreau0c630532016-10-21 17:52:58 +020011456 When possible, set-src preserves the original source port as long as the
11457 address family allows it, otherwise the source port is set to 0.
William Lallemand2e785f22016-05-25 01:48:42 +020011458
William Lallemand44be6402016-05-25 01:51:35 +020011459 - set-src-port <expr> :
11460 Is used to set the source port address to the value of specified
11461 expression.
11462
Cyril Bonté6c81d5f2018-10-17 00:14:51 +020011463 Arguments:
11464 <expr> Is a standard HAProxy expression formed by a sample-fetch
11465 followed by some converters.
William Lallemand44be6402016-05-25 01:51:35 +020011466
11467 Example:
William Lallemand44be6402016-05-25 01:51:35 +020011468 tcp-request connection set-src-port int(4000)
11469
Willy Tarreau0c630532016-10-21 17:52:58 +020011470 When possible, set-src-port preserves the original source address as long
11471 as the address family supports a port, otherwise it forces the source
11472 address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +020011473
William Lallemand13e9b0c2016-05-25 02:34:07 +020011474 - set-dst <expr> :
11475 Is used to set the destination IP address to the value of specified
11476 expression. Useful if you want to mask IP for privacy in log.
11477 If you want to provide an IP from a HTTP header use "http-request
11478 set-dst". If you want to connect to the new address/port, use
11479 '0.0.0.0:0' as a server address in the backend.
11480
11481 <expr> Is a standard HAProxy expression formed by a sample-fetch
11482 followed by some converters.
11483
11484 Example:
11485
11486 tcp-request connection set-dst dst,ipmask(24)
11487 tcp-request connection set-dst ipv4(10.0.0.1)
11488
Willy Tarreau0c630532016-10-21 17:52:58 +020011489 When possible, set-dst preserves the original destination port as long as
11490 the address family allows it, otherwise the destination port is set to 0.
11491
William Lallemand13e9b0c2016-05-25 02:34:07 +020011492 - set-dst-port <expr> :
11493 Is used to set the destination port address to the value of specified
11494 expression. If you want to connect to the new address/port, use
11495 '0.0.0.0:0' as a server address in the backend.
11496
11497
11498 <expr> Is a standard HAProxy expression formed by a sample-fetch
11499 followed by some converters.
11500
11501 Example:
11502
11503 tcp-request connection set-dst-port int(4000)
11504
Willy Tarreau0c630532016-10-21 17:52:58 +020011505 When possible, set-dst-port preserves the original destination address as
11506 long as the address family supports a port, otherwise it forces the
11507 destination address to IPv4 "0.0.0.0" before rewriting the port.
11508
Willy Tarreau2d392c22015-08-24 01:43:45 +020011509 - "silent-drop" :
11510 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +010011511 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +020011512 to prevent the client from being notified. The effect it then that the
11513 client still sees an established connection while there's none on
11514 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
11515 except that it doesn't use any local resource at all on the machine
11516 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +010011517 slow down stronger attackers. It is important to understand the impact
11518 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +020011519 client and HAProxy (firewalls, proxies, load balancers) will also keep
11520 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +010011521 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +020011522 TCP_REPAIR socket option is used to block the emission of a TCP
11523 reset. On other systems, the socket's TTL is reduced to 1 so that the
11524 TCP reset doesn't pass the first router, though it's still delivered to
11525 local networks. Do not use it unless you fully understand how it works.
11526
Willy Tarreaue9656522010-08-17 15:40:09 +020011527 Note that the "if/unless" condition is optional. If no condition is set on
11528 the action, it is simply performed unconditionally. That can be useful for
11529 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011530
Willy Tarreaue9656522010-08-17 15:40:09 +020011531 Example: accept all connections from white-listed hosts, reject too fast
11532 connection without counting them, and track accepted connections.
11533 This results in connection rate being capped from abusive sources.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011534
Willy Tarreaue9656522010-08-17 15:40:09 +020011535 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011536 tcp-request connection reject if { src_conn_rate gt 10 }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011537 tcp-request connection track-sc0 src
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011538
Willy Tarreaue9656522010-08-17 15:40:09 +020011539 Example: accept all connections from white-listed hosts, count all other
11540 connections and reject too fast ones. This results in abusive ones
11541 being blocked as long as they don't slow down.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011542
Willy Tarreaue9656522010-08-17 15:40:09 +020011543 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011544 tcp-request connection track-sc0 src
11545 tcp-request connection reject if { sc0_conn_rate gt 10 }
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011546
Willy Tarreau4f0d9192013-06-11 20:40:55 +020011547 Example: enable the PROXY protocol for traffic coming from all known proxies.
11548
11549 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
11550
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011551 See section 7 about ACL usage.
11552
Willy Tarreau4f614292016-10-21 17:49:36 +020011553 See also : "tcp-request session", "tcp-request content", "stick-table"
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011554
11555
Willy Tarreaue9656522010-08-17 15:40:09 +020011556tcp-request content <action> [{if | unless} <condition>]
11557 Perform an action on a new session depending on a layer 4-7 condition
Willy Tarreau62644772008-07-16 18:36:06 +020011558 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +020011559 no | yes | yes | yes
Willy Tarreaue9656522010-08-17 15:40:09 +020011560 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +020011561 <action> defines the action to perform if the condition applies. See
11562 below.
Willy Tarreau62644772008-07-16 18:36:06 +020011563
Willy Tarreaue9656522010-08-17 15:40:09 +020011564 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
Willy Tarreau62644772008-07-16 18:36:06 +020011565
Davor Ocelice9ed2812017-12-25 17:49:28 +010011566 A request's contents can be analyzed at an early stage of request processing
Willy Tarreaue9656522010-08-17 15:40:09 +020011567 called "TCP content inspection". During this stage, ACL-based rules are
11568 evaluated every time the request contents are updated, until either an
11569 "accept" or a "reject" rule matches, or the TCP request inspection delay
11570 expires with no matching rule.
Willy Tarreau62644772008-07-16 18:36:06 +020011571
Willy Tarreaue9656522010-08-17 15:40:09 +020011572 The first difference between these rules and "tcp-request connection" rules
11573 is that "tcp-request content" rules can make use of contents to take a
11574 decision. Most often, these decisions will consider a protocol recognition or
11575 validity. The second difference is that content-based rules can be used in
Willy Tarreauf3338342014-01-28 21:40:28 +010011576 both frontends and backends. In case of HTTP keep-alive with the client, all
11577 tcp-request content rules are evaluated again, so haproxy keeps a record of
11578 what sticky counters were assigned by a "tcp-request connection" versus a
11579 "tcp-request content" rule, and flushes all the content-related ones after
11580 processing an HTTP request, so that they may be evaluated again by the rules
11581 being evaluated again for the next request. This is of particular importance
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011582 when the rule tracks some L7 information or when it is conditioned by an
Willy Tarreauf3338342014-01-28 21:40:28 +010011583 L7-based ACL, since tracking may change between requests.
Willy Tarreau62644772008-07-16 18:36:06 +020011584
Willy Tarreaue9656522010-08-17 15:40:09 +020011585 Content-based rules are evaluated in their exact declaration order. If no
11586 rule matches or if there is no rule, the default action is to accept the
11587 contents. There is no specific limit to the number of rules which may be
11588 inserted.
Willy Tarreau62644772008-07-16 18:36:06 +020011589
Thierry FOURNIER236657b2015-08-19 08:25:14 +020011590 Several types of actions are supported :
Willy Tarreau18bf01e2014-06-13 16:18:52 +020011591 - accept : the request is accepted
Baptiste Assmann333939c2019-01-21 08:34:50 +010011592 - do-resolve: perform a DNS resolution
Willy Tarreau18bf01e2014-06-13 16:18:52 +020011593 - reject : the request is rejected and the connection is closed
11594 - capture : the specified sample expression is captured
Patrick Hemmer268a7072018-05-11 12:52:31 -040011595 - set-priority-class <expr> | set-priority-offset <expr>
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011596 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +020011597 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +010011598 - sc-inc-gpc1(<sc-id>)
Cédric Dufour0d7712d2019-11-06 18:38:53 +010011599 - sc-set-gpt0(<sc-id>) { <int> | <expr> }
Baptiste Assmanne1afd4f2019-04-18 16:21:13 +020011600 - set-dst <expr>
11601 - set-dst-port <expr>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011602 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +010011603 - unset-var(<var-name>)
Willy Tarreau2d392c22015-08-24 01:43:45 +020011604 - silent-drop
Davor Ocelice9ed2812017-12-25 17:49:28 +010011605 - send-spoe-group <engine-name> <group-name>
Christopher Faulet579d83b2019-11-22 15:34:17 +010011606 - use-service <service-name>
Willy Tarreau62644772008-07-16 18:36:06 +020011607
Willy Tarreaue9656522010-08-17 15:40:09 +020011608 They have the same meaning as their counter-parts in "tcp-request connection"
11609 so please refer to that section for a complete description.
Baptiste Assmann333939c2019-01-21 08:34:50 +010011610 For "do-resolve" action, please check the "http-request do-resolve"
11611 configuration section.
Willy Tarreau62644772008-07-16 18:36:06 +020011612
Willy Tarreauf3338342014-01-28 21:40:28 +010011613 While there is nothing mandatory about it, it is recommended to use the
11614 track-sc0 in "tcp-request connection" rules, track-sc1 for "tcp-request
11615 content" rules in the frontend, and track-sc2 for "tcp-request content"
11616 rules in the backend, because that makes the configuration more readable
11617 and easier to troubleshoot, but this is just a guideline and all counters
11618 may be used everywhere.
Willy Tarreau62644772008-07-16 18:36:06 +020011619
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011620 Note that the "if/unless" condition is optional. If no condition is set on
Willy Tarreaue9656522010-08-17 15:40:09 +020011621 the action, it is simply performed unconditionally. That can be useful for
11622 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau62644772008-07-16 18:36:06 +020011623
Christopher Faulet2079a4a2020-10-02 11:48:57 +020011624 Note also that it is recommended to use a "tcp-request session" rule to track
11625 information that does *not* depend on Layer 7 contents, especially for HTTP
11626 frontends. Some HTTP processing are performed at the session level and may
11627 lead to an early rejection of the requests. Thus, the tracking at the content
11628 level may be disturbed in such case. A warning is emitted during startup to
11629 prevent, as far as possible, such unreliable usage.
11630
Willy Tarreaue9656522010-08-17 15:40:09 +020011631 It is perfectly possible to match layer 7 contents with "tcp-request content"
Christopher Faulet7ea509e2020-10-02 11:38:46 +020011632 rules from a TCP proxy, since HTTP-specific ACL matches are able to
11633 preliminarily parse the contents of a buffer before extracting the required
11634 data. If the buffered contents do not parse as a valid HTTP message, then the
11635 ACL does not match. The parser which is involved there is exactly the same
11636 as for all other HTTP processing, so there is no risk of parsing something
11637 differently. In an HTTP frontend or an HTTP backend, it is guaranteed that
11638 HTTP contents will always be immediately present when the rule is evaluated
11639 first because the HTTP parsing is performed in the early stages of the
11640 connection processing, at the session level. But for such proxies, using
11641 "http-request" rules is much more natural and recommended.
Willy Tarreau62644772008-07-16 18:36:06 +020011642
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011643 Tracking layer7 information is also possible provided that the information
Willy Tarreau4d54c7c2014-09-16 15:48:15 +020011644 are present when the rule is processed. The rule processing engine is able to
11645 wait until the inspect delay expires when the data to be tracked is not yet
11646 available.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011647
Baptiste Assmanne1afd4f2019-04-18 16:21:13 +020011648 The "set-dst" and "set-dst-port" are used to set respectively the destination
11649 IP and port. More information on how to use it at "http-request set-dst".
11650
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011651 The "set-var" is used to set the content of a variable. The variable is
Willy Tarreau4f614292016-10-21 17:49:36 +020011652 declared inline. For "tcp-request session" rules, only session-level
11653 variables can be used, without any layer7 contents.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011654
Daniel Schneller0b547052016-03-21 20:46:57 +010011655 <var-name> The name of the variable starts with an indication about
11656 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010011657 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010011658 "sess" : the variable is shared with the whole session
11659 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011660 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +010011661 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011662 processing
Daniel Schneller0b547052016-03-21 20:46:57 +010011663 "res" : the variable is shared only during response
11664 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011665 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +010011666 The name may only contain characters 'a-z', 'A-Z', '0-9',
11667 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011668
11669 <expr> Is a standard HAProxy expression formed by a sample-fetch
11670 followed by some converters.
11671
Christopher Faulet85d79c92016-11-09 16:54:56 +010011672 The "unset-var" is used to unset a variable. See above for details about
11673 <var-name>.
11674
Patrick Hemmer268a7072018-05-11 12:52:31 -040011675 The "set-priority-class" is used to set the queue priority class of the
11676 current request. The value must be a sample expression which converts to an
11677 integer in the range -2047..2047. Results outside this range will be
11678 truncated. The priority class determines the order in which queued requests
11679 are processed. Lower values have higher priority.
11680
11681 The "set-priority-offset" is used to set the queue priority timestamp offset
11682 of the current request. The value must be a sample expression which converts
11683 to an integer in the range -524287..524287. Results outside this range will be
11684 truncated. When a request is queued, it is ordered first by the priority
11685 class, then by the current timestamp adjusted by the given offset in
11686 milliseconds. Lower values have higher priority.
11687 Note that the resulting timestamp is is only tracked with enough precision for
11688 524,287ms (8m44s287ms). If the request is queued long enough to where the
11689 adjusted timestamp exceeds this value, it will be misidentified as highest
11690 priority. Thus it is important to set "timeout queue" to a value, where when
11691 combined with the offset, does not exceed this limit.
11692
Christopher Faulet76c09ef2017-09-21 11:03:52 +020011693 The "send-spoe-group" is used to trigger sending of a group of SPOE
11694 messages. To do so, the SPOE engine used to send messages must be defined, as
11695 well as the SPOE group to send. Of course, the SPOE engine must refer to an
11696 existing SPOE filter. If not engine name is provided on the SPOE filter line,
11697 the SPOE agent name must be used.
11698
11699 <engine-name> The SPOE engine name.
11700
11701 <group-name> The SPOE group name as specified in the engine configuration.
11702
Christopher Faulet579d83b2019-11-22 15:34:17 +010011703 The "use-service" is used to executes a TCP service which will reply to the
11704 request and stop the evaluation of the rules. This service may choose to
11705 reply by sending any valid response or it may immediately close the
11706 connection without sending anything. Outside natives services, it is possible
11707 to write your own services in Lua. No further "tcp-request" rules are
11708 evaluated.
11709
11710 Example:
11711 tcp-request content use-service lua.deny { src -f /etc/haproxy/blacklist.lst }
11712
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011713 Example:
11714
11715 tcp-request content set-var(sess.my_var) src
Christopher Faulet85d79c92016-11-09 16:54:56 +010011716 tcp-request content unset-var(sess.my_var2)
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011717
Willy Tarreau62644772008-07-16 18:36:06 +020011718 Example:
Willy Tarreaue9656522010-08-17 15:40:09 +020011719 # Accept HTTP requests containing a Host header saying "example.com"
11720 # and reject everything else.
11721 acl is_host_com hdr(Host) -i example.com
11722 tcp-request inspect-delay 30s
Willy Tarreauc0239e02012-04-16 14:42:55 +020011723 tcp-request content accept if is_host_com
Willy Tarreaue9656522010-08-17 15:40:09 +020011724 tcp-request content reject
11725
11726 Example:
Willy Tarreau62644772008-07-16 18:36:06 +020011727 # reject SMTP connection if client speaks first
11728 tcp-request inspect-delay 30s
11729 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011730 tcp-request content reject if content_present
Willy Tarreau62644772008-07-16 18:36:06 +020011731
11732 # Forward HTTPS connection only if client speaks
11733 tcp-request inspect-delay 30s
11734 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011735 tcp-request content accept if content_present
Willy Tarreaue9656522010-08-17 15:40:09 +020011736 tcp-request content reject
11737
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011738 Example:
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030011739 # Track the last IP(stick-table type string) from X-Forwarded-For
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011740 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +020011741 tcp-request content track-sc0 hdr(x-forwarded-for,-1)
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030011742 # Or track the last IP(stick-table type ip|ipv6) from X-Forwarded-For
11743 tcp-request content track-sc0 req.hdr_ip(x-forwarded-for,-1)
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011744
11745 Example:
11746 # track request counts per "base" (concatenation of Host+URL)
11747 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +020011748 tcp-request content track-sc0 base table req-rate
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011749
Willy Tarreaue9656522010-08-17 15:40:09 +020011750 Example: track per-frontend and per-backend counters, block abusers at the
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030011751 frontend when the backend detects abuse(and marks gpc0).
Willy Tarreaue9656522010-08-17 15:40:09 +020011752
11753 frontend http
Davor Ocelice9ed2812017-12-25 17:49:28 +010011754 # Use General Purpose Counter 0 in SC0 as a global abuse counter
Willy Tarreaue9656522010-08-17 15:40:09 +020011755 # protecting all our sites
11756 stick-table type ip size 1m expire 5m store gpc0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011757 tcp-request connection track-sc0 src
11758 tcp-request connection reject if { sc0_get_gpc0 gt 0 }
Willy Tarreaue9656522010-08-17 15:40:09 +020011759 ...
11760 use_backend http_dynamic if { path_end .php }
11761
11762 backend http_dynamic
11763 # if a source makes too fast requests to this dynamic site (tracked
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011764 # by SC1), block it globally in the frontend.
Willy Tarreaue9656522010-08-17 15:40:09 +020011765 stick-table type ip size 1m expire 5m store http_req_rate(10s)
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011766 acl click_too_fast sc1_http_req_rate gt 10
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030011767 acl mark_as_abuser sc0_inc_gpc0(http) gt 0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011768 tcp-request content track-sc1 src
Willy Tarreaue9656522010-08-17 15:40:09 +020011769 tcp-request content reject if click_too_fast mark_as_abuser
Willy Tarreau62644772008-07-16 18:36:06 +020011770
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011771 See section 7 about ACL usage.
Willy Tarreau62644772008-07-16 18:36:06 +020011772
Jarno Huuskonen95b012b2017-04-06 13:59:14 +030011773 See also : "tcp-request connection", "tcp-request session",
11774 "tcp-request inspect-delay", and "http-request".
Willy Tarreau62644772008-07-16 18:36:06 +020011775
11776
11777tcp-request inspect-delay <timeout>
11778 Set the maximum allowed time to wait for data during content inspection
11779 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +020011780 no | yes | yes | yes
Willy Tarreau62644772008-07-16 18:36:06 +020011781 Arguments :
11782 <timeout> is the timeout value specified in milliseconds by default, but
11783 can be in any other unit if the number is suffixed by the unit,
11784 as explained at the top of this document.
11785
11786 People using haproxy primarily as a TCP relay are often worried about the
11787 risk of passing any type of protocol to a server without any analysis. In
11788 order to be able to analyze the request contents, we must first withhold
11789 the data then analyze them. This statement simply enables withholding of
11790 data for at most the specified amount of time.
11791
Willy Tarreaufb356202010-08-03 14:02:05 +020011792 TCP content inspection applies very early when a connection reaches a
11793 frontend, then very early when the connection is forwarded to a backend. This
11794 means that a connection may experience a first delay in the frontend and a
11795 second delay in the backend if both have tcp-request rules.
11796
Willy Tarreau62644772008-07-16 18:36:06 +020011797 Note that when performing content inspection, haproxy will evaluate the whole
11798 rules for every new chunk which gets in, taking into account the fact that
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011799 those data are partial. If no rule matches before the aforementioned delay,
Willy Tarreau62644772008-07-16 18:36:06 +020011800 a last check is performed upon expiration, this time considering that the
Willy Tarreaud869b242009-03-15 14:43:58 +010011801 contents are definitive. If no delay is set, haproxy will not wait at all
11802 and will immediately apply a verdict based on the available information.
11803 Obviously this is unlikely to be very useful and might even be racy, so such
11804 setups are not recommended.
Willy Tarreau62644772008-07-16 18:36:06 +020011805
11806 As soon as a rule matches, the request is released and continues as usual. If
11807 the timeout is reached and no rule matches, the default policy will be to let
11808 it pass through unaffected.
11809
11810 For most protocols, it is enough to set it to a few seconds, as most clients
11811 send the full request immediately upon connection. Add 3 or more seconds to
11812 cover TCP retransmits but that's all. For some protocols, it may make sense
Willy Tarreaud72758d2010-01-12 10:42:19 +010011813 to use large values, for instance to ensure that the client never talks
Davor Ocelice9ed2812017-12-25 17:49:28 +010011814 before the server (e.g. SMTP), or to wait for a client to talk before passing
11815 data to the server (e.g. SSL). Note that the client timeout must cover at
Willy Tarreaub824b002010-09-29 16:36:16 +020011816 least the inspection delay, otherwise it will expire first. If the client
11817 closes the connection or if the buffer is full, the delay immediately expires
11818 since the contents will not be able to change anymore.
Willy Tarreau62644772008-07-16 18:36:06 +020011819
Willy Tarreau55165fe2009-05-10 12:02:55 +020011820 See also : "tcp-request content accept", "tcp-request content reject",
Willy Tarreau62644772008-07-16 18:36:06 +020011821 "timeout client".
11822
11823
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011824tcp-response content <action> [{if | unless} <condition>]
11825 Perform an action on a session response depending on a layer 4-7 condition
11826 May be used in sections : defaults | frontend | listen | backend
11827 no | no | yes | yes
11828 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +020011829 <action> defines the action to perform if the condition applies. See
11830 below.
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011831
11832 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
11833
Davor Ocelice9ed2812017-12-25 17:49:28 +010011834 Response contents can be analyzed at an early stage of response processing
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011835 called "TCP content inspection". During this stage, ACL-based rules are
11836 evaluated every time the response contents are updated, until either an
Willy Tarreaucc1e04b2013-09-11 23:20:29 +020011837 "accept", "close" or a "reject" rule matches, or a TCP response inspection
11838 delay is set and expires with no matching rule.
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011839
11840 Most often, these decisions will consider a protocol recognition or validity.
11841
11842 Content-based rules are evaluated in their exact declaration order. If no
11843 rule matches or if there is no rule, the default action is to accept the
11844 contents. There is no specific limit to the number of rules which may be
11845 inserted.
11846
Thierry FOURNIER236657b2015-08-19 08:25:14 +020011847 Several types of actions are supported :
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011848 - accept :
11849 accepts the response if the condition is true (when used with "if")
11850 or false (when used with "unless"). The first such rule executed ends
11851 the rules evaluation.
11852
Willy Tarreaucc1e04b2013-09-11 23:20:29 +020011853 - close :
11854 immediately closes the connection with the server if the condition is
11855 true (when used with "if"), or false (when used with "unless"). The
11856 first such rule executed ends the rules evaluation. The main purpose of
11857 this action is to force a connection to be finished between a client
11858 and a server after an exchange when the application protocol expects
11859 some long time outs to elapse first. The goal is to eliminate idle
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011860 connections which take significant resources on servers with certain
Willy Tarreaucc1e04b2013-09-11 23:20:29 +020011861 protocols.
11862
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011863 - reject :
11864 rejects the response if the condition is true (when used with "if")
11865 or false (when used with "unless"). The first such rule executed ends
Jamie Gloudonaaa21002012-08-25 00:18:33 -040011866 the rules evaluation. Rejected session are immediately closed.
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011867
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011868 - set-var(<var-name>) <expr>
11869 Sets a variable.
11870
Christopher Faulet85d79c92016-11-09 16:54:56 +010011871 - unset-var(<var-name>)
11872 Unsets a variable.
11873
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +020011874 - sc-inc-gpc0(<sc-id>):
11875 This action increments the GPC0 counter according to the sticky
11876 counter designated by <sc-id>. If an error occurs, this action fails
11877 silently and the actions evaluation continues.
11878
Frédéric Lécaille6778b272018-01-29 15:22:53 +010011879 - sc-inc-gpc1(<sc-id>):
11880 This action increments the GPC1 counter according to the sticky
11881 counter designated by <sc-id>. If an error occurs, this action fails
11882 silently and the actions evaluation continues.
11883
Cédric Dufour0d7712d2019-11-06 18:38:53 +010011884 - sc-set-gpt0(<sc-id>) { <int> | <expr> }
11885 This action sets the 32-bit unsigned GPT0 tag according to the sticky
11886 counter designated by <sc-id> and the value of <int>/<expr>. The
11887 expected result is a boolean. If an error occurs, this action silently
11888 fails and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +020011889
Willy Tarreau2d392c22015-08-24 01:43:45 +020011890 - "silent-drop" :
11891 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +010011892 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +020011893 to prevent the client from being notified. The effect it then that the
11894 client still sees an established connection while there's none on
11895 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
11896 except that it doesn't use any local resource at all on the machine
11897 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +010011898 slow down stronger attackers. It is important to understand the impact
11899 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +020011900 client and HAProxy (firewalls, proxies, load balancers) will also keep
11901 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +010011902 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +020011903 TCP_REPAIR socket option is used to block the emission of a TCP
11904 reset. On other systems, the socket's TTL is reduced to 1 so that the
11905 TCP reset doesn't pass the first router, though it's still delivered to
11906 local networks. Do not use it unless you fully understand how it works.
11907
Christopher Faulet76c09ef2017-09-21 11:03:52 +020011908 - send-spoe-group <engine-name> <group-name>
11909 Send a group of SPOE messages.
11910
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011911 Note that the "if/unless" condition is optional. If no condition is set on
11912 the action, it is simply performed unconditionally. That can be useful for
11913 for changing the default action to a reject.
11914
Jamie Gloudonaaa21002012-08-25 00:18:33 -040011915 It is perfectly possible to match layer 7 contents with "tcp-response
11916 content" rules, but then it is important to ensure that a full response has
11917 been buffered, otherwise no contents will match. In order to achieve this,
11918 the best solution involves detecting the HTTP protocol during the inspection
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011919 period.
11920
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011921 The "set-var" is used to set the content of a variable. The variable is
11922 declared inline.
11923
Daniel Schneller0b547052016-03-21 20:46:57 +010011924 <var-name> The name of the variable starts with an indication about
11925 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010011926 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010011927 "sess" : the variable is shared with the whole session
11928 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011929 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +010011930 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011931 processing
Daniel Schneller0b547052016-03-21 20:46:57 +010011932 "res" : the variable is shared only during response
11933 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011934 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +010011935 The name may only contain characters 'a-z', 'A-Z', '0-9',
11936 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011937
11938 <expr> Is a standard HAProxy expression formed by a sample-fetch
11939 followed by some converters.
11940
11941 Example:
11942
11943 tcp-request content set-var(sess.my_var) src
11944
Christopher Faulet85d79c92016-11-09 16:54:56 +010011945 The "unset-var" is used to unset a variable. See above for details about
11946 <var-name>.
11947
11948 Example:
11949
11950 tcp-request content unset-var(sess.my_var)
11951
Christopher Faulet76c09ef2017-09-21 11:03:52 +020011952 The "send-spoe-group" is used to trigger sending of a group of SPOE
11953 messages. To do so, the SPOE engine used to send messages must be defined, as
11954 well as the SPOE group to send. Of course, the SPOE engine must refer to an
11955 existing SPOE filter. If not engine name is provided on the SPOE filter line,
11956 the SPOE agent name must be used.
11957
11958 <engine-name> The SPOE engine name.
11959
11960 <group-name> The SPOE group name as specified in the engine configuration.
11961
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011962 See section 7 about ACL usage.
11963
11964 See also : "tcp-request content", "tcp-response inspect-delay"
11965
11966
Willy Tarreau4f614292016-10-21 17:49:36 +020011967tcp-request session <action> [{if | unless} <condition>]
11968 Perform an action on a validated session depending on a layer 5 condition
11969 May be used in sections : defaults | frontend | listen | backend
11970 no | yes | yes | no
11971 Arguments :
11972 <action> defines the action to perform if the condition applies. See
11973 below.
11974
11975 <condition> is a standard layer5-only ACL-based condition (see section 7).
11976
Davor Ocelice9ed2812017-12-25 17:49:28 +010011977 Once a session is validated, (i.e. after all handshakes have been completed),
Willy Tarreau4f614292016-10-21 17:49:36 +020011978 it is possible to evaluate some conditions to decide whether this session
11979 must be accepted or dropped or have its counters tracked. Those conditions
11980 cannot make use of any data contents because no buffers are allocated yet and
11981 the processing cannot wait at this stage. The main use case it to copy some
11982 early information into variables (since variables are accessible in the
11983 session), or to keep track of some information collected after the handshake,
11984 such as SSL-level elements (SNI, ciphers, client cert's CN) or information
Davor Ocelice9ed2812017-12-25 17:49:28 +010011985 from the PROXY protocol header (e.g. track a source forwarded this way). The
Willy Tarreau4f614292016-10-21 17:49:36 +020011986 extracted information can thus be copied to a variable or tracked using
11987 "track-sc" rules. Of course it is also possible to decide to accept/reject as
11988 with other rulesets. Most operations performed here could also be performed
11989 in "tcp-request content" rules, except that in HTTP these rules are evaluated
11990 for each new request, and that might not always be acceptable. For example a
11991 rule might increment a counter on each evaluation. It would also be possible
11992 that a country is resolved by geolocation from the source IP address,
11993 assigned to a session-wide variable, then the source address rewritten from
11994 an HTTP header for all requests. If some contents need to be inspected in
11995 order to take the decision, the "tcp-request content" statements must be used
11996 instead.
11997
11998 The "tcp-request session" rules are evaluated in their exact declaration
11999 order. If no rule matches or if there is no rule, the default action is to
12000 accept the incoming session. There is no specific limit to the number of
12001 rules which may be inserted.
12002
12003 Several types of actions are supported :
12004 - accept : the request is accepted
12005 - reject : the request is rejected and the connection is closed
12006 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
12007 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +010012008 - sc-inc-gpc1(<sc-id>)
Cédric Dufour0d7712d2019-11-06 18:38:53 +010012009 - sc-set-gpt0(<sc-id>) { <int> | <expr> }
Christopher Faulet80c8acc2021-06-23 12:19:25 +020012010 - set-dst <expr>
12011 - set-dst-port <expr>
12012 - set-src <expr>
12013 - set-src-port <expr>
Willy Tarreau4f614292016-10-21 17:49:36 +020012014 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +010012015 - unset-var(<var-name>)
Willy Tarreau4f614292016-10-21 17:49:36 +020012016 - silent-drop
12017
12018 These actions have the same meaning as their respective counter-parts in
12019 "tcp-request connection" and "tcp-request content", so please refer to these
12020 sections for a complete description.
12021
12022 Note that the "if/unless" condition is optional. If no condition is set on
12023 the action, it is simply performed unconditionally. That can be useful for
12024 "track-sc*" actions as well as for changing the default action to a reject.
12025
12026 Example: track the original source address by default, or the one advertised
12027 in the PROXY protocol header for connection coming from the local
12028 proxies. The first connection-level rule enables receipt of the
12029 PROXY protocol for these ones, the second rule tracks whatever
12030 address we decide to keep after optional decoding.
12031
12032 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
12033 tcp-request session track-sc0 src
12034
12035 Example: accept all sessions from white-listed hosts, reject too fast
12036 sessions without counting them, and track accepted sessions.
12037 This results in session rate being capped from abusive sources.
12038
12039 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
12040 tcp-request session reject if { src_sess_rate gt 10 }
12041 tcp-request session track-sc0 src
12042
12043 Example: accept all sessions from white-listed hosts, count all other
12044 sessions and reject too fast ones. This results in abusive ones
12045 being blocked as long as they don't slow down.
12046
12047 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
12048 tcp-request session track-sc0 src
12049 tcp-request session reject if { sc0_sess_rate gt 10 }
12050
12051 See section 7 about ACL usage.
12052
12053 See also : "tcp-request connection", "tcp-request content", "stick-table"
12054
12055
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012056tcp-response inspect-delay <timeout>
12057 Set the maximum allowed time to wait for a response during content inspection
12058 May be used in sections : defaults | frontend | listen | backend
12059 no | no | yes | yes
12060 Arguments :
12061 <timeout> is the timeout value specified in milliseconds by default, but
12062 can be in any other unit if the number is suffixed by the unit,
12063 as explained at the top of this document.
12064
12065 See also : "tcp-response content", "tcp-request inspect-delay".
12066
12067
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010012068timeout check <timeout>
12069 Set additional check timeout, but only after a connection has been already
12070 established.
12071
12072 May be used in sections: defaults | frontend | listen | backend
12073 yes | no | yes | yes
12074 Arguments:
12075 <timeout> is the timeout value specified in milliseconds by default, but
12076 can be in any other unit if the number is suffixed by the unit,
12077 as explained at the top of this document.
12078
12079 If set, haproxy uses min("timeout connect", "inter") as a connect timeout
12080 for check and "timeout check" as an additional read timeout. The "min" is
Davor Ocelice9ed2812017-12-25 17:49:28 +010012081 used so that people running with *very* long "timeout connect" (e.g. those
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010012082 who needed this due to the queue or tarpit) do not slow down their checks.
Willy Tarreaud7550a22010-02-10 05:10:19 +010012083 (Please also note that there is no valid reason to have such long connect
12084 timeouts, because "timeout queue" and "timeout tarpit" can always be used to
12085 avoid that).
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010012086
12087 If "timeout check" is not set haproxy uses "inter" for complete check
12088 timeout (connect + read) exactly like all <1.3.15 version.
12089
12090 In most cases check request is much simpler and faster to handle than normal
12091 requests and people may want to kick out laggy servers so this timeout should
Willy Tarreau41a340d2008-01-22 12:25:31 +010012092 be smaller than "timeout server".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010012093
12094 This parameter is specific to backends, but can be specified once for all in
12095 "defaults" sections. This is in fact one of the easiest solutions not to
12096 forget about it.
12097
Willy Tarreau41a340d2008-01-22 12:25:31 +010012098 See also: "timeout connect", "timeout queue", "timeout server",
12099 "timeout tarpit".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010012100
12101
Willy Tarreau0ba27502007-12-24 16:55:16 +010012102timeout client <timeout>
Willy Tarreau0ba27502007-12-24 16:55:16 +010012103 Set the maximum inactivity time on the client side.
12104 May be used in sections : defaults | frontend | listen | backend
12105 yes | yes | yes | no
12106 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010012107 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010012108 can be in any other unit if the number is suffixed by the unit,
12109 as explained at the top of this document.
12110
12111 The inactivity timeout applies when the client is expected to acknowledge or
12112 send data. In HTTP mode, this timeout is particularly important to consider
12113 during the first phase, when the client sends the request, and during the
Baptiste Assmann2e1941e2016-03-06 23:24:12 +010012114 response while it is reading data sent by the server. That said, for the
12115 first phase, it is preferable to set the "timeout http-request" to better
12116 protect HAProxy from Slowloris like attacks. The value is specified in
12117 milliseconds by default, but can be in any other unit if the number is
Willy Tarreau0ba27502007-12-24 16:55:16 +010012118 suffixed by the unit, as specified at the top of this document. In TCP mode
12119 (and to a lesser extent, in HTTP mode), it is highly recommended that the
12120 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010012121 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +010012122 losses by specifying timeouts that are slightly above multiples of 3 seconds
Davor Ocelice9ed2812017-12-25 17:49:28 +010012123 (e.g. 4 or 5 seconds). If some long-lived sessions are mixed with short-lived
12124 sessions (e.g. WebSocket and HTTP), it's worth considering "timeout tunnel",
Willy Tarreau05cdd962014-05-10 14:30:07 +020012125 which overrides "timeout client" and "timeout server" for tunnels, as well as
12126 "timeout client-fin" for half-closed connections.
Willy Tarreau0ba27502007-12-24 16:55:16 +010012127
12128 This parameter is specific to frontends, but can be specified once for all in
12129 "defaults" sections. This is in fact one of the easiest solutions not to
12130 forget about it. An unspecified timeout results in an infinite timeout, which
12131 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050012132 during startup because it may result in accumulation of expired sessions in
Willy Tarreau0ba27502007-12-24 16:55:16 +010012133 the system if the system's timeouts are not configured either.
12134
Willy Tarreau95c4e142017-11-26 12:18:55 +010012135 This also applies to HTTP/2 connections, which will be closed with GOAWAY.
Lukas Tribus75df9d72017-11-24 19:05:12 +010012136
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020012137 See also : "timeout server", "timeout tunnel", "timeout http-request".
Willy Tarreau0ba27502007-12-24 16:55:16 +010012138
Willy Tarreau0ba27502007-12-24 16:55:16 +010012139
Willy Tarreau05cdd962014-05-10 14:30:07 +020012140timeout client-fin <timeout>
12141 Set the inactivity timeout on the client side for half-closed connections.
12142 May be used in sections : defaults | frontend | listen | backend
12143 yes | yes | yes | no
12144 Arguments :
12145 <timeout> is the timeout value specified in milliseconds by default, but
12146 can be in any other unit if the number is suffixed by the unit,
12147 as explained at the top of this document.
12148
12149 The inactivity timeout applies when the client is expected to acknowledge or
12150 send data while one direction is already shut down. This timeout is different
12151 from "timeout client" in that it only applies to connections which are closed
12152 in one direction. This is particularly useful to avoid keeping connections in
12153 FIN_WAIT state for too long when clients do not disconnect cleanly. This
12154 problem is particularly common long connections such as RDP or WebSocket.
12155 Note that this timeout can override "timeout tunnel" when a connection shuts
Willy Tarreau599391a2017-11-24 10:16:00 +010012156 down in one direction. It is applied to idle HTTP/2 connections once a GOAWAY
12157 frame was sent, often indicating an expectation that the connection quickly
12158 ends.
Willy Tarreau05cdd962014-05-10 14:30:07 +020012159
12160 This parameter is specific to frontends, but can be specified once for all in
12161 "defaults" sections. By default it is not set, so half-closed connections
12162 will use the other timeouts (timeout.client or timeout.tunnel).
12163
12164 See also : "timeout client", "timeout server-fin", and "timeout tunnel".
12165
12166
Willy Tarreau0ba27502007-12-24 16:55:16 +010012167timeout connect <timeout>
Willy Tarreau0ba27502007-12-24 16:55:16 +010012168 Set the maximum time to wait for a connection attempt to a server to succeed.
12169 May be used in sections : defaults | frontend | listen | backend
12170 yes | no | yes | yes
12171 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010012172 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010012173 can be in any other unit if the number is suffixed by the unit,
12174 as explained at the top of this document.
12175
12176 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010012177 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +010012178 cover one or several TCP packet losses by specifying timeouts that are
Davor Ocelice9ed2812017-12-25 17:49:28 +010012179 slightly above multiples of 3 seconds (e.g. 4 or 5 seconds). By default, the
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010012180 connect timeout also presets both queue and tarpit timeouts to the same value
12181 if these have not been specified.
Willy Tarreau0ba27502007-12-24 16:55:16 +010012182
12183 This parameter is specific to backends, but can be specified once for all in
12184 "defaults" sections. This is in fact one of the easiest solutions not to
12185 forget about it. An unspecified timeout results in an infinite timeout, which
12186 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050012187 during startup because it may result in accumulation of failed sessions in
Willy Tarreau0ba27502007-12-24 16:55:16 +010012188 the system if the system's timeouts are not configured either.
12189
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020012190 See also: "timeout check", "timeout queue", "timeout server", "timeout tarpit".
Willy Tarreau0ba27502007-12-24 16:55:16 +010012191
Willy Tarreau0ba27502007-12-24 16:55:16 +010012192
Willy Tarreaub16a5742010-01-10 14:46:16 +010012193timeout http-keep-alive <timeout>
12194 Set the maximum allowed time to wait for a new HTTP request to appear
12195 May be used in sections : defaults | frontend | listen | backend
12196 yes | yes | yes | yes
12197 Arguments :
12198 <timeout> is the timeout value specified in milliseconds by default, but
12199 can be in any other unit if the number is suffixed by the unit,
12200 as explained at the top of this document.
12201
12202 By default, the time to wait for a new request in case of keep-alive is set
12203 by "timeout http-request". However this is not always convenient because some
12204 people want very short keep-alive timeouts in order to release connections
12205 faster, and others prefer to have larger ones but still have short timeouts
12206 once the request has started to present itself.
12207
12208 The "http-keep-alive" timeout covers these needs. It will define how long to
12209 wait for a new HTTP request to start coming after a response was sent. Once
12210 the first byte of request has been seen, the "http-request" timeout is used
12211 to wait for the complete request to come. Note that empty lines prior to a
12212 new request do not refresh the timeout and are not counted as a new request.
12213
12214 There is also another difference between the two timeouts : when a connection
12215 expires during timeout http-keep-alive, no error is returned, the connection
12216 just closes. If the connection expires in "http-request" while waiting for a
12217 connection to complete, a HTTP 408 error is returned.
12218
12219 In general it is optimal to set this value to a few tens to hundreds of
12220 milliseconds, to allow users to fetch all objects of a page at once but
Davor Ocelice9ed2812017-12-25 17:49:28 +010012221 without waiting for further clicks. Also, if set to a very small value (e.g.
Willy Tarreaub16a5742010-01-10 14:46:16 +010012222 1 millisecond) it will probably only accept pipelined requests but not the
12223 non-pipelined ones. It may be a nice trade-off for very large sites running
Patrick Mézard2382ad62010-05-09 10:43:32 +020012224 with tens to hundreds of thousands of clients.
Willy Tarreaub16a5742010-01-10 14:46:16 +010012225
12226 If this parameter is not set, the "http-request" timeout applies, and if both
12227 are not set, "timeout client" still applies at the lower level. It should be
12228 set in the frontend to take effect, unless the frontend is in TCP mode, in
12229 which case the HTTP backend's timeout will be used.
12230
Willy Tarreau95c4e142017-11-26 12:18:55 +010012231 When using HTTP/2 "timeout client" is applied instead. This is so we can keep
12232 using short keep-alive timeouts in HTTP/1.1 while using longer ones in HTTP/2
Lukas Tribus75df9d72017-11-24 19:05:12 +010012233 (where we only have one connection per client and a connection setup).
12234
Willy Tarreaub16a5742010-01-10 14:46:16 +010012235 See also : "timeout http-request", "timeout client".
12236
12237
Willy Tarreau036fae02008-01-06 13:24:40 +010012238timeout http-request <timeout>
12239 Set the maximum allowed time to wait for a complete HTTP request
12240 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaucd7afc02009-07-12 10:03:17 +020012241 yes | yes | yes | yes
Willy Tarreau036fae02008-01-06 13:24:40 +010012242 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010012243 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau036fae02008-01-06 13:24:40 +010012244 can be in any other unit if the number is suffixed by the unit,
12245 as explained at the top of this document.
12246
12247 In order to offer DoS protection, it may be required to lower the maximum
12248 accepted time to receive a complete HTTP request without affecting the client
12249 timeout. This helps protecting against established connections on which
12250 nothing is sent. The client timeout cannot offer a good protection against
12251 this abuse because it is an inactivity timeout, which means that if the
12252 attacker sends one character every now and then, the timeout will not
12253 trigger. With the HTTP request timeout, no matter what speed the client
Willy Tarreau2705a612014-05-23 17:38:34 +020012254 types, the request will be aborted if it does not complete in time. When the
12255 timeout expires, an HTTP 408 response is sent to the client to inform it
12256 about the problem, and the connection is closed. The logs will report
12257 termination codes "cR". Some recent browsers are having problems with this
Davor Ocelice9ed2812017-12-25 17:49:28 +010012258 standard, well-documented behavior, so it might be needed to hide the 408
Willy Tarreau0f228a02015-05-01 15:37:53 +020012259 code using "option http-ignore-probes" or "errorfile 408 /dev/null". See
12260 more details in the explanations of the "cR" termination code in section 8.5.
Willy Tarreau036fae02008-01-06 13:24:40 +010012261
Baptiste Assmanneccdf432015-10-28 13:49:01 +010012262 By default, this timeout only applies to the header part of the request,
12263 and not to any data. As soon as the empty line is received, this timeout is
12264 not used anymore. When combined with "option http-buffer-request", this
12265 timeout also applies to the body of the request..
12266 It is used again on keep-alive connections to wait for a second
Willy Tarreaub16a5742010-01-10 14:46:16 +010012267 request if "timeout http-keep-alive" is not set.
Willy Tarreau036fae02008-01-06 13:24:40 +010012268
12269 Generally it is enough to set it to a few seconds, as most clients send the
12270 full request immediately upon connection. Add 3 or more seconds to cover TCP
Davor Ocelice9ed2812017-12-25 17:49:28 +010012271 retransmits but that's all. Setting it to very low values (e.g. 50 ms) will
Willy Tarreau036fae02008-01-06 13:24:40 +010012272 generally work on local networks as long as there are no packet losses. This
12273 will prevent people from sending bare HTTP requests using telnet.
12274
12275 If this parameter is not set, the client timeout still applies between each
Willy Tarreaucd7afc02009-07-12 10:03:17 +020012276 chunk of the incoming request. It should be set in the frontend to take
12277 effect, unless the frontend is in TCP mode, in which case the HTTP backend's
12278 timeout will be used.
Willy Tarreau036fae02008-01-06 13:24:40 +010012279
Willy Tarreau0f228a02015-05-01 15:37:53 +020012280 See also : "errorfile", "http-ignore-probes", "timeout http-keep-alive", and
Baptiste Assmanneccdf432015-10-28 13:49:01 +010012281 "timeout client", "option http-buffer-request".
Willy Tarreau036fae02008-01-06 13:24:40 +010012282
Willy Tarreau844e3c52008-01-11 16:28:18 +010012283
12284timeout queue <timeout>
12285 Set the maximum time to wait in the queue for a connection slot to be free
12286 May be used in sections : defaults | frontend | listen | backend
12287 yes | no | yes | yes
12288 Arguments :
12289 <timeout> is the timeout value specified in milliseconds by default, but
12290 can be in any other unit if the number is suffixed by the unit,
12291 as explained at the top of this document.
12292
12293 When a server's maxconn is reached, connections are left pending in a queue
12294 which may be server-specific or global to the backend. In order not to wait
12295 indefinitely, a timeout is applied to requests pending in the queue. If the
12296 timeout is reached, it is considered that the request will almost never be
12297 served, so it is dropped and a 503 error is returned to the client.
12298
12299 The "timeout queue" statement allows to fix the maximum time for a request to
12300 be left pending in a queue. If unspecified, the same value as the backend's
12301 connection timeout ("timeout connect") is used, for backwards compatibility
12302 with older versions with no "timeout queue" parameter.
12303
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020012304 See also : "timeout connect".
Willy Tarreau844e3c52008-01-11 16:28:18 +010012305
12306
12307timeout server <timeout>
Willy Tarreau844e3c52008-01-11 16:28:18 +010012308 Set the maximum inactivity time on the server side.
12309 May be used in sections : defaults | frontend | listen | backend
12310 yes | no | yes | yes
12311 Arguments :
12312 <timeout> is the timeout value specified in milliseconds by default, but
12313 can be in any other unit if the number is suffixed by the unit,
12314 as explained at the top of this document.
12315
12316 The inactivity timeout applies when the server is expected to acknowledge or
12317 send data. In HTTP mode, this timeout is particularly important to consider
12318 during the first phase of the server's response, when it has to send the
12319 headers, as it directly represents the server's processing time for the
12320 request. To find out what value to put there, it's often good to start with
12321 what would be considered as unacceptable response times, then check the logs
12322 to observe the response time distribution, and adjust the value accordingly.
12323
12324 The value is specified in milliseconds by default, but can be in any other
12325 unit if the number is suffixed by the unit, as specified at the top of this
12326 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
12327 recommended that the client timeout remains equal to the server timeout in
12328 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010012329 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +010012330 packet losses by specifying timeouts that are slightly above multiples of 3
Davor Ocelice9ed2812017-12-25 17:49:28 +010012331 seconds (e.g. 4 or 5 seconds minimum). If some long-lived sessions are mixed
12332 with short-lived sessions (e.g. WebSocket and HTTP), it's worth considering
Willy Tarreauce887fd2012-05-12 12:50:00 +020012333 "timeout tunnel", which overrides "timeout client" and "timeout server" for
12334 tunnels.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012335
12336 This parameter is specific to backends, but can be specified once for all in
12337 "defaults" sections. This is in fact one of the easiest solutions not to
12338 forget about it. An unspecified timeout results in an infinite timeout, which
12339 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050012340 during startup because it may result in accumulation of expired sessions in
Willy Tarreau844e3c52008-01-11 16:28:18 +010012341 the system if the system's timeouts are not configured either.
12342
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020012343 See also : "timeout client" and "timeout tunnel".
Willy Tarreau844e3c52008-01-11 16:28:18 +010012344
Willy Tarreau05cdd962014-05-10 14:30:07 +020012345
12346timeout server-fin <timeout>
12347 Set the inactivity timeout on the server side for half-closed connections.
12348 May be used in sections : defaults | frontend | listen | backend
12349 yes | no | yes | yes
12350 Arguments :
12351 <timeout> is the timeout value specified in milliseconds by default, but
12352 can be in any other unit if the number is suffixed by the unit,
12353 as explained at the top of this document.
12354
12355 The inactivity timeout applies when the server is expected to acknowledge or
12356 send data while one direction is already shut down. This timeout is different
12357 from "timeout server" in that it only applies to connections which are closed
12358 in one direction. This is particularly useful to avoid keeping connections in
12359 FIN_WAIT state for too long when a remote server does not disconnect cleanly.
12360 This problem is particularly common long connections such as RDP or WebSocket.
12361 Note that this timeout can override "timeout tunnel" when a connection shuts
12362 down in one direction. This setting was provided for completeness, but in most
12363 situations, it should not be needed.
12364
12365 This parameter is specific to backends, but can be specified once for all in
12366 "defaults" sections. By default it is not set, so half-closed connections
12367 will use the other timeouts (timeout.server or timeout.tunnel).
12368
12369 See also : "timeout client-fin", "timeout server", and "timeout tunnel".
12370
Willy Tarreau844e3c52008-01-11 16:28:18 +010012371
12372timeout tarpit <timeout>
Cyril Bonté78caf842010-03-10 22:41:43 +010012373 Set the duration for which tarpitted connections will be maintained
Willy Tarreau844e3c52008-01-11 16:28:18 +010012374 May be used in sections : defaults | frontend | listen | backend
12375 yes | yes | yes | yes
12376 Arguments :
12377 <timeout> is the tarpit duration specified in milliseconds by default, but
12378 can be in any other unit if the number is suffixed by the unit,
12379 as explained at the top of this document.
12380
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020012381 When a connection is tarpitted using "http-request tarpit", it is maintained
12382 open with no activity for a certain amount of time, then closed. "timeout
12383 tarpit" defines how long it will be maintained open.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012384
12385 The value is specified in milliseconds by default, but can be in any other
12386 unit if the number is suffixed by the unit, as specified at the top of this
12387 document. If unspecified, the same value as the backend's connection timeout
12388 ("timeout connect") is used, for backwards compatibility with older versions
Cyril Bonté78caf842010-03-10 22:41:43 +010012389 with no "timeout tarpit" parameter.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012390
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020012391 See also : "timeout connect".
Willy Tarreau844e3c52008-01-11 16:28:18 +010012392
12393
Willy Tarreauce887fd2012-05-12 12:50:00 +020012394timeout tunnel <timeout>
12395 Set the maximum inactivity time on the client and server side for tunnels.
12396 May be used in sections : defaults | frontend | listen | backend
12397 yes | no | yes | yes
12398 Arguments :
12399 <timeout> is the timeout value specified in milliseconds by default, but
12400 can be in any other unit if the number is suffixed by the unit,
12401 as explained at the top of this document.
12402
Jamie Gloudonaaa21002012-08-25 00:18:33 -040012403 The tunnel timeout applies when a bidirectional connection is established
Willy Tarreauce887fd2012-05-12 12:50:00 +020012404 between a client and a server, and the connection remains inactive in both
12405 directions. This timeout supersedes both the client and server timeouts once
12406 the connection becomes a tunnel. In TCP, this timeout is used as soon as no
Davor Ocelice9ed2812017-12-25 17:49:28 +010012407 analyzer remains attached to either connection (e.g. tcp content rules are
12408 accepted). In HTTP, this timeout is used when a connection is upgraded (e.g.
Willy Tarreauce887fd2012-05-12 12:50:00 +020012409 when switching to the WebSocket protocol, or forwarding a CONNECT request
12410 to a proxy), or after the first response when no keepalive/close option is
12411 specified.
12412
Willy Tarreau05cdd962014-05-10 14:30:07 +020012413 Since this timeout is usually used in conjunction with long-lived connections,
12414 it usually is a good idea to also set "timeout client-fin" to handle the
12415 situation where a client suddenly disappears from the net and does not
12416 acknowledge a close, or sends a shutdown and does not acknowledge pending
12417 data anymore. This can happen in lossy networks where firewalls are present,
12418 and is detected by the presence of large amounts of sessions in a FIN_WAIT
12419 state.
12420
Willy Tarreauce887fd2012-05-12 12:50:00 +020012421 The value is specified in milliseconds by default, but can be in any other
12422 unit if the number is suffixed by the unit, as specified at the top of this
12423 document. Whatever the expected normal idle time, it is a good practice to
12424 cover at least one or several TCP packet losses by specifying timeouts that
Davor Ocelice9ed2812017-12-25 17:49:28 +010012425 are slightly above multiples of 3 seconds (e.g. 4 or 5 seconds minimum).
Willy Tarreauce887fd2012-05-12 12:50:00 +020012426
12427 This parameter is specific to backends, but can be specified once for all in
12428 "defaults" sections. This is in fact one of the easiest solutions not to
12429 forget about it.
12430
12431 Example :
12432 defaults http
12433 option http-server-close
12434 timeout connect 5s
12435 timeout client 30s
Willy Tarreau05cdd962014-05-10 14:30:07 +020012436 timeout client-fin 30s
Willy Tarreauce887fd2012-05-12 12:50:00 +020012437 timeout server 30s
12438 timeout tunnel 1h # timeout to use with WebSocket and CONNECT
12439
Willy Tarreau05cdd962014-05-10 14:30:07 +020012440 See also : "timeout client", "timeout client-fin", "timeout server".
Willy Tarreauce887fd2012-05-12 12:50:00 +020012441
12442
Willy Tarreau844e3c52008-01-11 16:28:18 +010012443transparent (deprecated)
12444 Enable client-side transparent proxying
12445 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +010012446 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +010012447 Arguments : none
12448
12449 This keyword was introduced in order to provide layer 7 persistence to layer
12450 3 load balancers. The idea is to use the OS's ability to redirect an incoming
12451 connection for a remote address to a local process (here HAProxy), and let
12452 this process know what address was initially requested. When this option is
12453 used, sessions without cookies will be forwarded to the original destination
12454 IP address of the incoming request (which should match that of another
12455 equipment), while requests with cookies will still be forwarded to the
12456 appropriate server.
12457
12458 The "transparent" keyword is deprecated, use "option transparent" instead.
12459
12460 Note that contrary to a common belief, this option does NOT make HAProxy
12461 present the client's IP to the server when establishing the connection.
12462
Willy Tarreau844e3c52008-01-11 16:28:18 +010012463 See also: "option transparent"
12464
William Lallemanda73203e2012-03-12 12:48:57 +010012465unique-id-format <string>
12466 Generate a unique ID for each request.
12467 May be used in sections : defaults | frontend | listen | backend
12468 yes | yes | yes | no
12469 Arguments :
12470 <string> is a log-format string.
12471
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012472 This keyword creates a ID for each request using the custom log format. A
12473 unique ID is useful to trace a request passing through many components of
12474 a complex infrastructure. The newly created ID may also be logged using the
12475 %ID tag the log-format string.
William Lallemanda73203e2012-03-12 12:48:57 +010012476
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012477 The format should be composed from elements that are guaranteed to be
12478 unique when combined together. For instance, if multiple haproxy instances
12479 are involved, it might be important to include the node name. It is often
12480 needed to log the incoming connection's source and destination addresses
12481 and ports. Note that since multiple requests may be performed over the same
12482 connection, including a request counter may help differentiate them.
12483 Similarly, a timestamp may protect against a rollover of the counter.
12484 Logging the process ID will avoid collisions after a service restart.
William Lallemanda73203e2012-03-12 12:48:57 +010012485
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012486 It is recommended to use hexadecimal notation for many fields since it
12487 makes them more compact and saves space in logs.
William Lallemanda73203e2012-03-12 12:48:57 +010012488
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012489 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010012490
Julien Vehentf21be322014-03-07 08:27:34 -050012491 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010012492
12493 will generate:
12494
12495 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
12496
12497 See also: "unique-id-header"
12498
12499unique-id-header <name>
12500 Add a unique ID header in the HTTP request.
12501 May be used in sections : defaults | frontend | listen | backend
12502 yes | yes | yes | no
12503 Arguments :
12504 <name> is the name of the header.
12505
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012506 Add a unique-id header in the HTTP request sent to the server, using the
12507 unique-id-format. It can't work if the unique-id-format doesn't exist.
William Lallemanda73203e2012-03-12 12:48:57 +010012508
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012509 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010012510
Julien Vehentf21be322014-03-07 08:27:34 -050012511 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010012512 unique-id-header X-Unique-ID
12513
12514 will generate:
12515
12516 X-Unique-ID: 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
12517
12518 See also: "unique-id-format"
Willy Tarreau844e3c52008-01-11 16:28:18 +010012519
Willy Tarreauf51658d2014-04-23 01:21:56 +020012520use_backend <backend> [{if | unless} <condition>]
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020012521 Switch to a specific backend if/unless an ACL-based condition is matched.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012522 May be used in sections : defaults | frontend | listen | backend
12523 no | yes | yes | no
12524 Arguments :
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010012525 <backend> is the name of a valid backend or "listen" section, or a
12526 "log-format" string resolving to a backend name.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012527
Willy Tarreauf51658d2014-04-23 01:21:56 +020012528 <condition> is a condition composed of ACLs, as described in section 7. If
12529 it is omitted, the rule is unconditionally applied.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012530
12531 When doing content-switching, connections arrive on a frontend and are then
12532 dispatched to various backends depending on a number of conditions. The
12533 relation between the conditions and the backends is described with the
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020012534 "use_backend" keyword. While it is normally used with HTTP processing, it can
Davor Ocelice9ed2812017-12-25 17:49:28 +010012535 also be used in pure TCP, either without content using stateless ACLs (e.g.
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020012536 source address validation) or combined with a "tcp-request" rule to wait for
12537 some payload.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012538
12539 There may be as many "use_backend" rules as desired. All of these rules are
12540 evaluated in their declaration order, and the first one which matches will
12541 assign the backend.
12542
12543 In the first form, the backend will be used if the condition is met. In the
12544 second form, the backend will be used if the condition is not met. If no
12545 condition is valid, the backend defined with "default_backend" will be used.
12546 If no default backend is defined, either the servers in the same section are
12547 used (in case of a "listen" section) or, in case of a frontend, no server is
12548 used and a 503 service unavailable response is returned.
12549
Willy Tarreau51aecc72009-07-12 09:47:04 +020012550 Note that it is possible to switch from a TCP frontend to an HTTP backend. In
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012551 this case, either the frontend has already checked that the protocol is HTTP,
Willy Tarreau51aecc72009-07-12 09:47:04 +020012552 and backend processing will immediately follow, or the backend will wait for
12553 a complete HTTP request to get in. This feature is useful when a frontend
12554 must decode several protocols on a unique port, one of them being HTTP.
12555
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010012556 When <backend> is a simple name, it is resolved at configuration time, and an
12557 error is reported if the specified backend does not exist. If <backend> is
12558 a log-format string instead, no check may be done at configuration time, so
12559 the backend name is resolved dynamically at run time. If the resulting
12560 backend name does not correspond to any valid backend, no other rule is
12561 evaluated, and the default_backend directive is applied instead. Note that
12562 when using dynamic backend names, it is highly recommended to use a prefix
12563 that no other backend uses in order to ensure that an unauthorized backend
12564 cannot be forced from the request.
12565
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012566 It is worth mentioning that "use_backend" rules with an explicit name are
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010012567 used to detect the association between frontends and backends to compute the
12568 backend's "fullconn" setting. This cannot be done for dynamic names.
12569
12570 See also: "default_backend", "tcp-request", "fullconn", "log-format", and
12571 section 7 about ACLs.
Willy Tarreaud72758d2010-01-12 10:42:19 +010012572
Christopher Fauletb30b3102019-09-12 23:03:09 +020012573use-fcgi-app <name>
12574 Defines the FastCGI application to use for the backend.
12575 May be used in sections : defaults | frontend | listen | backend
12576 no | no | yes | yes
12577 Arguments :
12578 <name> is the name of the FastCGI application to use.
12579
12580 See section 10.1 about FastCGI application setup for details.
Willy Tarreau036fae02008-01-06 13:24:40 +010012581
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012582use-server <server> if <condition>
12583use-server <server> unless <condition>
12584 Only use a specific server if/unless an ACL-based condition is matched.
12585 May be used in sections : defaults | frontend | listen | backend
12586 no | no | yes | yes
12587 Arguments :
Jerome Magnin824186b2020-03-29 09:37:12 +020012588 <server> is the name of a valid server in the same backend section
12589 or a "log-format" string resolving to a server name.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012590
12591 <condition> is a condition composed of ACLs, as described in section 7.
12592
12593 By default, connections which arrive to a backend are load-balanced across
12594 the available servers according to the configured algorithm, unless a
12595 persistence mechanism such as a cookie is used and found in the request.
12596
12597 Sometimes it is desirable to forward a particular request to a specific
12598 server without having to declare a dedicated backend for this server. This
12599 can be achieved using the "use-server" rules. These rules are evaluated after
12600 the "redirect" rules and before evaluating cookies, and they have precedence
12601 on them. There may be as many "use-server" rules as desired. All of these
12602 rules are evaluated in their declaration order, and the first one which
12603 matches will assign the server.
12604
12605 If a rule designates a server which is down, and "option persist" is not used
12606 and no force-persist rule was validated, it is ignored and evaluation goes on
12607 with the next rules until one matches.
12608
12609 In the first form, the server will be used if the condition is met. In the
12610 second form, the server will be used if the condition is not met. If no
12611 condition is valid, the processing continues and the server will be assigned
12612 according to other persistence mechanisms.
12613
12614 Note that even if a rule is matched, cookie processing is still performed but
12615 does not assign the server. This allows prefixed cookies to have their prefix
12616 stripped.
12617
12618 The "use-server" statement works both in HTTP and TCP mode. This makes it
12619 suitable for use with content-based inspection. For instance, a server could
Lukas Tribusa267b5d2020-07-19 00:25:06 +020012620 be selected in a farm according to the TLS SNI field when using protocols with
Alex192600b2021-06-05 13:23:08 +020012621 implicit TLS (also see "req.ssl_sni"). And if these servers have their weight
Lukas Tribusa267b5d2020-07-19 00:25:06 +020012622 set to zero, they will not be used for other traffic.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012623
12624 Example :
12625 # intercept incoming TLS requests based on the SNI field
Alex192600b2021-06-05 13:23:08 +020012626 use-server www if { req.ssl_sni -i www.example.com }
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012627 server www 192.168.0.1:443 weight 0
Alex192600b2021-06-05 13:23:08 +020012628 use-server mail if { req.ssl_sni -i mail.example.com }
Lukas Tribusa267b5d2020-07-19 00:25:06 +020012629 server mail 192.168.0.1:465 weight 0
Alex192600b2021-06-05 13:23:08 +020012630 use-server imap if { req.ssl_sni -i imap.example.com }
Lukas Tribus98a3e3f2017-03-26 12:55:35 +000012631 server imap 192.168.0.1:993 weight 0
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012632 # all the rest is forwarded to this server
12633 server default 192.168.0.2:443 check
12634
Jerome Magnin824186b2020-03-29 09:37:12 +020012635 When <server> is a simple name, it is checked against existing servers in the
12636 configuration and an error is reported if the specified server does not exist.
12637 If it is a log-format, no check is performed when parsing the configuration,
12638 and if we can't resolve a valid server name at runtime but the use-server rule
Ilya Shipitsin11057a32020-06-21 21:18:27 +050012639 was conditioned by an ACL returning true, no other use-server rule is applied
Jerome Magnin824186b2020-03-29 09:37:12 +020012640 and we fall back to load balancing.
12641
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012642 See also: "use_backend", section 5 about server and section 7 about ACLs.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012643
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012644
Davor Ocelice9ed2812017-12-25 17:49:28 +0100126455. Bind and server options
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012646--------------------------
12647
12648The "bind", "server" and "default-server" keywords support a number of settings
12649depending on some build options and on the system HAProxy was built on. These
12650settings generally each consist in one word sometimes followed by a value,
12651written on the same line as the "bind" or "server" line. All these options are
12652described in this section.
12653
12654
126555.1. Bind options
12656-----------------
12657
12658The "bind" keyword supports a certain number of settings which are all passed
12659as arguments on the same line. The order in which those arguments appear makes
12660no importance, provided that they appear after the bind address. All of these
12661parameters are optional. Some of them consist in a single words (booleans),
12662while other ones expect a value after them. In this case, the value must be
12663provided immediately after the setting name.
12664
12665The currently supported settings are the following ones.
12666
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010012667accept-netscaler-cip <magic number>
12668 Enforces the use of the NetScaler Client IP insertion protocol over any
12669 connection accepted by any of the TCP sockets declared on the same line. The
12670 NetScaler Client IP insertion protocol dictates the layer 3/4 addresses of
12671 the incoming connection to be used everywhere an address is used, with the
12672 only exception of "tcp-request connection" rules which will only see the
12673 real connection address. Logs will reflect the addresses indicated in the
12674 protocol, unless it is violated, in which case the real address will still
12675 be used. This keyword combined with support from external components can be
12676 used as an efficient and reliable alternative to the X-Forwarded-For
Bertrand Jacquin90759682016-06-06 15:35:39 +010012677 mechanism which is not always reliable and not even always usable. See also
12678 "tcp-request connection expect-netscaler-cip" for a finer-grained setting of
12679 which client is allowed to use the protocol.
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010012680
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012681accept-proxy
12682 Enforces the use of the PROXY protocol over any connection accepted by any of
Willy Tarreau77992672014-06-14 11:06:17 +020012683 the sockets declared on the same line. Versions 1 and 2 of the PROXY protocol
12684 are supported and correctly detected. The PROXY protocol dictates the layer
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012685 3/4 addresses of the incoming connection to be used everywhere an address is
12686 used, with the only exception of "tcp-request connection" rules which will
12687 only see the real connection address. Logs will reflect the addresses
12688 indicated in the protocol, unless it is violated, in which case the real
Davor Ocelice9ed2812017-12-25 17:49:28 +010012689 address will still be used. This keyword combined with support from external
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012690 components can be used as an efficient and reliable alternative to the
12691 X-Forwarded-For mechanism which is not always reliable and not even always
Willy Tarreau4f0d9192013-06-11 20:40:55 +020012692 usable. See also "tcp-request connection expect-proxy" for a finer-grained
12693 setting of which client is allowed to use the protocol.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012694
Olivier Houchardc2aae742017-09-22 18:26:28 +020012695allow-0rtt
Bertrand Jacquina25282b2018-08-14 00:56:13 +010012696 Allow receiving early data when using TLSv1.3. This is disabled by default,
Olivier Houchard69752962019-01-08 15:35:32 +010012697 due to security considerations. Because it is vulnerable to replay attacks,
John Roeslerfb2fce12019-07-10 15:45:51 -050012698 you should only allow if for requests that are safe to replay, i.e. requests
Olivier Houchard69752962019-01-08 15:35:32 +010012699 that are idempotent. You can use the "wait-for-handshake" action for any
12700 request that wouldn't be safe with early data.
Olivier Houchardc2aae742017-09-22 18:26:28 +020012701
Willy Tarreauab861d32013-04-02 02:30:41 +020012702alpn <protocols>
12703 This enables the TLS ALPN extension and advertises the specified protocol
12704 list as supported on top of ALPN. The protocol list consists in a comma-
12705 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
John Roeslerfb2fce12019-07-10 15:45:51 -050012706 quotes). This requires that the SSL library is built with support for TLS
Willy Tarreauab861d32013-04-02 02:30:41 +020012707 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
Willy Tarreau95c4e142017-11-26 12:18:55 +010012708 initial NPN extension. ALPN is required to enable HTTP/2 on an HTTP frontend.
12709 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
12710 now obsolete NPN extension. At the time of writing this, most browsers still
12711 support both ALPN and NPN for HTTP/2 so a fallback to NPN may still work for
12712 a while. But ALPN must be used whenever possible. If both HTTP/2 and HTTP/1.1
12713 are expected to be supported, both versions can be advertised, in order of
12714 preference, like below :
12715
12716 bind :443 ssl crt pub.pem alpn h2,http/1.1
Willy Tarreauab861d32013-04-02 02:30:41 +020012717
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012718backlog <backlog>
Willy Tarreaue2711c72019-02-27 15:39:41 +010012719 Sets the socket's backlog to this value. If unspecified or 0, the frontend's
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012720 backlog is used instead, which generally defaults to the maxconn value.
12721
Emmanuel Hocdete7f2b732017-01-09 16:15:54 +010012722curves <curves>
12723 This setting is only available when support for OpenSSL was built in. It sets
12724 the string describing the list of elliptic curves algorithms ("curve suite")
12725 that are negotiated during the SSL/TLS handshake with ECDHE. The format of the
12726 string is a colon-delimited list of curve name.
12727 Example: "X25519:P-256" (without quote)
12728 When "curves" is set, "ecdhe" parameter is ignored.
12729
Emeric Brun7fb34422012-09-28 15:26:15 +020012730ecdhe <named curve>
12731 This setting is only available when support for OpenSSL was built in. It sets
Emeric Brun6924ef82013-03-06 14:08:53 +010012732 the named curve (RFC 4492) used to generate ECDH ephemeral keys. By default,
12733 used named curve is prime256v1.
Emeric Brun7fb34422012-09-28 15:26:15 +020012734
Emeric Brunfd33a262012-10-11 16:28:27 +020012735ca-file <cafile>
Emeric Brun1a073b42012-09-28 17:07:34 +020012736 This setting is only available when support for OpenSSL was built in. It
12737 designates a PEM file from which to load CA certificates used to verify
12738 client's certificate.
12739
Emeric Brunb6dc9342012-09-28 17:55:37 +020012740ca-ignore-err [all|<errorID>,...]
12741 This setting is only available when support for OpenSSL was built in.
12742 Sets a comma separated list of errorIDs to ignore during verify at depth > 0.
12743 If set to 'all', all errors are ignored. SSL handshake is not aborted if an
12744 error is ignored.
12745
Christopher Faulet31af49d2015-06-09 17:29:50 +020012746ca-sign-file <cafile>
12747 This setting is only available when support for OpenSSL was built in. It
12748 designates a PEM file containing both the CA certificate and the CA private
12749 key used to create and sign server's certificates. This is a mandatory
12750 setting when the dynamic generation of certificates is enabled. See
12751 'generate-certificates' for details.
12752
Bertrand Jacquind4d0a232016-11-13 16:37:12 +000012753ca-sign-pass <passphrase>
Christopher Faulet31af49d2015-06-09 17:29:50 +020012754 This setting is only available when support for OpenSSL was built in. It is
12755 the CA private key passphrase. This setting is optional and used only when
12756 the dynamic generation of certificates is enabled. See
12757 'generate-certificates' for details.
12758
Emmanuel Hocdet842e94e2019-12-16 16:39:17 +010012759ca-verify-file <cafile>
12760 This setting designates a PEM file from which to load CA certificates used to
12761 verify client's certificate. It designates CA certificates which must not be
12762 included in CA names sent in server hello message. Typically, "ca-file" must
12763 be defined with intermediate certificates, and "ca-verify-file" with
12764 certificates to ending the chain, like root CA.
12765
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012766ciphers <ciphers>
12767 This setting is only available when support for OpenSSL was built in. It sets
12768 the string describing the list of cipher algorithms ("cipher suite") that are
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +000012769 negotiated during the SSL/TLS handshake up to TLSv1.2. The format of the
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000012770 string is defined in "man 1 ciphers" from OpenSSL man pages. For background
Dirkjan Bussink415150f2018-09-14 11:14:21 +020012771 information and recommendations see e.g.
12772 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
12773 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
12774 cipher configuration, please check the "ciphersuites" keyword.
12775
12776ciphersuites <ciphersuites>
12777 This setting is only available when support for OpenSSL was built in and
12778 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the string describing
12779 the list of cipher algorithms ("cipher suite") that are negotiated during the
12780 TLSv1.3 handshake. The format of the string is defined in "man 1 ciphers" from
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000012781 OpenSSL man pages under the "ciphersuites" section. For cipher configuration
12782 for TLSv1.2 and earlier, please check the "ciphers" keyword.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012783
Emeric Brunfd33a262012-10-11 16:28:27 +020012784crl-file <crlfile>
Emeric Brun1a073b42012-09-28 17:07:34 +020012785 This setting is only available when support for OpenSSL was built in. It
12786 designates a PEM file from which to load certificate revocation list used
12787 to verify client's certificate.
12788
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012789crt <cert>
Alex Davies0fbf0162013-03-02 16:04:50 +000012790 This setting is only available when support for OpenSSL was built in. It
12791 designates a PEM file containing both the required certificates and any
12792 associated private keys. This file can be built by concatenating multiple
12793 PEM files into one (e.g. cat cert.pem key.pem > combined.pem). If your CA
12794 requires an intermediate certificate, this can also be concatenated into this
Emmanuel Hocdet70df7bf2019-01-04 11:08:20 +010012795 file. Intermediate certificate can also be shared in a directory via
12796 "issuers-chain-path" directive.
Alex Davies0fbf0162013-03-02 16:04:50 +000012797
William Lallemand4c5adbf2020-02-24 14:23:22 +010012798 If the file does not contain a private key, HAProxy will try to load
12799 the key at the same path suffixed by a ".key".
12800
Alex Davies0fbf0162013-03-02 16:04:50 +000012801 If the OpenSSL used supports Diffie-Hellman, parameters present in this file
12802 are loaded.
12803
12804 If a directory name is used instead of a PEM file, then all files found in
William Lallemand3f25ae32020-02-24 16:30:12 +010012805 that directory will be loaded in alphabetic order unless their name ends
12806 with '.key', '.issuer', '.ocsp' or '.sctl' (reserved extensions). This
12807 directive may be specified multiple times in order to load certificates from
12808 multiple files or directories. The certificates will be presented to clients
12809 who provide a valid TLS Server Name Indication field matching one of their
12810 CN or alt subjects. Wildcards are supported, where a wildcard character '*'
12811 is used instead of the first hostname component (e.g. *.example.org matches
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010012812 www.example.org but not www.sub.example.org).
Alex Davies0fbf0162013-03-02 16:04:50 +000012813
12814 If no SNI is provided by the client or if the SSL library does not support
12815 TLS extensions, or if the client provides an SNI hostname which does not
12816 match any certificate, then the first loaded certificate will be presented.
12817 This means that when loading certificates from a directory, it is highly
Cyril Bonté3180f7b2015-01-25 00:16:08 +010012818 recommended to load the default one first as a file or to ensure that it will
12819 always be the first one in the directory.
Alex Davies0fbf0162013-03-02 16:04:50 +000012820
Emeric Brune032bfa2012-09-28 13:01:45 +020012821 Note that the same cert may be loaded multiple times without side effects.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012822
Davor Ocelice9ed2812017-12-25 17:49:28 +010012823 Some CAs (such as GoDaddy) offer a drop down list of server types that do not
Alex Davies0fbf0162013-03-02 16:04:50 +000012824 include HAProxy when obtaining a certificate. If this happens be sure to
Davor Ocelice9ed2812017-12-25 17:49:28 +010012825 choose a web server that the CA believes requires an intermediate CA (for
12826 GoDaddy, selection Apache Tomcat will get the correct bundle, but many
Alex Davies0fbf0162013-03-02 16:04:50 +000012827 others, e.g. nginx, result in a wrong bundle that will not work for some
12828 clients).
12829
Emeric Brun4147b2e2014-06-16 18:36:30 +020012830 For each PEM file, haproxy checks for the presence of file at the same path
12831 suffixed by ".ocsp". If such file is found, support for the TLS Certificate
12832 Status Request extension (also known as "OCSP stapling") is automatically
12833 enabled. The content of this file is optional. If not empty, it must contain
12834 a valid OCSP Response in DER format. In order to be valid an OCSP Response
12835 must comply with the following rules: it has to indicate a good status,
12836 it has to be a single response for the certificate of the PEM file, and it
12837 has to be valid at the moment of addition. If these rules are not respected
12838 the OCSP Response is ignored and a warning is emitted. In order to identify
12839 which certificate an OCSP Response applies to, the issuer's certificate is
12840 necessary. If the issuer's certificate is not found in the PEM file, it will
12841 be loaded from a file at the same path as the PEM file suffixed by ".issuer"
12842 if it exists otherwise it will fail with an error.
12843
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010012844 For each PEM file, haproxy also checks for the presence of file at the same
12845 path suffixed by ".sctl". If such file is found, support for Certificate
12846 Transparency (RFC6962) TLS extension is enabled. The file must contain a
12847 valid Signed Certificate Timestamp List, as described in RFC. File is parsed
12848 to check basic syntax, but no signatures are verified.
12849
yanbzhu6c25e9e2016-01-05 12:52:02 -050012850 There are cases where it is desirable to support multiple key types, e.g. RSA
12851 and ECDSA in the cipher suites offered to the clients. This allows clients
12852 that support EC certificates to be able to use EC ciphers, while
12853 simultaneously supporting older, RSA only clients.
yanbzhud19630c2015-12-14 15:10:25 -050012854
William Lallemandf9ff3ec2020-10-02 17:57:44 +020012855 To achieve this, OpenSSL 1.1.1 is required, you can configure this behavior
12856 by providing one crt entry per certificate type, or by configuring a "cert
12857 bundle" like it was required before HAProxy 1.8. See "ssl-load-extra-files".
yanbzhud19630c2015-12-14 15:10:25 -050012858
Emeric Brunb6dc9342012-09-28 17:55:37 +020012859crt-ignore-err <errors>
Alex Davies0fbf0162013-03-02 16:04:50 +000012860 This setting is only available when support for OpenSSL was built in. Sets a
Davor Ocelice9ed2812017-12-25 17:49:28 +010012861 comma separated list of errorIDs to ignore during verify at depth == 0. If
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012862 set to 'all', all errors are ignored. SSL handshake is not aborted if an error
Alex Davies0fbf0162013-03-02 16:04:50 +000012863 is ignored.
Emeric Brunb6dc9342012-09-28 17:55:37 +020012864
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010012865crt-list <file>
12866 This setting is only available when support for OpenSSL was built in. It
Emmanuel Hocdet98263292016-12-29 18:26:15 +010012867 designates a list of PEM file with an optional ssl configuration and a SNI
12868 filter per certificate, with the following format for each line :
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010012869
Emmanuel Hocdet98263292016-12-29 18:26:15 +010012870 <crtfile> [\[<sslbindconf> ...\]] [[!]<snifilter> ...]
12871
William Lallemand5d036392020-06-30 16:11:36 +020012872 sslbindconf supports "allow-0rtt", "alpn", "ca-file", "ca-verify-file",
12873 "ciphers", "ciphersuites", "crl-file", "curves", "ecdhe", "no-ca-names",
12874 "npn", "verify" configuration. With BoringSSL and Openssl >= 1.1.1
12875 "ssl-min-ver" and "ssl-max-ver" are also supported. It overrides the
12876 configuration set in bind line for the certificate.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010012877
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +020012878 Wildcards are supported in the SNI filter. Negative filter are also supported,
Joao Moraiscaaa46c2020-11-21 07:42:20 -030012879 useful in combination with a wildcard filter to exclude a particular SNI, or
12880 after the first certificate to exclude a pattern from its CN or Subject Alt
12881 Name (SAN). The certificates will be presented to clients who provide a valid
12882 TLS Server Name Indication field matching one of the SNI filters. If no SNI
12883 filter is specified, the CN and SAN are used. This directive may be specified
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +020012884 multiple times. See the "crt" option for more information. The default
12885 certificate is still needed to meet OpenSSL expectations. If it is not used,
12886 the 'strict-sni' option may be used.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010012887
William Lallemandf9ff3ec2020-10-02 17:57:44 +020012888 Multi-cert bundling (see "ssl-load-extra-files") is supported with crt-list,
12889 as long as only the base name is given in the crt-list. SNI filter will do
12890 the same work on all bundled certificates.
yanbzhud19630c2015-12-14 15:10:25 -050012891
William Lallemand7c26ed72020-06-03 17:34:48 +020012892 Empty lines as well as lines beginning with a hash ('#') will be ignored.
12893
Joao Morais5ae6bfc2020-11-24 08:24:30 -030012894 The first declared certificate of a bind line is used as the default
12895 certificate, either from crt or crt-list option, which haproxy should use in
12896 the TLS handshake if no other certificate matches. This certificate will also
12897 be used if the provided SNI matches its CN or SAN, even if a matching SNI
12898 filter is found on any crt-list. The SNI filter !* can be used after the first
12899 declared certificate to not include its CN and SAN in the SNI tree, so it will
12900 never match except if no other certificate matches. This way the first
12901 declared certificate act as a fallback.
Joao Moraiscaaa46c2020-11-21 07:42:20 -030012902
Emmanuel Hocdet98263292016-12-29 18:26:15 +010012903 crt-list file example:
Joao Moraiscaaa46c2020-11-21 07:42:20 -030012904 cert1.pem !*
William Lallemand7c26ed72020-06-03 17:34:48 +020012905 # comment
Emmanuel Hocdet05942112017-02-20 16:11:50 +010012906 cert2.pem [alpn h2,http/1.1]
Emmanuel Hocdet98263292016-12-29 18:26:15 +010012907 certW.pem *.domain.tld !secure.domain.tld
Emmanuel Hocdet05942112017-02-20 16:11:50 +010012908 certS.pem [curves X25519:P-256 ciphers ECDHE-ECDSA-AES256-GCM-SHA384] secure.domain.tld
Emmanuel Hocdet98263292016-12-29 18:26:15 +010012909
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012910defer-accept
12911 Is an optional keyword which is supported only on certain Linux kernels. It
12912 states that a connection will only be accepted once some data arrive on it,
12913 or at worst after the first retransmit. This should be used only on protocols
Davor Ocelice9ed2812017-12-25 17:49:28 +010012914 for which the client talks first (e.g. HTTP). It can slightly improve
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012915 performance by ensuring that most of the request is already available when
12916 the connection is accepted. On the other hand, it will not be able to detect
12917 connections which don't talk. It is important to note that this option is
12918 broken in all kernels up to 2.6.31, as the connection is never accepted until
12919 the client talks. This can cause issues with front firewalls which would see
12920 an established connection while the proxy will only see it in SYN_RECV. This
12921 option is only supported on TCPv4/TCPv6 sockets and ignored by other ones.
12922
William Lallemandf6975e92017-05-26 17:42:10 +020012923expose-fd listeners
12924 This option is only usable with the stats socket. It gives your stats socket
12925 the capability to pass listeners FD to another HAProxy process.
William Lallemande202b1e2017-06-01 17:38:56 +020012926 During a reload with the master-worker mode, the process is automatically
12927 reexecuted adding -x and one of the stats socket with this option.
Davor Ocelice9ed2812017-12-25 17:49:28 +010012928 See also "-x" in the management guide.
William Lallemandf6975e92017-05-26 17:42:10 +020012929
Emeric Brun2cb7ae52012-10-05 14:14:21 +020012930force-sslv3
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012931 This option enforces use of SSLv3 only on SSL connections instantiated from
Emeric Brun2cb7ae52012-10-05 14:14:21 +020012932 this listener. SSLv3 is generally less expensive than the TLS counterparts
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012933 for high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012934 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020012935
12936force-tlsv10
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012937 This option enforces use of TLSv1.0 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012938 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012939 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020012940
12941force-tlsv11
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012942 This option enforces use of TLSv1.1 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012943 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012944 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020012945
12946force-tlsv12
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012947 This option enforces use of TLSv1.2 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012948 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012949 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020012950
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020012951force-tlsv13
12952 This option enforces use of TLSv1.3 only on SSL connections instantiated from
12953 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012954 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020012955
Christopher Faulet31af49d2015-06-09 17:29:50 +020012956generate-certificates
12957 This setting is only available when support for OpenSSL was built in. It
12958 enables the dynamic SSL certificates generation. A CA certificate and its
12959 private key are necessary (see 'ca-sign-file'). When HAProxy is configured as
12960 a transparent forward proxy, SSL requests generate errors because of a common
12961 name mismatch on the certificate presented to the client. With this option
12962 enabled, HAProxy will try to forge a certificate using the SNI hostname
12963 indicated by the client. This is done only if no certificate matches the SNI
12964 hostname (see 'crt-list'). If an error occurs, the default certificate is
12965 used, else the 'strict-sni' option is set.
12966 It can also be used when HAProxy is configured as a reverse proxy to ease the
12967 deployment of an architecture with many backends.
12968
12969 Creating a SSL certificate is an expensive operation, so a LRU cache is used
12970 to store forged certificates (see 'tune.ssl.ssl-ctx-cache-size'). It
Davor Ocelice9ed2812017-12-25 17:49:28 +010012971 increases the HAProxy's memory footprint to reduce latency when the same
Christopher Faulet31af49d2015-06-09 17:29:50 +020012972 certificate is used many times.
12973
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012974gid <gid>
12975 Sets the group of the UNIX sockets to the designated system gid. It can also
12976 be set by default in the global section's "unix-bind" statement. Note that
12977 some platforms simply ignore this. This setting is equivalent to the "group"
12978 setting except that the group ID is used instead of its name. This setting is
12979 ignored by non UNIX sockets.
12980
12981group <group>
12982 Sets the group of the UNIX sockets to the designated system group. It can
12983 also be set by default in the global section's "unix-bind" statement. Note
12984 that some platforms simply ignore this. This setting is equivalent to the
12985 "gid" setting except that the group name is used instead of its gid. This
12986 setting is ignored by non UNIX sockets.
12987
12988id <id>
12989 Fixes the socket ID. By default, socket IDs are automatically assigned, but
12990 sometimes it is more convenient to fix them to ease monitoring. This value
12991 must be strictly positive and unique within the listener/frontend. This
12992 option can only be used when defining only a single socket.
12993
12994interface <interface>
Lukas Tribusfce2e962013-02-12 22:13:19 +010012995 Restricts the socket to a specific interface. When specified, only packets
12996 received from that particular interface are processed by the socket. This is
12997 currently only supported on Linux. The interface must be a primary system
12998 interface, not an aliased interface. It is also possible to bind multiple
12999 frontends to the same address if they are bound to different interfaces. Note
13000 that binding to a network interface requires root privileges. This parameter
Jérôme Magnin61275192018-02-07 11:39:58 +010013001 is only compatible with TCPv4/TCPv6 sockets. When specified, return traffic
13002 uses the same interface as inbound traffic, and its associated routing table,
13003 even if there are explicit routes through different interfaces configured.
13004 This can prove useful to address asymmetric routing issues when the same
13005 client IP addresses need to be able to reach frontends hosted on different
13006 interfaces.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013007
Willy Tarreauabb175f2012-09-24 12:43:26 +020013008level <level>
13009 This setting is used with the stats sockets only to restrict the nature of
13010 the commands that can be issued on the socket. It is ignored by other
13011 sockets. <level> can be one of :
Davor Ocelice9ed2812017-12-25 17:49:28 +010013012 - "user" is the least privileged level; only non-sensitive stats can be
Willy Tarreauabb175f2012-09-24 12:43:26 +020013013 read, and no change is allowed. It would make sense on systems where it
13014 is not easy to restrict access to the socket.
13015 - "operator" is the default level and fits most common uses. All data can
Davor Ocelice9ed2812017-12-25 17:49:28 +010013016 be read, and only non-sensitive changes are permitted (e.g. clear max
Willy Tarreauabb175f2012-09-24 12:43:26 +020013017 counters).
Davor Ocelice9ed2812017-12-25 17:49:28 +010013018 - "admin" should be used with care, as everything is permitted (e.g. clear
Willy Tarreauabb175f2012-09-24 12:43:26 +020013019 all counters).
13020
Andjelko Iharosc4df59e2017-07-20 11:59:48 +020013021severity-output <format>
13022 This setting is used with the stats sockets only to configure severity
13023 level output prepended to informational feedback messages. Severity
13024 level of messages can range between 0 and 7, conforming to syslog
13025 rfc5424. Valid and successful socket commands requesting data
13026 (i.e. "show map", "get acl foo" etc.) will never have a severity level
13027 prepended. It is ignored by other sockets. <format> can be one of :
13028 - "none" (default) no severity level is prepended to feedback messages.
13029 - "number" severity level is prepended as a number.
13030 - "string" severity level is prepended as a string following the
13031 rfc5424 convention.
13032
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013033maxconn <maxconn>
13034 Limits the sockets to this number of concurrent connections. Extraneous
13035 connections will remain in the system's backlog until a connection is
13036 released. If unspecified, the limit will be the same as the frontend's
13037 maxconn. Note that in case of port ranges or multiple addresses, the same
13038 value will be applied to each socket. This setting enables different
13039 limitations on expensive sockets, for instance SSL entries which may easily
13040 eat all memory.
13041
13042mode <mode>
13043 Sets the octal mode used to define access permissions on the UNIX socket. It
13044 can also be set by default in the global section's "unix-bind" statement.
13045 Note that some platforms simply ignore this. This setting is ignored by non
13046 UNIX sockets.
13047
13048mss <maxseg>
13049 Sets the TCP Maximum Segment Size (MSS) value to be advertised on incoming
13050 connections. This can be used to force a lower MSS for certain specific
13051 ports, for instance for connections passing through a VPN. Note that this
13052 relies on a kernel feature which is theoretically supported under Linux but
13053 was buggy in all versions prior to 2.6.28. It may or may not work on other
13054 operating systems. It may also not change the advertised value but change the
13055 effective size of outgoing segments. The commonly advertised value for TCPv4
13056 over Ethernet networks is 1460 = 1500(MTU) - 40(IP+TCP). If this value is
13057 positive, it will be used as the advertised MSS. If it is negative, it will
13058 indicate by how much to reduce the incoming connection's advertised MSS for
13059 outgoing segments. This parameter is only compatible with TCP v4/v6 sockets.
13060
13061name <name>
13062 Sets an optional name for these sockets, which will be reported on the stats
13063 page.
13064
Willy Tarreaud72f0f32015-10-13 14:50:22 +020013065namespace <name>
13066 On Linux, it is possible to specify which network namespace a socket will
13067 belong to. This directive makes it possible to explicitly bind a listener to
13068 a namespace different from the default one. Please refer to your operating
13069 system's documentation to find more details about network namespaces.
13070
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013071nice <nice>
13072 Sets the 'niceness' of connections initiated from the socket. Value must be
13073 in the range -1024..1024 inclusive, and defaults to zero. Positive values
13074 means that such connections are more friendly to others and easily offer
13075 their place in the scheduler. On the opposite, negative values mean that
13076 connections want to run with a higher priority than others. The difference
13077 only happens under high loads when the system is close to saturation.
13078 Negative values are appropriate for low-latency or administration services,
13079 and high values are generally recommended for CPU intensive tasks such as SSL
13080 processing or bulk transfers which are less sensible to latency. For example,
13081 it may make sense to use a positive value for an SMTP socket and a negative
13082 one for an RDP socket.
13083
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020013084no-ca-names
13085 This setting is only available when support for OpenSSL was built in. It
13086 prevents from send CA names in server hello message when ca-file is used.
Emmanuel Hocdet842e94e2019-12-16 16:39:17 +010013087 Use "ca-verify-file" instead of "ca-file" with "no-ca-names".
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020013088
Emeric Brun9b3009b2012-10-05 11:55:06 +020013089no-sslv3
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013090 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013091 disables support for SSLv3 on any sockets instantiated from the listener when
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013092 SSL is supported. Note that SSLv2 is forced disabled in the code and cannot
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013093 be enabled using any configuration option. This option is also available on
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013094 global statement "ssl-default-bind-options". Use "ssl-min-ver" and
13095 "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013096
Emeric Brun90ad8722012-10-02 14:00:59 +020013097no-tls-tickets
13098 This setting is only available when support for OpenSSL was built in. It
13099 disables the stateless session resumption (RFC 5077 TLS Ticket
13100 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013101 session resumption is more expensive in CPU usage. This option is also
13102 available on global statement "ssl-default-bind-options".
Lukas Tribusbdb386d2020-03-10 00:56:09 +010013103 The TLS ticket mechanism is only used up to TLS 1.2.
13104 Forward Secrecy is compromised with TLS tickets, unless ticket keys
13105 are periodically rotated (via reload or by using "tls-ticket-keys").
Emeric Brun90ad8722012-10-02 14:00:59 +020013106
Emeric Brun9b3009b2012-10-05 11:55:06 +020013107no-tlsv10
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013108 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013109 disables support for TLSv1.0 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020013110 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013111 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013112 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
13113 and "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013114
Emeric Brun9b3009b2012-10-05 11:55:06 +020013115no-tlsv11
Emeric Brunf5da4932012-09-28 19:42:54 +020013116 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013117 disables support for TLSv1.1 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020013118 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013119 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013120 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
13121 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020013122
Emeric Brun9b3009b2012-10-05 11:55:06 +020013123no-tlsv12
Emeric Brunf5da4932012-09-28 19:42:54 +020013124 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013125 disables support for TLSv1.2 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020013126 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013127 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013128 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
13129 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020013130
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020013131no-tlsv13
13132 This setting is only available when support for OpenSSL was built in. It
13133 disables support for TLSv1.3 on any sockets instantiated from the listener
13134 when SSL is supported. Note that SSLv2 is forced disabled in the code and
13135 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013136 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
13137 and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020013138
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020013139npn <protocols>
13140 This enables the NPN TLS extension and advertises the specified protocol list
13141 as supported on top of NPN. The protocol list consists in a comma-delimited
13142 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
John Roeslerfb2fce12019-07-10 15:45:51 -050013143 This requires that the SSL library is built with support for TLS extensions
Willy Tarreauab861d32013-04-02 02:30:41 +020013144 enabled (check with haproxy -vv). Note that the NPN extension has been
Willy Tarreau95c4e142017-11-26 12:18:55 +010013145 replaced with the ALPN extension (see the "alpn" keyword), though this one is
13146 only available starting with OpenSSL 1.0.2. If HTTP/2 is desired on an older
13147 version of OpenSSL, NPN might still be used as most clients still support it
13148 at the time of writing this. It is possible to enable both NPN and ALPN
13149 though it probably doesn't make any sense out of testing.
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020013150
Lukas Tribus53ae85c2017-05-04 15:45:40 +000013151prefer-client-ciphers
13152 Use the client's preference when selecting the cipher suite, by default
13153 the server's preference is enforced. This option is also available on
13154 global statement "ssl-default-bind-options".
Lukas Tribus926594f2018-05-18 17:55:57 +020013155 Note that with OpenSSL >= 1.1.1 ChaCha20-Poly1305 is reprioritized anyway
13156 (without setting this option), if a ChaCha20-Poly1305 cipher is at the top of
13157 the client cipher list.
Lukas Tribus53ae85c2017-05-04 15:45:40 +000013158
Christopher Fauletc644fa92017-11-23 22:44:11 +010013159process <process-set>[/<thread-set>]
Willy Tarreaua36b3242019-02-02 13:14:34 +010013160 This restricts the list of processes or threads on which this listener is
Christopher Fauletc644fa92017-11-23 22:44:11 +010013161 allowed to run. It does not enforce any process but eliminates those which do
Davor Ocelice9ed2812017-12-25 17:49:28 +010013162 not match. If the frontend uses a "bind-process" setting, the intersection
Christopher Fauletc644fa92017-11-23 22:44:11 +010013163 between the two is applied. If in the end the listener is not allowed to run
13164 on any remaining process, a warning is emitted, and the listener will either
13165 run on the first process of the listener if a single process was specified,
13166 or on all of its processes if multiple processes were specified. If a thread
Davor Ocelice9ed2812017-12-25 17:49:28 +010013167 set is specified, it limits the threads allowed to process incoming
Willy Tarreaua36b3242019-02-02 13:14:34 +010013168 connections for this listener, for the the process set. If multiple processes
13169 and threads are configured, a warning is emitted, as it either results from a
13170 configuration error or a misunderstanding of these models. For the unlikely
13171 case where several ranges are needed, this directive may be repeated.
13172 <process-set> and <thread-set> must use the format
Christopher Fauletc644fa92017-11-23 22:44:11 +010013173
13174 all | odd | even | number[-[number]]
13175
13176 Ranges can be partially defined. The higher bound can be omitted. In such
13177 case, it is replaced by the corresponding maximum value. The main purpose of
13178 this directive is to be used with the stats sockets and have one different
13179 socket per process. The second purpose is to have multiple bind lines sharing
13180 the same IP:port but not the same process in a listener, so that the system
13181 can distribute the incoming connections into multiple queues and allow a
13182 smoother inter-process load balancing. Currently Linux 3.9 and above is known
13183 for supporting this. See also "bind-process" and "nbproc".
Willy Tarreau6ae1ba62014-05-07 19:01:58 +020013184
Christopher Fauleta717b992018-04-10 14:43:00 +020013185proto <name>
13186 Forces the multiplexer's protocol to use for the incoming connections. It
13187 must be compatible with the mode of the frontend (TCP or HTTP). It must also
13188 be usable on the frontend side. The list of available protocols is reported
13189 in haproxy -vv.
Daniel Corbett67a82712020-07-06 23:01:19 -040013190 Idea behind this option is to bypass the selection of the best multiplexer's
Christopher Fauleta717b992018-04-10 14:43:00 +020013191 protocol for all connections instantiated from this listening socket. For
Joseph Herlant71b4b152018-11-13 16:55:16 -080013192 instance, it is possible to force the http/2 on clear TCP by specifying "proto
Christopher Fauleta717b992018-04-10 14:43:00 +020013193 h2" on the bind line.
13194
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013195ssl
13196 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013197 enables SSL deciphering on connections instantiated from this listener. A
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013198 certificate is necessary (see "crt" above). All contents in the buffers will
13199 appear in clear text, so that ACLs and HTTP processing will only have access
Emmanuel Hocdetbd695fe2017-05-15 15:53:41 +020013200 to deciphered contents. SSLv3 is disabled per default, use "ssl-min-ver SSLv3"
13201 to enable it.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013202
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013203ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
13204 This option enforces use of <version> or lower on SSL connections instantiated
William Lallemand50df1cb2020-06-02 10:52:24 +020013205 from this listener. Using this setting without "ssl-min-ver" can be
13206 ambiguous because the default ssl-min-ver value could change in future HAProxy
13207 versions. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013208 "ssl-default-bind-options". See also "ssl-min-ver".
13209
13210ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
William Lallemand50df1cb2020-06-02 10:52:24 +020013211 This option enforces use of <version> or upper on SSL connections
13212 instantiated from this listener. The default value is "TLSv1.2". This option
13213 is also available on global statement "ssl-default-bind-options".
13214 See also "ssl-max-ver".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013215
Emmanuel Hocdet65623372013-01-24 17:17:15 +010013216strict-sni
13217 This setting is only available when support for OpenSSL was built in. The
13218 SSL/TLS negotiation is allow only if the client provided an SNI which match
13219 a certificate. The default certificate is not used.
13220 See the "crt" option for more information.
13221
Willy Tarreau2af207a2015-02-04 00:45:58 +010013222tcp-ut <delay>
Tim Düsterhus4896c442016-11-29 02:15:19 +010013223 Sets the TCP User Timeout for all incoming connections instantiated from this
Willy Tarreau2af207a2015-02-04 00:45:58 +010013224 listening socket. This option is available on Linux since version 2.6.37. It
13225 allows haproxy to configure a timeout for sockets which contain data not
Davor Ocelice9ed2812017-12-25 17:49:28 +010013226 receiving an acknowledgment for the configured delay. This is especially
Willy Tarreau2af207a2015-02-04 00:45:58 +010013227 useful on long-lived connections experiencing long idle periods such as
13228 remote terminals or database connection pools, where the client and server
13229 timeouts must remain high to allow a long period of idle, but where it is
13230 important to detect that the client has disappeared in order to release all
13231 resources associated with its connection (and the server's session). The
13232 argument is a delay expressed in milliseconds by default. This only works
13233 for regular TCP connections, and is ignored for other protocols.
13234
Willy Tarreau1c862c52012-10-05 16:21:00 +020013235tfo
Lukas Tribus0defb902013-02-13 23:35:39 +010013236 Is an optional keyword which is supported only on Linux kernels >= 3.7. It
Willy Tarreau1c862c52012-10-05 16:21:00 +020013237 enables TCP Fast Open on the listening socket, which means that clients which
13238 support this feature will be able to send a request and receive a response
13239 during the 3-way handshake starting from second connection, thus saving one
13240 round-trip after the first connection. This only makes sense with protocols
13241 that use high connection rates and where each round trip matters. This can
13242 possibly cause issues with many firewalls which do not accept data on SYN
13243 packets, so this option should only be enabled once well tested. This option
Lukas Tribus0999f762013-04-02 16:43:24 +020013244 is only supported on TCPv4/TCPv6 sockets and ignored by other ones. You may
13245 need to build HAProxy with USE_TFO=1 if your libc doesn't define
13246 TCP_FASTOPEN.
Willy Tarreau1c862c52012-10-05 16:21:00 +020013247
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010013248tls-ticket-keys <keyfile>
13249 Sets the TLS ticket keys file to load the keys from. The keys need to be 48
Emeric Brun9e754772019-01-10 17:51:55 +010013250 or 80 bytes long, depending if aes128 or aes256 is used, encoded with base64
13251 with one line per key (ex. openssl rand 80 | openssl base64 -A | xargs echo).
13252 The first key determines the key length used for next keys: you can't mix
13253 aes128 and aes256 keys. Number of keys is specified by the TLS_TICKETS_NO
13254 build option (default 3) and at least as many keys need to be present in
13255 the file. Last TLS_TICKETS_NO keys will be used for decryption and the
13256 penultimate one for encryption. This enables easy key rotation by just
13257 appending new key to the file and reloading the process. Keys must be
13258 periodically rotated (ex. every 12h) or Perfect Forward Secrecy is
13259 compromised. It is also a good idea to keep the keys off any permanent
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010013260 storage such as hard drives (hint: use tmpfs and don't swap those files).
13261 Lifetime hint can be changed using tune.ssl.timeout.
13262
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013263transparent
13264 Is an optional keyword which is supported only on certain Linux kernels. It
13265 indicates that the addresses will be bound even if they do not belong to the
13266 local machine, and that packets targeting any of these addresses will be
13267 intercepted just as if the addresses were locally configured. This normally
13268 requires that IP forwarding is enabled. Caution! do not use this with the
13269 default address '*', as it would redirect any traffic for the specified port.
13270 This keyword is available only when HAProxy is built with USE_LINUX_TPROXY=1.
13271 This parameter is only compatible with TCPv4 and TCPv6 sockets, depending on
13272 kernel version. Some distribution kernels include backports of the feature,
13273 so check for support with your vendor.
13274
Willy Tarreau77e3af92012-11-24 15:07:23 +010013275v4v6
13276 Is an optional keyword which is supported only on most recent systems
13277 including Linux kernels >= 2.4.21. It is used to bind a socket to both IPv4
13278 and IPv6 when it uses the default address. Doing so is sometimes necessary
13279 on systems which bind to IPv6 only by default. It has no effect on non-IPv6
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013280 sockets, and is overridden by the "v6only" option.
Willy Tarreau77e3af92012-11-24 15:07:23 +010013281
Willy Tarreau9b6700f2012-11-24 11:55:28 +010013282v6only
13283 Is an optional keyword which is supported only on most recent systems
13284 including Linux kernels >= 2.4.21. It is used to bind a socket to IPv6 only
13285 when it uses the default address. Doing so is sometimes preferred to doing it
Willy Tarreau77e3af92012-11-24 15:07:23 +010013286 system-wide as it is per-listener. It has no effect on non-IPv6 sockets and
13287 has precedence over the "v4v6" option.
Willy Tarreau9b6700f2012-11-24 11:55:28 +010013288
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013289uid <uid>
13290 Sets the owner of the UNIX sockets to the designated system uid. It can also
13291 be set by default in the global section's "unix-bind" statement. Note that
13292 some platforms simply ignore this. This setting is equivalent to the "user"
13293 setting except that the user numeric ID is used instead of its name. This
13294 setting is ignored by non UNIX sockets.
13295
13296user <user>
13297 Sets the owner of the UNIX sockets to the designated system user. It can also
13298 be set by default in the global section's "unix-bind" statement. Note that
13299 some platforms simply ignore this. This setting is equivalent to the "uid"
13300 setting except that the user name is used instead of its uid. This setting is
13301 ignored by non UNIX sockets.
13302
Emeric Brun1a073b42012-09-28 17:07:34 +020013303verify [none|optional|required]
13304 This setting is only available when support for OpenSSL was built in. If set
13305 to 'none', client certificate is not requested. This is the default. In other
13306 cases, a client certificate is requested. If the client does not provide a
13307 certificate after the request and if 'verify' is set to 'required', then the
13308 handshake is aborted, while it would have succeeded if set to 'optional'. The
Emeric Brunfd33a262012-10-11 16:28:27 +020013309 certificate provided by the client is always verified using CAs from
13310 'ca-file' and optional CRLs from 'crl-file'. On verify failure the handshake
13311 is aborted, regardless of the 'verify' option, unless the error code exactly
13312 matches one of those listed with 'ca-ignore-err' or 'crt-ignore-err'.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020013313
Willy Tarreaub6205fd2012-09-24 12:27:33 +0200133145.2. Server and default-server options
Cyril Bontéf0c60612010-02-06 14:44:47 +010013315------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020013316
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010013317The "server" and "default-server" keywords support a certain number of settings
13318which are all passed as arguments on the server line. The order in which those
13319arguments appear does not count, and they are all optional. Some of those
13320settings are single words (booleans) while others expect one or several values
13321after them. In this case, the values must immediately follow the setting name.
13322Except default-server, all those settings must be specified after the server's
13323address if they are used:
Willy Tarreau6a06a402007-07-15 20:15:28 +020013324
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013325 server <name> <address>[:port] [settings ...]
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010013326 default-server [settings ...]
Willy Tarreau6a06a402007-07-15 20:15:28 +020013327
Frédéric Lécailled2376272017-03-21 18:52:12 +010013328Note that all these settings are supported both by "server" and "default-server"
13329keywords, except "id" which is only supported by "server".
13330
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013331The currently supported settings are the following ones.
Willy Tarreau0ba27502007-12-24 16:55:16 +010013332
Willy Tarreauceb4ac92012-04-28 00:41:46 +020013333addr <ipv4|ipv6>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013334 Using the "addr" parameter, it becomes possible to use a different IP address
Baptiste Assmann13f83532016-03-06 23:14:36 +010013335 to send health-checks or to probe the agent-check. On some servers, it may be
13336 desirable to dedicate an IP address to specific component able to perform
13337 complex tests which are more suitable to health-checks than the application.
13338 This parameter is ignored if the "check" parameter is not set. See also the
13339 "port" parameter.
Willy Tarreau6a06a402007-07-15 20:15:28 +020013340
Simon Hormand60d6912013-11-25 10:46:36 +090013341agent-check
13342 Enable an auxiliary agent check which is run independently of a regular
Willy Tarreau81f5d942013-12-09 20:51:51 +010013343 health check. An agent health check is performed by making a TCP connection
Willy Tarreau7a0139e2018-12-16 08:42:56 +010013344 to the port set by the "agent-port" parameter and reading an ASCII string
13345 terminated by the first '\r' or '\n' met. The string is made of a series of
13346 words delimited by spaces, tabs or commas in any order, each consisting of :
Simon Hormand60d6912013-11-25 10:46:36 +090013347
Willy Tarreau81f5d942013-12-09 20:51:51 +010013348 - An ASCII representation of a positive integer percentage, e.g. "75%".
Simon Hormand60d6912013-11-25 10:46:36 +090013349 Values in this format will set the weight proportional to the initial
Willy Tarreauc5af3a62014-10-07 15:27:33 +020013350 weight of a server as configured when haproxy starts. Note that a zero
13351 weight is reported on the stats page as "DRAIN" since it has the same
13352 effect on the server (it's removed from the LB farm).
Simon Hormand60d6912013-11-25 10:46:36 +090013353
Davor Ocelice9ed2812017-12-25 17:49:28 +010013354 - The string "maxconn:" followed by an integer (no space between). Values
13355 in this format will set the maxconn of a server. The maximum number of
13356 connections advertised needs to be multiplied by the number of load
13357 balancers and different backends that use this health check to get the
13358 total number of connections the server might receive. Example: maxconn:30
Nenad Merdanovic174dd372016-04-24 23:10:06 +020013359
Willy Tarreau81f5d942013-12-09 20:51:51 +010013360 - The word "ready". This will turn the server's administrative state to the
Davor Ocelice9ed2812017-12-25 17:49:28 +010013361 READY mode, thus canceling any DRAIN or MAINT state
Simon Hormand60d6912013-11-25 10:46:36 +090013362
Willy Tarreau81f5d942013-12-09 20:51:51 +010013363 - The word "drain". This will turn the server's administrative state to the
13364 DRAIN mode, thus it will not accept any new connections other than those
13365 that are accepted via persistence.
Simon Hormand60d6912013-11-25 10:46:36 +090013366
Willy Tarreau81f5d942013-12-09 20:51:51 +010013367 - The word "maint". This will turn the server's administrative state to the
13368 MAINT mode, thus it will not accept any new connections at all, and health
13369 checks will be stopped.
Simon Hormand60d6912013-11-25 10:46:36 +090013370
William Dauchyf8e795c2020-09-26 13:35:51 +020013371 - The words "down", "fail", or "stopped", optionally followed by a
Willy Tarreau81f5d942013-12-09 20:51:51 +010013372 description string after a sharp ('#'). All of these mark the server's
13373 operating state as DOWN, but since the word itself is reported on the stats
13374 page, the difference allows an administrator to know if the situation was
13375 expected or not : the service may intentionally be stopped, may appear up
Davor Ocelice9ed2812017-12-25 17:49:28 +010013376 but fail some validity tests, or may be seen as down (e.g. missing process,
Willy Tarreau81f5d942013-12-09 20:51:51 +010013377 or port not responding).
Simon Hormand60d6912013-11-25 10:46:36 +090013378
Willy Tarreau81f5d942013-12-09 20:51:51 +010013379 - The word "up" sets back the server's operating state as UP if health checks
13380 also report that the service is accessible.
Simon Hormand60d6912013-11-25 10:46:36 +090013381
Willy Tarreau81f5d942013-12-09 20:51:51 +010013382 Parameters which are not advertised by the agent are not changed. For
13383 example, an agent might be designed to monitor CPU usage and only report a
13384 relative weight and never interact with the operating status. Similarly, an
13385 agent could be designed as an end-user interface with 3 radio buttons
13386 allowing an administrator to change only the administrative state. However,
13387 it is important to consider that only the agent may revert its own actions,
13388 so if a server is set to DRAIN mode or to DOWN state using the agent, the
13389 agent must implement the other equivalent actions to bring the service into
13390 operations again.
Simon Hormand60d6912013-11-25 10:46:36 +090013391
Simon Horman2f1f9552013-11-25 10:46:37 +090013392 Failure to connect to the agent is not considered an error as connectivity
13393 is tested by the regular health check which is enabled by the "check"
Willy Tarreau81f5d942013-12-09 20:51:51 +010013394 parameter. Warning though, it is not a good idea to stop an agent after it
13395 reports "down", since only an agent reporting "up" will be able to turn the
13396 server up again. Note that the CLI on the Unix stats socket is also able to
Willy Tarreau989222a2016-01-15 10:26:26 +010013397 force an agent's result in order to work around a bogus agent if needed.
Simon Horman2f1f9552013-11-25 10:46:37 +090013398
Willy Tarreau81f5d942013-12-09 20:51:51 +010013399 Requires the "agent-port" parameter to be set. See also the "agent-inter"
Frédéric Lécailled2376272017-03-21 18:52:12 +010013400 and "no-agent-check" parameters.
Simon Hormand60d6912013-11-25 10:46:36 +090013401
James Brown55f9ff12015-10-21 18:19:05 -070013402agent-send <string>
13403 If this option is specified, haproxy will send the given string (verbatim)
13404 to the agent server upon connection. You could, for example, encode
13405 the backend name into this string, which would enable your agent to send
13406 different responses based on the backend. Make sure to include a '\n' if
13407 you want to terminate your request with a newline.
13408
Simon Hormand60d6912013-11-25 10:46:36 +090013409agent-inter <delay>
13410 The "agent-inter" parameter sets the interval between two agent checks
13411 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
13412
13413 Just as with every other time-based parameter, it may be entered in any
13414 other explicit unit among { us, ms, s, m, h, d }. The "agent-inter"
13415 parameter also serves as a timeout for agent checks "timeout check" is
13416 not set. In order to reduce "resonance" effects when multiple servers are
13417 hosted on the same hardware, the agent and health checks of all servers
13418 are started with a small time offset between them. It is also possible to
13419 add some random noise in the agent and health checks interval using the
13420 global "spread-checks" keyword. This makes sense for instance when a lot
13421 of backends use the same servers.
13422
13423 See also the "agent-check" and "agent-port" parameters.
13424
Misiek768d8602017-01-09 09:52:43 +010013425agent-addr <addr>
13426 The "agent-addr" parameter sets address for agent check.
13427
13428 You can offload agent-check to another target, so you can make single place
13429 managing status and weights of servers defined in haproxy in case you can't
13430 make self-aware and self-managing services. You can specify both IP or
13431 hostname, it will be resolved.
13432
Simon Hormand60d6912013-11-25 10:46:36 +090013433agent-port <port>
13434 The "agent-port" parameter sets the TCP port used for agent checks.
13435
13436 See also the "agent-check" and "agent-inter" parameters.
13437
Olivier Houchard8cb2d2e2019-05-06 18:58:48 +020013438allow-0rtt
13439 Allow sending early data to the server when using TLS 1.3.
Olivier Houchard22c9b442019-05-06 19:01:04 +020013440 Note that early data will be sent only if the client used early data, or
13441 if the backend uses "retry-on" with the "0rtt-rejected" keyword.
Olivier Houchard8cb2d2e2019-05-06 18:58:48 +020013442
Olivier Houchardc7566002018-11-20 23:33:50 +010013443alpn <protocols>
13444 This enables the TLS ALPN extension and advertises the specified protocol
13445 list as supported on top of ALPN. The protocol list consists in a comma-
13446 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
John Roeslerfb2fce12019-07-10 15:45:51 -050013447 quotes). This requires that the SSL library is built with support for TLS
Olivier Houchardc7566002018-11-20 23:33:50 +010013448 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
13449 initial NPN extension. ALPN is required to connect to HTTP/2 servers.
13450 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
13451 now obsolete NPN extension.
13452 If both HTTP/2 and HTTP/1.1 are expected to be supported, both versions can
13453 be advertised, in order of preference, like below :
13454
13455 server 127.0.0.1:443 ssl crt pub.pem alpn h2,http/1.1
13456
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013457backup
13458 When "backup" is present on a server line, the server is only used in load
13459 balancing when all other non-backup servers are unavailable. Requests coming
13460 with a persistence cookie referencing the server will always be served
13461 though. By default, only the first operational backup server is used, unless
Frédéric Lécailled2376272017-03-21 18:52:12 +010013462 the "allbackups" option is set in the backend. See also the "no-backup" and
13463 "allbackups" options.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013464
Emeric Brunef42d922012-10-11 16:11:36 +020013465ca-file <cafile>
13466 This setting is only available when support for OpenSSL was built in. It
13467 designates a PEM file from which to load CA certificates used to verify
13468 server's certificate.
13469
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013470check
Jerome Magnin90702bc2020-04-26 14:23:04 +020013471 This option enables health checks on a server:
13472 - when not set, no health checking is performed, and the server is always
13473 considered available.
13474 - when set and no other check method is configured, the server is considered
13475 available when a connection can be established at the highest configured
13476 transport layer. This means TCP by default, or SSL/TLS when "ssl" or
13477 "check-ssl" are set, both possibly combined with connection prefixes such
13478 as a PROXY protocol header when "send-proxy" or "check-send-proxy" are
13479 set.
13480 - when set and an application-level health check is defined, the
13481 application-level exchanges are performed on top of the configured
13482 transport layer and the server is considered available if all of the
13483 exchanges succeed.
13484
13485 By default, health checks are performed on the same address and port as
13486 configured on the server, using the same encapsulation parameters (SSL/TLS,
13487 proxy-protocol header, etc... ). It is possible to change the destination
13488 address using "addr" and the port using "port". When done, it is assumed the
13489 server isn't checked on the service port, and configured encapsulation
Ilya Shipitsin4329a9a2020-05-05 21:17:10 +050013490 parameters are not reused. One must explicitly set "check-send-proxy" to send
Jerome Magnin90702bc2020-04-26 14:23:04 +020013491 connection headers, "check-ssl" to use SSL/TLS.
13492
13493 When "sni" or "alpn" are set on the server line, their value is not used for
13494 health checks and one must use "check-sni" or "check-alpn".
13495
13496 The default source address for health check traffic is the same as the one
13497 defined in the backend. It can be changed with the "source" keyword.
13498
13499 The interval between checks can be set using the "inter" keyword, and the
13500 "rise" and "fall" keywords can be used to define how many successful or
13501 failed health checks are required to flag a server available or not
13502 available.
13503
13504 Optional application-level health checks can be configured with "option
13505 httpchk", "option mysql-check" "option smtpchk", "option pgsql-check",
13506 "option ldap-check", or "option redis-check".
13507
13508 Example:
13509 # simple tcp check
13510 backend foo
13511 server s1 192.168.0.1:80 check
13512 # this does a tcp connect + tls handshake
13513 backend foo
13514 server s1 192.168.0.1:443 ssl check
13515 # simple tcp check is enough for check success
13516 backend foo
13517 option tcp-check
13518 tcp-check connect
13519 server s1 192.168.0.1:443 ssl check
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013520
Willy Tarreau6c16adc2012-10-05 00:04:16 +020013521check-send-proxy
13522 This option forces emission of a PROXY protocol line with outgoing health
13523 checks, regardless of whether the server uses send-proxy or not for the
13524 normal traffic. By default, the PROXY protocol is enabled for health checks
13525 if it is already enabled for normal traffic and if no "port" nor "addr"
13526 directive is present. However, if such a directive is present, the
13527 "check-send-proxy" option needs to be used to force the use of the
13528 protocol. See also the "send-proxy" option for more information.
13529
Olivier Houchard92150142018-12-21 19:47:01 +010013530check-alpn <protocols>
13531 Defines which protocols to advertise with ALPN. The protocol list consists in
13532 a comma-delimited list of protocol names, for instance: "http/1.1,http/1.0"
13533 (without quotes). If it is not set, the server ALPN is used.
13534
Christopher Fauletedc6ed92020-04-23 16:27:59 +020013535check-proto <name>
13536 Forces the multiplexer's protocol to use for the server's health-check
13537 connections. It must be compatible with the health-check type (TCP or
13538 HTTP). It must also be usable on the backend side. The list of available
13539 protocols is reported in haproxy -vv.
Daniel Corbett67a82712020-07-06 23:01:19 -040013540 Idea behind this option is to bypass the selection of the best multiplexer's
Christopher Fauletedc6ed92020-04-23 16:27:59 +020013541 protocol for health-check connections established to this server.
13542 If not defined, the server one will be used, if set.
13543
Jérôme Magninae9bb762018-12-09 16:08:26 +010013544check-sni <sni>
Olivier Houchard9130a962017-10-17 17:33:43 +020013545 This option allows you to specify the SNI to be used when doing health checks
Jérôme Magninae9bb762018-12-09 16:08:26 +010013546 over SSL. It is only possible to use a string to set <sni>. If you want to
13547 set a SNI for proxied traffic, see "sni".
Olivier Houchard9130a962017-10-17 17:33:43 +020013548
Willy Tarreau763a95b2012-10-04 23:15:39 +020013549check-ssl
13550 This option forces encryption of all health checks over SSL, regardless of
13551 whether the server uses SSL or not for the normal traffic. This is generally
13552 used when an explicit "port" or "addr" directive is specified and SSL health
13553 checks are not inherited. It is important to understand that this option
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013554 inserts an SSL transport layer below the checks, so that a simple TCP connect
Willy Tarreau763a95b2012-10-04 23:15:39 +020013555 check becomes an SSL connect, which replaces the old ssl-hello-chk. The most
13556 common use is to send HTTPS checks by combining "httpchk" with SSL checks.
Davor Ocelice9ed2812017-12-25 17:49:28 +010013557 All SSL settings are common to health checks and traffic (e.g. ciphers).
Frédéric Lécailled2376272017-03-21 18:52:12 +010013558 See the "ssl" option for more information and "no-check-ssl" to disable
13559 this option.
Willy Tarreau763a95b2012-10-04 23:15:39 +020013560
Alexander Liu2a54bb72019-05-22 19:44:48 +080013561check-via-socks4
John Roeslerfb2fce12019-07-10 15:45:51 -050013562 This option enables outgoing health checks using upstream socks4 proxy. By
Alexander Liu2a54bb72019-05-22 19:44:48 +080013563 default, the health checks won't go through socks tunnel even it was enabled
13564 for normal traffic.
13565
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020013566ciphers <ciphers>
Dirkjan Bussink415150f2018-09-14 11:14:21 +020013567 This setting is only available when support for OpenSSL was built in. This
13568 option sets the string describing the list of cipher algorithms that is
13569 negotiated during the SSL/TLS handshake with the server. The format of the
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000013570 string is defined in "man 1 ciphers" from OpenSSL man pages. For background
13571 information and recommendations see e.g.
13572 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
13573 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
13574 cipher configuration, please check the "ciphersuites" keyword.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020013575
Dirkjan Bussink415150f2018-09-14 11:14:21 +020013576ciphersuites <ciphersuites>
13577 This setting is only available when support for OpenSSL was built in and
13578 OpenSSL 1.1.1 or later was used to build HAProxy. This option sets the string
13579 describing the list of cipher algorithms that is negotiated during the TLS
13580 1.3 handshake with the server. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000013581 "man 1 ciphers" from OpenSSL man pages under the "ciphersuites" section.
13582 For cipher configuration for TLSv1.2 and earlier, please check the "ciphers"
13583 keyword.
Dirkjan Bussink415150f2018-09-14 11:14:21 +020013584
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013585cookie <value>
13586 The "cookie" parameter sets the cookie value assigned to the server to
13587 <value>. This value will be checked in incoming requests, and the first
13588 operational server possessing the same value will be selected. In return, in
13589 cookie insertion or rewrite modes, this value will be assigned to the cookie
13590 sent to the client. There is nothing wrong in having several servers sharing
13591 the same cookie value, and it is in fact somewhat common between normal and
13592 backup servers. See also the "cookie" keyword in backend section.
13593
Emeric Brunef42d922012-10-11 16:11:36 +020013594crl-file <crlfile>
13595 This setting is only available when support for OpenSSL was built in. It
13596 designates a PEM file from which to load certificate revocation list used
13597 to verify server's certificate.
13598
Emeric Bruna7aa3092012-10-26 12:58:00 +020013599crt <cert>
13600 This setting is only available when support for OpenSSL was built in.
13601 It designates a PEM file from which to load both a certificate and the
13602 associated private key. This file can be built by concatenating both PEM
13603 files into one. This certificate will be sent if the server send a client
13604 certificate request.
13605
Willy Tarreau96839092010-03-29 10:02:24 +020013606disabled
13607 The "disabled" keyword starts the server in the "disabled" state. That means
13608 that it is marked down in maintenance mode, and no connection other than the
13609 ones allowed by persist mode will reach it. It is very well suited to setup
13610 new servers, because normal traffic will never reach them, while it is still
13611 possible to test the service by making use of the force-persist mechanism.
Frédéric Lécailled2376272017-03-21 18:52:12 +010013612 See also "enabled" setting.
Willy Tarreau96839092010-03-29 10:02:24 +020013613
Frédéric Lécailled2376272017-03-21 18:52:12 +010013614enabled
13615 This option may be used as 'server' setting to reset any 'disabled'
13616 setting which would have been inherited from 'default-server' directive as
13617 default value.
13618 It may also be used as 'default-server' setting to reset any previous
13619 'default-server' 'disabled' setting.
Willy Tarreau96839092010-03-29 10:02:24 +020013620
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013621error-limit <count>
Willy Tarreau983e01e2010-01-11 18:42:06 +010013622 If health observing is enabled, the "error-limit" parameter specifies the
13623 number of consecutive errors that triggers event selected by the "on-error"
13624 option. By default it is set to 10 consecutive errors.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010013625
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013626 See also the "check", "error-limit" and "on-error".
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010013627
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013628fall <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013629 The "fall" parameter states that a server will be considered as dead after
13630 <count> consecutive unsuccessful health checks. This value defaults to 3 if
13631 unspecified. See also the "check", "inter" and "rise" parameters.
13632
Emeric Brun8694b9a2012-10-05 14:39:07 +020013633force-sslv3
13634 This option enforces use of SSLv3 only when SSL is used to communicate with
13635 the server. SSLv3 is generally less expensive than the TLS counterparts for
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013636 high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013637 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020013638
13639force-tlsv10
13640 This option enforces use of TLSv1.0 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013641 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013642 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020013643
13644force-tlsv11
13645 This option enforces use of TLSv1.1 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013646 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013647 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020013648
13649force-tlsv12
13650 This option enforces use of TLSv1.2 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013651 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013652 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020013653
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020013654force-tlsv13
13655 This option enforces use of TLSv1.3 only when SSL is used to communicate with
13656 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013657 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020013658
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013659id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +020013660 Set a persistent ID for the server. This ID must be positive and unique for
13661 the proxy. An unused ID will automatically be assigned if unset. The first
13662 assigned value will be 1. This ID is currently only returned in statistics.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013663
Willy Tarreau6a031d12016-11-07 19:42:35 +010013664init-addr {last | libc | none | <ip>},[...]*
13665 Indicate in what order the server's address should be resolved upon startup
13666 if it uses an FQDN. Attempts are made to resolve the address by applying in
Davor Ocelice9ed2812017-12-25 17:49:28 +010013667 turn each of the methods mentioned in the comma-delimited list. The first
Willy Tarreau6a031d12016-11-07 19:42:35 +010013668 method which succeeds is used. If the end of the list is reached without
13669 finding a working method, an error is thrown. Method "last" suggests to pick
13670 the address which appears in the state file (see "server-state-file"). Method
13671 "libc" uses the libc's internal resolver (gethostbyname() or getaddrinfo()
13672 depending on the operating system and build options). Method "none"
13673 specifically indicates that the server should start without any valid IP
13674 address in a down state. It can be useful to ignore some DNS issues upon
13675 startup, waiting for the situation to get fixed later. Finally, an IP address
13676 (IPv4 or IPv6) may be provided. It can be the currently known address of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010013677 server (e.g. filled by a configuration generator), or the address of a dummy
Willy Tarreau6a031d12016-11-07 19:42:35 +010013678 server used to catch old sessions and present them with a decent error
13679 message for example. When the "first" load balancing algorithm is used, this
13680 IP address could point to a fake server used to trigger the creation of new
13681 instances on the fly. This option defaults to "last,libc" indicating that the
13682 previous address found in the state file (if any) is used first, otherwise
13683 the libc's resolver is used. This ensures continued compatibility with the
Davor Ocelice9ed2812017-12-25 17:49:28 +010013684 historic behavior.
Willy Tarreau6a031d12016-11-07 19:42:35 +010013685
13686 Example:
13687 defaults
13688 # never fail on address resolution
13689 default-server init-addr last,libc,none
13690
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013691inter <delay>
13692fastinter <delay>
13693downinter <delay>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013694 The "inter" parameter sets the interval between two consecutive health checks
13695 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
13696 It is also possible to use "fastinter" and "downinter" to optimize delays
13697 between checks depending on the server state :
13698
Pieter Baauw44fc9df2015-09-17 21:30:46 +020013699 Server state | Interval used
13700 ----------------------------------------+----------------------------------
13701 UP 100% (non-transitional) | "inter"
13702 ----------------------------------------+----------------------------------
13703 Transitionally UP (going down "fall"), | "fastinter" if set,
13704 Transitionally DOWN (going up "rise"), | "inter" otherwise.
13705 or yet unchecked. |
13706 ----------------------------------------+----------------------------------
13707 DOWN 100% (non-transitional) | "downinter" if set,
13708 | "inter" otherwise.
13709 ----------------------------------------+----------------------------------
Willy Tarreaud72758d2010-01-12 10:42:19 +010013710
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013711 Just as with every other time-based parameter, they can be entered in any
13712 other explicit unit among { us, ms, s, m, h, d }. The "inter" parameter also
13713 serves as a timeout for health checks sent to servers if "timeout check" is
13714 not set. In order to reduce "resonance" effects when multiple servers are
Simon Hormand60d6912013-11-25 10:46:36 +090013715 hosted on the same hardware, the agent and health checks of all servers
13716 are started with a small time offset between them. It is also possible to
13717 add some random noise in the agent and health checks interval using the
13718 global "spread-checks" keyword. This makes sense for instance when a lot
13719 of backends use the same servers.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013720
Emeric Brun97556472020-05-30 01:42:45 +020013721log-proto <logproto>
13722 The "log-proto" specifies the protocol used to forward event messages to
13723 a server configured in a ring section. Possible values are "legacy"
13724 and "octet-count" corresponding respectively to "Non-transparent-framing"
13725 and "Octet counting" in rfc6587. "legacy" is the default.
13726
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013727maxconn <maxconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013728 The "maxconn" parameter specifies the maximal number of concurrent
13729 connections that will be sent to this server. If the number of incoming
Tim Duesterhuscefbbd92019-11-27 22:35:27 +010013730 concurrent connections goes higher than this value, they will be queued,
13731 waiting for a slot to be released. This parameter is very important as it can
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013732 save fragile servers from going down under extreme loads. If a "minconn"
13733 parameter is specified, the limit becomes dynamic. The default value is "0"
13734 which means unlimited. See also the "minconn" and "maxqueue" parameters, and
13735 the backend's "fullconn" keyword.
13736
Tim Duesterhuscefbbd92019-11-27 22:35:27 +010013737 In HTTP mode this parameter limits the number of concurrent requests instead
13738 of the number of connections. Multiple requests might be multiplexed over a
13739 single TCP connection to the server. As an example if you specify a maxconn
13740 of 50 you might see between 1 and 50 actual server connections, but no more
13741 than 50 concurrent requests.
13742
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013743maxqueue <maxqueue>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013744 The "maxqueue" parameter specifies the maximal number of connections which
13745 will wait in the queue for this server. If this limit is reached, next
13746 requests will be redispatched to other servers instead of indefinitely
13747 waiting to be served. This will break persistence but may allow people to
Willy Tarreau8ae8c482020-10-22 17:19:07 +020013748 quickly re-log in when the server they try to connect to is dying. Some load
13749 balancing algorithms such as leastconn take this into account and accept to
13750 add requests into a server's queue up to this value if it is explicitly set
13751 to a value greater than zero, which often allows to better smooth the load
13752 when dealing with single-digit maxconn values. The default value is "0" which
13753 means the queue is unlimited. See also the "maxconn" and "minconn" parameters
13754 and "balance leastconn".
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013755
Willy Tarreau9c538e02019-01-23 10:21:49 +010013756max-reuse <count>
13757 The "max-reuse" argument indicates the HTTP connection processors that they
13758 should not reuse a server connection more than this number of times to send
13759 new requests. Permitted values are -1 (the default), which disables this
13760 limit, or any positive value. Value zero will effectively disable keep-alive.
13761 This is only used to work around certain server bugs which cause them to leak
13762 resources over time. The argument is not necessarily respected by the lower
13763 layers as there might be technical limitations making it impossible to
13764 enforce. At least HTTP/2 connections to servers will respect it.
13765
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013766minconn <minconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013767 When the "minconn" parameter is set, the maxconn limit becomes a dynamic
13768 limit following the backend's load. The server will always accept at least
13769 <minconn> connections, never more than <maxconn>, and the limit will be on
13770 the ramp between both values when the backend has less than <fullconn>
13771 concurrent connections. This makes it possible to limit the load on the
13772 server during normal loads, but push it further for important loads without
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010013773 overloading the server during exceptional loads. See also the "maxconn"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013774 and "maxqueue" parameters, as well as the "fullconn" backend keyword.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010013775
Willy Tarreaud72f0f32015-10-13 14:50:22 +020013776namespace <name>
13777 On Linux, it is possible to specify which network namespace a socket will
13778 belong to. This directive makes it possible to explicitly bind a server to
13779 a namespace different from the default one. Please refer to your operating
13780 system's documentation to find more details about network namespaces.
13781
Frédéric Lécailled2376272017-03-21 18:52:12 +010013782no-agent-check
13783 This option may be used as "server" setting to reset any "agent-check"
13784 setting which would have been inherited from "default-server" directive as
13785 default value.
13786 It may also be used as "default-server" setting to reset any previous
13787 "default-server" "agent-check" setting.
13788
13789no-backup
13790 This option may be used as "server" setting to reset any "backup"
13791 setting which would have been inherited from "default-server" directive as
13792 default value.
13793 It may also be used as "default-server" setting to reset any previous
13794 "default-server" "backup" setting.
13795
13796no-check
13797 This option may be used as "server" setting to reset any "check"
13798 setting which would have been inherited from "default-server" directive as
13799 default value.
13800 It may also be used as "default-server" setting to reset any previous
13801 "default-server" "check" setting.
13802
13803no-check-ssl
13804 This option may be used as "server" setting to reset any "check-ssl"
13805 setting which would have been inherited from "default-server" directive as
13806 default value.
13807 It may also be used as "default-server" setting to reset any previous
13808 "default-server" "check-ssl" setting.
13809
Frédéric Lécailled2376272017-03-21 18:52:12 +010013810no-send-proxy
13811 This option may be used as "server" setting to reset any "send-proxy"
13812 setting which would have been inherited from "default-server" directive as
13813 default value.
13814 It may also be used as "default-server" setting to reset any previous
13815 "default-server" "send-proxy" setting.
13816
13817no-send-proxy-v2
13818 This option may be used as "server" setting to reset any "send-proxy-v2"
13819 setting which would have been inherited from "default-server" directive as
13820 default value.
13821 It may also be used as "default-server" setting to reset any previous
13822 "default-server" "send-proxy-v2" setting.
13823
13824no-send-proxy-v2-ssl
13825 This option may be used as "server" setting to reset any "send-proxy-v2-ssl"
13826 setting which would have been inherited from "default-server" directive as
13827 default value.
13828 It may also be used as "default-server" setting to reset any previous
13829 "default-server" "send-proxy-v2-ssl" setting.
13830
13831no-send-proxy-v2-ssl-cn
13832 This option may be used as "server" setting to reset any "send-proxy-v2-ssl-cn"
13833 setting which would have been inherited from "default-server" directive as
13834 default value.
13835 It may also be used as "default-server" setting to reset any previous
13836 "default-server" "send-proxy-v2-ssl-cn" setting.
13837
13838no-ssl
13839 This option may be used as "server" setting to reset any "ssl"
13840 setting which would have been inherited from "default-server" directive as
13841 default value.
13842 It may also be used as "default-server" setting to reset any previous
13843 "default-server" "ssl" setting.
13844
Willy Tarreau2a3fb1c2015-02-05 16:47:07 +010013845no-ssl-reuse
13846 This option disables SSL session reuse when SSL is used to communicate with
13847 the server. It will force the server to perform a full handshake for every
13848 new connection. It's probably only useful for benchmarking, troubleshooting,
13849 and for paranoid users.
13850
Emeric Brun9b3009b2012-10-05 11:55:06 +020013851no-sslv3
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020013852 This option disables support for SSLv3 when SSL is used to communicate with
13853 the server. Note that SSLv2 is disabled in the code and cannot be enabled
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013854 using any configuration option. Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020013855
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020013856 Supported in default-server: No
13857
Emeric Brunf9c5c472012-10-11 15:28:34 +020013858no-tls-tickets
13859 This setting is only available when support for OpenSSL was built in. It
13860 disables the stateless session resumption (RFC 5077 TLS Ticket
13861 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013862 session resumption is more expensive in CPU usage for servers. This option
13863 is also available on global statement "ssl-default-server-options".
Lukas Tribusbdb386d2020-03-10 00:56:09 +010013864 The TLS ticket mechanism is only used up to TLS 1.2.
13865 Forward Secrecy is compromised with TLS tickets, unless ticket keys
13866 are periodically rotated (via reload or by using "tls-ticket-keys").
Frédéric Lécailled2376272017-03-21 18:52:12 +010013867 See also "tls-tickets".
Emeric Brunf9c5c472012-10-11 15:28:34 +020013868
Emeric Brun9b3009b2012-10-05 11:55:06 +020013869no-tlsv10
Emeric Brun8694b9a2012-10-05 14:39:07 +020013870 This option disables support for TLSv1.0 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020013871 the server. Note that SSLv2 is disabled in the code and cannot be enabled
13872 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013873 often makes sense to disable it when communicating with local servers. This
13874 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013875 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020013876
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020013877 Supported in default-server: No
13878
Emeric Brun9b3009b2012-10-05 11:55:06 +020013879no-tlsv11
Emeric Brun8694b9a2012-10-05 14:39:07 +020013880 This option disables support for TLSv1.1 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020013881 the server. Note that SSLv2 is disabled in the code and cannot be enabled
13882 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013883 often makes sense to disable it when communicating with local servers. This
13884 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013885 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020013886
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020013887 Supported in default-server: No
13888
Emeric Brun9b3009b2012-10-05 11:55:06 +020013889no-tlsv12
Emeric Brun8694b9a2012-10-05 14:39:07 +020013890 This option disables support for TLSv1.2 when SSL is used to communicate with
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020013891 the server. Note that SSLv2 is disabled in the code and cannot be enabled
13892 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013893 often makes sense to disable it when communicating with local servers. This
13894 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013895 Use "ssl-min-ver" and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020013896
13897 Supported in default-server: No
13898
13899no-tlsv13
13900 This option disables support for TLSv1.3 when SSL is used to communicate with
13901 the server. Note that SSLv2 is disabled in the code and cannot be enabled
13902 using any configuration option. TLSv1 is more expensive than SSLv3 so it
13903 often makes sense to disable it when communicating with local servers. This
13904 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013905 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020013906
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020013907 Supported in default-server: No
13908
Frédéric Lécailled2376272017-03-21 18:52:12 +010013909no-verifyhost
13910 This option may be used as "server" setting to reset any "verifyhost"
13911 setting which would have been inherited from "default-server" directive as
13912 default value.
13913 It may also be used as "default-server" setting to reset any previous
13914 "default-server" "verifyhost" setting.
Willy Tarreau763a95b2012-10-04 23:15:39 +020013915
Frédéric Lécaille1b9423d2019-07-04 14:19:06 +020013916no-tfo
13917 This option may be used as "server" setting to reset any "tfo"
13918 setting which would have been inherited from "default-server" directive as
13919 default value.
13920 It may also be used as "default-server" setting to reset any previous
13921 "default-server" "tfo" setting.
13922
Simon Hormanfa461682011-06-25 09:39:49 +090013923non-stick
13924 Never add connections allocated to this sever to a stick-table.
13925 This may be used in conjunction with backup to ensure that
13926 stick-table persistence is disabled for backup servers.
13927
Olivier Houchardc7566002018-11-20 23:33:50 +010013928npn <protocols>
13929 This enables the NPN TLS extension and advertises the specified protocol list
13930 as supported on top of NPN. The protocol list consists in a comma-delimited
13931 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
John Roeslerfb2fce12019-07-10 15:45:51 -050013932 This requires that the SSL library is built with support for TLS extensions
Olivier Houchardc7566002018-11-20 23:33:50 +010013933 enabled (check with haproxy -vv). Note that the NPN extension has been
13934 replaced with the ALPN extension (see the "alpn" keyword), though this one is
13935 only available starting with OpenSSL 1.0.2.
13936
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010013937observe <mode>
13938 This option enables health adjusting based on observing communication with
13939 the server. By default this functionality is disabled and enabling it also
13940 requires to enable health checks. There are two supported modes: "layer4" and
13941 "layer7". In layer4 mode, only successful/unsuccessful tcp connections are
13942 significant. In layer7, which is only allowed for http proxies, responses
13943 received from server are verified, like valid/wrong http code, unparsable
Willy Tarreau150d1462012-03-10 08:19:02 +010013944 headers, a timeout, etc. Valid status codes include 100 to 499, 501 and 505.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010013945
13946 See also the "check", "on-error" and "error-limit".
13947
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013948on-error <mode>
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010013949 Select what should happen when enough consecutive errors are detected.
13950 Currently, four modes are available:
13951 - fastinter: force fastinter
13952 - fail-check: simulate a failed check, also forces fastinter (default)
13953 - sudden-death: simulate a pre-fatal failed health check, one more failed
13954 check will mark a server down, forces fastinter
13955 - mark-down: mark the server immediately down and force fastinter
13956
13957 See also the "check", "observe" and "error-limit".
13958
Simon Hormane0d1bfb2011-06-21 14:34:58 +090013959on-marked-down <action>
13960 Modify what occurs when a server is marked down.
13961 Currently one action is available:
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070013962 - shutdown-sessions: Shutdown peer sessions. When this setting is enabled,
13963 all connections to the server are immediately terminated when the server
13964 goes down. It might be used if the health check detects more complex cases
13965 than a simple connection status, and long timeouts would cause the service
13966 to remain unresponsive for too long a time. For instance, a health check
13967 might detect that a database is stuck and that there's no chance to reuse
13968 existing connections anymore. Connections killed this way are logged with
13969 a 'D' termination code (for "Down").
Simon Hormane0d1bfb2011-06-21 14:34:58 +090013970
13971 Actions are disabled by default
13972
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070013973on-marked-up <action>
13974 Modify what occurs when a server is marked up.
13975 Currently one action is available:
13976 - shutdown-backup-sessions: Shutdown sessions on all backup servers. This is
13977 done only if the server is not in backup state and if it is not disabled
13978 (it must have an effective weight > 0). This can be used sometimes to force
13979 an active server to take all the traffic back after recovery when dealing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013980 with long sessions (e.g. LDAP, SQL, ...). Doing this can cause more trouble
13981 than it tries to solve (e.g. incomplete transactions), so use this feature
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070013982 with extreme care. Sessions killed because a server comes up are logged
13983 with an 'U' termination code (for "Up").
13984
13985 Actions are disabled by default
13986
Willy Tarreau2f3f4d32020-07-01 07:43:51 +020013987pool-low-conn <max>
13988 Set a low threshold on the number of idling connections for a server, below
13989 which a thread will not try to steal a connection from another thread. This
13990 can be useful to improve CPU usage patterns in scenarios involving many very
13991 fast servers, in order to ensure all threads will keep a few idle connections
13992 all the time instead of letting them accumulate over one thread and migrating
13993 them from thread to thread. Typical values of twice the number of threads
13994 seem to show very good performance already with sub-millisecond response
13995 times. The default is zero, indicating that any idle connection can be used
13996 at any time. It is the recommended setting for normal use. This only applies
13997 to connections that can be shared according to the same principles as those
Willy Tarreau37e07892021-02-19 11:45:22 +010013998 applying to "http-reuse". In case connection sharing between threads would
13999 be disabled via "tune.idle-pool.shared", it can become very important to use
14000 this setting to make sure each thread always has a few connections, or the
14001 connection reuse rate will decrease as thread count increases.
Willy Tarreau2f3f4d32020-07-01 07:43:51 +020014002
Olivier Houchard006e3102018-12-10 18:30:32 +010014003pool-max-conn <max>
14004 Set the maximum number of idling connections for a server. -1 means unlimited
14005 connections, 0 means no idle connections. The default is -1. When idle
14006 connections are enabled, orphaned idle connections which do not belong to any
14007 client session anymore are moved to a dedicated pool so that they remain
14008 usable by future clients. This only applies to connections that can be shared
14009 according to the same principles as those applying to "http-reuse".
14010
Olivier Houchardb7b3faa2018-12-14 18:15:36 +010014011pool-purge-delay <delay>
14012 Sets the delay to start purging idle connections. Each <delay> interval, half
Olivier Houcharda56eebf2019-03-19 16:44:02 +010014013 of the idle connections are closed. 0 means we don't keep any idle connection.
Willy Tarreaufb553652019-06-04 14:06:31 +020014014 The default is 5s.
Olivier Houchardb7b3faa2018-12-14 18:15:36 +010014015
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014016port <port>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014017 Using the "port" parameter, it becomes possible to use a different port to
14018 send health-checks. On some servers, it may be desirable to dedicate a port
14019 to a specific component able to perform complex tests which are more suitable
14020 to health-checks than the application. It is common to run a simple script in
14021 inetd for instance. This parameter is ignored if the "check" parameter is not
14022 set. See also the "addr" parameter.
14023
Christopher Faulet8ed0a3e2018-04-10 14:45:45 +020014024proto <name>
Christopher Faulet8ed0a3e2018-04-10 14:45:45 +020014025 Forces the multiplexer's protocol to use for the outgoing connections to this
14026 server. It must be compatible with the mode of the backend (TCP or HTTP). It
14027 must also be usable on the backend side. The list of available protocols is
14028 reported in haproxy -vv.
Daniel Corbett67a82712020-07-06 23:01:19 -040014029 Idea behind this option is to bypass the selection of the best multiplexer's
Christopher Faulet8ed0a3e2018-04-10 14:45:45 +020014030 protocol for all connections established to this server.
14031
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014032redir <prefix>
14033 The "redir" parameter enables the redirection mode for all GET and HEAD
14034 requests addressing this server. This means that instead of having HAProxy
14035 forward the request to the server, it will send an "HTTP 302" response with
14036 the "Location" header composed of this prefix immediately followed by the
14037 requested URI beginning at the leading '/' of the path component. That means
14038 that no trailing slash should be used after <prefix>. All invalid requests
14039 will be rejected, and all non-GET or HEAD requests will be normally served by
14040 the server. Note that since the response is completely forged, no header
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010014041 mangling nor cookie insertion is possible in the response. However, cookies in
Davor Ocelice9ed2812017-12-25 17:49:28 +010014042 requests are still analyzed, making this solution completely usable to direct
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014043 users to a remote location in case of local disaster. Main use consists in
14044 increasing bandwidth for static servers by having the clients directly
14045 connect to them. Note: never use a relative location here, it would cause a
14046 loop between the client and HAProxy!
14047
14048 Example : server srv1 192.168.1.1:80 redir http://image1.mydomain.com check
14049
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014050rise <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014051 The "rise" parameter states that a server will be considered as operational
14052 after <count> consecutive successful health checks. This value defaults to 2
14053 if unspecified. See also the "check", "inter" and "fall" parameters.
14054
Baptiste Assmann8e2d9432018-06-22 15:04:43 +020014055resolve-opts <option>,<option>,...
14056 Comma separated list of options to apply to DNS resolution linked to this
14057 server.
14058
14059 Available options:
14060
14061 * allow-dup-ip
14062 By default, HAProxy prevents IP address duplication in a backend when DNS
14063 resolution at runtime is in operation.
14064 That said, for some cases, it makes sense that two servers (in the same
14065 backend, being resolved by the same FQDN) have the same IP address.
14066 For such case, simply enable this option.
14067 This is the opposite of prevent-dup-ip.
14068
Daniel Corbettf8716912019-11-17 09:48:56 -050014069 * ignore-weight
14070 Ignore any weight that is set within an SRV record. This is useful when
14071 you would like to control the weights using an alternate method, such as
14072 using an "agent-check" or through the runtime api.
14073
Baptiste Assmann8e2d9432018-06-22 15:04:43 +020014074 * prevent-dup-ip
14075 Ensure HAProxy's default behavior is enforced on a server: prevent re-using
14076 an IP address already set to a server in the same backend and sharing the
14077 same fqdn.
14078 This is the opposite of allow-dup-ip.
14079
14080 Example:
14081 backend b_myapp
14082 default-server init-addr none resolvers dns
14083 server s1 myapp.example.com:80 check resolve-opts allow-dup-ip
14084 server s2 myapp.example.com:81 check resolve-opts allow-dup-ip
14085
14086 With the option allow-dup-ip set:
14087 * if the nameserver returns a single IP address, then both servers will use
14088 it
14089 * If the nameserver returns 2 IP addresses, then each server will pick up a
14090 different address
14091
14092 Default value: not set
14093
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014094resolve-prefer <family>
14095 When DNS resolution is enabled for a server and multiple IP addresses from
14096 different families are returned, HAProxy will prefer using an IP address
14097 from the family mentioned in the "resolve-prefer" parameter.
14098 Available families: "ipv4" and "ipv6"
14099
Baptiste Assmannc4aabae2015-08-04 22:43:06 +020014100 Default value: ipv6
14101
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020014102 Example:
14103
14104 server s1 app1.domain.com:80 resolvers mydns resolve-prefer ipv6
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014105
Thierry Fournierac88cfe2016-02-17 22:05:30 +010014106resolve-net <network>[,<network[,...]]
John Roeslerfb2fce12019-07-10 15:45:51 -050014107 This option prioritizes the choice of an ip address matching a network. This is
Thierry Fournierac88cfe2016-02-17 22:05:30 +010014108 useful with clouds to prefer a local ip. In some cases, a cloud high
Tim Düsterhus4896c442016-11-29 02:15:19 +010014109 availability service can be announced with many ip addresses on many
Davor Ocelice9ed2812017-12-25 17:49:28 +010014110 different datacenters. The latency between datacenter is not negligible, so
14111 this patch permits to prefer a local datacenter. If no address matches the
Thierry Fournierac88cfe2016-02-17 22:05:30 +010014112 configured network, another address is selected.
14113
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020014114 Example:
14115
14116 server s1 app1.domain.com:80 resolvers mydns resolve-net 10.0.0.0/8
Thierry Fournierac88cfe2016-02-17 22:05:30 +010014117
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014118resolvers <id>
14119 Points to an existing "resolvers" section to resolve current server's
14120 hostname.
14121
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020014122 Example:
14123
14124 server s1 app1.domain.com:80 check resolvers mydns
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014125
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020014126 See also section 5.3
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014127
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010014128send-proxy
14129 The "send-proxy" parameter enforces use of the PROXY protocol over any
14130 connection established to this server. The PROXY protocol informs the other
14131 end about the layer 3/4 addresses of the incoming connection, so that it can
14132 know the client's address or the public address it accessed to, whatever the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010014133 upper layer protocol. For connections accepted by an "accept-proxy" or
14134 "accept-netscaler-cip" listener, the advertised address will be used. Only
14135 TCPv4 and TCPv6 address families are supported. Other families such as
14136 Unix sockets, will report an UNKNOWN family. Servers using this option can
14137 fully be chained to another instance of haproxy listening with an
14138 "accept-proxy" setting. This setting must not be used if the server isn't
14139 aware of the protocol. When health checks are sent to the server, the PROXY
14140 protocol is automatically used when this option is set, unless there is an
14141 explicit "port" or "addr" directive, in which case an explicit
14142 "check-send-proxy" directive would also be needed to use the PROXY protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010014143 See also the "no-send-proxy" option of this section and "accept-proxy" and
14144 "accept-netscaler-cip" option of the "bind" keyword.
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010014145
David Safb76832014-05-08 23:42:08 -040014146send-proxy-v2
14147 The "send-proxy-v2" parameter enforces use of the PROXY protocol version 2
14148 over any connection established to this server. The PROXY protocol informs
14149 the other end about the layer 3/4 addresses of the incoming connection, so
14150 that it can know the client's address or the public address it accessed to,
Emmanuel Hocdet404d9782017-10-24 10:55:14 +020014151 whatever the upper layer protocol. It also send ALPN information if an alpn
14152 have been negotiated. This setting must not be used if the server isn't aware
14153 of this version of the protocol. See also the "no-send-proxy-v2" option of
14154 this section and send-proxy" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040014155
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010014156proxy-v2-options <option>[,<option>]*
Tim Duesterhuscf6e0c82020-03-13 12:34:24 +010014157 The "proxy-v2-options" parameter add options to send in PROXY protocol
14158 version 2 when "send-proxy-v2" is used. Options available are:
14159
14160 - ssl : See also "send-proxy-v2-ssl".
14161 - cert-cn : See also "send-proxy-v2-ssl-cn".
14162 - ssl-cipher: Name of the used cipher.
14163 - cert-sig : Signature algorithm of the used certificate.
14164 - cert-key : Key algorithm of the used certificate
14165 - authority : Host name value passed by the client (only SNI from a TLS
14166 connection is supported).
14167 - crc32c : Checksum of the PROXYv2 header.
14168 - unique-id : Send a unique ID generated using the frontend's
14169 "unique-id-format" within the PROXYv2 header.
14170 This unique-id is primarily meant for "mode tcp". It can
14171 lead to unexpected results in "mode http", because the
14172 generated unique ID is also used for the first HTTP request
14173 within a Keep-Alive connection.
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010014174
David Safb76832014-05-08 23:42:08 -040014175send-proxy-v2-ssl
14176 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
14177 2 over any connection established to this server. The PROXY protocol informs
14178 the other end about the layer 3/4 addresses of the incoming connection, so
14179 that it can know the client's address or the public address it accessed to,
14180 whatever the upper layer protocol. In addition, the SSL information extension
14181 of the PROXY protocol is added to the PROXY protocol header. This setting
14182 must not be used if the server isn't aware of this version of the protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010014183 See also the "no-send-proxy-v2-ssl" option of this section and the
14184 "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040014185
14186send-proxy-v2-ssl-cn
14187 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
14188 2 over any connection established to this server. The PROXY protocol informs
14189 the other end about the layer 3/4 addresses of the incoming connection, so
14190 that it can know the client's address or the public address it accessed to,
14191 whatever the upper layer protocol. In addition, the SSL information extension
14192 of the PROXY protocol, along along with the Common Name from the subject of
14193 the client certificate (if any), is added to the PROXY protocol header. This
14194 setting must not be used if the server isn't aware of this version of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010014195 protocol. See also the "no-send-proxy-v2-ssl-cn" option of this section and
14196 the "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040014197
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014198slowstart <start_time_in_ms>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014199 The "slowstart" parameter for a server accepts a value in milliseconds which
14200 indicates after how long a server which has just come back up will run at
14201 full speed. Just as with every other time-based parameter, it can be entered
14202 in any other explicit unit among { us, ms, s, m, h, d }. The speed grows
14203 linearly from 0 to 100% during this time. The limitation applies to two
14204 parameters :
14205
14206 - maxconn: the number of connections accepted by the server will grow from 1
14207 to 100% of the usual dynamic limit defined by (minconn,maxconn,fullconn).
14208
14209 - weight: when the backend uses a dynamic weighted algorithm, the weight
14210 grows linearly from 1 to 100%. In this case, the weight is updated at every
14211 health-check. For this reason, it is important that the "inter" parameter
14212 is smaller than the "slowstart", in order to maximize the number of steps.
14213
14214 The slowstart never applies when haproxy starts, otherwise it would cause
14215 trouble to running servers. It only applies when a server has been previously
14216 seen as failed.
14217
Willy Tarreau732eac42015-07-09 11:40:25 +020014218sni <expression>
14219 The "sni" parameter evaluates the sample fetch expression, converts it to a
14220 string and uses the result as the host name sent in the SNI TLS extension to
14221 the server. A typical use case is to send the SNI received from the client in
14222 a bridged HTTPS scenario, using the "ssl_fc_sni" sample fetch for the
Willy Tarreau2ab88672017-07-05 18:23:03 +020014223 expression, though alternatives such as req.hdr(host) can also make sense. If
14224 "verify required" is set (which is the recommended setting), the resulting
Willy Tarreauad92a9a2017-07-28 11:38:41 +020014225 name will also be matched against the server certificate's names. See the
Jérôme Magninb36a6d22018-12-09 16:03:40 +010014226 "verify" directive for more details. If you want to set a SNI for health
14227 checks, see the "check-sni" directive for more details.
Willy Tarreau732eac42015-07-09 11:40:25 +020014228
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020014229source <addr>[:<pl>[-<ph>]] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +020014230source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020014231source <addr>[:<pl>[-<ph>]] [interface <name>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014232 The "source" parameter sets the source address which will be used when
14233 connecting to the server. It follows the exact same parameters and principle
14234 as the backend "source" keyword, except that it only applies to the server
14235 referencing it. Please consult the "source" keyword for details.
14236
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020014237 Additionally, the "source" statement on a server line allows one to specify a
14238 source port range by indicating the lower and higher bounds delimited by a
14239 dash ('-'). Some operating systems might require a valid IP address when a
14240 source port range is specified. It is permitted to have the same IP/range for
14241 several servers. Doing so makes it possible to bypass the maximum of 64k
14242 total concurrent connections. The limit will then reach 64k connections per
14243 server.
14244
Lukas Tribus7d56c6d2016-09-13 09:51:15 +000014245 Since Linux 4.2/libc 2.23 IP_BIND_ADDRESS_NO_PORT is set for connections
14246 specifying the source address without port(s).
14247
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020014248ssl
Willy Tarreau44f65392013-06-25 07:56:20 +020014249 This option enables SSL ciphering on outgoing connections to the server. It
14250 is critical to verify server certificates using "verify" when using SSL to
14251 connect to servers, otherwise the communication is prone to trivial man in
14252 the-middle attacks rendering SSL useless. When this option is used, health
14253 checks are automatically sent in SSL too unless there is a "port" or an
14254 "addr" directive indicating the check should be sent to a different location.
Frédéric Lécailled2376272017-03-21 18:52:12 +010014255 See the "no-ssl" to disable "ssl" option and "check-ssl" option to force
14256 SSL health checks.
Willy Tarreau763a95b2012-10-04 23:15:39 +020014257
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020014258ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
14259 This option enforces use of <version> or lower when SSL is used to communicate
14260 with the server. This option is also available on global statement
14261 "ssl-default-server-options". See also "ssl-min-ver".
14262
14263ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
14264 This option enforces use of <version> or upper when SSL is used to communicate
14265 with the server. This option is also available on global statement
14266 "ssl-default-server-options". See also "ssl-max-ver".
14267
Frédéric Lécailled2376272017-03-21 18:52:12 +010014268ssl-reuse
14269 This option may be used as "server" setting to reset any "no-ssl-reuse"
14270 setting which would have been inherited from "default-server" directive as
14271 default value.
14272 It may also be used as "default-server" setting to reset any previous
14273 "default-server" "no-ssl-reuse" setting.
14274
14275stick
14276 This option may be used as "server" setting to reset any "non-stick"
14277 setting which would have been inherited from "default-server" directive as
14278 default value.
14279 It may also be used as "default-server" setting to reset any previous
14280 "default-server" "non-stick" setting.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020014281
Alexander Liu2a54bb72019-05-22 19:44:48 +080014282socks4 <addr>:<port>
John Roeslerfb2fce12019-07-10 15:45:51 -050014283 This option enables upstream socks4 tunnel for outgoing connections to the
Alexander Liu2a54bb72019-05-22 19:44:48 +080014284 server. Using this option won't force the health check to go via socks4 by
14285 default. You will have to use the keyword "check-via-socks4" to enable it.
14286
Willy Tarreau163d4622015-10-13 16:16:41 +020014287tcp-ut <delay>
14288 Sets the TCP User Timeout for all outgoing connections to this server. This
14289 option is available on Linux since version 2.6.37. It allows haproxy to
14290 configure a timeout for sockets which contain data not receiving an
Davor Ocelice9ed2812017-12-25 17:49:28 +010014291 acknowledgment for the configured delay. This is especially useful on
Willy Tarreau163d4622015-10-13 16:16:41 +020014292 long-lived connections experiencing long idle periods such as remote
14293 terminals or database connection pools, where the client and server timeouts
14294 must remain high to allow a long period of idle, but where it is important to
14295 detect that the server has disappeared in order to release all resources
14296 associated with its connection (and the client's session). One typical use
14297 case is also to force dead server connections to die when health checks are
14298 too slow or during a soft reload since health checks are then disabled. The
14299 argument is a delay expressed in milliseconds by default. This only works for
14300 regular TCP connections, and is ignored for other protocols.
14301
Willy Tarreau034c88c2017-01-23 23:36:45 +010014302tfo
14303 This option enables using TCP fast open when connecting to servers, on
14304 systems that support it (currently only the Linux kernel >= 4.11).
14305 See the "tfo" bind option for more information about TCP fast open.
14306 Please note that when using tfo, you should also use the "conn-failure",
14307 "empty-response" and "response-timeout" keywords for "retry-on", or haproxy
Frédéric Lécaille1b9423d2019-07-04 14:19:06 +020014308 won't be able to retry the connection on failure. See also "no-tfo".
Willy Tarreau034c88c2017-01-23 23:36:45 +010014309
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014310track [<proxy>/]<server>
Willy Tarreau32091232014-05-16 13:52:00 +020014311 This option enables ability to set the current state of the server by tracking
14312 another one. It is possible to track a server which itself tracks another
14313 server, provided that at the end of the chain, a server has health checks
14314 enabled. If <proxy> is omitted the current one is used. If disable-on-404 is
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014315 used, it has to be enabled on both proxies.
14316
Frédéric Lécailled2376272017-03-21 18:52:12 +010014317tls-tickets
14318 This option may be used as "server" setting to reset any "no-tls-tickets"
14319 setting which would have been inherited from "default-server" directive as
14320 default value.
Lukas Tribusbdb386d2020-03-10 00:56:09 +010014321 The TLS ticket mechanism is only used up to TLS 1.2.
14322 Forward Secrecy is compromised with TLS tickets, unless ticket keys
14323 are periodically rotated (via reload or by using "tls-ticket-keys").
Frédéric Lécailled2376272017-03-21 18:52:12 +010014324 It may also be used as "default-server" setting to reset any previous
Bjoern Jacke5ab7eb62020-02-13 14:16:16 +010014325 "default-server" "no-tls-tickets" setting.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014326
Emeric Brunef42d922012-10-11 16:11:36 +020014327verify [none|required]
14328 This setting is only available when support for OpenSSL was built in. If set
Emeric Brun850efd52014-01-29 12:24:34 +010014329 to 'none', server certificate is not verified. In the other case, The
Willy Tarreauad92a9a2017-07-28 11:38:41 +020014330 certificate provided by the server is verified using CAs from 'ca-file' and
14331 optional CRLs from 'crl-file' after having checked that the names provided in
Davor Ocelice9ed2812017-12-25 17:49:28 +010014332 the certificate's subject and subjectAlternateNames attributes match either
Willy Tarreauad92a9a2017-07-28 11:38:41 +020014333 the name passed using the "sni" directive, or if not provided, the static
14334 host name passed using the "verifyhost" directive. When no name is found, the
14335 certificate's names are ignored. For this reason, without SNI it's important
14336 to use "verifyhost". On verification failure the handshake is aborted. It is
14337 critically important to verify server certificates when using SSL to connect
14338 to servers, otherwise the communication is prone to trivial man-in-the-middle
14339 attacks rendering SSL totally useless. Unless "ssl_server_verify" appears in
14340 the global section, "verify" is set to "required" by default.
Emeric Brunef42d922012-10-11 16:11:36 +020014341
Evan Broderbe554312013-06-27 00:05:25 -070014342verifyhost <hostname>
14343 This setting is only available when support for OpenSSL was built in, and
Willy Tarreauad92a9a2017-07-28 11:38:41 +020014344 only takes effect if 'verify required' is also specified. This directive sets
14345 a default static hostname to check the server's certificate against when no
14346 SNI was used to connect to the server. If SNI is not used, this is the only
14347 way to enable hostname verification. This static hostname, when set, will
14348 also be used for health checks (which cannot provide an SNI value). If none
14349 of the hostnames in the certificate match the specified hostname, the
14350 handshake is aborted. The hostnames in the server-provided certificate may
14351 include wildcards. See also "verify", "sni" and "no-verifyhost" options.
Evan Broderbe554312013-06-27 00:05:25 -070014352
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014353weight <weight>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014354 The "weight" parameter is used to adjust the server's weight relative to
14355 other servers. All servers will receive a load proportional to their weight
14356 relative to the sum of all weights, so the higher the weight, the higher the
Willy Tarreau6704d672009-06-15 10:56:05 +020014357 load. The default weight is 1, and the maximal value is 256. A value of 0
14358 means the server will not participate in load-balancing but will still accept
14359 persistent connections. If this parameter is used to distribute the load
14360 according to server's capacity, it is recommended to start with values which
14361 can both grow and shrink, for instance between 10 and 100 to leave enough
14362 room above and below for later adjustments.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014363
14364
Cyril Bonté46175dd2015-07-02 22:45:32 +0200143655.3. Server IP address resolution using DNS
14366-------------------------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014367
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014368HAProxy allows using a host name on the server line to retrieve its IP address
14369using name servers. By default, HAProxy resolves the name when parsing the
Thayne McCombsdab4ba62021-01-07 21:24:41 -070014370configuration file, at startup and cache the result for the process's life.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014371This is not sufficient in some cases, such as in Amazon where a server's IP
14372can change after a reboot or an ELB Virtual IP can change based on current
14373workload.
14374This chapter describes how HAProxy can be configured to process server's name
14375resolution at run time.
14376Whether run time server name resolution has been enable or not, HAProxy will
14377carry on doing the first resolution when parsing the configuration.
14378
14379
Cyril Bonté46175dd2015-07-02 22:45:32 +0200143805.3.1. Global overview
14381----------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014382
14383As we've seen in introduction, name resolution in HAProxy occurs at two
14384different steps of the process life:
14385
14386 1. when starting up, HAProxy parses the server line definition and matches a
14387 host name. It uses libc functions to get the host name resolved. This
14388 resolution relies on /etc/resolv.conf file.
14389
Christopher Faulet67957bd2017-09-27 11:00:59 +020014390 2. at run time, HAProxy performs periodically name resolutions for servers
14391 requiring DNS resolutions.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014392
14393A few other events can trigger a name resolution at run time:
14394 - when a server's health check ends up in a connection timeout: this may be
14395 because the server has a new IP address. So we need to trigger a name
14396 resolution to know this new IP.
14397
Christopher Faulet67957bd2017-09-27 11:00:59 +020014398When using resolvers, the server name can either be a hostname, or a SRV label.
Davor Ocelice9ed2812017-12-25 17:49:28 +010014399HAProxy considers anything that starts with an underscore as a SRV label. If a
Christopher Faulet67957bd2017-09-27 11:00:59 +020014400SRV label is specified, then the corresponding SRV records will be retrieved
14401from the DNS server, and the provided hostnames will be used. The SRV label
14402will be checked periodically, and if any server are added or removed, haproxy
14403will automatically do the same.
Olivier Houchardecfa18d2017-08-07 17:30:03 +020014404
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014405A few things important to notice:
John Roeslerfb2fce12019-07-10 15:45:51 -050014406 - all the name servers are queried in the meantime. HAProxy will process the
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014407 first valid response.
14408
14409 - a resolution is considered as invalid (NX, timeout, refused), when all the
14410 servers return an error.
14411
14412
Cyril Bonté46175dd2015-07-02 22:45:32 +0200144135.3.2. The resolvers section
14414----------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014415
14416This section is dedicated to host information related to name resolution in
Christopher Faulet67957bd2017-09-27 11:00:59 +020014417HAProxy. There can be as many as resolvers section as needed. Each section can
14418contain many name servers.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014419
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014420When multiple name servers are configured in a resolvers section, then HAProxy
14421uses the first valid response. In case of invalid responses, only the last one
14422is treated. Purpose is to give the chance to a slow server to deliver a valid
14423answer after a fast faulty or outdated server.
14424
14425When each server returns a different error type, then only the last error is
Christopher Faulet67957bd2017-09-27 11:00:59 +020014426used by HAProxy. The following processing is applied on this error:
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014427
Christopher Faulet67957bd2017-09-27 11:00:59 +020014428 1. HAProxy retries the same DNS query with a new query type. The A queries are
14429 switch to AAAA or the opposite. SRV queries are not concerned here. Timeout
14430 errors are also excluded.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014431
Christopher Faulet67957bd2017-09-27 11:00:59 +020014432 2. When the fallback on the query type was done (or not applicable), HAProxy
14433 retries the original DNS query, with the preferred query type.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014434
Christopher Faulet67957bd2017-09-27 11:00:59 +020014435 3. HAProxy retries previous steps <resolve_retires> times. If no valid
14436 response is received after that, it stops the DNS resolution and reports
14437 the error.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014438
Christopher Faulet67957bd2017-09-27 11:00:59 +020014439For example, with 2 name servers configured in a resolvers section, the
14440following scenarios are possible:
14441
14442 - First response is valid and is applied directly, second response is
14443 ignored
14444
14445 - First response is invalid and second one is valid, then second response is
14446 applied
14447
14448 - First response is a NX domain and second one a truncated response, then
14449 HAProxy retries the query with a new type
14450
14451 - First response is a NX domain and second one is a timeout, then HAProxy
14452 retries the query with a new type
14453
14454 - Query timed out for both name servers, then HAProxy retries it with the
14455 same query type
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014456
Olivier Houcharda8c6db82017-07-06 18:46:47 +020014457As a DNS server may not answer all the IPs in one DNS request, haproxy keeps
14458a cache of previous answers, an answer will be considered obsolete after
Christopher Faulet67957bd2017-09-27 11:00:59 +020014459<hold obsolete> seconds without the IP returned.
Olivier Houcharda8c6db82017-07-06 18:46:47 +020014460
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014461
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014462resolvers <resolvers id>
Davor Ocelice9ed2812017-12-25 17:49:28 +010014463 Creates a new name server list labeled <resolvers id>
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014464
14465A resolvers section accept the following parameters:
14466
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020014467accepted_payload_size <nb>
Davor Ocelice9ed2812017-12-25 17:49:28 +010014468 Defines the maximum payload size accepted by HAProxy and announced to all the
Christopher Faulet67957bd2017-09-27 11:00:59 +020014469 name servers configured in this resolvers section.
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020014470 <nb> is in bytes. If not set, HAProxy announces 512. (minimal value defined
14471 by RFC 6891)
14472
Baptiste Assmann9d8dbbc2017-08-18 23:35:08 +020014473 Note: the maximum allowed value is 8192.
14474
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014475nameserver <id> <ip>:<port>
14476 DNS server description:
14477 <id> : label of the server, should be unique
14478 <ip> : IP address of the server
14479 <port> : port where the DNS service actually runs
14480
Ben Draut44e609b2018-05-29 15:40:08 -060014481parse-resolv-conf
14482 Adds all nameservers found in /etc/resolv.conf to this resolvers nameservers
14483 list. Ordered as if each nameserver in /etc/resolv.conf was individually
14484 placed in the resolvers section in place of this directive.
14485
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014486hold <status> <period>
14487 Defines <period> during which the last name resolution should be kept based
14488 on last resolution <status>
Baptiste Assmann987e16d2016-11-02 22:23:31 +010014489 <status> : last name resolution status. Acceptable values are "nx",
Olivier Houcharda8c6db82017-07-06 18:46:47 +020014490 "other", "refused", "timeout", "valid", "obsolete".
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014491 <period> : interval between two successive name resolution when the last
14492 answer was in <status>. It follows the HAProxy time format.
14493 <period> is in milliseconds by default.
14494
Baptiste Assmann686408b2017-08-18 10:15:42 +020014495 Default value is 10s for "valid", 0s for "obsolete" and 30s for others.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014496
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014497resolve_retries <nb>
14498 Defines the number <nb> of queries to send to resolve a server name before
14499 giving up.
14500 Default value: 3
14501
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014502 A retry occurs on name server timeout or when the full sequence of DNS query
14503 type failover is over and we need to start up from the default ANY query
14504 type.
14505
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014506timeout <event> <time>
14507 Defines timeouts related to name resolution
14508 <event> : the event on which the <time> timeout period applies to.
14509 events available are:
Frédéric Lécaille93d33162019-03-06 09:35:59 +010014510 - resolve : default time to trigger name resolutions when no
14511 other time applied.
Christopher Faulet67957bd2017-09-27 11:00:59 +020014512 Default value: 1s
14513 - retry : time between two DNS queries, when no valid response
Frédéric Lécaille93d33162019-03-06 09:35:59 +010014514 have been received.
Christopher Faulet67957bd2017-09-27 11:00:59 +020014515 Default value: 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014516 <time> : time related to the event. It follows the HAProxy time format.
14517 <time> is expressed in milliseconds.
14518
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020014519 Example:
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014520
14521 resolvers mydns
14522 nameserver dns1 10.0.0.1:53
14523 nameserver dns2 10.0.0.2:53
Ben Draut44e609b2018-05-29 15:40:08 -060014524 parse-resolv-conf
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014525 resolve_retries 3
Christopher Faulet67957bd2017-09-27 11:00:59 +020014526 timeout resolve 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014527 timeout retry 1s
Baptiste Assmann987e16d2016-11-02 22:23:31 +010014528 hold other 30s
14529 hold refused 30s
14530 hold nx 30s
14531 hold timeout 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014532 hold valid 10s
Olivier Houcharda8c6db82017-07-06 18:46:47 +020014533 hold obsolete 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014534
14535
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200145366. Cache
14537---------
14538
14539HAProxy provides a cache, which was designed to perform cache on small objects
14540(favicon, css...). This is a minimalist low-maintenance cache which runs in
14541RAM.
14542
14543The cache is based on a memory which is shared between processes and threads,
14544this memory is split in blocks of 1k.
14545
14546If an object is not used anymore, it can be deleted to store a new object
14547independently of its expiration date. The oldest objects are deleted first
14548when we try to allocate a new one.
14549
14550The cache uses a hash of the host header and the URI as the key.
14551
14552It's possible to view the status of a cache using the Unix socket command
14553"show cache" consult section 9.3 "Unix Socket commands" of Management Guide
14554for more details.
14555
14556When an object is delivered from the cache, the server name in the log is
14557replaced by "<CACHE>".
14558
14559
145606.1. Limitation
14561----------------
14562
14563The cache won't store and won't deliver objects in these cases:
14564
14565- If the response is not a 200
14566- If the response contains a Vary header
14567- If the Content-Length + the headers size is greater than "max-object-size"
14568- If the response is not cacheable
Remi Tricot-Le Bretonea8e0812020-11-26 15:51:29 +010014569- If the response does not have an explicit expiration time (s-maxage or max-age
14570 Cache-Control directives or Expires header) or a validator (ETag or Last-Modified
14571 headers)
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020014572
14573- If the request is not a GET
14574- If the HTTP version of the request is smaller than 1.1
14575- If the request contains an Authorization header
14576
14577
145786.2. Setup
14579-----------
14580
14581To setup a cache, you must define a cache section and use it in a proxy with
14582the corresponding http-request and response actions.
14583
14584
145856.2.1. Cache section
14586---------------------
14587
14588cache <name>
14589 Declare a cache section, allocate a shared cache memory named <name>, the
14590 size of cache is mandatory.
14591
14592total-max-size <megabytes>
14593 Define the size in RAM of the cache in megabytes. This size is split in
14594 blocks of 1kB which are used by the cache entries. Its maximum value is 4095.
14595
14596max-object-size <bytes>
14597 Define the maximum size of the objects to be cached. Must not be greater than
14598 an half of "total-max-size". If not set, it equals to a 256th of the cache size.
14599 All objects with sizes larger than "max-object-size" will not be cached.
14600
14601max-age <seconds>
14602 Define the maximum expiration duration. The expiration is set has the lowest
14603 value between the s-maxage or max-age (in this order) directive in the
14604 Cache-Control response header and this value. The default value is 60
14605 seconds, which means that you can't cache an object more than 60 seconds by
14606 default.
14607
14608
146096.2.2. Proxy section
14610---------------------
14611
14612http-request cache-use <name> [ { if | unless } <condition> ]
14613 Try to deliver a cached object from the cache <name>. This directive is also
14614 mandatory to store the cache as it calculates the cache hash. If you want to
14615 use a condition for both storage and delivering that's a good idea to put it
14616 after this one.
14617
14618http-response cache-store <name> [ { if | unless } <condition> ]
14619 Store an http-response within the cache. The storage of the response headers
14620 is done at this step, which means you can use others http-response actions
14621 to modify headers before or after the storage of the response. This action
14622 is responsible for the setup of the cache storage filter.
14623
14624
14625Example:
14626
14627 backend bck1
14628 mode http
14629
14630 http-request cache-use foobar
14631 http-response cache-store foobar
14632 server srv1 127.0.0.1:80
14633
14634 cache foobar
14635 total-max-size 4
14636 max-age 240
14637
14638
Willy Tarreau74ca5042013-06-11 23:12:07 +0200146397. Using ACLs and fetching samples
14640----------------------------------
14641
Davor Ocelice9ed2812017-12-25 17:49:28 +010014642HAProxy is capable of extracting data from request or response streams, from
Willy Tarreau74ca5042013-06-11 23:12:07 +020014643client or server information, from tables, environmental information etc...
14644The action of extracting such data is called fetching a sample. Once retrieved,
14645these samples may be used for various purposes such as a key to a stick-table,
14646but most common usages consist in matching them against predefined constant
14647data called patterns.
14648
14649
146507.1. ACL basics
14651---------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014652
14653The use of Access Control Lists (ACL) provides a flexible solution to perform
14654content switching and generally to take decisions based on content extracted
14655from the request, the response or any environmental status. The principle is
14656simple :
14657
Willy Tarreau74ca5042013-06-11 23:12:07 +020014658 - extract a data sample from a stream, table or the environment
Willy Tarreaue6b11e42013-11-26 19:02:32 +010014659 - optionally apply some format conversion to the extracted sample
Willy Tarreau74ca5042013-06-11 23:12:07 +020014660 - apply one or multiple pattern matching methods on this sample
14661 - perform actions only when a pattern matches the sample
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014662
Willy Tarreau74ca5042013-06-11 23:12:07 +020014663The actions generally consist in blocking a request, selecting a backend, or
14664adding a header.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014665
14666In order to define a test, the "acl" keyword is used. The syntax is :
14667
Willy Tarreau74ca5042013-06-11 23:12:07 +020014668 acl <aclname> <criterion> [flags] [operator] [<value>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014669
14670This creates a new ACL <aclname> or completes an existing one with new tests.
14671Those tests apply to the portion of request/response specified in <criterion>
14672and may be adjusted with optional flags [flags]. Some criteria also support
Willy Tarreaue6b11e42013-11-26 19:02:32 +010014673an operator which may be specified before the set of values. Optionally some
14674conversion operators may be applied to the sample, and they will be specified
14675as a comma-delimited list of keywords just after the first keyword. The values
14676are of the type supported by the criterion, and are separated by spaces.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014677
14678ACL names must be formed from upper and lower case letters, digits, '-' (dash),
14679'_' (underscore) , '.' (dot) and ':' (colon). ACL names are case-sensitive,
14680which means that "my_acl" and "My_Acl" are two different ACLs.
14681
14682There is no enforced limit to the number of ACLs. The unused ones do not affect
14683performance, they just consume a small amount of memory.
14684
Willy Tarreau74ca5042013-06-11 23:12:07 +020014685The criterion generally is the name of a sample fetch method, or one of its ACL
14686specific declinations. The default test method is implied by the output type of
14687this sample fetch method. The ACL declinations can describe alternate matching
Willy Tarreaue6b11e42013-11-26 19:02:32 +010014688methods of a same sample fetch method. The sample fetch methods are the only
14689ones supporting a conversion.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014690
14691Sample fetch methods return data which can be of the following types :
14692 - boolean
14693 - integer (signed or unsigned)
14694 - IPv4 or IPv6 address
14695 - string
14696 - data block
14697
Willy Tarreaue6b11e42013-11-26 19:02:32 +010014698Converters transform any of these data into any of these. For example, some
14699converters might convert a string to a lower-case string while other ones
14700would turn a string to an IPv4 address, or apply a netmask to an IP address.
14701The resulting sample is of the type of the last converter applied to the list,
14702which defaults to the type of the sample fetch method.
14703
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020014704Each sample or converter returns data of a specific type, specified with its
14705keyword in this documentation. When an ACL is declared using a standard sample
14706fetch method, certain types automatically involved a default matching method
14707which are summarized in the table below :
14708
14709 +---------------------+-----------------+
14710 | Sample or converter | Default |
14711 | output type | matching method |
14712 +---------------------+-----------------+
14713 | boolean | bool |
14714 +---------------------+-----------------+
14715 | integer | int |
14716 +---------------------+-----------------+
14717 | ip | ip |
14718 +---------------------+-----------------+
14719 | string | str |
14720 +---------------------+-----------------+
14721 | binary | none, use "-m" |
14722 +---------------------+-----------------+
14723
14724Note that in order to match a binary samples, it is mandatory to specify a
14725matching method, see below.
14726
Willy Tarreau74ca5042013-06-11 23:12:07 +020014727The ACL engine can match these types against patterns of the following types :
14728 - boolean
14729 - integer or integer range
14730 - IP address / network
14731 - string (exact, substring, suffix, prefix, subdir, domain)
14732 - regular expression
14733 - hex block
14734
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014735The following ACL flags are currently supported :
14736
Willy Tarreau2b5285d2010-05-09 23:45:24 +020014737 -i : ignore case during matching of all subsequent patterns.
14738 -f : load patterns from a file.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014739 -m : use a specific pattern matching method
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010014740 -n : forbid the DNS resolutions
Thierry FOURNIER9860c412014-01-29 14:23:29 +010014741 -M : load the file pointed by -f like a map file.
Thierry FOURNIER3534d882014-01-20 17:01:44 +010014742 -u : force the unique id of the ACL
Willy Tarreau6a06a402007-07-15 20:15:28 +020014743 -- : force end of flags. Useful when a string looks like one of the flags.
14744
Willy Tarreau74ca5042013-06-11 23:12:07 +020014745The "-f" flag is followed by the name of a file from which all lines will be
14746read as individual values. It is even possible to pass multiple "-f" arguments
14747if the patterns are to be loaded from multiple files. Empty lines as well as
14748lines beginning with a sharp ('#') will be ignored. All leading spaces and tabs
14749will be stripped. If it is absolutely necessary to insert a valid pattern
14750beginning with a sharp, just prefix it with a space so that it is not taken for
14751a comment. Depending on the data type and match method, haproxy may load the
14752lines into a binary tree, allowing very fast lookups. This is true for IPv4 and
14753exact string matching. In this case, duplicates will automatically be removed.
14754
Thierry FOURNIER9860c412014-01-29 14:23:29 +010014755The "-M" flag allows an ACL to use a map file. If this flag is set, the file is
14756parsed as two column file. The first column contains the patterns used by the
14757ACL, and the second column contain the samples. The sample can be used later by
14758a map. This can be useful in some rare cases where an ACL would just be used to
14759check for the existence of a pattern in a map before a mapping is applied.
14760
Thierry FOURNIER3534d882014-01-20 17:01:44 +010014761The "-u" flag forces the unique id of the ACL. This unique id is used with the
14762socket interface to identify ACL and dynamically change its values. Note that a
14763file is always identified by its name even if an id is set.
14764
Willy Tarreau74ca5042013-06-11 23:12:07 +020014765Also, note that the "-i" flag applies to subsequent entries and not to entries
14766loaded from files preceding it. For instance :
14767
14768 acl valid-ua hdr(user-agent) -f exact-ua.lst -i -f generic-ua.lst test
14769
14770In this example, each line of "exact-ua.lst" will be exactly matched against
14771the "user-agent" header of the request. Then each line of "generic-ua" will be
14772case-insensitively matched. Then the word "test" will be insensitively matched
14773as well.
14774
14775The "-m" flag is used to select a specific pattern matching method on the input
14776sample. All ACL-specific criteria imply a pattern matching method and generally
14777do not need this flag. However, this flag is useful with generic sample fetch
14778methods to describe how they're going to be matched against the patterns. This
14779is required for sample fetches which return data type for which there is no
Davor Ocelice9ed2812017-12-25 17:49:28 +010014780obvious matching method (e.g. string or binary). When "-m" is specified and
Willy Tarreau74ca5042013-06-11 23:12:07 +020014781followed by a pattern matching method name, this method is used instead of the
14782default one for the criterion. This makes it possible to match contents in ways
14783that were not initially planned, or with sample fetch methods which return a
14784string. The matching method also affects the way the patterns are parsed.
14785
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010014786The "-n" flag forbids the dns resolutions. It is used with the load of ip files.
14787By default, if the parser cannot parse ip address it considers that the parsed
14788string is maybe a domain name and try dns resolution. The flag "-n" disable this
14789resolution. It is useful for detecting malformed ip lists. Note that if the DNS
14790server is not reachable, the haproxy configuration parsing may last many minutes
John Roeslerfb2fce12019-07-10 15:45:51 -050014791waiting for the timeout. During this time no error messages are displayed. The
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010014792flag "-n" disable this behavior. Note also that during the runtime, this
14793function is disabled for the dynamic acl modifications.
14794
Willy Tarreau74ca5042013-06-11 23:12:07 +020014795There are some restrictions however. Not all methods can be used with all
14796sample fetch methods. Also, if "-m" is used in conjunction with "-f", it must
14797be placed first. The pattern matching method must be one of the following :
Willy Tarreau5adeda12013-03-31 22:13:34 +020014798
14799 - "found" : only check if the requested sample could be found in the stream,
14800 but do not compare it against any pattern. It is recommended not
Willy Tarreau74ca5042013-06-11 23:12:07 +020014801 to pass any pattern to avoid confusion. This matching method is
14802 particularly useful to detect presence of certain contents such
14803 as headers, cookies, etc... even if they are empty and without
14804 comparing them to anything nor counting them.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014805
14806 - "bool" : check the value as a boolean. It can only be applied to fetches
14807 which return a boolean or integer value, and takes no pattern.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014808 Value zero or false does not match, all other values do match.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014809
14810 - "int" : match the value as an integer. It can be used with integer and
Willy Tarreau74ca5042013-06-11 23:12:07 +020014811 boolean samples. Boolean false is integer 0, true is integer 1.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014812
14813 - "ip" : match the value as an IPv4 or IPv6 address. It is compatible
Willy Tarreau74ca5042013-06-11 23:12:07 +020014814 with IP address samples only, so it is implied and never needed.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014815
Davor Ocelice9ed2812017-12-25 17:49:28 +010014816 - "bin" : match the contents against a hexadecimal string representing a
Willy Tarreau5adeda12013-03-31 22:13:34 +020014817 binary sequence. This may be used with binary or string samples.
14818
14819 - "len" : match the sample's length as an integer. This may be used with
14820 binary or string samples.
14821
Willy Tarreau74ca5042013-06-11 23:12:07 +020014822 - "str" : exact match : match the contents against a string. This may be
14823 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014824
Willy Tarreau74ca5042013-06-11 23:12:07 +020014825 - "sub" : substring match : check that the contents contain at least one of
14826 the provided string patterns. This may be used with binary or
14827 string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014828
Willy Tarreau74ca5042013-06-11 23:12:07 +020014829 - "reg" : regex match : match the contents against a list of regular
14830 expressions. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014831
Willy Tarreau74ca5042013-06-11 23:12:07 +020014832 - "beg" : prefix match : check that the contents begin like the provided
14833 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014834
Willy Tarreau74ca5042013-06-11 23:12:07 +020014835 - "end" : suffix match : check that the contents end like the provided
14836 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014837
Willy Tarreau74ca5042013-06-11 23:12:07 +020014838 - "dir" : subdir match : check that a slash-delimited portion of the
14839 contents exactly matches one of the provided string patterns.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014840 This may be used with binary or string samples.
14841
Willy Tarreau74ca5042013-06-11 23:12:07 +020014842 - "dom" : domain match : check that a dot-delimited portion of the contents
14843 exactly match one of the provided string patterns. This may be
14844 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014845
14846For example, to quickly detect the presence of cookie "JSESSIONID" in an HTTP
14847request, it is possible to do :
14848
14849 acl jsess_present cook(JSESSIONID) -m found
14850
14851In order to apply a regular expression on the 500 first bytes of data in the
14852buffer, one would use the following acl :
14853
14854 acl script_tag payload(0,500) -m reg -i <script>
14855
Willy Tarreaue6b11e42013-11-26 19:02:32 +010014856On systems where the regex library is much slower when using "-i", it is
14857possible to convert the sample to lowercase before matching, like this :
14858
14859 acl script_tag payload(0,500),lower -m reg <script>
14860
Willy Tarreau74ca5042013-06-11 23:12:07 +020014861All ACL-specific criteria imply a default matching method. Most often, these
14862criteria are composed by concatenating the name of the original sample fetch
14863method and the matching method. For example, "hdr_beg" applies the "beg" match
14864to samples retrieved using the "hdr" fetch method. Since all ACL-specific
14865criteria rely on a sample fetch method, it is always possible instead to use
14866the original sample fetch method and the explicit matching method using "-m".
Willy Tarreau2b5285d2010-05-09 23:45:24 +020014867
Willy Tarreau74ca5042013-06-11 23:12:07 +020014868If an alternate match is specified using "-m" on an ACL-specific criterion,
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014869the matching method is simply applied to the underlying sample fetch method.
14870For example, all ACLs below are exact equivalent :
Willy Tarreau2b5285d2010-05-09 23:45:24 +020014871
Willy Tarreau74ca5042013-06-11 23:12:07 +020014872 acl short_form hdr_beg(host) www.
14873 acl alternate1 hdr_beg(host) -m beg www.
14874 acl alternate2 hdr_dom(host) -m beg www.
14875 acl alternate3 hdr(host) -m beg www.
Willy Tarreau2b5285d2010-05-09 23:45:24 +020014876
Willy Tarreau2b5285d2010-05-09 23:45:24 +020014877
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020014878The table below summarizes the compatibility matrix between sample or converter
14879types and the pattern types to fetch against. It indicates for each compatible
14880combination the name of the matching method to be used, surrounded with angle
14881brackets ">" and "<" when the method is the default one and will work by
14882default without "-m".
Willy Tarreau0ba27502007-12-24 16:55:16 +010014883
Willy Tarreau74ca5042013-06-11 23:12:07 +020014884 +-------------------------------------------------+
14885 | Input sample type |
14886 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020014887 | pattern type | boolean | integer | ip | string | binary |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014888 +----------------------+---------+---------+---------+---------+---------+
14889 | none (presence only) | found | found | found | found | found |
14890 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020014891 | none (boolean value) |> bool <| bool | | bool | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014892 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020014893 | integer (value) | int |> int <| int | int | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014894 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010014895 | integer (length) | len | len | len | len | len |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014896 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020014897 | IP address | | |> ip <| ip | ip |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014898 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020014899 | exact string | str | str | str |> str <| str |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014900 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010014901 | prefix | beg | beg | beg | beg | beg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014902 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010014903 | suffix | end | end | end | end | end |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014904 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010014905 | substring | sub | sub | sub | sub | sub |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014906 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010014907 | subdir | dir | dir | dir | dir | dir |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014908 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010014909 | domain | dom | dom | dom | dom | dom |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014910 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010014911 | regex | reg | reg | reg | reg | reg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014912 +----------------------+---------+---------+---------+---------+---------+
14913 | hex block | | | | bin | bin |
14914 +----------------------+---------+---------+---------+---------+---------+
Willy Tarreau6a06a402007-07-15 20:15:28 +020014915
14916
Willy Tarreau74ca5042013-06-11 23:12:07 +0200149177.1.1. Matching booleans
14918------------------------
14919
14920In order to match a boolean, no value is needed and all values are ignored.
14921Boolean matching is used by default for all fetch methods of type "boolean".
14922When boolean matching is used, the fetched value is returned as-is, which means
14923that a boolean "true" will always match and a boolean "false" will never match.
14924
14925Boolean matching may also be enforced using "-m bool" on fetch methods which
14926return an integer value. Then, integer value 0 is converted to the boolean
14927"false" and all other values are converted to "true".
14928
Willy Tarreau6a06a402007-07-15 20:15:28 +020014929
Willy Tarreau74ca5042013-06-11 23:12:07 +0200149307.1.2. Matching integers
14931------------------------
14932
14933Integer matching applies by default to integer fetch methods. It can also be
14934enforced on boolean fetches using "-m int". In this case, "false" is converted
14935to the integer 0, and "true" is converted to the integer 1.
14936
14937Integer matching also supports integer ranges and operators. Note that integer
14938matching only applies to positive values. A range is a value expressed with a
14939lower and an upper bound separated with a colon, both of which may be omitted.
Willy Tarreau6a06a402007-07-15 20:15:28 +020014940
14941For instance, "1024:65535" is a valid range to represent a range of
14942unprivileged ports, and "1024:" would also work. "0:1023" is a valid
14943representation of privileged ports, and ":1023" would also work.
14944
Willy Tarreau62644772008-07-16 18:36:06 +020014945As a special case, some ACL functions support decimal numbers which are in fact
14946two integers separated by a dot. This is used with some version checks for
14947instance. All integer properties apply to those decimal numbers, including
14948ranges and operators.
14949
Willy Tarreau6a06a402007-07-15 20:15:28 +020014950For an easier usage, comparison operators are also supported. Note that using
Willy Tarreau0ba27502007-12-24 16:55:16 +010014951operators with ranges does not make much sense and is strongly discouraged.
14952Similarly, it does not make much sense to perform order comparisons with a set
14953of values.
Willy Tarreau6a06a402007-07-15 20:15:28 +020014954
Willy Tarreau0ba27502007-12-24 16:55:16 +010014955Available operators for integer matching are :
Willy Tarreau6a06a402007-07-15 20:15:28 +020014956
14957 eq : true if the tested value equals at least one value
14958 ge : true if the tested value is greater than or equal to at least one value
14959 gt : true if the tested value is greater than at least one value
14960 le : true if the tested value is less than or equal to at least one value
14961 lt : true if the tested value is less than at least one value
14962
Willy Tarreau0ba27502007-12-24 16:55:16 +010014963For instance, the following ACL matches any negative Content-Length header :
Willy Tarreau6a06a402007-07-15 20:15:28 +020014964
14965 acl negative-length hdr_val(content-length) lt 0
14966
Willy Tarreau62644772008-07-16 18:36:06 +020014967This one matches SSL versions between 3.0 and 3.1 (inclusive) :
14968
14969 acl sslv3 req_ssl_ver 3:3.1
14970
Willy Tarreau6a06a402007-07-15 20:15:28 +020014971
Willy Tarreau74ca5042013-06-11 23:12:07 +0200149727.1.3. Matching strings
14973-----------------------
14974
14975String matching applies to string or binary fetch methods, and exists in 6
14976different forms :
14977
14978 - exact match (-m str) : the extracted string must exactly match the
Davor Ocelice9ed2812017-12-25 17:49:28 +010014979 patterns;
Willy Tarreau74ca5042013-06-11 23:12:07 +020014980
14981 - substring match (-m sub) : the patterns are looked up inside the
Davor Ocelice9ed2812017-12-25 17:49:28 +010014982 extracted string, and the ACL matches if any of them is found inside;
Willy Tarreau74ca5042013-06-11 23:12:07 +020014983
14984 - prefix match (-m beg) : the patterns are compared with the beginning of
14985 the extracted string, and the ACL matches if any of them matches.
14986
14987 - suffix match (-m end) : the patterns are compared with the end of the
14988 extracted string, and the ACL matches if any of them matches.
14989
Baptiste Assmann33db6002016-03-06 23:32:10 +010014990 - subdir match (-m dir) : the patterns are looked up inside the extracted
Willy Tarreau74ca5042013-06-11 23:12:07 +020014991 string, delimited with slashes ("/"), and the ACL matches if any of them
14992 matches.
14993
14994 - domain match (-m dom) : the patterns are looked up inside the extracted
14995 string, delimited with dots ("."), and the ACL matches if any of them
14996 matches.
Willy Tarreau6a06a402007-07-15 20:15:28 +020014997
14998String matching applies to verbatim strings as they are passed, with the
14999exception of the backslash ("\") which makes it possible to escape some
15000characters such as the space. If the "-i" flag is passed before the first
15001string, then the matching will be performed ignoring the case. In order
15002to match the string "-i", either set it second, or pass the "--" flag
Willy Tarreau0ba27502007-12-24 16:55:16 +010015003before the first string. Same applies of course to match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020015004
Mathias Weiersmuellercb250fc2019-12-02 09:43:40 +010015005Do not use string matches for binary fetches which might contain null bytes
15006(0x00), as the comparison stops at the occurrence of the first null byte.
15007Instead, convert the binary fetch to a hex string with the hex converter first.
15008
15009Example:
15010 # matches if the string <tag> is present in the binary sample
15011 acl tag_found req.payload(0,0),hex -m sub 3C7461673E
15012
Willy Tarreau6a06a402007-07-15 20:15:28 +020015013
Willy Tarreau74ca5042013-06-11 23:12:07 +0200150147.1.4. Matching regular expressions (regexes)
15015---------------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020015016
15017Just like with string matching, regex matching applies to verbatim strings as
15018they are passed, with the exception of the backslash ("\") which makes it
15019possible to escape some characters such as the space. If the "-i" flag is
15020passed before the first regex, then the matching will be performed ignoring
15021the case. In order to match the string "-i", either set it second, or pass
Willy Tarreau0ba27502007-12-24 16:55:16 +010015022the "--" flag before the first string. Same principle applies of course to
15023match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020015024
15025
Willy Tarreau74ca5042013-06-11 23:12:07 +0200150267.1.5. Matching arbitrary data blocks
15027-------------------------------------
15028
15029It is possible to match some extracted samples against a binary block which may
15030not safely be represented as a string. For this, the patterns must be passed as
15031a series of hexadecimal digits in an even number, when the match method is set
15032to binary. Each sequence of two digits will represent a byte. The hexadecimal
15033digits may be used upper or lower case.
15034
15035Example :
15036 # match "Hello\n" in the input stream (\x48 \x65 \x6c \x6c \x6f \x0a)
15037 acl hello payload(0,6) -m bin 48656c6c6f0a
15038
15039
150407.1.6. Matching IPv4 and IPv6 addresses
15041---------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020015042
15043IPv4 addresses values can be specified either as plain addresses or with a
15044netmask appended, in which case the IPv4 address matches whenever it is
15045within the network. Plain addresses may also be replaced with a resolvable
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010015046host name, but this practice is generally discouraged as it makes it more
Willy Tarreau0ba27502007-12-24 16:55:16 +010015047difficult to read and debug configurations. If hostnames are used, you should
15048at least ensure that they are present in /etc/hosts so that the configuration
15049does not depend on any random DNS match at the moment the configuration is
15050parsed.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015051
Daniel Schnellereba56342016-04-13 00:26:52 +020015052The dotted IPv4 address notation is supported in both regular as well as the
15053abbreviated form with all-0-octets omitted:
15054
15055 +------------------+------------------+------------------+
15056 | Example 1 | Example 2 | Example 3 |
15057 +------------------+------------------+------------------+
15058 | 192.168.0.1 | 10.0.0.12 | 127.0.0.1 |
15059 | 192.168.1 | 10.12 | 127.1 |
15060 | 192.168.0.1/22 | 10.0.0.12/8 | 127.0.0.1/8 |
15061 | 192.168.1/22 | 10.12/8 | 127.1/8 |
15062 +------------------+------------------+------------------+
15063
15064Notice that this is different from RFC 4632 CIDR address notation in which
15065192.168.42/24 would be equivalent to 192.168.42.0/24.
15066
Willy Tarreauceb4ac92012-04-28 00:41:46 +020015067IPv6 may be entered in their usual form, with or without a netmask appended.
15068Only bit counts are accepted for IPv6 netmasks. In order to avoid any risk of
15069trouble with randomly resolved IP addresses, host names are never allowed in
15070IPv6 patterns.
15071
15072HAProxy is also able to match IPv4 addresses with IPv6 addresses in the
15073following situations :
15074 - tested address is IPv4, pattern address is IPv4, the match applies
15075 in IPv4 using the supplied mask if any.
15076 - tested address is IPv6, pattern address is IPv6, the match applies
15077 in IPv6 using the supplied mask if any.
15078 - tested address is IPv6, pattern address is IPv4, the match applies in IPv4
15079 using the pattern's mask if the IPv6 address matches with 2002:IPV4::,
15080 ::IPV4 or ::ffff:IPV4, otherwise it fails.
15081 - tested address is IPv4, pattern address is IPv6, the IPv4 address is first
15082 converted to IPv6 by prefixing ::ffff: in front of it, then the match is
15083 applied in IPv6 using the supplied IPv6 mask.
15084
Willy Tarreau74ca5042013-06-11 23:12:07 +020015085
150867.2. Using ACLs to form conditions
15087----------------------------------
15088
15089Some actions are only performed upon a valid condition. A condition is a
15090combination of ACLs with operators. 3 operators are supported :
15091
15092 - AND (implicit)
15093 - OR (explicit with the "or" keyword or the "||" operator)
15094 - Negation with the exclamation mark ("!")
Willy Tarreau6a06a402007-07-15 20:15:28 +020015095
Willy Tarreau74ca5042013-06-11 23:12:07 +020015096A condition is formed as a disjunctive form:
Willy Tarreau6a06a402007-07-15 20:15:28 +020015097
Willy Tarreau74ca5042013-06-11 23:12:07 +020015098 [!]acl1 [!]acl2 ... [!]acln { or [!]acl1 [!]acl2 ... [!]acln } ...
Willy Tarreaubef91e72013-03-31 23:14:46 +020015099
Willy Tarreau74ca5042013-06-11 23:12:07 +020015100Such conditions are generally used after an "if" or "unless" statement,
15101indicating when the condition will trigger the action.
Willy Tarreaubef91e72013-03-31 23:14:46 +020015102
Willy Tarreau74ca5042013-06-11 23:12:07 +020015103For instance, to block HTTP requests to the "*" URL with methods other than
15104"OPTIONS", as well as POST requests without content-length, and GET or HEAD
15105requests with a content-length greater than 0, and finally every request which
15106is not either GET/HEAD/POST/OPTIONS !
15107
15108 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030015109 http-request deny if HTTP_URL_STAR !METH_OPTIONS || METH_POST missing_cl
15110 http-request deny if METH_GET HTTP_CONTENT
15111 http-request deny unless METH_GET or METH_POST or METH_OPTIONS
Willy Tarreau74ca5042013-06-11 23:12:07 +020015112
15113To select a different backend for requests to static contents on the "www" site
15114and to every request on the "img", "video", "download" and "ftp" hosts :
15115
15116 acl url_static path_beg /static /images /img /css
15117 acl url_static path_end .gif .png .jpg .css .js
15118 acl host_www hdr_beg(host) -i www
15119 acl host_static hdr_beg(host) -i img. video. download. ftp.
15120
Davor Ocelice9ed2812017-12-25 17:49:28 +010015121 # now use backend "static" for all static-only hosts, and for static URLs
Willy Tarreau74ca5042013-06-11 23:12:07 +020015122 # of host "www". Use backend "www" for the rest.
15123 use_backend static if host_static or host_www url_static
15124 use_backend www if host_www
15125
15126It is also possible to form rules using "anonymous ACLs". Those are unnamed ACL
15127expressions that are built on the fly without needing to be declared. They must
15128be enclosed between braces, with a space before and after each brace (because
15129the braces must be seen as independent words). Example :
15130
15131 The following rule :
15132
15133 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030015134 http-request deny if METH_POST missing_cl
Willy Tarreau74ca5042013-06-11 23:12:07 +020015135
15136 Can also be written that way :
15137
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030015138 http-request deny if METH_POST { hdr_cnt(Content-length) eq 0 }
Willy Tarreau74ca5042013-06-11 23:12:07 +020015139
15140It is generally not recommended to use this construct because it's a lot easier
15141to leave errors in the configuration when written that way. However, for very
15142simple rules matching only one source IP address for instance, it can make more
15143sense to use them than to declare ACLs with random names. Another example of
15144good use is the following :
15145
15146 With named ACLs :
15147
15148 acl site_dead nbsrv(dynamic) lt 2
15149 acl site_dead nbsrv(static) lt 2
15150 monitor fail if site_dead
15151
15152 With anonymous ACLs :
15153
15154 monitor fail if { nbsrv(dynamic) lt 2 } || { nbsrv(static) lt 2 }
15155
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030015156See section 4.2 for detailed help on the "http-request deny" and "use_backend"
15157keywords.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015158
15159
151607.3. Fetching samples
15161---------------------
15162
15163Historically, sample fetch methods were only used to retrieve data to match
15164against patterns using ACLs. With the arrival of stick-tables, a new class of
15165sample fetch methods was created, most often sharing the same syntax as their
15166ACL counterpart. These sample fetch methods are also known as "fetches". As
15167of now, ACLs and fetches have converged. All ACL fetch methods have been made
15168available as fetch methods, and ACLs may use any sample fetch method as well.
15169
15170This section details all available sample fetch methods and their output type.
15171Some sample fetch methods have deprecated aliases that are used to maintain
15172compatibility with existing configurations. They are then explicitly marked as
15173deprecated and should not be used in new setups.
15174
15175The ACL derivatives are also indicated when available, with their respective
15176matching methods. These ones all have a well defined default pattern matching
15177method, so it is never necessary (though allowed) to pass the "-m" option to
15178indicate how the sample will be matched using ACLs.
15179
15180As indicated in the sample type versus matching compatibility matrix above,
15181when using a generic sample fetch method in an ACL, the "-m" option is
15182mandatory unless the sample type is one of boolean, integer, IPv4 or IPv6. When
15183the same keyword exists as an ACL keyword and as a standard fetch method, the
15184ACL engine will automatically pick the ACL-only one by default.
15185
15186Some of these keywords support one or multiple mandatory arguments, and one or
15187multiple optional arguments. These arguments are strongly typed and are checked
15188when the configuration is parsed so that there is no risk of running with an
Davor Ocelice9ed2812017-12-25 17:49:28 +010015189incorrect argument (e.g. an unresolved backend name). Fetch function arguments
15190are passed between parenthesis and are delimited by commas. When an argument
Willy Tarreau74ca5042013-06-11 23:12:07 +020015191is optional, it will be indicated below between square brackets ('[ ]'). When
15192all arguments are optional, the parenthesis may be omitted.
15193
15194Thus, the syntax of a standard sample fetch method is one of the following :
15195 - name
15196 - name(arg1)
15197 - name(arg1,arg2)
15198
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015199
152007.3.1. Converters
15201-----------------
15202
Willy Tarreaue6b11e42013-11-26 19:02:32 +010015203Sample fetch methods may be combined with transformations to be applied on top
15204of the fetched sample (also called "converters"). These combinations form what
15205is called "sample expressions" and the result is a "sample". Initially this
15206was only supported by "stick on" and "stick store-request" directives but this
Davor Ocelice9ed2812017-12-25 17:49:28 +010015207has now be extended to all places where samples may be used (ACLs, log-format,
Willy Tarreaue6b11e42013-11-26 19:02:32 +010015208unique-id-format, add-header, ...).
15209
15210These transformations are enumerated as a series of specific keywords after the
15211sample fetch method. These keywords may equally be appended immediately after
15212the fetch keyword's argument, delimited by a comma. These keywords can also
Davor Ocelice9ed2812017-12-25 17:49:28 +010015213support some arguments (e.g. a netmask) which must be passed in parenthesis.
Willy Tarreau0ba27502007-12-24 16:55:16 +010015214
Willy Tarreau97707872015-01-27 15:12:13 +010015215A certain category of converters are bitwise and arithmetic operators which
15216support performing basic operations on integers. Some bitwise operations are
15217supported (and, or, xor, cpl) and some arithmetic operations are supported
15218(add, sub, mul, div, mod, neg). Some comparators are provided (odd, even, not,
15219bool) which make it possible to report a match without having to write an ACL.
15220
Willy Tarreau74ca5042013-06-11 23:12:07 +020015221The currently available list of transformation keywords include :
Willy Tarreau0ba27502007-12-24 16:55:16 +010015222
Ben Shillitof25e8e52016-12-02 14:25:37 +00001522351d.single(<prop>[,<prop>*])
15224 Returns values for the properties requested as a string, where values are
15225 separated by the delimiter specified with "51degrees-property-separator".
15226 The device is identified using the User-Agent header passed to the
15227 converter. The function can be passed up to five property names, and if a
15228 property name can't be found, the value "NoData" is returned.
15229
15230 Example :
Davor Ocelice9ed2812017-12-25 17:49:28 +010015231 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request,
15232 # containing values for the three properties requested by using the
Ben Shillitof25e8e52016-12-02 14:25:37 +000015233 # User-Agent passed to the converter.
15234 frontend http-in
15235 bind *:8081
15236 default_backend servers
15237 http-request set-header X-51D-DeviceTypeMobileTablet \
15238 %[req.fhdr(User-Agent),51d.single(DeviceType,IsMobile,IsTablet)]
15239
Willy Tarreau97707872015-01-27 15:12:13 +010015240add(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015241 Adds <value> to the input value of type signed integer, and returns the
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015242 result as a signed integer. <value> can be a numeric value or a variable
Daniel Schneller0b547052016-03-21 20:46:57 +010015243 name. The name of the variable starts with an indication about its scope. The
15244 scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015245 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015246 "sess" : the variable is shared with the whole session
15247 "txn" : the variable is shared with the transaction (request and response)
15248 "req" : the variable is shared only during request processing
15249 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010015250 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015251 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010015252
Nenad Merdanovicc31499d2019-03-23 11:00:32 +010015253aes_gcm_dec(<bits>,<nonce>,<key>,<aead_tag>)
15254 Decrypts the raw byte input using the AES128-GCM, AES192-GCM or
15255 AES256-GCM algorithm, depending on the <bits> parameter. All other parameters
15256 need to be base64 encoded and the returned result is in raw byte format.
15257 If the <aead_tag> validation fails, the converter doesn't return any data.
15258 The <nonce>, <key> and <aead_tag> can either be strings or variables. This
15259 converter requires at least OpenSSL 1.0.1.
15260
15261 Example:
15262 http-response set-header X-Decrypted-Text %[var(txn.enc),\
15263 aes_gcm_dec(128,txn.nonce,Zm9vb2Zvb29mb29wZm9vbw==,txn.aead_tag)]
15264
Willy Tarreau97707872015-01-27 15:12:13 +010015265and(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015266 Performs a bitwise "AND" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015267 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010015268 numeric value or a variable name. The name of the variable starts with an
15269 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015270 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015271 "sess" : the variable is shared with the whole session
15272 "txn" : the variable is shared with the transaction (request and response)
15273 "req" : the variable is shared only during request processing
15274 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010015275 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015276 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010015277
Holger Just1bfc24b2017-05-06 00:56:53 +020015278b64dec
15279 Converts (decodes) a base64 encoded input string to its binary
15280 representation. It performs the inverse operation of base64().
15281
Emeric Brun53d1a982014-04-30 18:21:37 +020015282base64
15283 Converts a binary input sample to a base64 string. It is used to log or
Davor Ocelice9ed2812017-12-25 17:49:28 +010015284 transfer binary content in a way that can be reliably transferred (e.g.
Emeric Brun53d1a982014-04-30 18:21:37 +020015285 an SSL ID can be copied in a header).
15286
Willy Tarreau97707872015-01-27 15:12:13 +010015287bool
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015288 Returns a boolean TRUE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010015289 non-null, otherwise returns FALSE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010015290 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010015291 presence of a flag).
15292
Emeric Brun54c4ac82014-11-03 15:32:43 +010015293bytes(<offset>[,<length>])
15294 Extracts some bytes from an input binary sample. The result is a binary
15295 sample starting at an offset (in bytes) of the original sample and
Tim Düsterhus4896c442016-11-29 02:15:19 +010015296 optionally truncated at the given length.
Emeric Brun54c4ac82014-11-03 15:32:43 +010015297
Willy Tarreau280f42b2018-02-19 15:34:12 +010015298concat([<start>],[<var>],[<end>])
15299 Concatenates up to 3 fields after the current sample which is then turned to
15300 a string. The first one, <start>, is a constant string, that will be appended
15301 immediately after the existing sample. It may be omitted if not used. The
15302 second one, <var>, is a variable name. The variable will be looked up, its
15303 contents converted to a string, and it will be appended immediately after the
15304 <first> part. If the variable is not found, nothing is appended. It may be
15305 omitted as well. The third field, <end> is a constant string that will be
15306 appended after the variable. It may also be omitted. Together, these elements
15307 allow to concatenate variables with delimiters to an existing set of
15308 variables. This can be used to build new variables made of a succession of
Willy Tarreauef21fac2020-02-14 13:37:20 +010015309 other variables, such as colon-delimited values. If commas or closing
Daniel Corbett67a82712020-07-06 23:01:19 -040015310 parenthesis are needed as delimiters, they must be protected by quotes or
Willy Tarreauef21fac2020-02-14 13:37:20 +010015311 backslashes, themselves protected so that they are not stripped by the first
15312 level parser. See examples below.
Willy Tarreau280f42b2018-02-19 15:34:12 +010015313
15314 Example:
15315 tcp-request session set-var(sess.src) src
15316 tcp-request session set-var(sess.dn) ssl_c_s_dn
15317 tcp-request session set-var(txn.sig) str(),concat(<ip=,sess.ip,>),concat(<dn=,sess.dn,>)
Willy Tarreauef21fac2020-02-14 13:37:20 +010015318 tcp-request session set-var(txn.ipport) "str(),concat('addr=(',sess.ip),concat(',',sess.port,')')"
Willy Tarreau280f42b2018-02-19 15:34:12 +010015319 http-request set-header x-hap-sig %[var(txn.sig)]
15320
Willy Tarreau97707872015-01-27 15:12:13 +010015321cpl
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015322 Takes the input value of type signed integer, applies a ones-complement
15323 (flips all bits) and returns the result as an signed integer.
Willy Tarreau97707872015-01-27 15:12:13 +010015324
Willy Tarreau80599772015-01-20 19:35:24 +010015325crc32([<avalanche>])
15326 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32
15327 hash function. Optionally, it is possible to apply a full avalanche hash
15328 function to the output if the optional <avalanche> argument equals 1. This
15329 converter uses the same functions as used by the various hash-based load
15330 balancing algorithms, so it will provide exactly the same results. It is
15331 provided for compatibility with other software which want a CRC32 to be
15332 computed on some input keys, so it follows the most common implementation as
15333 found in Ethernet, Gzip, PNG, etc... It is slower than the other algorithms
15334 but may provide a better or at least less predictable distribution. It must
15335 not be used for security purposes as a 32-bit hash is trivial to break. See
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010015336 also "djb2", "sdbm", "wt6", "crc32c" and the "hash-type" directive.
15337
15338crc32c([<avalanche>])
15339 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32C
15340 hash function. Optionally, it is possible to apply a full avalanche hash
15341 function to the output if the optional <avalanche> argument equals 1. This
15342 converter uses the same functions as described in RFC4960, Appendix B [8].
15343 It is provided for compatibility with other software which want a CRC32C to be
15344 computed on some input keys. It is slower than the other algorithms and it must
15345 not be used for security purposes as a 32-bit hash is trivial to break. See
15346 also "djb2", "sdbm", "wt6", "crc32" and the "hash-type" directive.
Willy Tarreau80599772015-01-20 19:35:24 +010015347
Christopher Fauletea159d62020-04-01 16:21:44 +020015348cut_crlf
15349 Cuts the string representation of the input sample on the first carriage
15350 return ('\r') or newline ('\n') character found. Only the string length is
15351 updated.
15352
David Carlier29b3ca32015-09-25 14:09:21 +010015353da-csv-conv(<prop>[,<prop>*])
David Carlier4542b102015-06-01 13:54:29 +020015354 Asks the DeviceAtlas converter to identify the User Agent string passed on
15355 input, and to emit a string made of the concatenation of the properties
15356 enumerated in argument, delimited by the separator defined by the global
15357 keyword "deviceatlas-property-separator", or by default the pipe character
David Carlier840b0242016-03-16 10:09:55 +000015358 ('|'). There's a limit of 12 different properties imposed by the haproxy
David Carlier4542b102015-06-01 13:54:29 +020015359 configuration language.
15360
15361 Example:
15362 frontend www
Cyril Bonté307ee1e2015-09-28 23:16:06 +020015363 bind *:8881
15364 default_backend servers
David Carlier840b0242016-03-16 10:09:55 +000015365 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 +020015366
Willy Tarreau0851fd52019-12-17 10:07:25 +010015367debug([<prefix][,<destination>])
15368 This converter is used as debug tool. It takes a capture of the input sample
15369 and sends it to event sink <destination>, which may designate a ring buffer
15370 such as "buf0", as well as "stdout", or "stderr". Available sinks may be
15371 checked at run time by issuing "show events" on the CLI. When not specified,
15372 the output will be "buf0", which may be consulted via the CLI's "show events"
15373 command. An optional prefix <prefix> may be passed to help distinguish
15374 outputs from multiple expressions. It will then appear before the colon in
15375 the output message. The input sample is passed as-is on the output, so that
15376 it is safe to insert the debug converter anywhere in a chain, even with non-
15377 printable sample types.
15378
15379 Example:
15380 tcp-request connection track-sc0 src,debug(track-sc)
Thierry FOURNIER9687c772015-05-07 15:46:29 +020015381
Patrick Gansterer8e366512020-04-22 16:47:57 +020015382digest(<algorithm>)
15383 Converts a binary input sample to a message digest. The result is a binary
15384 sample. The <algorithm> must be an OpenSSL message digest name (e.g. sha256).
15385
15386 Please note that this converter is only available when haproxy has been
15387 compiled with USE_OPENSSL.
15388
Willy Tarreau97707872015-01-27 15:12:13 +010015389div(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015390 Divides the input value of type signed integer by <value>, and returns the
15391 result as an signed integer. If <value> is null, the largest unsigned
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015392 integer is returned (typically 2^63-1). <value> can be a numeric value or a
Daniel Schneller0b547052016-03-21 20:46:57 +010015393 variable name. The name of the variable starts with an indication about its
15394 scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015395 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015396 "sess" : the variable is shared with the whole session
15397 "txn" : the variable is shared with the transaction (request and response)
15398 "req" : the variable is shared only during request processing
15399 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010015400 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015401 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010015402
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015403djb2([<avalanche>])
15404 Hashes a binary input sample into an unsigned 32-bit quantity using the DJB2
15405 hash function. Optionally, it is possible to apply a full avalanche hash
15406 function to the output if the optional <avalanche> argument equals 1. This
15407 converter uses the same functions as used by the various hash-based load
15408 balancing algorithms, so it will provide exactly the same results. It is
15409 mostly intended for debugging, but can be used as a stick-table entry to
15410 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010015411 32-bit hash is trivial to break. See also "crc32", "sdbm", "wt6", "crc32c",
15412 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015413
Willy Tarreau97707872015-01-27 15:12:13 +010015414even
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015415 Returns a boolean TRUE if the input value of type signed integer is even
Willy Tarreau97707872015-01-27 15:12:13 +010015416 otherwise returns FALSE. It is functionally equivalent to "not,and(1),bool".
15417
Marcin Deranek9631a282018-04-16 14:30:46 +020015418field(<index>,<delimiters>[,<count>])
15419 Extracts the substring at the given index counting from the beginning
15420 (positive index) or from the end (negative index) considering given delimiters
15421 from an input string. Indexes start at 1 or -1 and delimiters are a string
15422 formatted list of chars. Optionally you can specify <count> of fields to
15423 extract (default: 1). Value of 0 indicates extraction of all remaining
15424 fields.
15425
15426 Example :
15427 str(f1_f2_f3__f5),field(5,_) # f5
15428 str(f1_f2_f3__f5),field(2,_,0) # f2_f3__f5
15429 str(f1_f2_f3__f5),field(2,_,2) # f2_f3
15430 str(f1_f2_f3__f5),field(-2,_,3) # f2_f3_
15431 str(f1_f2_f3__f5),field(-3,_,0) # f1_f2_f3
Emeric Brunf399b0d2014-11-03 17:07:03 +010015432
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015433hex
Davor Ocelice9ed2812017-12-25 17:49:28 +010015434 Converts a binary input sample to a hex string containing two hex digits per
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015435 input byte. It is used to log or transfer hex dumps of some binary input data
Davor Ocelice9ed2812017-12-25 17:49:28 +010015436 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 +020015437 header).
Thierry FOURNIER2f49d6d2014-03-12 15:01:52 +010015438
Dragan Dosen3f957b22017-10-24 09:27:34 +020015439hex2i
15440 Converts a hex string containing two hex digits per input byte to an
John Roeslerfb2fce12019-07-10 15:45:51 -050015441 integer. If the input value cannot be converted, then zero is returned.
Dragan Dosen3f957b22017-10-24 09:27:34 +020015442
Christopher Faulet4ccc12f2020-04-01 09:08:32 +020015443htonl
15444 Converts the input integer value to its 32-bit binary representation in the
15445 network byte order. Because sample fetches own signed 64-bit integer, when
15446 this converter is used, the input integer value is first casted to an
15447 unsigned 32-bit integer.
15448
Patrick Gansterer8e366512020-04-22 16:47:57 +020015449hmac(<algorithm>, <key>)
15450 Converts a binary input sample to a message authentication code with the given
15451 key. The result is a binary sample. The <algorithm> must be one of the
15452 registered OpenSSL message digest names (e.g. sha256). The <key> parameter must
15453 be base64 encoded and can either be a string or a variable.
15454
15455 Please note that this converter is only available when haproxy has been
15456 compiled with USE_OPENSSL.
15457
Cyril Bonté6bcd1822019-11-05 23:13:59 +010015458http_date([<offset],[<unit>])
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015459 Converts an integer supposed to contain a date since epoch to a string
15460 representing this date in a format suitable for use in HTTP header fields. If
Damien Claisseae6f1252019-10-30 15:57:28 +000015461 an offset value is specified, then it is added to the date before the
15462 conversion is operated. This is particularly useful to emit Date header fields,
15463 Expires values in responses when combined with a positive offset, or
15464 Last-Modified values when the offset is negative.
15465 If a unit value is specified, then consider the timestamp as either
15466 "s" for seconds (default behavior), "ms" for milliseconds, or "us" for
15467 microseconds since epoch. Offset is assumed to have the same unit as
15468 input timestamp.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015469
Tim Duesterhus3943e4f2020-09-11 14:25:23 +020015470iif(<true>,<false>)
15471 Returns the <true> string if the input value is true. Returns the <false>
15472 string otherwise.
15473
15474 Example:
Tim Duesterhus870713b2020-09-11 17:13:12 +020015475 http-request set-header x-forwarded-proto %[ssl_fc,iif(https,http)]
Tim Duesterhus3943e4f2020-09-11 14:25:23 +020015476
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015477in_table(<table>)
15478 Uses the string representation of the input sample to perform a look up in
15479 the specified table. If the key is not found in the table, a boolean false
15480 is returned. Otherwise a boolean true is returned. This can be used to verify
Davor Ocelice9ed2812017-12-25 17:49:28 +010015481 the presence of a certain key in a table tracking some elements (e.g. whether
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015482 or not a source IP address or an Authorization header was already seen).
15483
Tim Duesterhus1478aa72018-01-25 16:24:51 +010015484ipmask(<mask4>, [<mask6>])
15485 Apply a mask to an IP address, and use the result for lookups and storage.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020015486 This can be used to make all hosts within a certain mask to share the same
Tim Duesterhus1478aa72018-01-25 16:24:51 +010015487 table entries and as such use the same server. The mask4 can be passed in
15488 dotted form (e.g. 255.255.255.0) or in CIDR form (e.g. 24). The mask6 can
15489 be passed in quadruplet form (e.g. ffff:ffff::) or in CIDR form (e.g. 64).
15490 If no mask6 is given IPv6 addresses will fail to convert for backwards
15491 compatibility reasons.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020015492
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015493json([<input-code>])
Davor Ocelice9ed2812017-12-25 17:49:28 +010015494 Escapes the input string and produces an ASCII output string ready to use as a
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015495 JSON string. The converter tries to decode the input string according to the
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020015496 <input-code> parameter. It can be "ascii", "utf8", "utf8s", "utf8p" or
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015497 "utf8ps". The "ascii" decoder never fails. The "utf8" decoder detects 3 types
15498 of errors:
15499 - bad UTF-8 sequence (lone continuation byte, bad number of continuation
15500 bytes, ...)
15501 - invalid range (the decoded value is within a UTF-8 prohibited range),
15502 - code overlong (the value is encoded with more bytes than necessary).
15503
15504 The UTF-8 JSON encoding can produce a "too long value" error when the UTF-8
15505 character is greater than 0xffff because the JSON string escape specification
15506 only authorizes 4 hex digits for the value encoding. The UTF-8 decoder exists
15507 in 4 variants designated by a combination of two suffix letters : "p" for
15508 "permissive" and "s" for "silently ignore". The behaviors of the decoders
15509 are :
Davor Ocelice9ed2812017-12-25 17:49:28 +010015510 - "ascii" : never fails;
15511 - "utf8" : fails on any detected errors;
15512 - "utf8s" : never fails, but removes characters corresponding to errors;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015513 - "utf8p" : accepts and fixes the overlong errors, but fails on any other
Davor Ocelice9ed2812017-12-25 17:49:28 +010015514 error;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015515 - "utf8ps" : never fails, accepts and fixes the overlong errors, but removes
15516 characters corresponding to the other errors.
15517
15518 This converter is particularly useful for building properly escaped JSON for
Davor Ocelice9ed2812017-12-25 17:49:28 +010015519 logging to servers which consume JSON-formatted traffic logs.
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015520
15521 Example:
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015522 capture request header Host len 15
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020015523 capture request header user-agent len 150
15524 log-format '{"ip":"%[src]","user-agent":"%[capture.req.hdr(1),json(utf8s)]"}'
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015525
15526 Input request from client 127.0.0.1:
15527 GET / HTTP/1.0
15528 User-Agent: Very "Ugly" UA 1/2
15529
15530 Output log:
15531 {"ip":"127.0.0.1","user-agent":"Very \"Ugly\" UA 1\/2"}
15532
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015533language(<value>[,<default>])
15534 Returns the value with the highest q-factor from a list as extracted from the
15535 "accept-language" header using "req.fhdr". Values with no q-factor have a
15536 q-factor of 1. Values with a q-factor of 0 are dropped. Only values which
15537 belong to the list of semi-colon delimited <values> will be considered. The
15538 argument <value> syntax is "lang[;lang[;lang[;...]]]". If no value matches the
15539 given list and a default value is provided, it is returned. Note that language
15540 names may have a variant after a dash ('-'). If this variant is present in the
15541 list, it will be matched, but if it is not, only the base language is checked.
15542 The match is case-sensitive, and the output string is always one of those
Davor Ocelice9ed2812017-12-25 17:49:28 +010015543 provided in arguments. The ordering of arguments is meaningless, only the
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015544 ordering of the values in the request counts, as the first value among
15545 multiple sharing the same q-factor is used.
Thierry FOURNIERad903512014-04-11 17:51:01 +020015546
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015547 Example :
Thierry FOURNIERad903512014-04-11 17:51:01 +020015548
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015549 # this configuration switches to the backend matching a
15550 # given language based on the request :
Thierry FOURNIERad903512014-04-11 17:51:01 +020015551
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015552 acl es req.fhdr(accept-language),language(es;fr;en) -m str es
15553 acl fr req.fhdr(accept-language),language(es;fr;en) -m str fr
15554 acl en req.fhdr(accept-language),language(es;fr;en) -m str en
15555 use_backend spanish if es
15556 use_backend french if fr
15557 use_backend english if en
15558 default_backend choose_your_language
Thierry FOURNIERad903512014-04-11 17:51:01 +020015559
Willy Tarreau60a2ee72017-12-15 07:13:48 +010015560length
Etienne Carriereed0d24e2017-12-13 13:41:34 +010015561 Get the length of the string. This can only be placed after a string
15562 sample fetch function or after a transformation keyword returning a string
15563 type. The result is of type integer.
15564
Willy Tarreauffcb2e42014-07-10 16:29:08 +020015565lower
15566 Convert a string sample to lower case. This can only be placed after a string
15567 sample fetch function or after a transformation keyword returning a string
15568 type. The result is of type string.
15569
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020015570ltime(<format>[,<offset>])
15571 Converts an integer supposed to contain a date since epoch to a string
15572 representing this date in local time using a format defined by the <format>
15573 string using strftime(3). The purpose is to allow any date format to be used
15574 in logs. An optional <offset> in seconds may be applied to the input date
15575 (positive or negative). See the strftime() man page for the format supported
15576 by your operating system. See also the utime converter.
15577
15578 Example :
15579
15580 # Emit two colons, one with the local time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010015581 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020015582 log-format %[date,ltime(%Y%m%d%H%M%S)]\ %ci:%cp
15583
Christopher Faulet51fc9d12020-04-01 17:24:41 +020015584ltrim(<chars>)
15585 Skips any characters from <chars> from the beginning of the string
15586 representation of the input sample.
15587
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015588map(<map_file>[,<default_value>])
15589map_<match_type>(<map_file>[,<default_value>])
15590map_<match_type>_<output_type>(<map_file>[,<default_value>])
15591 Search the input value from <map_file> using the <match_type> matching method,
15592 and return the associated value converted to the type <output_type>. If the
15593 input value cannot be found in the <map_file>, the converter returns the
15594 <default_value>. If the <default_value> is not set, the converter fails and
15595 acts as if no input value could be fetched. If the <match_type> is not set, it
15596 defaults to "str". Likewise, if the <output_type> is not set, it defaults to
15597 "str". For convenience, the "map" keyword is an alias for "map_str" and maps a
15598 string to another string.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010015599
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015600 It is important to avoid overlapping between the keys : IP addresses and
15601 strings are stored in trees, so the first of the finest match will be used.
15602 Other keys are stored in lists, so the first matching occurrence will be used.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010015603
Tim Düsterhus4896c442016-11-29 02:15:19 +010015604 The following array contains the list of all map functions available sorted by
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015605 input type, match type and output type.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010015606
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015607 input type | match method | output type str | output type int | output type ip
15608 -----------+--------------+-----------------+-----------------+---------------
15609 str | str | map_str | map_str_int | map_str_ip
15610 -----------+--------------+-----------------+-----------------+---------------
Willy Tarreau787a4c02014-05-10 07:55:30 +020015611 str | beg | map_beg | map_beg_int | map_end_ip
15612 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015613 str | sub | map_sub | map_sub_int | map_sub_ip
15614 -----------+--------------+-----------------+-----------------+---------------
15615 str | dir | map_dir | map_dir_int | map_dir_ip
15616 -----------+--------------+-----------------+-----------------+---------------
15617 str | dom | map_dom | map_dom_int | map_dom_ip
15618 -----------+--------------+-----------------+-----------------+---------------
15619 str | end | map_end | map_end_int | map_end_ip
15620 -----------+--------------+-----------------+-----------------+---------------
Ruoshan Huang3c5e3742016-12-02 16:25:31 +080015621 str | reg | map_reg | map_reg_int | map_reg_ip
15622 -----------+--------------+-----------------+-----------------+---------------
15623 str | reg | map_regm | map_reg_int | map_reg_ip
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015624 -----------+--------------+-----------------+-----------------+---------------
15625 int | int | map_int | map_int_int | map_int_ip
15626 -----------+--------------+-----------------+-----------------+---------------
15627 ip | ip | map_ip | map_ip_int | map_ip_ip
15628 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010015629
Thierry Fournier8feaa662016-02-10 22:55:20 +010015630 The special map called "map_regm" expect matching zone in the regular
15631 expression and modify the output replacing back reference (like "\1") by
15632 the corresponding match text.
15633
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015634 The file contains one key + value per line. Lines which start with '#' are
15635 ignored, just like empty lines. Leading tabs and spaces are stripped. The key
15636 is then the first "word" (series of non-space/tabs characters), and the value
15637 is what follows this series of space/tab till the end of the line excluding
15638 trailing spaces/tabs.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010015639
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015640 Example :
15641
15642 # this is a comment and is ignored
15643 2.22.246.0/23 United Kingdom \n
15644 <-><-----------><--><------------><---->
15645 | | | | `- trailing spaces ignored
15646 | | | `---------- value
15647 | | `-------------------- middle spaces ignored
15648 | `---------------------------- key
15649 `------------------------------------ leading spaces ignored
15650
Willy Tarreau97707872015-01-27 15:12:13 +010015651mod(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015652 Divides the input value of type signed integer by <value>, and returns the
15653 remainder as an signed integer. If <value> is null, then zero is returned.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015654 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010015655 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015656 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015657 "sess" : the variable is shared with the whole session
15658 "txn" : the variable is shared with the transaction (request and response)
15659 "req" : the variable is shared only during request processing
15660 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010015661 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015662 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010015663
15664mul(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015665 Multiplies the input value of type signed integer by <value>, and returns
Thierry FOURNIER00c005c2015-07-08 01:10:21 +020015666 the product as an signed integer. In case of overflow, the largest possible
15667 value for the sign is returned so that the operation doesn't wrap around.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015668 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010015669 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015670 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015671 "sess" : the variable is shared with the whole session
15672 "txn" : the variable is shared with the transaction (request and response)
15673 "req" : the variable is shared only during request processing
15674 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010015675 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015676 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010015677
Nenad Merdanovicb7e7c472017-03-12 21:56:55 +010015678nbsrv
15679 Takes an input value of type string, interprets it as a backend name and
15680 returns the number of usable servers in that backend. Can be used in places
15681 where we want to look up a backend from a dynamic name, like a result of a
15682 map lookup.
15683
Willy Tarreau97707872015-01-27 15:12:13 +010015684neg
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015685 Takes the input value of type signed integer, computes the opposite value,
15686 and returns the remainder as an signed integer. 0 is identity. This operator
15687 is provided for reversed subtracts : in order to subtract the input from a
15688 constant, simply perform a "neg,add(value)".
Willy Tarreau97707872015-01-27 15:12:13 +010015689
15690not
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015691 Returns a boolean FALSE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010015692 non-null, otherwise returns TRUE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010015693 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010015694 absence of a flag).
15695
15696odd
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015697 Returns a boolean TRUE if the input value of type signed integer is odd
Willy Tarreau97707872015-01-27 15:12:13 +010015698 otherwise returns FALSE. It is functionally equivalent to "and(1),bool".
15699
15700or(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015701 Performs a bitwise "OR" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015702 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010015703 numeric value or a variable name. The name of the variable starts with an
15704 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015705 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015706 "sess" : the variable is shared with the whole session
15707 "txn" : the variable is shared with the transaction (request and response)
15708 "req" : the variable is shared only during request processing
15709 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010015710 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
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010015713protobuf(<field_number>,[<field_type>])
15714 This extracts the protocol buffers message field in raw mode of an input binary
15715 sample representation of a protocol buffer message with <field_number> as field
15716 number (dotted notation) if <field_type> is not present, or as an integer sample
15717 if this field is present (see also "ungrpc" below).
15718 The list of the authorized types is the following one: "int32", "int64", "uint32",
15719 "uint64", "sint32", "sint64", "bool", "enum" for the "varint" wire type 0
15720 "fixed64", "sfixed64", "double" for the 64bit wire type 1, "fixed32", "sfixed32",
15721 "float" for the wire type 5. Note that "string" is considered as a length-delimited
15722 type, so it does not require any <field_type> argument to be extracted.
15723 More information may be found here about the protocol buffers message field types:
15724 https://developers.google.com/protocol-buffers/docs/encoding
15725
Willy Tarreauc4dc3502015-01-23 20:39:28 +010015726regsub(<regex>,<subst>[,<flags>])
Willy Tarreau7eda8492015-01-20 19:47:06 +010015727 Applies a regex-based substitution to the input string. It does the same
15728 operation as the well-known "sed" utility with "s/<regex>/<subst>/". By
15729 default it will replace in the input string the first occurrence of the
15730 largest part matching the regular expression <regex> with the substitution
15731 string <subst>. It is possible to replace all occurrences instead by adding
15732 the flag "g" in the third argument <flags>. It is also possible to make the
15733 regex case insensitive by adding the flag "i" in <flags>. Since <flags> is a
15734 string, it is made up from the concatenation of all desired flags. Thus if
15735 both "i" and "g" are desired, using "gi" or "ig" will have the same effect.
Willy Tarreauef21fac2020-02-14 13:37:20 +010015736 The first use of this converter is to replace certain characters or sequence
15737 of characters with other ones.
15738
15739 It is highly recommended to enclose the regex part using protected quotes to
15740 improve clarity and never have a closing parenthesis from the regex mixed up
15741 with the parenthesis from the function. Just like in Bourne shell, the first
15742 level of quotes is processed when delimiting word groups on the line, a
15743 second level is usable for argument. It is recommended to use single quotes
15744 outside since these ones do not try to resolve backslashes nor dollar signs.
Willy Tarreau7eda8492015-01-20 19:47:06 +010015745
Willy Tarreaucd0d2ed2020-02-14 17:33:06 +010015746 Examples:
Willy Tarreau7eda8492015-01-20 19:47:06 +010015747
15748 # de-duplicate "/" in header "x-path".
15749 # input: x-path: /////a///b/c/xzxyz/
15750 # output: x-path: /a/b/c/xzxyz/
Willy Tarreauef21fac2020-02-14 13:37:20 +010015751 http-request set-header x-path "%[hdr(x-path),regsub('/+','/','g')]"
Willy Tarreau7eda8492015-01-20 19:47:06 +010015752
Willy Tarreaucd0d2ed2020-02-14 17:33:06 +010015753 # copy query string to x-query and drop all leading '?', ';' and '&'
15754 http-request set-header x-query "%[query,regsub([?;&]*,'')]"
15755
Jerome Magnin07e1e3c2020-02-16 19:20:19 +010015756 # capture groups and backreferences
15757 # both lines do the same.
Willy Tarreau465dc7d2020-10-08 18:05:56 +020015758 http-request redirect location %[url,'regsub("(foo|bar)([0-9]+)?","\2\1",i)']
Jerome Magnin07e1e3c2020-02-16 19:20:19 +010015759 http-request redirect location %[url,regsub(\"(foo|bar)([0-9]+)?\",\"\2\1\",i)]
15760
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020015761capture-req(<id>)
15762 Capture the string entry in the request slot <id> and returns the entry as
15763 is. If the slot doesn't exist, the capture fails silently.
15764
15765 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020015766 "http-response capture", "capture.req.hdr" and
15767 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020015768
15769capture-res(<id>)
15770 Capture the string entry in the response slot <id> and returns the entry as
15771 is. If the slot doesn't exist, the capture fails silently.
15772
15773 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020015774 "http-response capture", "capture.req.hdr" and
15775 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020015776
Christopher Faulet568415a2020-04-01 17:24:47 +020015777rtrim(<chars>)
15778 Skips any characters from <chars> from the end of the string representation
15779 of the input sample.
15780
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015781sdbm([<avalanche>])
15782 Hashes a binary input sample into an unsigned 32-bit quantity using the SDBM
15783 hash function. Optionally, it is possible to apply a full avalanche hash
15784 function to the output if the optional <avalanche> argument equals 1. This
15785 converter uses the same functions as used by the various hash-based load
15786 balancing algorithms, so it will provide exactly the same results. It is
15787 mostly intended for debugging, but can be used as a stick-table entry to
15788 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010015789 32-bit hash is trivial to break. See also "crc32", "djb2", "wt6", "crc32c",
15790 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015791
Tim Duesterhusf38175c2020-06-09 11:48:42 +020015792secure_memcmp(<var>)
15793 Compares the contents of <var> with the input value. Both values are treated
15794 as a binary string. Returns a boolean indicating whether both binary strings
15795 match.
15796
15797 If both binary strings have the same length then the comparison will be
15798 performed in constant time.
15799
15800 Please note that this converter is only available when haproxy has been
15801 compiled with USE_OPENSSL.
15802
15803 Example :
15804
15805 http-request set-var(txn.token) hdr(token)
15806 # Check whether the token sent by the client matches the secret token
15807 # value, without leaking the contents using a timing attack.
15808 acl token_given str(my_secret_token),secure_memcmp(txn.token)
15809
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020015810set-var(<var name>)
Davor Ocelice9ed2812017-12-25 17:49:28 +010015811 Sets a variable with the input content and returns the content on the output
15812 as-is. The variable keeps the value and the associated input type. The name of
15813 the variable starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015814 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015815 "sess" : the variable is shared with the whole session
15816 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020015817 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010015818 "req" : the variable is shared only during request processing,
15819 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020015820 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015821 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020015822
Dragan Dosen6e5a9ca2017-10-24 09:18:23 +020015823sha1
Tim Duesterhusd4376302019-06-17 12:41:44 +020015824 Converts a binary input sample to a SHA-1 digest. The result is a binary
Dragan Dosen6e5a9ca2017-10-24 09:18:23 +020015825 sample with length of 20 bytes.
15826
Tim Duesterhusd4376302019-06-17 12:41:44 +020015827sha2([<bits>])
15828 Converts a binary input sample to a digest in the SHA-2 family. The result
15829 is a binary sample with length of <bits>/8 bytes.
15830
15831 Valid values for <bits> are 224, 256, 384, 512, each corresponding to
15832 SHA-<bits>. The default value is 256.
15833
15834 Please note that this converter is only available when haproxy has been
15835 compiled with USE_OPENSSL.
15836
Nenad Merdanovic177adc92019-08-27 01:58:13 +020015837srv_queue
15838 Takes an input value of type string, either a server name or <backend>/<server>
15839 format and returns the number of queued sessions on that server. Can be used
15840 in places where we want to look up queued sessions from a dynamic name, like a
15841 cookie value (e.g. req.cook(SRVID),srv_queue) and then make a decision to break
15842 persistence or direct a request elsewhere.
15843
Tim Duesterhusca097c12018-04-27 21:18:45 +020015844strcmp(<var>)
15845 Compares the contents of <var> with the input value of type string. Returns
15846 the result as a signed integer compatible with strcmp(3): 0 if both strings
15847 are identical. A value less than 0 if the left string is lexicographically
15848 smaller than the right string or if the left string is shorter. A value greater
15849 than 0 otherwise (right string greater than left string or the right string is
15850 shorter).
15851
Tim Duesterhusf38175c2020-06-09 11:48:42 +020015852 See also the secure_memcmp converter if you need to compare two binary
15853 strings in constant time.
15854
Tim Duesterhusca097c12018-04-27 21:18:45 +020015855 Example :
15856
15857 http-request set-var(txn.host) hdr(host)
15858 # Check whether the client is attempting domain fronting.
15859 acl ssl_sni_http_host_match ssl_fc_sni,strcmp(txn.host) eq 0
15860
15861
Willy Tarreau97707872015-01-27 15:12:13 +010015862sub(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015863 Subtracts <value> from the input value of type signed integer, and returns
15864 the result as an signed integer. Note: in order to subtract the input from
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015865 a constant, simply perform a "neg,add(value)". <value> can be a numeric value
Daniel Schneller0b547052016-03-21 20:46:57 +010015866 or a variable name. The name of the variable starts with an indication about
15867 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015868 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015869 "sess" : the variable is shared with the whole session
15870 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015871 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010015872 "req" : the variable is shared only during request processing,
15873 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015874 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015875 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010015876
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015877table_bytes_in_rate(<table>)
15878 Uses the string representation of the input sample to perform a look up in
15879 the specified table. If the key is not found in the table, integer value zero
15880 is returned. Otherwise the converter returns the average client-to-server
15881 bytes rate associated with the input sample in the designated table, measured
15882 in amount of bytes over the period configured in the table. See also the
15883 sc_bytes_in_rate sample fetch keyword.
15884
15885
15886table_bytes_out_rate(<table>)
15887 Uses the string representation of the input sample to perform a look up in
15888 the specified table. If the key is not found in the table, integer value zero
15889 is returned. Otherwise the converter returns the average server-to-client
15890 bytes rate associated with the input sample in the designated table, measured
15891 in amount of bytes over the period configured in the table. See also the
15892 sc_bytes_out_rate sample fetch keyword.
15893
15894table_conn_cnt(<table>)
15895 Uses the string representation of the input sample to perform a look up in
15896 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010015897 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015898 connections associated with the input sample in the designated table. See
15899 also the sc_conn_cnt sample fetch keyword.
15900
15901table_conn_cur(<table>)
15902 Uses the string representation of the input sample to perform a look up in
15903 the specified table. If the key is not found in the table, integer value zero
15904 is returned. Otherwise the converter returns the current amount of concurrent
15905 tracked connections associated with the input sample in the designated table.
15906 See also the sc_conn_cur sample fetch keyword.
15907
15908table_conn_rate(<table>)
15909 Uses the string representation of the input sample to perform a look up in
15910 the specified table. If the key is not found in the table, integer value zero
15911 is returned. Otherwise the converter returns the average incoming connection
15912 rate associated with the input sample in the designated table. See also the
15913 sc_conn_rate sample fetch keyword.
15914
Thierry FOURNIER236657b2015-08-19 08:25:14 +020015915table_gpt0(<table>)
15916 Uses the string representation of the input sample to perform a look up in
15917 the specified table. If the key is not found in the table, boolean value zero
15918 is returned. Otherwise the converter returns the current value of the first
15919 general purpose tag associated with the input sample in the designated table.
15920 See also the sc_get_gpt0 sample fetch keyword.
15921
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015922table_gpc0(<table>)
15923 Uses the string representation of the input sample to perform a look up in
15924 the specified table. If the key is not found in the table, integer value zero
15925 is returned. Otherwise the converter returns the current value of the first
15926 general purpose counter associated with the input sample in the designated
15927 table. See also the sc_get_gpc0 sample fetch keyword.
15928
15929table_gpc0_rate(<table>)
15930 Uses the string representation of the input sample to perform a look up in
15931 the specified table. If the key is not found in the table, integer value zero
15932 is returned. Otherwise the converter returns the frequency which the gpc0
15933 counter was incremented over the configured period in the table, associated
15934 with the input sample in the designated table. See also the sc_get_gpc0_rate
15935 sample fetch keyword.
15936
Frédéric Lécaille6778b272018-01-29 15:22:53 +010015937table_gpc1(<table>)
15938 Uses the string representation of the input sample to perform a look up in
15939 the specified table. If the key is not found in the table, integer value zero
15940 is returned. Otherwise the converter returns the current value of the second
15941 general purpose counter associated with the input sample in the designated
15942 table. See also the sc_get_gpc1 sample fetch keyword.
15943
15944table_gpc1_rate(<table>)
15945 Uses the string representation of the input sample to perform a look up in
15946 the specified table. If the key is not found in the table, integer value zero
15947 is returned. Otherwise the converter returns the frequency which the gpc1
15948 counter was incremented over the configured period in the table, associated
15949 with the input sample in the designated table. See also the sc_get_gpc1_rate
15950 sample fetch keyword.
15951
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015952table_http_err_cnt(<table>)
15953 Uses the string representation of the input sample to perform a look up in
15954 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010015955 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015956 errors associated with the input sample in the designated table. See also the
15957 sc_http_err_cnt sample fetch keyword.
15958
15959table_http_err_rate(<table>)
15960 Uses the string representation of the input sample to perform a look up in
15961 the specified table. If the key is not found in the table, integer value zero
15962 is returned. Otherwise the average rate of HTTP errors associated with the
15963 input sample in the designated table, measured in amount of errors over the
15964 period configured in the table. See also the sc_http_err_rate sample fetch
15965 keyword.
15966
15967table_http_req_cnt(<table>)
15968 Uses the string representation of the input sample to perform a look up in
15969 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010015970 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015971 requests associated with the input sample in the designated table. See also
15972 the sc_http_req_cnt sample fetch keyword.
15973
15974table_http_req_rate(<table>)
15975 Uses the string representation of the input sample to perform a look up in
15976 the specified table. If the key is not found in the table, integer value zero
15977 is returned. Otherwise the average rate of HTTP requests associated with the
15978 input sample in the designated table, measured in amount of requests over the
15979 period configured in the table. See also the sc_http_req_rate sample fetch
15980 keyword.
15981
15982table_kbytes_in(<table>)
15983 Uses the string representation of the input sample to perform a look up in
15984 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010015985 is returned. Otherwise the converter returns the cumulative number of client-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015986 to-server data associated with the input sample in the designated table,
15987 measured in kilobytes. The test is currently performed on 32-bit integers,
15988 which limits values to 4 terabytes. See also the sc_kbytes_in sample fetch
15989 keyword.
15990
15991table_kbytes_out(<table>)
15992 Uses the string representation of the input sample to perform a look up in
15993 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010015994 is returned. Otherwise the converter returns the cumulative number of server-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015995 to-client data associated with the input sample in the designated table,
15996 measured in kilobytes. The test is currently performed on 32-bit integers,
15997 which limits values to 4 terabytes. See also the sc_kbytes_out sample fetch
15998 keyword.
15999
16000table_server_id(<table>)
16001 Uses the string representation of the input sample to perform a look up in
16002 the specified table. If the key is not found in the table, integer value zero
16003 is returned. Otherwise the converter returns the server ID associated with
16004 the input sample in the designated table. A server ID is associated to a
16005 sample by a "stick" rule when a connection to a server succeeds. A server ID
16006 zero means that no server is associated with this key.
16007
16008table_sess_cnt(<table>)
16009 Uses the string representation of the input sample to perform a look up in
16010 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010016011 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016012 sessions associated with the input sample in the designated table. Note that
16013 a session here refers to an incoming connection being accepted by the
16014 "tcp-request connection" rulesets. See also the sc_sess_cnt sample fetch
16015 keyword.
16016
16017table_sess_rate(<table>)
16018 Uses the string representation of the input sample to perform a look up in
16019 the specified table. If the key is not found in the table, integer value zero
16020 is returned. Otherwise the converter returns the average incoming session
16021 rate associated with the input sample in the designated table. Note that a
16022 session here refers to an incoming connection being accepted by the
16023 "tcp-request connection" rulesets. See also the sc_sess_rate sample fetch
16024 keyword.
16025
16026table_trackers(<table>)
16027 Uses the string representation of the input sample to perform a look up in
16028 the specified table. If the key is not found in the table, integer value zero
16029 is returned. Otherwise the converter returns the current amount of concurrent
16030 connections tracking the same key as the input sample in the designated
16031 table. It differs from table_conn_cur in that it does not rely on any stored
16032 information but on the table's reference count (the "use" value which is
16033 returned by "show table" on the CLI). This may sometimes be more suited for
16034 layer7 tracking. It can be used to tell a server how many concurrent
16035 connections there are from a given address for example. See also the
16036 sc_trackers sample fetch keyword.
16037
Willy Tarreauffcb2e42014-07-10 16:29:08 +020016038upper
16039 Convert a string sample to upper case. This can only be placed after a string
16040 sample fetch function or after a transformation keyword returning a string
16041 type. The result is of type string.
16042
Willy Tarreau62ba9ba2020-04-23 17:54:47 +020016043url_dec([<in_form>])
16044 Takes an url-encoded string provided as input and returns the decoded version
16045 as output. The input and the output are of type string. If the <in_form>
16046 argument is set to a non-zero integer value, the input string is assumed to
16047 be part of a form or query string and the '+' character will be turned into a
16048 space (' '). Otherwise this will only happen after a question mark indicating
16049 a query string ('?').
Thierry FOURNIER82ff3c92015-05-07 15:46:20 +020016050
William Dauchy55ed7c42021-01-06 23:39:50 +010016051url_enc([<enc_type>])
16052 Takes a string provided as input and returns the encoded version as output.
16053 The input and the output are of type string. By default the type of encoding
16054 is meant for `query` type. There is no other type supported for now but the
16055 optional argument is here for future changes.
16056
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010016057ungrpc(<field_number>,[<field_type>])
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010016058 This extracts the protocol buffers message field in raw mode of an input binary
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010016059 sample representation of a gRPC message with <field_number> as field number
16060 (dotted notation) if <field_type> is not present, or as an integer sample if this
16061 field is present.
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010016062 The list of the authorized types is the following one: "int32", "int64", "uint32",
16063 "uint64", "sint32", "sint64", "bool", "enum" for the "varint" wire type 0
16064 "fixed64", "sfixed64", "double" for the 64bit wire type 1, "fixed32", "sfixed32",
16065 "float" for the wire type 5. Note that "string" is considered as a length-delimited
Frédéric Lécaille93d33162019-03-06 09:35:59 +010016066 type, so it does not require any <field_type> argument to be extracted.
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010016067 More information may be found here about the protocol buffers message field types:
16068 https://developers.google.com/protocol-buffers/docs/encoding
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010016069
16070 Example:
16071 // with such a protocol buffer .proto file content adapted from
16072 // https://github.com/grpc/grpc/blob/master/examples/protos/route_guide.proto
16073
16074 message Point {
16075 int32 latitude = 1;
16076 int32 longitude = 2;
16077 }
16078
16079 message PPoint {
16080 Point point = 59;
16081 }
16082
16083 message Rectangle {
16084 // One corner of the rectangle.
16085 PPoint lo = 48;
16086 // The other corner of the rectangle.
16087 PPoint hi = 49;
16088 }
16089
Peter Gervaidf4c9d22020-06-11 18:05:11 +020016090 let's say a body request is made of a "Rectangle" object value (two PPoint
16091 protocol buffers messages), the four protocol buffers fields could be
16092 extracted with these "ungrpc" directives:
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010016093
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010016094 req.body,ungrpc(48.59.1,int32) # "latitude" of "lo" first PPoint
16095 req.body,ungrpc(48.59.2,int32) # "longitude" of "lo" first PPoint
John Roeslerfb2fce12019-07-10 15:45:51 -050016096 req.body,ungrpc(49.59.1,int32) # "latitude" of "hi" second PPoint
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010016097 req.body,ungrpc(49.59.2,int32) # "longitude" of "hi" second PPoint
16098
Peter Gervaidf4c9d22020-06-11 18:05:11 +020016099 We could also extract the intermediary 48.59 field as a binary sample as follows:
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010016100
Frédéric Lécaille93d33162019-03-06 09:35:59 +010016101 req.body,ungrpc(48.59)
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010016102
Peter Gervaidf4c9d22020-06-11 18:05:11 +020016103 As a gRPC message is always made of a gRPC header followed by protocol buffers
16104 messages, in the previous example the "latitude" of "lo" first PPoint
16105 could be extracted with these equivalent directives:
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010016106
16107 req.body,ungrpc(48.59),protobuf(1,int32)
16108 req.body,ungrpc(48),protobuf(59.1,int32)
16109 req.body,ungrpc(48),protobuf(59),protobuf(1,int32)
16110
Peter Gervaidf4c9d22020-06-11 18:05:11 +020016111 Note that the first convert must be "ungrpc", the remaining ones must be
16112 "protobuf" and only the last one may have or not a second argument to
16113 interpret the previous binary sample.
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010016114
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010016115
Christopher Faulet85d79c92016-11-09 16:54:56 +010016116unset-var(<var name>)
16117 Unsets a variable if the input content is defined. The name of the variable
16118 starts with an indication about its scope. The scopes allowed are:
16119 "proc" : the variable is shared with the whole process
16120 "sess" : the variable is shared with the whole session
16121 "txn" : the variable is shared with the transaction (request and
16122 response),
16123 "req" : the variable is shared only during request processing,
16124 "res" : the variable is shared only during response processing.
16125 This prefix is followed by a name. The separator is a '.'. The name may only
16126 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
16127
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020016128utime(<format>[,<offset>])
16129 Converts an integer supposed to contain a date since epoch to a string
16130 representing this date in UTC time using a format defined by the <format>
16131 string using strftime(3). The purpose is to allow any date format to be used
16132 in logs. An optional <offset> in seconds may be applied to the input date
16133 (positive or negative). See the strftime() man page for the format supported
16134 by your operating system. See also the ltime converter.
16135
16136 Example :
16137
16138 # Emit two colons, one with the UTC time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010016139 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020016140 log-format %[date,utime(%Y%m%d%H%M%S)]\ %ci:%cp
16141
Marcin Deranek9631a282018-04-16 14:30:46 +020016142word(<index>,<delimiters>[,<count>])
16143 Extracts the nth word counting from the beginning (positive index) or from
16144 the end (negative index) considering given delimiters from an input string.
16145 Indexes start at 1 or -1 and delimiters are a string formatted list of chars.
Jerome Magnin88209322020-01-28 13:33:44 +010016146 Delimiters at the beginning or end of the input string are ignored.
Marcin Deranek9631a282018-04-16 14:30:46 +020016147 Optionally you can specify <count> of words to extract (default: 1).
16148 Value of 0 indicates extraction of all remaining words.
16149
16150 Example :
16151 str(f1_f2_f3__f5),word(4,_) # f5
16152 str(f1_f2_f3__f5),word(2,_,0) # f2_f3__f5
16153 str(f1_f2_f3__f5),word(3,_,2) # f3__f5
16154 str(f1_f2_f3__f5),word(-2,_,3) # f1_f2_f3
16155 str(f1_f2_f3__f5),word(-3,_,0) # f1_f2
Jerome Magnin88209322020-01-28 13:33:44 +010016156 str(/f1/f2/f3/f4),word(1,/) # f1
Emeric Brunc9a0f6d2014-11-25 14:09:01 +010016157
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020016158wt6([<avalanche>])
16159 Hashes a binary input sample into an unsigned 32-bit quantity using the WT6
16160 hash function. Optionally, it is possible to apply a full avalanche hash
16161 function to the output if the optional <avalanche> argument equals 1. This
16162 converter uses the same functions as used by the various hash-based load
16163 balancing algorithms, so it will provide exactly the same results. It is
16164 mostly intended for debugging, but can be used as a stick-table entry to
16165 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010016166 32-bit hash is trivial to break. See also "crc32", "djb2", "sdbm", "crc32c",
16167 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020016168
Willy Tarreau97707872015-01-27 15:12:13 +010016169xor(<value>)
16170 Performs a bitwise "XOR" (exclusive OR) between <value> and the input value
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016171 of type signed integer, and returns the result as an signed integer.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016172 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010016173 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010016174 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010016175 "sess" : the variable is shared with the whole session
16176 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016177 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010016178 "req" : the variable is shared only during request processing,
16179 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016180 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010016181 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010016182
Thierry FOURNIER01e09742016-12-26 11:46:11 +010016183xxh32([<seed>])
16184 Hashes a binary input sample into an unsigned 32-bit quantity using the 32-bit
16185 variant of the XXHash hash function. This hash supports a seed which defaults
16186 to zero but a different value maybe passed as the <seed> argument. This hash
16187 is known to be very good and very fast so it can be used to hash URLs and/or
16188 URL parameters for use as stick-table keys to collect statistics with a low
16189 collision rate, though care must be taken as the algorithm is not considered
16190 as cryptographically secure.
16191
16192xxh64([<seed>])
16193 Hashes a binary input sample into a signed 64-bit quantity using the 64-bit
16194 variant of the XXHash hash function. This hash supports a seed which defaults
16195 to zero but a different value maybe passed as the <seed> argument. This hash
16196 is known to be very good and very fast so it can be used to hash URLs and/or
16197 URL parameters for use as stick-table keys to collect statistics with a low
16198 collision rate, though care must be taken as the algorithm is not considered
16199 as cryptographically secure.
16200
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010016201
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200162027.3.2. Fetching samples from internal states
Willy Tarreau74ca5042013-06-11 23:12:07 +020016203--------------------------------------------
16204
16205A first set of sample fetch methods applies to internal information which does
16206not even relate to any client information. These ones are sometimes used with
16207"monitor-fail" directives to report an internal status to external watchers.
16208The sample fetch methods described in this section are usable anywhere.
16209
16210always_false : boolean
16211 Always returns the boolean "false" value. It may be used with ACLs as a
16212 temporary replacement for another one when adjusting configurations.
16213
16214always_true : boolean
16215 Always returns the boolean "true" value. It may be used with ACLs as a
16216 temporary replacement for another one when adjusting configurations.
16217
16218avg_queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010016219 Returns the total number of queued connections of the designated backend
Willy Tarreau74ca5042013-06-11 23:12:07 +020016220 divided by the number of active servers. The current backend is used if no
16221 backend is specified. This is very similar to "queue" except that the size of
16222 the farm is considered, in order to give a more accurate measurement of the
16223 time it may take for a new connection to be processed. The main usage is with
16224 ACL to return a sorry page to new users when it becomes certain they will get
16225 a degraded service, or to pass to the backend servers in a header so that
16226 they decide to work in degraded mode or to disable some functions to speed up
16227 the processing a bit. Note that in the event there would not be any active
16228 server anymore, twice the number of queued connections would be considered as
16229 the measured value. This is a fair estimate, as we expect one server to get
16230 back soon anyway, but we still prefer to send new traffic to another backend
16231 if in better shape. See also the "queue", "be_conn", and "be_sess_rate"
16232 sample fetches.
Krzysztof Piotr Oledzki346f76d2010-01-12 21:59:30 +010016233
Willy Tarreau74ca5042013-06-11 23:12:07 +020016234be_conn([<backend>]) : integer
Willy Tarreaua36af912009-10-10 12:02:45 +020016235 Applies to the number of currently established connections on the backend,
16236 possibly including the connection being evaluated. If no backend name is
16237 specified, the current one is used. But it is also possible to check another
16238 backend. It can be used to use a specific farm when the nominal one is full.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040016239 See also the "fe_conn", "queue", "be_conn_free", and "be_sess_rate" criteria.
16240
16241be_conn_free([<backend>]) : integer
16242 Returns an integer value corresponding to the number of available connections
16243 across available servers in the backend. Queue slots are not included. Backup
16244 servers are also not included, unless all other servers are down. If no
16245 backend name is specified, the current one is used. But it is also possible
16246 to check another backend. It can be used to use a specific farm when the
Patrick Hemmer155e93e2018-06-14 18:01:35 -040016247 nominal one is full. See also the "be_conn", "connslots", and "srv_conn_free"
16248 criteria.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040016249
16250 OTHER CAVEATS AND NOTES: if any of the server maxconn, or maxqueue is 0
16251 (meaning unlimited), then this fetch clearly does not make sense, in which
16252 case the value returned will be -1.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016253
Willy Tarreau74ca5042013-06-11 23:12:07 +020016254be_sess_rate([<backend>]) : integer
16255 Returns an integer value corresponding to the sessions creation rate on the
16256 backend, in number of new sessions per second. This is used with ACLs to
16257 switch to an alternate backend when an expensive or fragile one reaches too
Davor Ocelice9ed2812017-12-25 17:49:28 +010016258 high a session rate, or to limit abuse of service (e.g. prevent sucking of an
Willy Tarreau74ca5042013-06-11 23:12:07 +020016259 online dictionary). It can also be useful to add this element to logs using a
16260 log-format directive.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016261
16262 Example :
16263 # Redirect to an error page if the dictionary is requested too often
16264 backend dynamic
16265 mode http
16266 acl being_scanned be_sess_rate gt 100
16267 redirect location /denied.html if being_scanned
Willy Tarreau0ba27502007-12-24 16:55:16 +010016268
Davor Ocelice9ed2812017-12-25 17:49:28 +010016269bin(<hex>) : bin
Thierry FOURNIERcc103292015-06-06 19:30:17 +020016270 Returns a binary chain. The input is the hexadecimal representation
16271 of the string.
16272
16273bool(<bool>) : bool
16274 Returns a boolean value. <bool> can be 'true', 'false', '1' or '0'.
16275 'false' and '0' are the same. 'true' and '1' are the same.
16276
Willy Tarreau74ca5042013-06-11 23:12:07 +020016277connslots([<backend>]) : integer
16278 Returns an integer value corresponding to the number of connection slots
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016279 still available in the backend, by totaling the maximum amount of
Willy Tarreau74ca5042013-06-11 23:12:07 +020016280 connections on all servers and the maximum queue size. This is probably only
16281 used with ACLs.
Tait Clarridge7896d522012-12-05 21:39:31 -050016282
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080016283 The basic idea here is to be able to measure the number of connection "slots"
Willy Tarreau55165fe2009-05-10 12:02:55 +020016284 still available (connection + queue), so that anything beyond that (intended
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080016285 usage; see "use_backend" keyword) can be redirected to a different backend.
16286
Willy Tarreau55165fe2009-05-10 12:02:55 +020016287 'connslots' = number of available server connection slots, + number of
16288 available server queue slots.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080016289
Willy Tarreaua36af912009-10-10 12:02:45 +020016290 Note that while "fe_conn" may be used, "connslots" comes in especially
Willy Tarreau55165fe2009-05-10 12:02:55 +020016291 useful when you have a case of traffic going to one single ip, splitting into
Willy Tarreau74ca5042013-06-11 23:12:07 +020016292 multiple backends (perhaps using ACLs to do name-based load balancing) and
Willy Tarreau55165fe2009-05-10 12:02:55 +020016293 you want to be able to differentiate between different backends, and their
Davor Ocelice9ed2812017-12-25 17:49:28 +010016294 available "connslots". Also, whereas "nbsrv" only measures servers that are
Willy Tarreau74ca5042013-06-11 23:12:07 +020016295 actually *down*, this fetch is more fine-grained and looks into the number of
Willy Tarreaua36af912009-10-10 12:02:45 +020016296 available connection slots as well. See also "queue" and "avg_queue".
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080016297
Willy Tarreau55165fe2009-05-10 12:02:55 +020016298 OTHER CAVEATS AND NOTES: at this point in time, the code does not take care
16299 of dynamic connections. Also, if any of the server maxconn, or maxqueue is 0,
Willy Tarreau74ca5042013-06-11 23:12:07 +020016300 then this fetch clearly does not make sense, in which case the value returned
Willy Tarreau55165fe2009-05-10 12:02:55 +020016301 will be -1.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080016302
Willy Tarreau70fe9442018-11-22 16:07:39 +010016303cpu_calls : integer
16304 Returns the number of calls to the task processing the stream or current
16305 request since it was allocated. This number is reset for each new request on
16306 the same connections in case of HTTP keep-alive. This value should usually be
16307 low and stable (around 2 calls for a typically simple request) but may become
16308 high if some processing (compression, caching or analysis) is performed. This
16309 is purely for performance monitoring purposes.
16310
16311cpu_ns_avg : integer
16312 Returns the average number of nanoseconds spent in each call to the task
16313 processing the stream or current request. This number is reset for each new
16314 request on the same connections in case of HTTP keep-alive. This value
16315 indicates the overall cost of processing the request or the connection for
16316 each call. There is no good nor bad value but the time spent in a call
16317 automatically causes latency for other processing (see lat_ns_avg below),
16318 and may affect other connection's apparent response time. Certain operations
16319 like compression, complex regex matching or heavy Lua operations may directly
16320 affect this value, and having it in the logs will make it easier to spot the
16321 faulty processing that needs to be fixed to recover decent performance.
16322 Note: this value is exactly cpu_ns_tot divided by cpu_calls.
16323
16324cpu_ns_tot : integer
16325 Returns the total number of nanoseconds spent in each call to the task
16326 processing the stream or current request. This number is reset for each new
16327 request on the same connections in case of HTTP keep-alive. This value
16328 indicates the overall cost of processing the request or the connection for
16329 each call. There is no good nor bad value but the time spent in a call
16330 automatically causes latency for other processing (see lat_ns_avg below),
16331 induces CPU costs on the machine, and may affect other connection's apparent
16332 response time. Certain operations like compression, complex regex matching or
16333 heavy Lua operations may directly affect this value, and having it in the
16334 logs will make it easier to spot the faulty processing that needs to be fixed
16335 to recover decent performance. The value may be artificially high due to a
16336 high cpu_calls count, for example when processing many HTTP chunks, and for
16337 this reason it is often preferred to log cpu_ns_avg instead.
16338
Cyril Bonté6bcd1822019-11-05 23:13:59 +010016339date([<offset>],[<unit>]) : integer
Willy Tarreau6236d3a2013-07-25 14:28:25 +020016340 Returns the current date as the epoch (number of seconds since 01/01/1970).
Damien Claisseae6f1252019-10-30 15:57:28 +000016341
16342 If an offset value is specified, then it is added to the current date before
16343 returning the value. This is particularly useful to compute relative dates,
16344 as both positive and negative offsets are allowed.
Willy Tarreau276fae92013-07-25 14:36:01 +020016345 It is useful combined with the http_date converter.
16346
Damien Claisseae6f1252019-10-30 15:57:28 +000016347 <unit> is facultative, and can be set to "s" for seconds (default behavior),
16348 "ms" for milliseconds or "us" for microseconds.
16349 If unit is set, return value is an integer reflecting either seconds,
16350 milliseconds or microseconds since epoch, plus offset.
16351 It is useful when a time resolution of less than a second is needed.
16352
Willy Tarreau276fae92013-07-25 14:36:01 +020016353 Example :
16354
16355 # set an expires header to now+1 hour in every response
16356 http-response set-header Expires %[date(3600),http_date]
Willy Tarreau6236d3a2013-07-25 14:28:25 +020016357
Damien Claisseae6f1252019-10-30 15:57:28 +000016358 # set an expires header to now+1 hour in every response, with
16359 # millisecond granularity
16360 http-response set-header Expires %[date(3600000,ms),http_date(0,ms)]
16361
Etienne Carrierea792a0a2018-01-17 13:43:24 +010016362date_us : integer
16363 Return the microseconds part of the date (the "second" part is returned by
16364 date sample). This sample is coherent with the date sample as it is comes
16365 from the same timeval structure.
16366
Willy Tarreaud716f9b2017-10-13 11:03:15 +020016367distcc_body(<token>[,<occ>]) : binary
16368 Parses a distcc message and returns the body associated to occurrence #<occ>
16369 of the token <token>. Occurrences start at 1, and when unspecified, any may
16370 match though in practice only the first one is checked for now. This can be
16371 used to extract file names or arguments in files built using distcc through
16372 haproxy. Please refer to distcc's protocol documentation for the complete
16373 list of supported tokens.
16374
16375distcc_param(<token>[,<occ>]) : integer
16376 Parses a distcc message and returns the parameter associated to occurrence
16377 #<occ> of the token <token>. Occurrences start at 1, and when unspecified,
16378 any may match though in practice only the first one is checked for now. This
16379 can be used to extract certain information such as the protocol version, the
16380 file size or the argument in files built using distcc through haproxy.
16381 Another use case consists in waiting for the start of the preprocessed file
16382 contents before connecting to the server to avoid keeping idle connections.
16383 Please refer to distcc's protocol documentation for the complete list of
16384 supported tokens.
16385
16386 Example :
16387 # wait up to 20s for the pre-processed file to be uploaded
16388 tcp-request inspect-delay 20s
16389 tcp-request content accept if { distcc_param(DOTI) -m found }
16390 # send large files to the big farm
16391 use_backend big_farm if { distcc_param(DOTI) gt 1000000 }
16392
Willy Tarreau595ec542013-06-12 21:34:28 +020016393env(<name>) : string
16394 Returns a string containing the value of environment variable <name>. As a
16395 reminder, environment variables are per-process and are sampled when the
16396 process starts. This can be useful to pass some information to a next hop
16397 server, or with ACLs to take specific action when the process is started a
16398 certain way.
16399
16400 Examples :
16401 # Pass the Via header to next hop with the local hostname in it
16402 http-request add-header Via 1.1\ %[env(HOSTNAME)]
16403
16404 # reject cookie-less requests when the STOP environment variable is set
16405 http-request deny if !{ cook(SESSIONID) -m found } { env(STOP) -m found }
16406
Willy Tarreau74ca5042013-06-11 23:12:07 +020016407fe_conn([<frontend>]) : integer
16408 Returns the number of currently established connections on the frontend,
Willy Tarreaud63335a2010-02-26 12:56:52 +010016409 possibly including the connection being evaluated. If no frontend name is
16410 specified, the current one is used. But it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020016411 frontend. It can be used to return a sorry page before hard-blocking, or to
16412 use a specific backend to drain new requests when the farm is considered
Davor Ocelice9ed2812017-12-25 17:49:28 +010016413 full. This is mostly used with ACLs but can also be used to pass some
Willy Tarreau74ca5042013-06-11 23:12:07 +020016414 statistics to servers in HTTP headers. See also the "dst_conn", "be_conn",
16415 "fe_sess_rate" fetches.
Willy Tarreaua36af912009-10-10 12:02:45 +020016416
Nenad Merdanovicad9a7e92016-10-03 04:57:37 +020016417fe_req_rate([<frontend>]) : integer
16418 Returns an integer value corresponding to the number of HTTP requests per
16419 second sent to a frontend. This number can differ from "fe_sess_rate" in
16420 situations where client-side keep-alive is enabled.
16421
Willy Tarreau74ca5042013-06-11 23:12:07 +020016422fe_sess_rate([<frontend>]) : integer
16423 Returns an integer value corresponding to the sessions creation rate on the
16424 frontend, in number of new sessions per second. This is used with ACLs to
16425 limit the incoming session rate to an acceptable range in order to prevent
16426 abuse of service at the earliest moment, for example when combined with other
16427 layer 4 ACLs in order to force the clients to wait a bit for the rate to go
16428 down below the limit. It can also be useful to add this element to logs using
16429 a log-format directive. See also the "rate-limit sessions" directive for use
16430 in frontends.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010016431
16432 Example :
16433 # This frontend limits incoming mails to 10/s with a max of 100
16434 # concurrent connections. We accept any connection below 10/s, and
16435 # force excess clients to wait for 100 ms. Since clients are limited to
16436 # 100 max, there cannot be more than 10 incoming mails per second.
16437 frontend mail
16438 bind :25
16439 mode tcp
16440 maxconn 100
16441 acl too_fast fe_sess_rate ge 10
16442 tcp-request inspect-delay 100ms
16443 tcp-request content accept if ! too_fast
16444 tcp-request content accept if WAIT_END
Willy Tarreaud72758d2010-01-12 10:42:19 +010016445
Nenad Merdanovic807a6e72017-03-12 22:00:00 +010016446hostname : string
16447 Returns the system hostname.
16448
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016449int(<integer>) : signed integer
16450 Returns a signed integer.
16451
Thierry FOURNIERcc103292015-06-06 19:30:17 +020016452ipv4(<ipv4>) : ipv4
16453 Returns an ipv4.
16454
16455ipv6(<ipv6>) : ipv6
16456 Returns an ipv6.
16457
Willy Tarreau70fe9442018-11-22 16:07:39 +010016458lat_ns_avg : integer
16459 Returns the average number of nanoseconds spent between the moment the task
16460 handling the stream is woken up and the moment it is effectively called. This
16461 number is reset for each new request on the same connections in case of HTTP
16462 keep-alive. This value indicates the overall latency inflicted to the current
16463 request by all other requests being processed in parallel, and is a direct
16464 indicator of perceived performance due to noisy neighbours. In order to keep
16465 the value low, it is possible to reduce the scheduler's run queue depth using
16466 "tune.runqueue-depth", to reduce the number of concurrent events processed at
16467 once using "tune.maxpollevents", to decrease the stream's nice value using
Willy Tarreaue7723bd2020-06-24 11:11:02 +020016468 the "nice" option on the "bind" lines or in the frontend, to enable low
16469 latency scheduling using "tune.sched.low-latency", or to look for other heavy
16470 requests in logs (those exhibiting large values of "cpu_ns_avg"), whose
16471 processing needs to be adjusted or fixed. Compression of large buffers could
16472 be a culprit, like heavy regex or long lists of regex. Note: this value is
16473 exactly lat_ns_tot divided by cpu_calls.
Willy Tarreau70fe9442018-11-22 16:07:39 +010016474
16475lat_ns_tot : integer
16476 Returns the total number of nanoseconds spent between the moment the task
16477 handling the stream is woken up and the moment it is effectively called. This
16478 number is reset for each new request on the same connections in case of HTTP
16479 keep-alive. This value indicates the overall latency inflicted to the current
16480 request by all other requests being processed in parallel, and is a direct
16481 indicator of perceived performance due to noisy neighbours. In order to keep
16482 the value low, it is possible to reduce the scheduler's run queue depth using
16483 "tune.runqueue-depth", to reduce the number of concurrent events processed at
16484 once using "tune.maxpollevents", to decrease the stream's nice value using
Willy Tarreaue7723bd2020-06-24 11:11:02 +020016485 the "nice" option on the "bind" lines or in the frontend, to enable low
16486 latency scheduling using "tune.sched.low-latency", or to look for other heavy
16487 requests in logs (those exhibiting large values of "cpu_ns_avg"), whose
16488 processing needs to be adjusted or fixed. Compression of large buffers could
16489 be a culprit, like heavy regex or long lists of regex. Note: while it
Willy Tarreau70fe9442018-11-22 16:07:39 +010016490 may intuitively seem that the total latency adds to a transfer time, it is
16491 almost never true because while a task waits for the CPU, network buffers
16492 continue to fill up and the next call will process more at once. The value
16493 may be artificially high due to a high cpu_calls count, for example when
16494 processing many HTTP chunks, and for this reason it is often preferred to log
16495 lat_ns_avg instead, which is a more relevant performance indicator.
16496
Thierry FOURNIERcc103292015-06-06 19:30:17 +020016497meth(<method>) : method
16498 Returns a method.
16499
Willy Tarreau0f30d262014-11-24 16:02:05 +010016500nbproc : integer
16501 Returns an integer value corresponding to the number of processes that were
16502 started (it equals the global "nbproc" setting). This is useful for logging
16503 and debugging purposes.
16504
Willy Tarreau74ca5042013-06-11 23:12:07 +020016505nbsrv([<backend>]) : integer
16506 Returns an integer value corresponding to the number of usable servers of
16507 either the current backend or the named backend. This is mostly used with
16508 ACLs but can also be useful when added to logs. This is normally used to
Willy Tarreaud63335a2010-02-26 12:56:52 +010016509 switch to an alternate backend when the number of servers is too low to
16510 to handle some load. It is useful to report a failure when combined with
16511 "monitor fail".
Willy Tarreau079ff0a2009-03-05 21:34:28 +010016512
Patrick Hemmerfabb24f2018-08-13 14:07:57 -040016513prio_class : integer
16514 Returns the priority class of the current session for http mode or connection
16515 for tcp mode. The value will be that set by the last call to "http-request
16516 set-priority-class" or "tcp-request content set-priority-class".
16517
16518prio_offset : integer
16519 Returns the priority offset of the current session for http mode or
16520 connection for tcp mode. The value will be that set by the last call to
16521 "http-request set-priority-offset" or "tcp-request content
16522 set-priority-offset".
16523
Willy Tarreau0f30d262014-11-24 16:02:05 +010016524proc : integer
16525 Returns an integer value corresponding to the position of the process calling
16526 the function, between 1 and global.nbproc. This is useful for logging and
16527 debugging purposes.
16528
Willy Tarreau74ca5042013-06-11 23:12:07 +020016529queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010016530 Returns the total number of queued connections of the designated backend,
16531 including all the connections in server queues. If no backend name is
16532 specified, the current one is used, but it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020016533 one. This is useful with ACLs or to pass statistics to backend servers. This
16534 can be used to take actions when queuing goes above a known level, generally
16535 indicating a surge of traffic or a massive slowdown on the servers. One
16536 possible action could be to reject new users but still accept old ones. See
16537 also the "avg_queue", "be_conn", and "be_sess_rate" fetches.
16538
Willy Tarreau84310e22014-02-14 11:59:04 +010016539rand([<range>]) : integer
16540 Returns a random integer value within a range of <range> possible values,
16541 starting at zero. If the range is not specified, it defaults to 2^32, which
16542 gives numbers between 0 and 4294967295. It can be useful to pass some values
16543 needed to take some routing decisions for example, or just for debugging
16544 purposes. This random must not be used for security purposes.
16545
Willy Tarreau74ca5042013-06-11 23:12:07 +020016546srv_conn([<backend>/]<server>) : integer
16547 Returns an integer value corresponding to the number of currently established
16548 connections on the designated server, possibly including the connection being
16549 evaluated. If <backend> is omitted, then the server is looked up in the
16550 current backend. It can be used to use a specific farm when one server is
16551 full, or to inform the server about our view of the number of active
Patrick Hemmer155e93e2018-06-14 18:01:35 -040016552 connections with it. See also the "fe_conn", "be_conn", "queue", and
16553 "srv_conn_free" fetch methods.
16554
16555srv_conn_free([<backend>/]<server>) : integer
16556 Returns an integer value corresponding to the number of available connections
16557 on the designated server, possibly including the connection being evaluated.
16558 The value does not include queue slots. If <backend> is omitted, then the
16559 server is looked up in the current backend. It can be used to use a specific
16560 farm when one server is full, or to inform the server about our view of the
16561 number of active connections with it. See also the "be_conn_free" and
16562 "srv_conn" fetch methods.
16563
16564 OTHER CAVEATS AND NOTES: If the server maxconn is 0, then this fetch clearly
16565 does not make sense, in which case the value returned will be -1.
Willy Tarreau74ca5042013-06-11 23:12:07 +020016566
16567srv_is_up([<backend>/]<server>) : boolean
16568 Returns true when the designated server is UP, and false when it is either
16569 DOWN or in maintenance mode. If <backend> is omitted, then the server is
16570 looked up in the current backend. It is mainly used to take action based on
Davor Ocelice9ed2812017-12-25 17:49:28 +010016571 an external status reported via a health check (e.g. a geographical site's
Willy Tarreau74ca5042013-06-11 23:12:07 +020016572 availability). Another possible use which is more of a hack consists in
16573 using dummy servers as boolean variables that can be enabled or disabled from
16574 the CLI, so that rules depending on those ACLs can be tweaked in realtime.
16575
Willy Tarreauff2b7af2017-10-13 11:46:26 +020016576srv_queue([<backend>/]<server>) : integer
16577 Returns an integer value corresponding to the number of connections currently
16578 pending in the designated server's queue. If <backend> is omitted, then the
16579 server is looked up in the current backend. It can sometimes be used together
16580 with the "use-server" directive to force to use a known faster server when it
16581 is not much loaded. See also the "srv_conn", "avg_queue" and "queue" sample
16582 fetch methods.
16583
Willy Tarreau74ca5042013-06-11 23:12:07 +020016584srv_sess_rate([<backend>/]<server>) : integer
16585 Returns an integer corresponding to the sessions creation rate on the
16586 designated server, in number of new sessions per second. If <backend> is
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016587 omitted, then the server is looked up in the current backend. This is mostly
Willy Tarreau74ca5042013-06-11 23:12:07 +020016588 used with ACLs but can make sense with logs too. This is used to switch to an
16589 alternate backend when an expensive or fragile one reaches too high a session
Davor Ocelice9ed2812017-12-25 17:49:28 +010016590 rate, or to limit abuse of service (e.g. prevent latent requests from
Willy Tarreau74ca5042013-06-11 23:12:07 +020016591 overloading servers).
16592
16593 Example :
16594 # Redirect to a separate back
16595 acl srv1_full srv_sess_rate(be1/srv1) gt 50
16596 acl srv2_full srv_sess_rate(be1/srv2) gt 50
16597 use_backend be2 if srv1_full or srv2_full
16598
Christopher Faulet1bea8652020-07-10 16:03:45 +020016599srv_iweight([<backend>/]<server>): integer
16600 Returns an integer corresponding to the server's initial weight. If <backend>
16601 is omitted, then the server is looked up in the current backend. See also
16602 "srv_weight" and "srv_uweight".
16603
16604srv_uweight([<backend>/]<server>): integer
16605 Returns an integer corresponding to the user visible server's weight. If
16606 <backend> is omitted, then the server is looked up in the current
16607 backend. See also "srv_weight" and "srv_iweight".
16608
16609srv_weight([<backend>/]<server>): integer
16610 Returns an integer corresponding to the current (or effective) server's
16611 weight. If <backend> is omitted, then the server is looked up in the current
16612 backend. See also "srv_iweight" and "srv_uweight".
16613
Willy Tarreau0f30d262014-11-24 16:02:05 +010016614stopping : boolean
16615 Returns TRUE if the process calling the function is currently stopping. This
16616 can be useful for logging, or for relaxing certain checks or helping close
16617 certain connections upon graceful shutdown.
16618
Thierry FOURNIERcc103292015-06-06 19:30:17 +020016619str(<string>) : string
16620 Returns a string.
16621
Willy Tarreau74ca5042013-06-11 23:12:07 +020016622table_avl([<table>]) : integer
16623 Returns the total number of available entries in the current proxy's
16624 stick-table or in the designated stick-table. See also table_cnt.
16625
16626table_cnt([<table>]) : integer
16627 Returns the total number of entries currently in use in the current proxy's
16628 stick-table or in the designated stick-table. See also src_conn_cnt and
16629 table_avl for other entry counting methods.
16630
Christopher Faulet34adb2a2017-11-21 21:45:38 +010016631thread : integer
16632 Returns an integer value corresponding to the position of the thread calling
16633 the function, between 0 and (global.nbthread-1). This is useful for logging
16634 and debugging purposes.
16635
Alexandar Lazic6f273212021-06-01 00:27:01 +020016636uuid([<version>]) : string
16637 Returns a UUID following the RFC4122 standard. If the version is not
16638 specified, a UUID version 4 (fully random) is returned.
16639 Currently, only version 4 is supported.
16640
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020016641var(<var-name>) : undefined
16642 Returns a variable with the stored type. If the variable is not set, the
Daniel Schneller0b547052016-03-21 20:46:57 +010016643 sample fetch fails. The name of the variable starts with an indication
16644 about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010016645 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010016646 "sess" : the variable is shared with the whole session
16647 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020016648 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010016649 "req" : the variable is shared only during request processing,
16650 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020016651 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010016652 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020016653
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200166547.3.3. Fetching samples at Layer 4
Willy Tarreau74ca5042013-06-11 23:12:07 +020016655----------------------------------
16656
16657The layer 4 usually describes just the transport layer which in haproxy is
16658closest to the connection, where no content is yet made available. The fetch
16659methods described here are usable as low as the "tcp-request connection" rule
16660sets unless they require some future information. Those generally include
16661TCP/IP addresses and ports, as well as elements from stick-tables related to
Willy Tarreau4d4149c2013-07-23 19:33:46 +020016662the incoming connection. For retrieving a value from a sticky counters, the
16663counter number can be explicitly set as 0, 1, or 2 using the pre-defined
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020016664"sc0_", "sc1_", or "sc2_" prefix. These three pre-defined prefixes can only be
16665used if MAX_SESS_STKCTR value does not exceed 3, otherwise the counter number
16666can be specified as the first integer argument when using the "sc_" prefix.
16667Starting from "sc_0" to "sc_N" where N is (MAX_SESS_STKCTR-1). An optional
16668table may be specified with the "sc*" form, in which case the currently
16669tracked key will be looked up into this alternate table instead of the table
16670currently being tracked.
Willy Tarreau74ca5042013-06-11 23:12:07 +020016671
Jérôme Magnin35e53a62019-01-16 14:38:37 +010016672bc_http_major : integer
Jérôme Magnin86577422018-12-07 09:03:11 +010016673 Returns the backend connection's HTTP major version encoding, which may be 1
16674 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
16675 encoding and not the version present in the request header.
16676
Willy Tarreau74ca5042013-06-11 23:12:07 +020016677be_id : integer
16678 Returns an integer containing the current backend's id. It can be used in
Christopher Fauletd1b44642020-04-30 09:51:15 +020016679 frontends with responses to check which backend processed the request. It can
16680 also be used in a tcp-check or an http-check ruleset.
Willy Tarreau74ca5042013-06-11 23:12:07 +020016681
Marcin Deranekd2471c22016-12-12 14:08:05 +010016682be_name : string
16683 Returns a string containing the current backend's name. It can be used in
Christopher Fauletd1b44642020-04-30 09:51:15 +020016684 frontends with responses to check which backend processed the request. It can
16685 also be used in a tcp-check or an http-check ruleset.
Marcin Deranekd2471c22016-12-12 14:08:05 +010016686
Willy Tarreau74ca5042013-06-11 23:12:07 +020016687dst : ip
16688 This is the destination IPv4 address of the connection on the client side,
16689 which is the address the client connected to. It can be useful when running
16690 in transparent mode. It is of type IP and works on both IPv4 and IPv6 tables.
16691 On IPv6 tables, IPv4 address is mapped to its IPv6 equivalent, according to
Willy Tarreau64ded3d2019-01-23 10:02:15 +010016692 RFC 4291. When the incoming connection passed through address translation or
16693 redirection involving connection tracking, the original destination address
16694 before the redirection will be reported. On Linux systems, the source and
16695 destination may seldom appear reversed if the nf_conntrack_tcp_loose sysctl
16696 is set, because a late response may reopen a timed out connection and switch
16697 what is believed to be the source and the destination.
Willy Tarreau74ca5042013-06-11 23:12:07 +020016698
16699dst_conn : integer
16700 Returns an integer value corresponding to the number of currently established
16701 connections on the same socket including the one being evaluated. It is
16702 normally used with ACLs but can as well be used to pass the information to
16703 servers in an HTTP header or in logs. It can be used to either return a sorry
16704 page before hard-blocking, or to use a specific backend to drain new requests
16705 when the socket is considered saturated. This offers the ability to assign
16706 different limits to different listening ports or addresses. See also the
16707 "fe_conn" and "be_conn" fetches.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016708
Willy Tarreau16e01562016-08-09 16:46:18 +020016709dst_is_local : boolean
16710 Returns true if the destination address of the incoming connection is local
16711 to the system, or false if the address doesn't exist on the system, meaning
16712 that it was intercepted in transparent mode. It can be useful to apply
16713 certain rules by default to forwarded traffic and other rules to the traffic
Davor Ocelice9ed2812017-12-25 17:49:28 +010016714 targeting the real address of the machine. For example the stats page could
Willy Tarreau16e01562016-08-09 16:46:18 +020016715 be delivered only on this address, or SSH access could be locally redirected.
16716 Please note that the check involves a few system calls, so it's better to do
16717 it only once per connection.
16718
Willy Tarreau74ca5042013-06-11 23:12:07 +020016719dst_port : integer
16720 Returns an integer value corresponding to the destination TCP port of the
16721 connection on the client side, which is the port the client connected to.
16722 This might be used when running in transparent mode, when assigning dynamic
16723 ports to some clients for a whole application session, to stick all users to
16724 a same server, or to pass the destination port information to a server using
16725 an HTTP header.
16726
Willy Tarreau60ca10a2017-08-18 15:26:54 +020016727fc_http_major : integer
16728 Reports the front connection's HTTP major version encoding, which may be 1
16729 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
16730 encoding and not on the version present in the request header.
16731
Geoff Simmons7185b782019-08-27 18:31:16 +020016732fc_pp_authority : string
16733 Returns the authority TLV sent by the client in the PROXY protocol header,
16734 if any.
16735
Tim Duesterhusd1b15b62020-03-13 12:34:23 +010016736fc_pp_unique_id : string
16737 Returns the unique ID TLV sent by the client in the PROXY protocol header,
16738 if any.
16739
Emeric Brun4f603012017-01-05 15:11:44 +010016740fc_rcvd_proxy : boolean
16741 Returns true if the client initiated the connection with a PROXY protocol
16742 header.
16743
Thierry Fournier / OZON.IO6310bef2016-07-24 20:16:50 +020016744fc_rtt(<unit>) : integer
16745 Returns the Round Trip Time (RTT) measured by the kernel for the client
16746 connection. <unit> is facultative, by default the unit is milliseconds. <unit>
16747 can be set to "ms" for milliseconds or "us" for microseconds. If the server
16748 connection is not established, if the connection is not TCP or if the
16749 operating system does not support TCP_INFO, for example Linux kernels before
16750 2.4, the sample fetch fails.
16751
16752fc_rttvar(<unit>) : integer
16753 Returns the Round Trip Time (RTT) variance measured by the kernel for the
16754 client connection. <unit> is facultative, by default the unit is milliseconds.
16755 <unit> can be set to "ms" for milliseconds or "us" for microseconds. If the
16756 server connection is not established, if the connection is not TCP or if the
16757 operating system does not support TCP_INFO, for example Linux kernels before
16758 2.4, the sample fetch fails.
16759
Christopher Fauletba0c53e2019-10-17 14:40:48 +020016760fc_unacked : integer
Joe Williams30fcd392016-08-10 07:06:44 -070016761 Returns the unacked counter measured by the kernel for the client connection.
16762 If the server connection is not established, if the connection is not TCP or
16763 if the operating system does not support TCP_INFO, for example Linux kernels
16764 before 2.4, the sample fetch fails.
16765
Christopher Fauletba0c53e2019-10-17 14:40:48 +020016766fc_sacked : integer
Joe Williams30fcd392016-08-10 07:06:44 -070016767 Returns the sacked counter measured by the kernel for the client connection.
16768 If the server connection is not established, if the connection is not TCP or
16769 if the operating system does not support TCP_INFO, for example Linux kernels
16770 before 2.4, the sample fetch fails.
16771
Christopher Fauletba0c53e2019-10-17 14:40:48 +020016772fc_retrans : integer
Joe Williams30fcd392016-08-10 07:06:44 -070016773 Returns the retransmits counter measured by the kernel for the client
16774 connection. If the server connection is not established, if the connection is
16775 not TCP or if the operating system does not support TCP_INFO, for example
16776 Linux kernels before 2.4, the sample fetch fails.
16777
Christopher Fauletba0c53e2019-10-17 14:40:48 +020016778fc_fackets : integer
Joe Williams30fcd392016-08-10 07:06:44 -070016779 Returns the fack counter measured by the kernel for the client
16780 connection. If the server connection is not established, if the connection is
16781 not TCP or if the operating system does not support TCP_INFO, for example
16782 Linux kernels before 2.4, the sample fetch fails.
16783
Christopher Fauletba0c53e2019-10-17 14:40:48 +020016784fc_lost : integer
Joe Williams30fcd392016-08-10 07:06:44 -070016785 Returns the lost counter measured by the kernel for the client
16786 connection. If the server connection is not established, if the connection is
16787 not TCP or if the operating system does not support TCP_INFO, for example
16788 Linux kernels before 2.4, the sample fetch fails.
16789
Christopher Fauletba0c53e2019-10-17 14:40:48 +020016790fc_reordering : integer
Joe Williams30fcd392016-08-10 07:06:44 -070016791 Returns the reordering counter measured by the kernel for the client
16792 connection. If the server connection is not established, if the connection is
16793 not TCP or if the operating system does not support TCP_INFO, for example
16794 Linux kernels before 2.4, the sample fetch fails.
16795
Marcin Deranek9a66dfb2018-04-13 14:37:50 +020016796fe_defbe : string
16797 Returns a string containing the frontend's default backend name. It can be
16798 used in frontends to check which backend will handle requests by default.
16799
Willy Tarreau74ca5042013-06-11 23:12:07 +020016800fe_id : integer
16801 Returns an integer containing the current frontend's id. It can be used in
Marcin Deranek6e413ed2016-12-13 12:40:01 +010016802 backends to check from which frontend it was called, or to stick all users
Willy Tarreau74ca5042013-06-11 23:12:07 +020016803 coming via a same frontend to the same server.
16804
Marcin Deranekd2471c22016-12-12 14:08:05 +010016805fe_name : string
16806 Returns a string containing the current frontend's name. It can be used in
16807 backends to check from which frontend it was called, or to stick all users
16808 coming via a same frontend to the same server.
16809
Cyril Bonté62ba8702014-04-22 23:52:25 +020016810sc_bytes_in_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016811sc0_bytes_in_rate([<table>]) : integer
16812sc1_bytes_in_rate([<table>]) : integer
16813sc2_bytes_in_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020016814 Returns the average client-to-server bytes rate from the currently tracked
16815 counters, measured in amount of bytes over the period configured in the
16816 table. See also src_bytes_in_rate.
16817
Cyril Bonté62ba8702014-04-22 23:52:25 +020016818sc_bytes_out_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016819sc0_bytes_out_rate([<table>]) : integer
16820sc1_bytes_out_rate([<table>]) : integer
16821sc2_bytes_out_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020016822 Returns the average server-to-client bytes rate from the currently tracked
16823 counters, measured in amount of bytes over the period configured in the
16824 table. See also src_bytes_out_rate.
16825
Cyril Bonté62ba8702014-04-22 23:52:25 +020016826sc_clr_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016827sc0_clr_gpc0([<table>]) : integer
16828sc1_clr_gpc0([<table>]) : integer
16829sc2_clr_gpc0([<table>]) : integer
Willy Tarreauf73cd112011-08-13 01:45:16 +020016830 Clears the first General Purpose Counter associated to the currently tracked
16831 counters, and returns its previous value. Before the first invocation, the
Willy Tarreau869948b2013-01-04 14:14:57 +010016832 stored value is zero, so first invocation will always return zero. This is
16833 typically used as a second ACL in an expression in order to mark a connection
16834 when a first ACL was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020016835
Jarno Huuskonen676f6222017-03-30 09:19:45 +030016836 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020016837 # block if 5 consecutive requests continue to come faster than 10 sess
16838 # per second, and reset the counter as soon as the traffic slows down.
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020016839 acl abuse sc0_http_req_rate gt 10
16840 acl kill sc0_inc_gpc0 gt 5
16841 acl save sc0_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020016842 tcp-request connection accept if !abuse save
16843 tcp-request connection reject if abuse kill
16844
Frédéric Lécaille6778b272018-01-29 15:22:53 +010016845sc_clr_gpc1(<ctr>[,<table>]) : integer
16846sc0_clr_gpc1([<table>]) : integer
16847sc1_clr_gpc1([<table>]) : integer
16848sc2_clr_gpc1([<table>]) : integer
16849 Clears the second General Purpose Counter associated to the currently tracked
16850 counters, and returns its previous value. Before the first invocation, the
16851 stored value is zero, so first invocation will always return zero. This is
16852 typically used as a second ACL in an expression in order to mark a connection
16853 when a first ACL was verified.
16854
Cyril Bonté62ba8702014-04-22 23:52:25 +020016855sc_conn_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016856sc0_conn_cnt([<table>]) : integer
16857sc1_conn_cnt([<table>]) : integer
16858sc2_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010016859 Returns the cumulative number of incoming connections from currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020016860 counters. See also src_conn_cnt.
16861
Cyril Bonté62ba8702014-04-22 23:52:25 +020016862sc_conn_cur(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016863sc0_conn_cur([<table>]) : integer
16864sc1_conn_cur([<table>]) : integer
16865sc2_conn_cur([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020016866 Returns the current amount of concurrent connections tracking the same
16867 tracked counters. This number is automatically incremented when tracking
16868 begins and decremented when tracking stops. See also src_conn_cur.
16869
Cyril Bonté62ba8702014-04-22 23:52:25 +020016870sc_conn_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016871sc0_conn_rate([<table>]) : integer
16872sc1_conn_rate([<table>]) : integer
16873sc2_conn_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020016874 Returns the average connection rate from the currently tracked counters,
16875 measured in amount of connections over the period configured in the table.
16876 See also src_conn_rate.
16877
Cyril Bonté62ba8702014-04-22 23:52:25 +020016878sc_get_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016879sc0_get_gpc0([<table>]) : integer
16880sc1_get_gpc0([<table>]) : integer
16881sc2_get_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020016882 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020016883 currently tracked counters. See also src_get_gpc0 and sc/sc0/sc1/sc2_inc_gpc0.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020016884
Frédéric Lécaille6778b272018-01-29 15:22:53 +010016885sc_get_gpc1(<ctr>[,<table>]) : integer
16886sc0_get_gpc1([<table>]) : integer
16887sc1_get_gpc1([<table>]) : integer
16888sc2_get_gpc1([<table>]) : integer
16889 Returns the value of the second General Purpose Counter associated to the
16890 currently tracked counters. See also src_get_gpc1 and sc/sc0/sc1/sc2_inc_gpc1.
16891
Thierry FOURNIER236657b2015-08-19 08:25:14 +020016892sc_get_gpt0(<ctr>[,<table>]) : integer
16893sc0_get_gpt0([<table>]) : integer
16894sc1_get_gpt0([<table>]) : integer
16895sc2_get_gpt0([<table>]) : integer
16896 Returns the value of the first General Purpose Tag associated to the
16897 currently tracked counters. See also src_get_gpt0.
16898
Cyril Bonté62ba8702014-04-22 23:52:25 +020016899sc_gpc0_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016900sc0_gpc0_rate([<table>]) : integer
16901sc1_gpc0_rate([<table>]) : integer
16902sc2_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020016903 Returns the average increment rate of the first General Purpose Counter
16904 associated to the currently tracked counters. It reports the frequency
16905 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020016906 src_gpc0_rate, sc/sc0/sc1/sc2_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
16907 that the "gpc0_rate" counter must be stored in the stick-table for a value to
16908 be returned, as "gpc0" only holds the event count.
Willy Tarreaue9656522010-08-17 15:40:09 +020016909
Frédéric Lécaille6778b272018-01-29 15:22:53 +010016910sc_gpc1_rate(<ctr>[,<table>]) : integer
16911sc0_gpc1_rate([<table>]) : integer
16912sc1_gpc1_rate([<table>]) : integer
16913sc2_gpc1_rate([<table>]) : integer
16914 Returns the average increment rate of the second General Purpose Counter
16915 associated to the currently tracked counters. It reports the frequency
16916 which the gpc1 counter was incremented over the configured period. See also
16917 src_gpcA_rate, sc/sc0/sc1/sc2_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
16918 that the "gpc1_rate" counter must be stored in the stick-table for a value to
16919 be returned, as "gpc1" only holds the event count.
16920
Cyril Bonté62ba8702014-04-22 23:52:25 +020016921sc_http_err_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016922sc0_http_err_cnt([<table>]) : integer
16923sc1_http_err_cnt([<table>]) : integer
16924sc2_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010016925 Returns the cumulative number of HTTP errors from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020016926 counters. This includes the both request errors and 4xx error responses.
16927 See also src_http_err_cnt.
16928
Cyril Bonté62ba8702014-04-22 23:52:25 +020016929sc_http_err_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016930sc0_http_err_rate([<table>]) : integer
16931sc1_http_err_rate([<table>]) : integer
16932sc2_http_err_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020016933 Returns the average rate of HTTP errors from the currently tracked counters,
16934 measured in amount of errors over the period configured in the table. This
16935 includes the both request errors and 4xx error responses. See also
16936 src_http_err_rate.
16937
Cyril Bonté62ba8702014-04-22 23:52:25 +020016938sc_http_req_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016939sc0_http_req_cnt([<table>]) : integer
16940sc1_http_req_cnt([<table>]) : integer
16941sc2_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010016942 Returns the cumulative number of HTTP requests from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020016943 counters. This includes every started request, valid or not. See also
16944 src_http_req_cnt.
16945
Cyril Bonté62ba8702014-04-22 23:52:25 +020016946sc_http_req_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016947sc0_http_req_rate([<table>]) : integer
16948sc1_http_req_rate([<table>]) : integer
16949sc2_http_req_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020016950 Returns the average rate of HTTP requests from the currently tracked
16951 counters, measured in amount of requests over the period configured in
16952 the table. This includes every started request, valid or not. See also
16953 src_http_req_rate.
16954
Cyril Bonté62ba8702014-04-22 23:52:25 +020016955sc_inc_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016956sc0_inc_gpc0([<table>]) : integer
16957sc1_inc_gpc0([<table>]) : integer
16958sc2_inc_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020016959 Increments the first General Purpose Counter associated to the currently
Willy Tarreau869948b2013-01-04 14:14:57 +010016960 tracked counters, and returns its new value. Before the first invocation,
16961 the stored value is zero, so first invocation will increase it to 1 and will
16962 return 1. This is typically used as a second ACL in an expression in order
16963 to mark a connection when a first ACL was verified :
Willy Tarreaue9656522010-08-17 15:40:09 +020016964
Jarno Huuskonen676f6222017-03-30 09:19:45 +030016965 Example:
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020016966 acl abuse sc0_http_req_rate gt 10
16967 acl kill sc0_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020016968 tcp-request connection reject if abuse kill
16969
Frédéric Lécaille6778b272018-01-29 15:22:53 +010016970sc_inc_gpc1(<ctr>[,<table>]) : integer
16971sc0_inc_gpc1([<table>]) : integer
16972sc1_inc_gpc1([<table>]) : integer
16973sc2_inc_gpc1([<table>]) : integer
16974 Increments the second General Purpose Counter associated to the currently
16975 tracked counters, and returns its new value. Before the first invocation,
16976 the stored value is zero, so first invocation will increase it to 1 and will
16977 return 1. This is typically used as a second ACL in an expression in order
16978 to mark a connection when a first ACL was verified.
16979
Cyril Bonté62ba8702014-04-22 23:52:25 +020016980sc_kbytes_in(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016981sc0_kbytes_in([<table>]) : integer
16982sc1_kbytes_in([<table>]) : integer
16983sc2_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020016984 Returns the total amount of client-to-server data from the currently tracked
16985 counters, measured in kilobytes. The test is currently performed on 32-bit
16986 integers, which limits values to 4 terabytes. See also src_kbytes_in.
Willy Tarreaue9656522010-08-17 15:40:09 +020016987
Cyril Bonté62ba8702014-04-22 23:52:25 +020016988sc_kbytes_out(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016989sc0_kbytes_out([<table>]) : integer
16990sc1_kbytes_out([<table>]) : integer
16991sc2_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020016992 Returns the total amount of server-to-client data from the currently tracked
16993 counters, measured in kilobytes. The test is currently performed on 32-bit
16994 integers, which limits values to 4 terabytes. See also src_kbytes_out.
Willy Tarreaue9656522010-08-17 15:40:09 +020016995
Cyril Bonté62ba8702014-04-22 23:52:25 +020016996sc_sess_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016997sc0_sess_cnt([<table>]) : integer
16998sc1_sess_cnt([<table>]) : integer
16999sc2_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017000 Returns the cumulative number of incoming connections that were transformed
Willy Tarreaue9656522010-08-17 15:40:09 +020017001 into sessions, which means that they were accepted by a "tcp-request
17002 connection" rule, from the currently tracked counters. A backend may count
17003 more sessions than connections because each connection could result in many
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017004 backend sessions if some HTTP keep-alive is performed over the connection
Willy Tarreaue9656522010-08-17 15:40:09 +020017005 with the client. See also src_sess_cnt.
17006
Cyril Bonté62ba8702014-04-22 23:52:25 +020017007sc_sess_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017008sc0_sess_rate([<table>]) : integer
17009sc1_sess_rate([<table>]) : integer
17010sc2_sess_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017011 Returns the average session rate from the currently tracked counters,
17012 measured in amount of sessions over the period configured in the table. A
17013 session is a connection that got past the early "tcp-request connection"
17014 rules. A backend may count more sessions than connections because each
17015 connection could result in many backend sessions if some HTTP keep-alive is
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017016 performed over the connection with the client. See also src_sess_rate.
Willy Tarreaue9656522010-08-17 15:40:09 +020017017
Cyril Bonté62ba8702014-04-22 23:52:25 +020017018sc_tracked(<ctr>[,<table>]) : boolean
Willy Tarreau0f791d42013-07-23 19:56:43 +020017019sc0_tracked([<table>]) : boolean
17020sc1_tracked([<table>]) : boolean
17021sc2_tracked([<table>]) : boolean
Willy Tarreau6f1615f2013-06-03 15:15:22 +020017022 Returns true if the designated session counter is currently being tracked by
17023 the current session. This can be useful when deciding whether or not we want
17024 to set some values in a header passed to the server.
17025
Cyril Bonté62ba8702014-04-22 23:52:25 +020017026sc_trackers(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017027sc0_trackers([<table>]) : integer
17028sc1_trackers([<table>]) : integer
17029sc2_trackers([<table>]) : integer
Willy Tarreau2406db42012-12-09 12:16:43 +010017030 Returns the current amount of concurrent connections tracking the same
17031 tracked counters. This number is automatically incremented when tracking
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020017032 begins and decremented when tracking stops. It differs from sc0_conn_cur in
Willy Tarreau2406db42012-12-09 12:16:43 +010017033 that it does not rely on any stored information but on the table's reference
17034 count (the "use" value which is returned by "show table" on the CLI). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020017035 may sometimes be more suited for layer7 tracking. It can be used to tell a
17036 server how many concurrent connections there are from a given address for
17037 example.
Willy Tarreau2406db42012-12-09 12:16:43 +010017038
Willy Tarreau74ca5042013-06-11 23:12:07 +020017039so_id : integer
17040 Returns an integer containing the current listening socket's id. It is useful
17041 in frontends involving many "bind" lines, or to stick all users coming via a
17042 same socket to the same server.
Willy Tarreaud63335a2010-02-26 12:56:52 +010017043
Jerome Magnineb421b22020-03-27 22:08:40 +010017044so_name : string
17045 Returns a string containing the current listening socket's name, as defined
17046 with name on a "bind" line. It can serve the same purposes as so_id but with
17047 strings instead of integers.
17048
Willy Tarreau74ca5042013-06-11 23:12:07 +020017049src : ip
Davor Ocelice9ed2812017-12-25 17:49:28 +010017050 This is the source IPv4 address of the client of the session. It is of type
Willy Tarreau74ca5042013-06-11 23:12:07 +020017051 IP and works on both IPv4 and IPv6 tables. On IPv6 tables, IPv4 addresses are
17052 mapped to their IPv6 equivalent, according to RFC 4291. Note that it is the
17053 TCP-level source address which is used, and not the address of a client
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010017054 behind a proxy. However if the "accept-proxy" or "accept-netscaler-cip" bind
17055 directive is used, it can be the address of a client behind another
17056 PROXY-protocol compatible component for all rule sets except
Willy Tarreau64ded3d2019-01-23 10:02:15 +010017057 "tcp-request connection" which sees the real address. When the incoming
17058 connection passed through address translation or redirection involving
17059 connection tracking, the original destination address before the redirection
17060 will be reported. On Linux systems, the source and destination may seldom
17061 appear reversed if the nf_conntrack_tcp_loose sysctl is set, because a late
17062 response may reopen a timed out connection and switch what is believed to be
17063 the source and the destination.
Willy Tarreaud63335a2010-02-26 12:56:52 +010017064
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010017065 Example:
17066 # add an HTTP header in requests with the originating address' country
17067 http-request set-header X-Country %[src,map_ip(geoip.lst)]
17068
Willy Tarreau74ca5042013-06-11 23:12:07 +020017069src_bytes_in_rate([<table>]) : integer
17070 Returns the average bytes rate from the incoming connection's source address
17071 in the current proxy's stick-table or in the designated stick-table, measured
17072 in amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017073 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_in_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017074
Willy Tarreau74ca5042013-06-11 23:12:07 +020017075src_bytes_out_rate([<table>]) : integer
17076 Returns the average bytes rate to the incoming connection's source address in
17077 the current proxy's stick-table or in the designated stick-table, measured in
Willy Tarreauc9705a12010-07-27 20:05:50 +020017078 amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017079 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_out_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017080
Willy Tarreau74ca5042013-06-11 23:12:07 +020017081src_clr_gpc0([<table>]) : integer
17082 Clears the first General Purpose Counter associated to the incoming
17083 connection's source address in the current proxy's stick-table or in the
17084 designated stick-table, and returns its previous value. If the address is not
17085 found, an entry is created and 0 is returned. This is typically used as a
17086 second ACL in an expression in order to mark a connection when a first ACL
17087 was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020017088
Jarno Huuskonen676f6222017-03-30 09:19:45 +030017089 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020017090 # block if 5 consecutive requests continue to come faster than 10 sess
17091 # per second, and reset the counter as soon as the traffic slows down.
17092 acl abuse src_http_req_rate gt 10
17093 acl kill src_inc_gpc0 gt 5
Willy Tarreau869948b2013-01-04 14:14:57 +010017094 acl save src_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020017095 tcp-request connection accept if !abuse save
17096 tcp-request connection reject if abuse kill
17097
Frédéric Lécaille6778b272018-01-29 15:22:53 +010017098src_clr_gpc1([<table>]) : integer
17099 Clears the second General Purpose Counter associated to the incoming
17100 connection's source address in the current proxy's stick-table or in the
17101 designated stick-table, and returns its previous value. If the address is not
17102 found, an entry is created and 0 is returned. This is typically used as a
17103 second ACL in an expression in order to mark a connection when a first ACL
17104 was verified.
17105
Willy Tarreau74ca5042013-06-11 23:12:07 +020017106src_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017107 Returns the cumulative number of connections initiated from the current
Willy Tarreau74ca5042013-06-11 23:12:07 +020017108 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020017109 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017110 See also sc/sc0/sc1/sc2_conn_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017111
Willy Tarreau74ca5042013-06-11 23:12:07 +020017112src_conn_cur([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020017113 Returns the current amount of concurrent connections initiated from the
Willy Tarreau74ca5042013-06-11 23:12:07 +020017114 current incoming connection's source address in the current proxy's
17115 stick-table or in the designated stick-table. If the address is not found,
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017116 zero is returned. See also sc/sc0/sc1/sc2_conn_cur.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017117
Willy Tarreau74ca5042013-06-11 23:12:07 +020017118src_conn_rate([<table>]) : integer
17119 Returns the average connection rate from the incoming connection's source
17120 address in the current proxy's stick-table or in the designated stick-table,
17121 measured in amount of connections over the period configured in the table. If
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017122 the address is not found, zero is returned. See also sc/sc0/sc1/sc2_conn_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017123
Willy Tarreau74ca5042013-06-11 23:12:07 +020017124src_get_gpc0([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020017125 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau74ca5042013-06-11 23:12:07 +020017126 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020017127 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017128 See also sc/sc0/sc1/sc2_get_gpc0 and src_inc_gpc0.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017129
Frédéric Lécaille6778b272018-01-29 15:22:53 +010017130src_get_gpc1([<table>]) : integer
17131 Returns the value of the second General Purpose Counter associated to the
17132 incoming connection's source address in the current proxy's stick-table or in
17133 the designated stick-table. If the address is not found, zero is returned.
17134 See also sc/sc0/sc1/sc2_get_gpc1 and src_inc_gpc1.
17135
Thierry FOURNIER236657b2015-08-19 08:25:14 +020017136src_get_gpt0([<table>]) : integer
17137 Returns the value of the first General Purpose Tag associated to the
17138 incoming connection's source address in the current proxy's stick-table or in
17139 the designated stick-table. If the address is not found, zero is returned.
17140 See also sc/sc0/sc1/sc2_get_gpt0.
17141
Willy Tarreau74ca5042013-06-11 23:12:07 +020017142src_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020017143 Returns the average increment rate of the first General Purpose Counter
Willy Tarreau74ca5042013-06-11 23:12:07 +020017144 associated to the incoming connection's source address in the current proxy's
Willy Tarreauba2ffd12013-05-29 15:54:14 +020017145 stick-table or in the designated stick-table. It reports the frequency
17146 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017147 sc/sc0/sc1/sc2_gpc0_rate, src_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
17148 that the "gpc0_rate" counter must be stored in the stick-table for a value to
17149 be returned, as "gpc0" only holds the event count.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020017150
Frédéric Lécaille6778b272018-01-29 15:22:53 +010017151src_gpc1_rate([<table>]) : integer
17152 Returns the average increment rate of the second General Purpose Counter
17153 associated to the incoming connection's source address in the current proxy's
17154 stick-table or in the designated stick-table. It reports the frequency
17155 which the gpc1 counter was incremented over the configured period. See also
17156 sc/sc0/sc1/sc2_gpc1_rate, src_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
17157 that the "gpc1_rate" counter must be stored in the stick-table for a value to
17158 be returned, as "gpc1" only holds the event count.
17159
Willy Tarreau74ca5042013-06-11 23:12:07 +020017160src_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017161 Returns the cumulative number of HTTP errors from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020017162 source address in the current proxy's stick-table or in the designated
Willy Tarreauc9705a12010-07-27 20:05:50 +020017163 stick-table. This includes the both request errors and 4xx error responses.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017164 See also sc/sc0/sc1/sc2_http_err_cnt. If the address is not found, zero is
Willy Tarreau74ca5042013-06-11 23:12:07 +020017165 returned.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017166
Willy Tarreau74ca5042013-06-11 23:12:07 +020017167src_http_err_rate([<table>]) : integer
17168 Returns the average rate of HTTP errors from the incoming connection's source
17169 address in the current proxy's stick-table or in the designated stick-table,
17170 measured in amount of errors over the period configured in the table. This
17171 includes the both request errors and 4xx error responses. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017172 not found, zero is returned. See also sc/sc0/sc1/sc2_http_err_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017173
Willy Tarreau74ca5042013-06-11 23:12:07 +020017174src_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017175 Returns the cumulative number of HTTP requests from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020017176 source address in the current proxy's stick-table or in the designated stick-
17177 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017178 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017179
Willy Tarreau74ca5042013-06-11 23:12:07 +020017180src_http_req_rate([<table>]) : integer
17181 Returns the average rate of HTTP requests from the incoming connection's
17182 source address in the current proxy's stick-table or in the designated stick-
17183 table, measured in amount of requests over the period configured in the
Willy Tarreauc9705a12010-07-27 20:05:50 +020017184 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017185 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017186
Willy Tarreau74ca5042013-06-11 23:12:07 +020017187src_inc_gpc0([<table>]) : integer
17188 Increments the first General Purpose Counter associated to the incoming
17189 connection's source address in the current proxy's stick-table or in the
17190 designated stick-table, and returns its new value. If the address is not
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020017191 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc0.
Willy Tarreau74ca5042013-06-11 23:12:07 +020017192 This is typically used as a second ACL in an expression in order to mark a
17193 connection when a first ACL was verified :
Willy Tarreauc9705a12010-07-27 20:05:50 +020017194
Jarno Huuskonen676f6222017-03-30 09:19:45 +030017195 Example:
Willy Tarreauc9705a12010-07-27 20:05:50 +020017196 acl abuse src_http_req_rate gt 10
Willy Tarreau869948b2013-01-04 14:14:57 +010017197 acl kill src_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020017198 tcp-request connection reject if abuse kill
Willy Tarreauc9705a12010-07-27 20:05:50 +020017199
Frédéric Lécaille6778b272018-01-29 15:22:53 +010017200src_inc_gpc1([<table>]) : integer
17201 Increments the second General Purpose Counter associated to the incoming
17202 connection's source address in the current proxy's stick-table or in the
17203 designated stick-table, and returns its new value. If the address is not
17204 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc1.
17205 This is typically used as a second ACL in an expression in order to mark a
17206 connection when a first ACL was verified.
17207
Willy Tarreau16e01562016-08-09 16:46:18 +020017208src_is_local : boolean
17209 Returns true if the source address of the incoming connection is local to the
17210 system, or false if the address doesn't exist on the system, meaning that it
17211 comes from a remote machine. Note that UNIX addresses are considered local.
17212 It can be useful to apply certain access restrictions based on where the
Davor Ocelice9ed2812017-12-25 17:49:28 +010017213 client comes from (e.g. require auth or https for remote machines). Please
Willy Tarreau16e01562016-08-09 16:46:18 +020017214 note that the check involves a few system calls, so it's better to do it only
17215 once per connection.
17216
Willy Tarreau74ca5042013-06-11 23:12:07 +020017217src_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020017218 Returns the total amount of data received from the incoming connection's
17219 source address in the current proxy's stick-table or in the designated
17220 stick-table, measured in kilobytes. If the address is not found, zero is
17221 returned. The test is currently performed on 32-bit integers, which limits
17222 values to 4 terabytes. See also sc/sc0/sc1/sc2_kbytes_in.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017223
Willy Tarreau74ca5042013-06-11 23:12:07 +020017224src_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020017225 Returns the total amount of data sent to the incoming connection's source
17226 address in the current proxy's stick-table or in the designated stick-table,
17227 measured in kilobytes. If the address is not found, zero is returned. The
17228 test is currently performed on 32-bit integers, which limits values to 4
17229 terabytes. See also sc/sc0/sc1/sc2_kbytes_out.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020017230
Willy Tarreau74ca5042013-06-11 23:12:07 +020017231src_port : integer
17232 Returns an integer value corresponding to the TCP source port of the
17233 connection on the client side, which is the port the client connected from.
17234 Usage of this function is very limited as modern protocols do not care much
17235 about source ports nowadays.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010017236
Willy Tarreau74ca5042013-06-11 23:12:07 +020017237src_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017238 Returns the cumulative number of connections initiated from the incoming
Willy Tarreauc9705a12010-07-27 20:05:50 +020017239 connection's source IPv4 address in the current proxy's stick-table or in the
17240 designated stick-table, that were transformed into sessions, which means that
17241 they were accepted by "tcp-request" rules. If the address is not found, zero
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017242 is returned. See also sc/sc0/sc1/sc2_sess_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017243
Willy Tarreau74ca5042013-06-11 23:12:07 +020017244src_sess_rate([<table>]) : integer
17245 Returns the average session rate from the incoming connection's source
17246 address in the current proxy's stick-table or in the designated stick-table,
17247 measured in amount of sessions over the period configured in the table. A
17248 session is a connection that went past the early "tcp-request" rules. If the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017249 address is not found, zero is returned. See also sc/sc0/sc1/sc2_sess_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017250
Willy Tarreau74ca5042013-06-11 23:12:07 +020017251src_updt_conn_cnt([<table>]) : integer
17252 Creates or updates the entry associated to the incoming connection's source
17253 address in the current proxy's stick-table or in the designated stick-table.
17254 This table must be configured to store the "conn_cnt" data type, otherwise
17255 the match will be ignored. The current count is incremented by one, and the
17256 expiration timer refreshed. The updated count is returned, so this match
17257 can't return zero. This was used to reject service abusers based on their
17258 source address. Note: it is recommended to use the more complete "track-sc*"
17259 actions in "tcp-request" rules instead.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020017260
17261 Example :
17262 # This frontend limits incoming SSH connections to 3 per 10 second for
17263 # each source address, and rejects excess connections until a 10 second
17264 # silence is observed. At most 20 addresses are tracked.
17265 listen ssh
17266 bind :22
17267 mode tcp
17268 maxconn 100
Willy Tarreauc9705a12010-07-27 20:05:50 +020017269 stick-table type ip size 20 expire 10s store conn_cnt
Willy Tarreau74ca5042013-06-11 23:12:07 +020017270 tcp-request content reject if { src_updt_conn_cnt gt 3 }
Willy Tarreaua975b8f2010-06-05 19:13:27 +020017271 server local 127.0.0.1:22
17272
Willy Tarreau74ca5042013-06-11 23:12:07 +020017273srv_id : integer
17274 Returns an integer containing the server's id when processing the response.
17275 While it's almost only used with ACLs, it may be used for logging or
Christopher Fauletd1b44642020-04-30 09:51:15 +020017276 debugging. It can also be used in a tcp-check or an http-check ruleset.
Hervé COMMOWICKdaa824e2011-08-05 12:09:44 +020017277
vkill1dfd1652019-10-30 16:58:14 +080017278srv_name : string
17279 Returns a string containing the server's name when processing the response.
17280 While it's almost only used with ACLs, it may be used for logging or
Christopher Fauletd1b44642020-04-30 09:51:15 +020017281 debugging. It can also be used in a tcp-check or an http-check ruleset.
vkill1dfd1652019-10-30 16:58:14 +080017282
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200172837.3.4. Fetching samples at Layer 5
Willy Tarreau74ca5042013-06-11 23:12:07 +020017284----------------------------------
Willy Tarreau0b1cd942010-05-16 22:18:27 +020017285
Willy Tarreau74ca5042013-06-11 23:12:07 +020017286The layer 5 usually describes just the session layer which in haproxy is
17287closest to the session once all the connection handshakes are finished, but
17288when no content is yet made available. The fetch methods described here are
17289usable as low as the "tcp-request content" rule sets unless they require some
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017290future information. Those generally include the results of SSL negotiations.
Willy Tarreauc735a072011-03-29 00:57:02 +020017291
Ben Shillitof25e8e52016-12-02 14:25:37 +00001729251d.all(<prop>[,<prop>*]) : string
17293 Returns values for the properties requested as a string, where values are
17294 separated by the delimiter specified with "51degrees-property-separator".
17295 The device is identified using all the important HTTP headers from the
17296 request. The function can be passed up to five property names, and if a
17297 property name can't be found, the value "NoData" is returned.
17298
17299 Example :
17300 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request
17301 # containing the three properties requested using all relevant headers from
17302 # the request.
17303 frontend http-in
17304 bind *:8081
17305 default_backend servers
17306 http-request set-header X-51D-DeviceTypeMobileTablet \
17307 %[51d.all(DeviceType,IsMobile,IsTablet)]
17308
Emeric Brun645ae792014-04-30 14:21:06 +020017309ssl_bc : boolean
17310 Returns true when the back connection was made via an SSL/TLS transport
17311 layer and is locally deciphered. This means the outgoing connection was made
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017312 other a server with the "ssl" option. It can be used in a tcp-check or an
17313 http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017314
17315ssl_bc_alg_keysize : integer
17316 Returns the symmetric cipher key size supported in bits when the outgoing
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017317 connection was made over an SSL/TLS transport layer. It can be used in a
17318 tcp-check or an http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017319
Olivier Houchard6b77f492018-11-22 18:18:29 +010017320ssl_bc_alpn : string
17321 This extracts the Application Layer Protocol Negotiation field from an
17322 outgoing connection made via a TLS transport layer.
Michael Prokop4438c602019-05-24 10:25:45 +020017323 The result is a string containing the protocol name negotiated with the
Olivier Houchard6b77f492018-11-22 18:18:29 +010017324 server. The SSL library must have been built with support for TLS
17325 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
17326 not advertised unless the "alpn" keyword on the "server" line specifies a
17327 protocol list. Also, nothing forces the server to pick a protocol from this
17328 list, any other one may be requested. The TLS ALPN extension is meant to
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017329 replace the TLS NPN extension. See also "ssl_bc_npn". It can be used in a
17330 tcp-check or an http-check ruleset.
Olivier Houchard6b77f492018-11-22 18:18:29 +010017331
Emeric Brun645ae792014-04-30 14:21:06 +020017332ssl_bc_cipher : string
17333 Returns the name of the used cipher when the outgoing connection was made
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017334 over an SSL/TLS transport layer. It can be used in a tcp-check or an
17335 http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017336
Patrick Hemmer65674662019-06-04 08:13:03 -040017337ssl_bc_client_random : binary
17338 Returns the client random of the back connection when the incoming connection
17339 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
17340 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017341 It can be used in a tcp-check or an http-check ruleset.
Patrick Hemmer65674662019-06-04 08:13:03 -040017342
Emeric Brun74f7ffa2018-02-19 16:14:12 +010017343ssl_bc_is_resumed : boolean
17344 Returns true when the back connection was made over an SSL/TLS transport
17345 layer and the newly created SSL session was resumed using a cached
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017346 session or a TLS ticket. It can be used in a tcp-check or an http-check
17347 ruleset.
Emeric Brun74f7ffa2018-02-19 16:14:12 +010017348
Olivier Houchard6b77f492018-11-22 18:18:29 +010017349ssl_bc_npn : string
17350 This extracts the Next Protocol Negotiation field from an outgoing connection
17351 made via a TLS transport layer. The result is a string containing the
Michael Prokop4438c602019-05-24 10:25:45 +020017352 protocol name negotiated with the server . The SSL library must have been
Olivier Houchard6b77f492018-11-22 18:18:29 +010017353 built with support for TLS extensions enabled (check haproxy -vv). Note that
17354 the TLS NPN extension is not advertised unless the "npn" keyword on the
17355 "server" line specifies a protocol list. Also, nothing forces the server to
17356 pick a protocol from this list, any other one may be used. Please note that
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017357 the TLS NPN extension was replaced with ALPN. It can be used in a tcp-check
17358 or an http-check ruleset.
Olivier Houchard6b77f492018-11-22 18:18:29 +010017359
Emeric Brun645ae792014-04-30 14:21:06 +020017360ssl_bc_protocol : string
17361 Returns the name of the used protocol when the outgoing connection was made
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017362 over an SSL/TLS transport layer. It can be used in a tcp-check or an
17363 http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017364
Emeric Brunb73a9b02014-04-30 18:49:19 +020017365ssl_bc_unique_id : binary
Emeric Brun645ae792014-04-30 14:21:06 +020017366 When the outgoing connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020017367 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017368 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64". It
17369 can be used in a tcp-check or an http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017370
Patrick Hemmer65674662019-06-04 08:13:03 -040017371ssl_bc_server_random : binary
17372 Returns the server random of the back connection when the incoming connection
17373 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
17374 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017375 It can be used in a tcp-check or an http-check ruleset.
Patrick Hemmer65674662019-06-04 08:13:03 -040017376
Emeric Brun645ae792014-04-30 14:21:06 +020017377ssl_bc_session_id : binary
17378 Returns the SSL ID of the back connection when the outgoing connection was
17379 made over an SSL/TLS transport layer. It is useful to log if we want to know
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017380 if session was reused or not. It can be used in a tcp-check or an http-check
17381 ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017382
Patrick Hemmere0275472018-04-28 19:15:51 -040017383ssl_bc_session_key : binary
17384 Returns the SSL session master key of the back connection when the outgoing
17385 connection was made over an SSL/TLS transport layer. It is useful to decrypt
17386 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017387 BoringSSL. It can be used in a tcp-check or an http-check ruleset.
Patrick Hemmere0275472018-04-28 19:15:51 -040017388
Emeric Brun645ae792014-04-30 14:21:06 +020017389ssl_bc_use_keysize : integer
17390 Returns the symmetric cipher key size used in bits when the outgoing
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017391 connection was made over an SSL/TLS transport layer. It can be used in a
17392 tcp-check or an http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017393
Willy Tarreau74ca5042013-06-11 23:12:07 +020017394ssl_c_ca_err : integer
17395 When the incoming connection was made over an SSL/TLS transport layer,
17396 returns the ID of the first error detected during verification of the client
17397 certificate at depth > 0, or 0 if no error was encountered during this
17398 verification process. Please refer to your SSL library's documentation to
17399 find the exhaustive list of error codes.
Willy Tarreauc735a072011-03-29 00:57:02 +020017400
Willy Tarreau74ca5042013-06-11 23:12:07 +020017401ssl_c_ca_err_depth : integer
17402 When the incoming connection was made over an SSL/TLS transport layer,
17403 returns the depth in the CA chain of the first error detected during the
17404 verification of the client certificate. If no error is encountered, 0 is
17405 returned.
Willy Tarreau0ba27502007-12-24 16:55:16 +010017406
Christopher Faulet7a507632020-11-06 12:10:33 +010017407ssl_c_chain_der : binary
William Dauchya598b502020-08-06 18:11:38 +020017408 Returns the DER formatted chain certificate presented by the client when the
17409 incoming connection was made over an SSL/TLS transport layer. When used for
17410 an ACL, the value(s) to match against can be passed in hexadecimal form. One
17411 can parse the result with any lib accepting ASN.1 DER data. It currentlly
17412 does not support resumed sessions.
17413
Christopher Faulet7a507632020-11-06 12:10:33 +010017414ssl_c_der : binary
17415 Returns the DER formatted certificate presented by the client when the
17416 incoming connection was made over an SSL/TLS transport layer. When used for
17417 an ACL, the value(s) to match against can be passed in hexadecimal form.
17418
Willy Tarreau74ca5042013-06-11 23:12:07 +020017419ssl_c_err : integer
17420 When the incoming connection was made over an SSL/TLS transport layer,
17421 returns the ID of the first error detected during verification at depth 0, or
17422 0 if no error was encountered during this verification process. Please refer
17423 to your SSL library's documentation to find the exhaustive list of error
17424 codes.
Willy Tarreau62644772008-07-16 18:36:06 +020017425
Elliot Otchet71f82972020-01-15 08:12:14 -050017426ssl_c_i_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020017427 When the incoming connection was made over an SSL/TLS transport layer,
17428 returns the full distinguished name of the issuer of the certificate
17429 presented by the client when no <entry> is specified, or the value of the
17430 first given entry found from the beginning of the DN. If a positive/negative
17431 occurrence number is specified as the optional second argument, it returns
17432 the value of the nth given entry value from the beginning/end of the DN.
17433 For instance, "ssl_c_i_dn(OU,2)" the second organization unit, and
17434 "ssl_c_i_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050017435 The <format> parameter allows you to receive the DN suitable for
17436 consumption by different protocols. Currently supported is rfc2253 for
17437 LDAP v3.
17438 If you'd like to modify the format only you can specify an empty string
17439 and zero for the first two parameters. Example: ssl_c_i_dn(,0,rfc2253)
Willy Tarreau62644772008-07-16 18:36:06 +020017440
Willy Tarreau74ca5042013-06-11 23:12:07 +020017441ssl_c_key_alg : string
17442 Returns the name of the algorithm used to generate the key of the certificate
17443 presented by the client when the incoming connection was made over an SSL/TLS
17444 transport layer.
Willy Tarreau62644772008-07-16 18:36:06 +020017445
Willy Tarreau74ca5042013-06-11 23:12:07 +020017446ssl_c_notafter : string
17447 Returns the end date presented by the client as a formatted string
17448 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
17449 transport layer.
Emeric Brunbede3d02009-06-30 17:54:00 +020017450
Willy Tarreau74ca5042013-06-11 23:12:07 +020017451ssl_c_notbefore : string
17452 Returns the start date presented by the client as a formatted string
17453 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
17454 transport layer.
Willy Tarreaub6672b52011-12-12 17:23:41 +010017455
Elliot Otchet71f82972020-01-15 08:12:14 -050017456ssl_c_s_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020017457 When the incoming connection was made over an SSL/TLS transport layer,
17458 returns the full distinguished name of the subject of the certificate
17459 presented by the client when no <entry> is specified, or the value of the
17460 first given entry found from the beginning of the DN. If a positive/negative
17461 occurrence number is specified as the optional second argument, it returns
17462 the value of the nth given entry value from the beginning/end of the DN.
17463 For instance, "ssl_c_s_dn(OU,2)" the second organization unit, and
17464 "ssl_c_s_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050017465 The <format> parameter allows you to receive the DN suitable for
17466 consumption by different protocols. Currently supported is rfc2253 for
17467 LDAP v3.
17468 If you'd like to modify the format only you can specify an empty string
17469 and zero for the first two parameters. Example: ssl_c_s_dn(,0,rfc2253)
Willy Tarreaub6672b52011-12-12 17:23:41 +010017470
Willy Tarreau74ca5042013-06-11 23:12:07 +020017471ssl_c_serial : binary
17472 Returns the serial of the certificate presented by the client when the
17473 incoming connection was made over an SSL/TLS transport layer. When used for
17474 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun2525b6b2012-10-18 15:59:43 +020017475
Willy Tarreau74ca5042013-06-11 23:12:07 +020017476ssl_c_sha1 : binary
17477 Returns the SHA-1 fingerprint of the certificate presented by the client when
17478 the incoming connection was made over an SSL/TLS transport layer. This can be
17479 used to stick a client to a server, or to pass this information to a server.
Willy Tarreau2d0caa32014-07-02 19:01:22 +020017480 Note that the output is binary, so if you want to pass that signature to the
17481 server, you need to encode it in hex or base64, such as in the example below:
17482
Jarno Huuskonen676f6222017-03-30 09:19:45 +030017483 Example:
Willy Tarreau2d0caa32014-07-02 19:01:22 +020017484 http-request set-header X-SSL-Client-SHA1 %[ssl_c_sha1,hex]
Emeric Brun2525b6b2012-10-18 15:59:43 +020017485
Willy Tarreau74ca5042013-06-11 23:12:07 +020017486ssl_c_sig_alg : string
17487 Returns the name of the algorithm used to sign the certificate presented by
17488 the client when the incoming connection was made over an SSL/TLS transport
17489 layer.
Emeric Brun87855892012-10-17 17:39:35 +020017490
Willy Tarreau74ca5042013-06-11 23:12:07 +020017491ssl_c_used : boolean
17492 Returns true if current SSL session uses a client certificate even if current
17493 connection uses SSL session resumption. See also "ssl_fc_has_crt".
Emeric Brun7f56e742012-10-19 18:15:40 +020017494
Willy Tarreau74ca5042013-06-11 23:12:07 +020017495ssl_c_verify : integer
17496 Returns the verify result error ID when the incoming connection was made over
17497 an SSL/TLS transport layer, otherwise zero if no error is encountered. Please
17498 refer to your SSL library's documentation for an exhaustive list of error
17499 codes.
Emeric Brunce5ad802012-10-22 14:11:22 +020017500
Willy Tarreau74ca5042013-06-11 23:12:07 +020017501ssl_c_version : integer
17502 Returns the version of the certificate presented by the client when the
17503 incoming connection was made over an SSL/TLS transport layer.
Emeric Brunce5ad802012-10-22 14:11:22 +020017504
Emeric Brun43e79582014-10-29 19:03:26 +010017505ssl_f_der : binary
17506 Returns the DER formatted certificate presented by the frontend when the
17507 incoming connection was made over an SSL/TLS transport layer. When used for
17508 an ACL, the value(s) to match against can be passed in hexadecimal form.
17509
Elliot Otchet71f82972020-01-15 08:12:14 -050017510ssl_f_i_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020017511 When the incoming connection was made over an SSL/TLS transport layer,
17512 returns the full distinguished name of the issuer of the certificate
17513 presented by the frontend when no <entry> is specified, or the value of the
17514 first given entry found from the beginning of the DN. If a positive/negative
Emeric Brun87855892012-10-17 17:39:35 +020017515 occurrence number is specified as the optional second argument, it returns
Willy Tarreau74ca5042013-06-11 23:12:07 +020017516 the value of the nth given entry value from the beginning/end of the DN.
17517 For instance, "ssl_f_i_dn(OU,2)" the second organization unit, and
17518 "ssl_f_i_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050017519 The <format> parameter allows you to receive the DN suitable for
17520 consumption by different protocols. Currently supported is rfc2253 for
17521 LDAP v3.
17522 If you'd like to modify the format only you can specify an empty string
17523 and zero for the first two parameters. Example: ssl_f_i_dn(,0,rfc2253)
Emeric Brun87855892012-10-17 17:39:35 +020017524
Willy Tarreau74ca5042013-06-11 23:12:07 +020017525ssl_f_key_alg : string
17526 Returns the name of the algorithm used to generate the key of the certificate
17527 presented by the frontend when the incoming connection was made over an
17528 SSL/TLS transport layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020017529
Willy Tarreau74ca5042013-06-11 23:12:07 +020017530ssl_f_notafter : string
17531 Returns the end date presented by the frontend as a formatted string
17532 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
17533 transport layer.
Emeric Brun2525b6b2012-10-18 15:59:43 +020017534
Willy Tarreau74ca5042013-06-11 23:12:07 +020017535ssl_f_notbefore : string
17536 Returns the start date presented by the frontend as a formatted string
17537 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
17538 transport layer.
Emeric Brun87855892012-10-17 17:39:35 +020017539
Elliot Otchet71f82972020-01-15 08:12:14 -050017540ssl_f_s_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020017541 When the incoming connection was made over an SSL/TLS transport layer,
17542 returns the full distinguished name of the subject of the certificate
17543 presented by the frontend when no <entry> is specified, or the value of the
17544 first given entry found from the beginning of the DN. If a positive/negative
17545 occurrence number is specified as the optional second argument, it returns
17546 the value of the nth given entry value from the beginning/end of the DN.
17547 For instance, "ssl_f_s_dn(OU,2)" the second organization unit, and
17548 "ssl_f_s_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050017549 The <format> parameter allows you to receive the DN suitable for
17550 consumption by different protocols. Currently supported is rfc2253 for
17551 LDAP v3.
17552 If you'd like to modify the format only you can specify an empty string
17553 and zero for the first two parameters. Example: ssl_f_s_dn(,0,rfc2253)
Emeric Brunce5ad802012-10-22 14:11:22 +020017554
Willy Tarreau74ca5042013-06-11 23:12:07 +020017555ssl_f_serial : binary
17556 Returns the serial of the certificate presented by the frontend when the
17557 incoming connection was made over an SSL/TLS transport layer. When used for
17558 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun87855892012-10-17 17:39:35 +020017559
Emeric Brun55f4fa82014-04-30 17:11:25 +020017560ssl_f_sha1 : binary
17561 Returns the SHA-1 fingerprint of the certificate presented by the frontend
17562 when the incoming connection was made over an SSL/TLS transport layer. This
17563 can be used to know which certificate was chosen using SNI.
17564
Willy Tarreau74ca5042013-06-11 23:12:07 +020017565ssl_f_sig_alg : string
17566 Returns the name of the algorithm used to sign the certificate presented by
17567 the frontend when the incoming connection was made over an SSL/TLS transport
17568 layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020017569
Willy Tarreau74ca5042013-06-11 23:12:07 +020017570ssl_f_version : integer
17571 Returns the version of the certificate presented by the frontend when the
17572 incoming connection was made over an SSL/TLS transport layer.
17573
17574ssl_fc : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020017575 Returns true when the front connection was made via an SSL/TLS transport
17576 layer and is locally deciphered. This means it has matched a socket declared
17577 with a "bind" line having the "ssl" option.
17578
Willy Tarreau74ca5042013-06-11 23:12:07 +020017579 Example :
17580 # This passes "X-Proto: https" to servers when client connects over SSL
17581 listen http-https
17582 bind :80
17583 bind :443 ssl crt /etc/haproxy.pem
17584 http-request add-header X-Proto https if { ssl_fc }
17585
17586ssl_fc_alg_keysize : integer
17587 Returns the symmetric cipher key size supported in bits when the incoming
17588 connection was made over an SSL/TLS transport layer.
17589
17590ssl_fc_alpn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017591 This extracts the Application Layer Protocol Negotiation field from an
Willy Tarreau74ca5042013-06-11 23:12:07 +020017592 incoming connection made via a TLS transport layer and locally deciphered by
17593 haproxy. The result is a string containing the protocol name advertised by
17594 the client. The SSL library must have been built with support for TLS
17595 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
17596 not advertised unless the "alpn" keyword on the "bind" line specifies a
17597 protocol list. Also, nothing forces the client to pick a protocol from this
17598 list, any other one may be requested. The TLS ALPN extension is meant to
17599 replace the TLS NPN extension. See also "ssl_fc_npn".
17600
Willy Tarreau74ca5042013-06-11 23:12:07 +020017601ssl_fc_cipher : string
17602 Returns the name of the used cipher when the incoming connection was made
17603 over an SSL/TLS transport layer.
Willy Tarreauab861d32013-04-02 02:30:41 +020017604
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010017605ssl_fc_cipherlist_bin : binary
17606 Returns the binary form of the client hello cipher list. The maximum returned
17607 value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010017608 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010017609
17610ssl_fc_cipherlist_hex : string
17611 Returns the binary form of the client hello cipher list encoded as
17612 hexadecimal. The maximum returned value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010017613 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010017614
17615ssl_fc_cipherlist_str : string
17616 Returns the decoded text form of the client hello cipher list. The maximum
17617 number of ciphers returned is according with the value of
17618 "tune.ssl.capture-cipherlist-size". Note that this sample-fetch is only
Davor Ocelice9ed2812017-12-25 17:49:28 +010017619 available with OpenSSL >= 1.0.2. If the function is not enabled, this
Emmanuel Hocdetddcde192017-09-01 17:32:08 +020017620 sample-fetch returns the hash like "ssl_fc_cipherlist_xxh".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010017621
17622ssl_fc_cipherlist_xxh : integer
17623 Returns a xxh64 of the cipher list. This hash can be return only is the value
17624 "tune.ssl.capture-cipherlist-size" is set greater than 0, however the hash
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010017625 take in account all the data of the cipher list.
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010017626
Patrick Hemmer65674662019-06-04 08:13:03 -040017627ssl_fc_client_random : binary
17628 Returns the client random of the front connection when the incoming connection
17629 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
17630 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
17631
William Lallemand7d42ef52020-07-06 11:41:30 +020017632ssl_fc_client_early_traffic_secret : string
17633 Return the CLIENT_EARLY_TRAFFIC_SECRET as an hexadecimal string for the
17634 front connection when the incoming connection was made over a TLS 1.3
17635 transport layer.
17636 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
17637 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
17638 activated with "tune.ssl.keylog on" in the global section. See also
17639 "tune.ssl.keylog"
17640
17641ssl_fc_client_handshake_traffic_secret : string
17642 Return the CLIENT_HANDSHAKE_TRAFFIC_SECRET as an hexadecimal string for the
17643 front connection when the incoming connection was made over a TLS 1.3
17644 transport layer.
17645 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
17646 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
17647 activated with "tune.ssl.keylog on" in the global section. See also
17648 "tune.ssl.keylog"
17649
17650ssl_fc_client_traffic_secret_0 : string
17651 Return the CLIENT_TRAFFIC_SECRET_0 as an hexadecimal string for the
17652 front connection when the incoming connection was made over a TLS 1.3
17653 transport layer.
17654 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
17655 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
17656 activated with "tune.ssl.keylog on" in the global section. See also
17657 "tune.ssl.keylog"
17658
17659ssl_fc_exporter_secret : string
17660 Return the EXPORTER_SECRET as an hexadecimal string for the
17661 front connection when the incoming connection was made over a TLS 1.3
17662 transport layer.
17663 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
17664 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
17665 activated with "tune.ssl.keylog on" in the global section. See also
17666 "tune.ssl.keylog"
17667
17668ssl_fc_early_exporter_secret : string
17669 Return the EARLY_EXPORTER_SECRET as an hexadecimal string for the
17670 front connection when the incoming connection was made over an TLS 1.3
17671 transport layer.
17672 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
17673 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
17674 activated with "tune.ssl.keylog on" in the global section. See also
17675 "tune.ssl.keylog"
17676
Willy Tarreau74ca5042013-06-11 23:12:07 +020017677ssl_fc_has_crt : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020017678 Returns true if a client certificate is present in an incoming connection over
17679 SSL/TLS transport layer. Useful if 'verify' statement is set to 'optional'.
Emeric Brun9143d372012-12-20 15:44:16 +010017680 Note: on SSL session resumption with Session ID or TLS ticket, client
17681 certificate is not present in the current connection but may be retrieved
17682 from the cache or the ticket. So prefer "ssl_c_used" if you want to check if
17683 current SSL session uses a client certificate.
Emeric Brun2525b6b2012-10-18 15:59:43 +020017684
Olivier Houchardccaa7de2017-10-02 11:51:03 +020017685ssl_fc_has_early : boolean
17686 Returns true if early data were sent, and the handshake didn't happen yet. As
17687 it has security implications, it is useful to be able to refuse those, or
17688 wait until the handshake happened.
17689
Willy Tarreau74ca5042013-06-11 23:12:07 +020017690ssl_fc_has_sni : boolean
17691 This checks for the presence of a Server Name Indication TLS extension (SNI)
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020017692 in an incoming connection was made over an SSL/TLS transport layer. Returns
17693 true when the incoming connection presents a TLS SNI field. This requires
John Roeslerfb2fce12019-07-10 15:45:51 -050017694 that the SSL library is built with support for TLS extensions enabled (check
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020017695 haproxy -vv).
Willy Tarreau7875d092012-09-10 08:20:03 +020017696
Nenad Merdanovic1516fe32016-05-17 03:31:21 +020017697ssl_fc_is_resumed : boolean
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020017698 Returns true if the SSL/TLS session has been resumed through the use of
Jérôme Magnin4a326cb2018-01-15 14:01:17 +010017699 SSL session cache or TLS tickets on an incoming connection over an SSL/TLS
17700 transport layer.
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020017701
Willy Tarreau74ca5042013-06-11 23:12:07 +020017702ssl_fc_npn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017703 This extracts the Next Protocol Negotiation field from an incoming connection
Willy Tarreau74ca5042013-06-11 23:12:07 +020017704 made via a TLS transport layer and locally deciphered by haproxy. The result
17705 is a string containing the protocol name advertised by the client. The SSL
17706 library must have been built with support for TLS extensions enabled (check
17707 haproxy -vv). Note that the TLS NPN extension is not advertised unless the
17708 "npn" keyword on the "bind" line specifies a protocol list. Also, nothing
17709 forces the client to pick a protocol from this list, any other one may be
17710 requested. Please note that the TLS NPN extension was replaced with ALPN.
Willy Tarreaua33c6542012-10-15 13:19:06 +020017711
Willy Tarreau74ca5042013-06-11 23:12:07 +020017712ssl_fc_protocol : string
17713 Returns the name of the used protocol when the incoming connection was made
17714 over an SSL/TLS transport layer.
Willy Tarreau7875d092012-09-10 08:20:03 +020017715
Emeric Brunb73a9b02014-04-30 18:49:19 +020017716ssl_fc_unique_id : binary
David Sc1ad52e2014-04-08 18:48:47 -040017717 When the incoming connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020017718 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
17719 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
David Sc1ad52e2014-04-08 18:48:47 -040017720
William Lallemand7d42ef52020-07-06 11:41:30 +020017721ssl_fc_server_handshake_traffic_secret : string
17722 Return the SERVER_HANDSHAKE_TRAFFIC_SECRET as an hexadecimal string for the
17723 front connection when the incoming connection was made over a TLS 1.3
17724 transport layer.
17725 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
17726 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
17727 activated with "tune.ssl.keylog on" in the global section. See also
17728 "tune.ssl.keylog"
17729
17730ssl_fc_server_traffic_secret_0 : string
17731 Return the SERVER_TRAFFIC_SECRET_0 as an hexadecimal string for the
17732 front connection when the incoming connection was made over an TLS 1.3
17733 transport layer.
17734 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
17735 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
17736 activated with "tune.ssl.keylog on" in the global section. See also
17737 "tune.ssl.keylog"
17738
Patrick Hemmer65674662019-06-04 08:13:03 -040017739ssl_fc_server_random : binary
17740 Returns the server random of the front connection when the incoming connection
17741 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
17742 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
17743
Willy Tarreau74ca5042013-06-11 23:12:07 +020017744ssl_fc_session_id : binary
17745 Returns the SSL ID of the front connection when the incoming connection was
17746 made over an SSL/TLS transport layer. It is useful to stick a given client to
17747 a server. It is important to note that some browsers refresh their session ID
17748 every few minutes.
Willy Tarreau7875d092012-09-10 08:20:03 +020017749
Patrick Hemmere0275472018-04-28 19:15:51 -040017750ssl_fc_session_key : binary
17751 Returns the SSL session master key of the front connection when the incoming
17752 connection was made over an SSL/TLS transport layer. It is useful to decrypt
17753 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
17754 BoringSSL.
17755
17756
Willy Tarreau74ca5042013-06-11 23:12:07 +020017757ssl_fc_sni : string
17758 This extracts the Server Name Indication TLS extension (SNI) field from an
17759 incoming connection made via an SSL/TLS transport layer and locally
17760 deciphered by haproxy. The result (when present) typically is a string
17761 matching the HTTPS host name (253 chars or less). The SSL library must have
17762 been built with support for TLS extensions enabled (check haproxy -vv).
17763
Alex192600b2021-06-05 13:23:08 +020017764 This fetch is different from "req.ssl_sni" above in that it applies to the
Willy Tarreau74ca5042013-06-11 23:12:07 +020017765 connection being deciphered by haproxy and not to SSL contents being blindly
17766 forwarded. See also "ssl_fc_sni_end" and "ssl_fc_sni_reg" below. This
John Roeslerfb2fce12019-07-10 15:45:51 -050017767 requires that the SSL library is built with support for TLS extensions
Cyril Bonté9c1eb1e2012-10-09 22:45:34 +020017768 enabled (check haproxy -vv).
Willy Tarreau62644772008-07-16 18:36:06 +020017769
Willy Tarreau74ca5042013-06-11 23:12:07 +020017770 ACL derivatives :
Willy Tarreau74ca5042013-06-11 23:12:07 +020017771 ssl_fc_sni_end : suffix match
17772 ssl_fc_sni_reg : regex match
Emeric Brun589fcad2012-10-16 14:13:26 +020017773
Willy Tarreau74ca5042013-06-11 23:12:07 +020017774ssl_fc_use_keysize : integer
17775 Returns the symmetric cipher key size used in bits when the incoming
17776 connection was made over an SSL/TLS transport layer.
Willy Tarreaub6fb4202008-07-20 11:18:28 +020017777
William Lallemandbfa3e812020-06-25 20:07:18 +020017778ssl_s_der : binary
17779 Returns the DER formatted certificate presented by the server when the
17780 outgoing connection was made over an SSL/TLS transport layer. When used for
17781 an ACL, the value(s) to match against can be passed in hexadecimal form.
17782
William Dauchya598b502020-08-06 18:11:38 +020017783ssl_s_chain_der : binary
17784 Returns the DER formatted chain certificate presented by the server when the
17785 outgoing connection was made over an SSL/TLS transport layer. When used for
17786 an ACL, the value(s) to match against can be passed in hexadecimal form. One
17787 can parse the result with any lib accepting ASN.1 DER data. It currentlly
17788 does not support resumed sessions.
17789
William Lallemandbfa3e812020-06-25 20:07:18 +020017790ssl_s_key_alg : string
17791 Returns the name of the algorithm used to generate the key of the certificate
17792 presented by the server when the outgoing connection was made over an
17793 SSL/TLS transport layer.
17794
17795ssl_s_notafter : string
17796 Returns the end date presented by the server as a formatted string
17797 YYMMDDhhmmss[Z] when the outgoing connection was made over an SSL/TLS
17798 transport layer.
17799
17800ssl_s_notbefore : string
17801 Returns the start date presented by the server as a formatted string
17802 YYMMDDhhmmss[Z] when the outgoing connection was made over an SSL/TLS
17803 transport layer.
17804
17805ssl_s_i_dn([<entry>[,<occ>[,<format>]]]) : string
17806 When the outgoing connection was made over an SSL/TLS transport layer,
17807 returns the full distinguished name of the issuer of the certificate
17808 presented by the server when no <entry> is specified, or the value of the
17809 first given entry found from the beginning of the DN. If a positive/negative
17810 occurrence number is specified as the optional second argument, it returns
17811 the value of the nth given entry value from the beginning/end of the DN.
William Lallemand8f600c82020-06-26 09:55:06 +020017812 For instance, "ssl_s_i_dn(OU,2)" the second organization unit, and
17813 "ssl_s_i_dn(CN)" retrieves the common name.
William Lallemandbfa3e812020-06-25 20:07:18 +020017814 The <format> parameter allows you to receive the DN suitable for
17815 consumption by different protocols. Currently supported is rfc2253 for
17816 LDAP v3.
17817 If you'd like to modify the format only you can specify an empty string
17818 and zero for the first two parameters. Example: ssl_s_i_dn(,0,rfc2253)
17819
17820ssl_s_s_dn([<entry>[,<occ>[,<format>]]]) : string
17821 When the outgoing connection was made over an SSL/TLS transport layer,
17822 returns the full distinguished name of the subject of the certificate
17823 presented by the server when no <entry> is specified, or the value of the
17824 first given entry found from the beginning of the DN. If a positive/negative
17825 occurrence number is specified as the optional second argument, it returns
17826 the value of the nth given entry value from the beginning/end of the DN.
William Lallemand8f600c82020-06-26 09:55:06 +020017827 For instance, "ssl_s_s_dn(OU,2)" the second organization unit, and
17828 "ssl_s_s_dn(CN)" retrieves the common name.
William Lallemandbfa3e812020-06-25 20:07:18 +020017829 The <format> parameter allows you to receive the DN suitable for
17830 consumption by different protocols. Currently supported is rfc2253 for
17831 LDAP v3.
17832 If you'd like to modify the format only you can specify an empty string
17833 and zero for the first two parameters. Example: ssl_s_s_dn(,0,rfc2253)
17834
17835ssl_s_serial : binary
17836 Returns the serial of the certificate presented by the server when the
17837 outgoing connection was made over an SSL/TLS transport layer. When used for
17838 an ACL, the value(s) to match against can be passed in hexadecimal form.
17839
17840ssl_s_sha1 : binary
17841 Returns the SHA-1 fingerprint of the certificate presented by the server
17842 when the outgoing connection was made over an SSL/TLS transport layer. This
17843 can be used to know which certificate was chosen using SNI.
17844
17845ssl_s_sig_alg : string
17846 Returns the name of the algorithm used to sign the certificate presented by
17847 the server when the outgoing connection was made over an SSL/TLS transport
17848 layer.
17849
17850ssl_s_version : integer
17851 Returns the version of the certificate presented by the server when the
17852 outgoing connection was made over an SSL/TLS transport layer.
Willy Tarreaub6fb4202008-07-20 11:18:28 +020017853
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200178547.3.5. Fetching samples from buffer contents (Layer 6)
Willy Tarreau74ca5042013-06-11 23:12:07 +020017855------------------------------------------------------
Willy Tarreaub6fb4202008-07-20 11:18:28 +020017856
Willy Tarreau74ca5042013-06-11 23:12:07 +020017857Fetching samples from buffer contents is a bit different from the previous
17858sample fetches above because the sampled data are ephemeral. These data can
17859only be used when they're available and will be lost when they're forwarded.
17860For this reason, samples fetched from buffer contents during a request cannot
17861be used in a response for example. Even while the data are being fetched, they
17862can change. Sometimes it is necessary to set some delays or combine multiple
17863sample fetch methods to ensure that the expected data are complete and usable,
17864for example through TCP request content inspection. Please see the "tcp-request
17865content" keyword for more detailed information on the subject.
Willy Tarreau62644772008-07-16 18:36:06 +020017866
Willy Tarreau74ca5042013-06-11 23:12:07 +020017867payload(<offset>,<length>) : binary (deprecated)
Davor Ocelice9ed2812017-12-25 17:49:28 +010017868 This is an alias for "req.payload" when used in the context of a request (e.g.
Willy Tarreau74ca5042013-06-11 23:12:07 +020017869 "stick on", "stick match"), and for "res.payload" when used in the context of
17870 a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010017871
Willy Tarreau74ca5042013-06-11 23:12:07 +020017872payload_lv(<offset1>,<length>[,<offset2>]) : binary (deprecated)
17873 This is an alias for "req.payload_lv" when used in the context of a request
Davor Ocelice9ed2812017-12-25 17:49:28 +010017874 (e.g. "stick on", "stick match"), and for "res.payload_lv" when used in the
Willy Tarreau74ca5042013-06-11 23:12:07 +020017875 context of a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010017876
Willy Tarreau74ca5042013-06-11 23:12:07 +020017877req.len : integer
17878req_len : integer (deprecated)
17879 Returns an integer value corresponding to the number of bytes present in the
17880 request buffer. This is mostly used in ACL. It is important to understand
17881 that this test does not return false as long as the buffer is changing. This
17882 means that a check with equality to zero will almost always immediately match
17883 at the beginning of the session, while a test for more data will wait for
17884 that data to come in and return false only when haproxy is certain that no
17885 more data will come in. This test was designed to be used with TCP request
17886 content inspection.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020017887
Willy Tarreau74ca5042013-06-11 23:12:07 +020017888req.payload(<offset>,<length>) : binary
17889 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020017890 in the request buffer. As a special case, if the <length> argument is zero,
17891 the the whole buffer from <offset> to the end is extracted. This can be used
17892 with ACLs in order to check for the presence of some content in a buffer at
17893 any location.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020017894
Willy Tarreau74ca5042013-06-11 23:12:07 +020017895 ACL alternatives :
17896 payload(<offset>,<length>) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020017897
Willy Tarreau74ca5042013-06-11 23:12:07 +020017898req.payload_lv(<offset1>,<length>[,<offset2>]) : binary
17899 This extracts a binary block whose size is specified at <offset1> for <length>
17900 bytes, and which starts at <offset2> if specified or just after the length in
17901 the request buffer. The <offset2> parameter also supports relative offsets if
17902 prepended with a '+' or '-' sign.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020017903
Willy Tarreau74ca5042013-06-11 23:12:07 +020017904 ACL alternatives :
17905 payload_lv(<offset1>,<length>[,<offset2>]) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020017906
Willy Tarreau74ca5042013-06-11 23:12:07 +020017907 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020017908
Willy Tarreau74ca5042013-06-11 23:12:07 +020017909req.proto_http : boolean
17910req_proto_http : boolean (deprecated)
17911 Returns true when data in the request buffer look like HTTP and correctly
17912 parses as such. It is the same parser as the common HTTP request parser which
17913 is used so there should be no surprises. The test does not match until the
17914 request is complete, failed or timed out. This test may be used to report the
17915 protocol in TCP logs, but the biggest use is to block TCP request analysis
17916 until a complete HTTP request is present in the buffer, for example to track
17917 a header.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020017918
Willy Tarreau74ca5042013-06-11 23:12:07 +020017919 Example:
17920 # track request counts per "base" (concatenation of Host+URL)
17921 tcp-request inspect-delay 10s
17922 tcp-request content reject if !HTTP
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020017923 tcp-request content track-sc0 base table req-rate
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020017924
Willy Tarreau74ca5042013-06-11 23:12:07 +020017925req.rdp_cookie([<name>]) : string
17926rdp_cookie([<name>]) : string (deprecated)
17927 When the request buffer looks like the RDP protocol, extracts the RDP cookie
17928 <name>, or any cookie if unspecified. The parser only checks for the first
17929 cookie, as illustrated in the RDP protocol specification. The cookie name is
17930 case insensitive. Generally the "MSTS" cookie name will be used, as it can
17931 contain the user name of the client connecting to the server if properly
17932 configured on the client. The "MSTSHASH" cookie is often used as well for
17933 session stickiness to servers.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020017934
Willy Tarreau74ca5042013-06-11 23:12:07 +020017935 This differs from "balance rdp-cookie" in that any balancing algorithm may be
17936 used and thus the distribution of clients to backend servers is not linked to
17937 a hash of the RDP cookie. It is envisaged that using a balancing algorithm
17938 such as "balance roundrobin" or "balance leastconn" will lead to a more even
17939 distribution of clients to backend servers than the hash used by "balance
17940 rdp-cookie".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020017941
Willy Tarreau74ca5042013-06-11 23:12:07 +020017942 ACL derivatives :
17943 req_rdp_cookie([<name>]) : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020017944
Willy Tarreau74ca5042013-06-11 23:12:07 +020017945 Example :
17946 listen tse-farm
17947 bind 0.0.0.0:3389
17948 # wait up to 5s for an RDP cookie in the request
17949 tcp-request inspect-delay 5s
17950 tcp-request content accept if RDP_COOKIE
17951 # apply RDP cookie persistence
17952 persist rdp-cookie
17953 # Persist based on the mstshash cookie
17954 # This is only useful makes sense if
17955 # balance rdp-cookie is not used
17956 stick-table type string size 204800
17957 stick on req.rdp_cookie(mstshash)
17958 server srv1 1.1.1.1:3389
17959 server srv1 1.1.1.2:3389
Willy Tarreau04aa6a92012-04-06 18:57:55 +020017960
Willy Tarreau74ca5042013-06-11 23:12:07 +020017961 See also : "balance rdp-cookie", "persist rdp-cookie", "tcp-request" and the
17962 "req_rdp_cookie" ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020017963
Willy Tarreau74ca5042013-06-11 23:12:07 +020017964req.rdp_cookie_cnt([name]) : integer
17965rdp_cookie_cnt([name]) : integer (deprecated)
17966 Tries to parse the request buffer as RDP protocol, then returns an integer
17967 corresponding to the number of RDP cookies found. If an optional cookie name
17968 is passed, only cookies matching this name are considered. This is mostly
17969 used in ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020017970
Willy Tarreau74ca5042013-06-11 23:12:07 +020017971 ACL derivatives :
17972 req_rdp_cookie_cnt([<name>]) : integer match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020017973
Alex Zorin4afdd132018-12-30 13:56:28 +110017974req.ssl_alpn : string
17975 Returns a string containing the values of the Application-Layer Protocol
17976 Negotiation (ALPN) TLS extension (RFC7301), sent by the client within the SSL
17977 ClientHello message. Note that this only applies to raw contents found in the
17978 request buffer and not to the contents deciphered via an SSL data layer, so
17979 this will not work with "bind" lines having the "ssl" option. This is useful
17980 in ACL to make a routing decision based upon the ALPN preferences of a TLS
Jarno Huuskonene504f812019-01-03 07:56:49 +020017981 client, like in the example below. See also "ssl_fc_alpn".
Alex Zorin4afdd132018-12-30 13:56:28 +110017982
17983 Examples :
17984 # Wait for a client hello for at most 5 seconds
17985 tcp-request inspect-delay 5s
17986 tcp-request content accept if { req_ssl_hello_type 1 }
Jarno Huuskonene504f812019-01-03 07:56:49 +020017987 use_backend bk_acme if { req.ssl_alpn acme-tls/1 }
Alex Zorin4afdd132018-12-30 13:56:28 +110017988 default_backend bk_default
17989
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020017990req.ssl_ec_ext : boolean
17991 Returns a boolean identifying if client sent the Supported Elliptic Curves
17992 Extension as defined in RFC4492, section 5.1. within the SSL ClientHello
Cyril Bonté307ee1e2015-09-28 23:16:06 +020017993 message. This can be used to present ECC compatible clients with EC
17994 certificate and to use RSA for all others, on the same IP address. Note that
17995 this only applies to raw contents found in the request buffer and not to
17996 contents deciphered via an SSL data layer, so this will not work with "bind"
17997 lines having the "ssl" option.
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020017998
Willy Tarreau74ca5042013-06-11 23:12:07 +020017999req.ssl_hello_type : integer
18000req_ssl_hello_type : integer (deprecated)
18001 Returns an integer value containing the type of the SSL hello message found
18002 in the request buffer if the buffer contains data that parse as a complete
18003 SSL (v3 or superior) client hello message. Note that this only applies to raw
18004 contents found in the request buffer and not to contents deciphered via an
18005 SSL data layer, so this will not work with "bind" lines having the "ssl"
18006 option. This is mostly used in ACL to detect presence of an SSL hello message
18007 that is supposed to contain an SSL session ID usable for stickiness.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018008
Willy Tarreau74ca5042013-06-11 23:12:07 +020018009req.ssl_sni : string
18010req_ssl_sni : string (deprecated)
18011 Returns a string containing the value of the Server Name TLS extension sent
18012 by a client in a TLS stream passing through the request buffer if the buffer
18013 contains data that parse as a complete SSL (v3 or superior) client hello
18014 message. Note that this only applies to raw contents found in the request
18015 buffer and not to contents deciphered via an SSL data layer, so this will not
Lukas Tribusa267b5d2020-07-19 00:25:06 +020018016 work with "bind" lines having the "ssl" option. This will only work for actual
18017 implicit TLS based protocols like HTTPS (443), IMAPS (993), SMTPS (465),
18018 however it will not work for explicit TLS based protocols, like SMTP (25/587)
18019 or IMAP (143). SNI normally contains the name of the host the client tries to
18020 connect to (for recent browsers). SNI is useful for allowing or denying access
18021 to certain hosts when SSL/TLS is used by the client. This test was designed to
18022 be used with TCP request content inspection. If content switching is needed,
18023 it is recommended to first wait for a complete client hello (type 1), like in
18024 the example below. See also "ssl_fc_sni".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018025
Willy Tarreau74ca5042013-06-11 23:12:07 +020018026 ACL derivatives :
Alex192600b2021-06-05 13:23:08 +020018027 req.ssl_sni : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018028
Willy Tarreau74ca5042013-06-11 23:12:07 +020018029 Examples :
18030 # Wait for a client hello for at most 5 seconds
18031 tcp-request inspect-delay 5s
18032 tcp-request content accept if { req_ssl_hello_type 1 }
Alex192600b2021-06-05 13:23:08 +020018033 use_backend bk_allow if { req.ssl_sni -f allowed_sites }
Willy Tarreau74ca5042013-06-11 23:12:07 +020018034 default_backend bk_sorry_page
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018035
Pradeep Jindalbb2acf52015-09-29 10:12:57 +053018036req.ssl_st_ext : integer
18037 Returns 0 if the client didn't send a SessionTicket TLS Extension (RFC5077)
18038 Returns 1 if the client sent SessionTicket TLS Extension
18039 Returns 2 if the client also sent non-zero length TLS SessionTicket
18040 Note that this only applies to raw contents found in the request buffer and
18041 not to contents deciphered via an SSL data layer, so this will not work with
18042 "bind" lines having the "ssl" option. This can for example be used to detect
18043 whether the client sent a SessionTicket or not and stick it accordingly, if
18044 no SessionTicket then stick on SessionID or don't stick as there's no server
18045 side state is there when SessionTickets are in use.
18046
Willy Tarreau74ca5042013-06-11 23:12:07 +020018047req.ssl_ver : integer
18048req_ssl_ver : integer (deprecated)
18049 Returns an integer value containing the version of the SSL/TLS protocol of a
18050 stream present in the request buffer. Both SSLv2 hello messages and SSLv3
18051 messages are supported. TLSv1 is announced as SSL version 3.1. The value is
18052 composed of the major version multiplied by 65536, added to the minor
18053 version. Note that this only applies to raw contents found in the request
18054 buffer and not to contents deciphered via an SSL data layer, so this will not
18055 work with "bind" lines having the "ssl" option. The ACL version of the test
Davor Ocelice9ed2812017-12-25 17:49:28 +010018056 matches against a decimal notation in the form MAJOR.MINOR (e.g. 3.1). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020018057 fetch is mostly used in ACL.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018058
Willy Tarreau74ca5042013-06-11 23:12:07 +020018059 ACL derivatives :
18060 req_ssl_ver : decimal match
Willy Tarreaud63335a2010-02-26 12:56:52 +010018061
Willy Tarreau47e8eba2013-09-11 23:28:46 +020018062res.len : integer
18063 Returns an integer value corresponding to the number of bytes present in the
18064 response buffer. This is mostly used in ACL. It is important to understand
18065 that this test does not return false as long as the buffer is changing. This
18066 means that a check with equality to zero will almost always immediately match
18067 at the beginning of the session, while a test for more data will wait for
18068 that data to come in and return false only when haproxy is certain that no
18069 more data will come in. This test was designed to be used with TCP response
Christopher Faulete596d182020-05-05 17:46:34 +020018070 content inspection. But it may also be used in tcp-check based expect rules.
Willy Tarreau47e8eba2013-09-11 23:28:46 +020018071
Willy Tarreau74ca5042013-06-11 23:12:07 +020018072res.payload(<offset>,<length>) : binary
18073 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020018074 in the response buffer. As a special case, if the <length> argument is zero,
Christopher Faulete596d182020-05-05 17:46:34 +020018075 the whole buffer from <offset> to the end is extracted. This can be used
Willy Tarreau00f00842013-08-02 11:07:32 +020018076 with ACLs in order to check for the presence of some content in a buffer at
Christopher Faulete596d182020-05-05 17:46:34 +020018077 any location. It may also be used in tcp-check based expect rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018078
Willy Tarreau74ca5042013-06-11 23:12:07 +020018079res.payload_lv(<offset1>,<length>[,<offset2>]) : binary
18080 This extracts a binary block whose size is specified at <offset1> for <length>
18081 bytes, and which starts at <offset2> if specified or just after the length in
18082 the response buffer. The <offset2> parameter also supports relative offsets
Christopher Faulete596d182020-05-05 17:46:34 +020018083 if prepended with a '+' or '-' sign. It may also be used in tcp-check based
18084 expect rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018085
Willy Tarreau74ca5042013-06-11 23:12:07 +020018086 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018087
Willy Tarreau971f7b62015-09-29 14:06:59 +020018088res.ssl_hello_type : integer
18089rep_ssl_hello_type : integer (deprecated)
18090 Returns an integer value containing the type of the SSL hello message found
18091 in the response buffer if the buffer contains data that parses as a complete
18092 SSL (v3 or superior) hello message. Note that this only applies to raw
18093 contents found in the response buffer and not to contents deciphered via an
18094 SSL data layer, so this will not work with "server" lines having the "ssl"
18095 option. This is mostly used in ACL to detect presence of an SSL hello message
18096 that is supposed to contain an SSL session ID usable for stickiness.
18097
Willy Tarreau74ca5042013-06-11 23:12:07 +020018098wait_end : boolean
18099 This fetch either returns true when the inspection period is over, or does
18100 not fetch. It is only used in ACLs, in conjunction with content analysis to
Davor Ocelice9ed2812017-12-25 17:49:28 +010018101 avoid returning a wrong verdict early. It may also be used to delay some
Willy Tarreau74ca5042013-06-11 23:12:07 +020018102 actions, such as a delayed reject for some special addresses. Since it either
18103 stops the rules evaluation or immediately returns true, it is recommended to
Davor Ocelice9ed2812017-12-25 17:49:28 +010018104 use this acl as the last one in a rule. Please note that the default ACL
Willy Tarreau74ca5042013-06-11 23:12:07 +020018105 "WAIT_END" is always usable without prior declaration. This test was designed
18106 to be used with TCP request content inspection.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018107
Willy Tarreau74ca5042013-06-11 23:12:07 +020018108 Examples :
18109 # delay every incoming request by 2 seconds
18110 tcp-request inspect-delay 2s
18111 tcp-request content accept if WAIT_END
Willy Tarreaud63335a2010-02-26 12:56:52 +010018112
Willy Tarreau74ca5042013-06-11 23:12:07 +020018113 # don't immediately tell bad guys they are rejected
18114 tcp-request inspect-delay 10s
18115 acl goodguys src 10.0.0.0/24
18116 acl badguys src 10.0.1.0/24
18117 tcp-request content accept if goodguys
18118 tcp-request content reject if badguys WAIT_END
18119 tcp-request content reject
18120
18121
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200181227.3.6. Fetching HTTP samples (Layer 7)
Willy Tarreau74ca5042013-06-11 23:12:07 +020018123--------------------------------------
18124
18125It is possible to fetch samples from HTTP contents, requests and responses.
18126This application layer is also called layer 7. It is only possible to fetch the
18127data in this section when a full HTTP request or response has been parsed from
18128its respective request or response buffer. This is always the case with all
18129HTTP specific rules and for sections running with "mode http". When using TCP
18130content inspection, it may be necessary to support an inspection delay in order
18131to let the request or response come in first. These fetches may require a bit
18132more CPU resources than the layer 4 ones, but not much since the request and
18133response are indexed.
18134
18135base : string
18136 This returns the concatenation of the first Host header and the path part of
18137 the request, which starts at the first slash and ends before the question
18138 mark. It can be useful in virtual hosted environments to detect URL abuses as
18139 well as to improve shared caches efficiency. Using this with a limited size
18140 stick table also allows one to collect statistics about most commonly
18141 requested objects by host/path. With ACLs it can allow simple content
18142 switching rules involving the host and the path at the same time, such as
18143 "www.example.com/favicon.ico". See also "path" and "uri".
18144
18145 ACL derivatives :
18146 base : exact string match
18147 base_beg : prefix match
18148 base_dir : subdir match
18149 base_dom : domain match
18150 base_end : suffix match
18151 base_len : length match
18152 base_reg : regex match
18153 base_sub : substring match
18154
18155base32 : integer
18156 This returns a 32-bit hash of the value returned by the "base" fetch method
18157 above. This is useful to track per-URL activity on high traffic sites without
18158 having to store all URLs. Instead a shorter hash is stored, saving a lot of
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020018159 memory. The output type is an unsigned integer. The hash function used is
18160 SDBM with full avalanche on the output. Technically, base32 is exactly equal
18161 to "base,sdbm(1)".
Willy Tarreau74ca5042013-06-11 23:12:07 +020018162
18163base32+src : binary
18164 This returns the concatenation of the base32 fetch above and the src fetch
18165 below. The resulting type is of type binary, with a size of 8 or 20 bytes
18166 depending on the source address family. This can be used to track per-IP,
18167 per-URL counters.
18168
William Lallemand65ad6e12014-01-31 15:08:02 +010018169capture.req.hdr(<idx>) : string
18170 This extracts the content of the header captured by the "capture request
18171 header", idx is the position of the capture keyword in the configuration.
18172 The first entry is an index of 0. See also: "capture request header".
18173
18174capture.req.method : string
18175 This extracts the METHOD of an HTTP request. It can be used in both request
18176 and response. Unlike "method", it can be used in both request and response
18177 because it's allocated.
18178
18179capture.req.uri : string
18180 This extracts the request's URI, which starts at the first slash and ends
18181 before the first space in the request (without the host part). Unlike "path"
18182 and "url", it can be used in both request and response because it's
18183 allocated.
18184
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020018185capture.req.ver : string
18186 This extracts the request's HTTP version and returns either "HTTP/1.0" or
18187 "HTTP/1.1". Unlike "req.ver", it can be used in both request, response, and
18188 logs because it relies on a persistent flag.
18189
William Lallemand65ad6e12014-01-31 15:08:02 +010018190capture.res.hdr(<idx>) : string
18191 This extracts the content of the header captured by the "capture response
18192 header", idx is the position of the capture keyword in the configuration.
18193 The first entry is an index of 0.
18194 See also: "capture response header"
18195
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020018196capture.res.ver : string
18197 This extracts the response's HTTP version and returns either "HTTP/1.0" or
18198 "HTTP/1.1". Unlike "res.ver", it can be used in logs because it relies on a
18199 persistent flag.
18200
Willy Tarreaua5910cc2015-05-02 00:46:08 +020018201req.body : binary
Christopher Fauletaf4dc4c2020-05-05 17:33:25 +020018202 This returns the HTTP request's available body as a block of data. It is
18203 recommended to use "option http-buffer-request" to be sure to wait, as much
18204 as possible, for the request's body.
Willy Tarreaua5910cc2015-05-02 00:46:08 +020018205
Thierry FOURNIER9826c772015-05-20 15:50:54 +020018206req.body_param([<name>) : string
18207 This fetch assumes that the body of the POST request is url-encoded. The user
18208 can check if the "content-type" contains the value
18209 "application/x-www-form-urlencoded". This extracts the first occurrence of the
18210 parameter <name> in the body, which ends before '&'. The parameter name is
18211 case-sensitive. If no name is given, any parameter will match, and the first
18212 one will be returned. The result is a string corresponding to the value of the
18213 parameter <name> as presented in the request body (no URL decoding is
18214 performed). Note that the ACL version of this fetch iterates over multiple
18215 parameters and will iteratively report all parameters values if no name is
18216 given.
18217
Willy Tarreaua5910cc2015-05-02 00:46:08 +020018218req.body_len : integer
18219 This returns the length of the HTTP request's available body in bytes. It may
18220 be lower than the advertised length if the body is larger than the buffer. It
Christopher Fauletaf4dc4c2020-05-05 17:33:25 +020018221 is recommended to use "option http-buffer-request" to be sure to wait, as
18222 much as possible, for the request's body.
Willy Tarreaua5910cc2015-05-02 00:46:08 +020018223
18224req.body_size : integer
18225 This returns the advertised length of the HTTP request's body in bytes. It
Christopher Fauletaf4dc4c2020-05-05 17:33:25 +020018226 will represent the advertised Content-Length header, or the size of the
18227 available data in case of chunked encoding.
Willy Tarreaua5910cc2015-05-02 00:46:08 +020018228
Willy Tarreau74ca5042013-06-11 23:12:07 +020018229req.cook([<name>]) : string
18230cook([<name>]) : string (deprecated)
18231 This extracts the last occurrence of the cookie name <name> on a "Cookie"
18232 header line from the request, and returns its value as string. If no name is
18233 specified, the first cookie value is returned. When used with ACLs, all
18234 matching cookies are evaluated. Spaces around the name and the value are
18235 ignored as requested by the Cookie header specification (RFC6265). The cookie
18236 name is case-sensitive. Empty cookies are valid, so an empty cookie may very
18237 well return an empty value if it is present. Use the "found" match to detect
18238 presence. Use the res.cook() variant for response cookies sent by the server.
18239
18240 ACL derivatives :
18241 cook([<name>]) : exact string match
18242 cook_beg([<name>]) : prefix match
18243 cook_dir([<name>]) : subdir match
18244 cook_dom([<name>]) : domain match
18245 cook_end([<name>]) : suffix match
18246 cook_len([<name>]) : length match
18247 cook_reg([<name>]) : regex match
18248 cook_sub([<name>]) : substring match
Willy Tarreaud63335a2010-02-26 12:56:52 +010018249
Willy Tarreau74ca5042013-06-11 23:12:07 +020018250req.cook_cnt([<name>]) : integer
18251cook_cnt([<name>]) : integer (deprecated)
18252 Returns an integer value representing the number of occurrences of the cookie
18253 <name> in the request, or all cookies if <name> is not specified.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018254
Willy Tarreau74ca5042013-06-11 23:12:07 +020018255req.cook_val([<name>]) : integer
18256cook_val([<name>]) : integer (deprecated)
18257 This extracts the last occurrence of the cookie name <name> on a "Cookie"
18258 header line from the request, and converts its value to an integer which is
18259 returned. If no name is specified, the first cookie value is returned. When
18260 used in ACLs, all matching names are iterated over until a value matches.
Willy Tarreau0e698542011-09-16 08:32:32 +020018261
Willy Tarreau74ca5042013-06-11 23:12:07 +020018262cookie([<name>]) : string (deprecated)
18263 This extracts the last occurrence of the cookie name <name> on a "Cookie"
18264 header line from the request, or a "Set-Cookie" header from the response, and
18265 returns its value as a string. A typical use is to get multiple clients
18266 sharing a same profile use the same server. This can be similar to what
Willy Tarreau294d0f02015-08-10 19:40:12 +020018267 "appsession" did with the "request-learn" statement, but with support for
Willy Tarreau74ca5042013-06-11 23:12:07 +020018268 multi-peer synchronization and state keeping across restarts. If no name is
18269 specified, the first cookie value is returned. This fetch should not be used
18270 anymore and should be replaced by req.cook() or res.cook() instead as it
18271 ambiguously uses the direction based on the context where it is used.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018272
Willy Tarreau74ca5042013-06-11 23:12:07 +020018273hdr([<name>[,<occ>]]) : string
18274 This is equivalent to req.hdr() when used on requests, and to res.hdr() when
18275 used on responses. Please refer to these respective fetches for more details.
18276 In case of doubt about the fetch direction, please use the explicit ones.
18277 Note that contrary to the hdr() sample fetch method, the hdr_* ACL keywords
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030018278 unambiguously apply to the request headers.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018279
Willy Tarreau74ca5042013-06-11 23:12:07 +020018280req.fhdr(<name>[,<occ>]) : string
Tim Duesterhus7fb56282021-01-23 17:50:21 +010018281 This returns the full value of the last occurrence of header <name> in an
18282 HTTP request. It differs from req.hdr() in that any commas present in the
18283 value are returned and are not used as delimiters. This is sometimes useful
18284 with headers such as User-Agent.
18285
18286 When used from an ACL, all occurrences are iterated over until a match is
18287 found.
18288
Willy Tarreau74ca5042013-06-11 23:12:07 +020018289 Optionally, a specific occurrence might be specified as a position number.
18290 Positive values indicate a position from the first occurrence, with 1 being
18291 the first one. Negative values indicate positions relative to the last one,
Tim Duesterhus7fb56282021-01-23 17:50:21 +010018292 with -1 being the last one.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018293
Willy Tarreau74ca5042013-06-11 23:12:07 +020018294req.fhdr_cnt([<name>]) : integer
18295 Returns an integer value representing the number of occurrences of request
18296 header field name <name>, or the total number of header fields if <name> is
Tim Duesterhus7fb56282021-01-23 17:50:21 +010018297 not specified. Like req.fhdr() it differs from res.hdr_cnt() by not splitting
18298 headers at commas.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018299
Willy Tarreau74ca5042013-06-11 23:12:07 +020018300req.hdr([<name>[,<occ>]]) : string
Tim Duesterhus7fb56282021-01-23 17:50:21 +010018301 This returns the last comma-separated value of the header <name> in an HTTP
18302 request. The fetch considers any comma as a delimiter for distinct values.
18303 This is useful if you need to process headers that are defined to be a list
18304 of values, such as Accept, or X-Forwarded-For. If full-line headers are
18305 desired instead, use req.fhdr(). Please carefully check RFC 7231 to know how
18306 certain headers are supposed to be parsed. Also, some of them are case
18307 insensitive (e.g. Connection).
18308
18309 When used from an ACL, all occurrences are iterated over until a match is
18310 found.
18311
Willy Tarreau74ca5042013-06-11 23:12:07 +020018312 Optionally, a specific occurrence might be specified as a position number.
18313 Positive values indicate a position from the first occurrence, with 1 being
18314 the first one. Negative values indicate positions relative to the last one,
Tim Duesterhus7fb56282021-01-23 17:50:21 +010018315 with -1 being the last one.
18316
18317 A typical use is with the X-Forwarded-For header once converted to IP,
18318 associated with an IP stick-table.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018319
Willy Tarreau74ca5042013-06-11 23:12:07 +020018320 ACL derivatives :
18321 hdr([<name>[,<occ>]]) : exact string match
18322 hdr_beg([<name>[,<occ>]]) : prefix match
18323 hdr_dir([<name>[,<occ>]]) : subdir match
18324 hdr_dom([<name>[,<occ>]]) : domain match
18325 hdr_end([<name>[,<occ>]]) : suffix match
18326 hdr_len([<name>[,<occ>]]) : length match
18327 hdr_reg([<name>[,<occ>]]) : regex match
18328 hdr_sub([<name>[,<occ>]]) : substring match
18329
18330req.hdr_cnt([<name>]) : integer
18331hdr_cnt([<header>]) : integer (deprecated)
18332 Returns an integer value representing the number of occurrences of request
18333 header field name <name>, or the total number of header field values if
Tim Duesterhus7fb56282021-01-23 17:50:21 +010018334 <name> is not specified. Like req.hdr() it counts each comma separated
18335 part of the header's value. If counting of full-line headers is desired,
18336 then req.fhdr_cnt() should be used instead.
18337
18338 With ACLs, it can be used to detect presence, absence or abuse of a specific
18339 header, as well as to block request smuggling attacks by rejecting requests
18340 which contain more than one of certain headers.
18341
18342 Refer to req.hdr() for more information on header matching.
Willy Tarreau74ca5042013-06-11 23:12:07 +020018343
18344req.hdr_ip([<name>[,<occ>]]) : ip
18345hdr_ip([<name>[,<occ>]]) : ip (deprecated)
18346 This extracts the last occurrence of header <name> in an HTTP request,
18347 converts it to an IPv4 or IPv6 address and returns this address. When used
18348 with ACLs, all occurrences are checked, and if <name> is omitted, every value
Willy Tarreau7e2e49e2021-03-25 14:12:29 +010018349 of every header is checked. The parser strictly adheres to the format
18350 described in RFC7239, with the extension that IPv4 addresses may optionally
18351 be followed by a colon (':') and a valid decimal port number (0 to 65535),
18352 which will be silently dropped. All other forms will not match and will
18353 cause the address to be ignored.
Tim Duesterhus7fb56282021-01-23 17:50:21 +010018354
18355 The <occ> parameter is processed as with req.hdr().
18356
18357 A typical use is with the X-Forwarded-For and X-Client-IP headers.
Willy Tarreau74ca5042013-06-11 23:12:07 +020018358
18359req.hdr_val([<name>[,<occ>]]) : integer
18360hdr_val([<name>[,<occ>]]) : integer (deprecated)
18361 This extracts the last occurrence of header <name> in an HTTP request, and
18362 converts it to an integer value. When used with ACLs, all occurrences are
18363 checked, and if <name> is omitted, every value of every header is checked.
Tim Duesterhus7fb56282021-01-23 17:50:21 +010018364
18365 The <occ> parameter is processed as with req.hdr().
18366
18367 A typical use is with the X-Forwarded-For header.
Willy Tarreau74ca5042013-06-11 23:12:07 +020018368
Christopher Faulet26167842020-11-24 17:13:24 +010018369req.hdrs : string
18370 Returns the current request headers as string including the last empty line
18371 separating headers from the request body. The last empty line can be used to
18372 detect a truncated header block. This sample fetch is useful for some SPOE
18373 headers analyzers and for advanced logging.
18374
18375req.hdrs_bin : binary
18376 Returns the current request headers contained in preparsed binary form. This
18377 is useful for offloading some processing with SPOE. Each string is described
18378 by a length followed by the number of bytes indicated in the length. The
18379 length is represented using the variable integer encoding detailed in the
18380 SPOE documentation. The end of the list is marked by a couple of empty header
18381 names and values (length of 0 for both).
18382
18383 *(<str:header-name><str:header-value>)<empty string><empty string>
Frédéric Lécailleec891192019-02-26 15:02:35 +010018384
Christopher Faulet26167842020-11-24 17:13:24 +010018385 int: refer to the SPOE documentation for the encoding
18386 str: <int:length><bytes>
Frédéric Lécailleec891192019-02-26 15:02:35 +010018387
Willy Tarreau74ca5042013-06-11 23:12:07 +020018388http_auth(<userlist>) : boolean
18389 Returns a boolean indicating whether the authentication data received from
18390 the client match a username & password stored in the specified userlist. This
18391 fetch function is not really useful outside of ACLs. Currently only http
18392 basic auth is supported.
18393
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010018394http_auth_group(<userlist>) : string
18395 Returns a string corresponding to the user name found in the authentication
18396 data received from the client if both the user name and password are valid
18397 according to the specified userlist. The main purpose is to use it in ACLs
18398 where it is then checked whether the user belongs to any group within a list.
Willy Tarreau74ca5042013-06-11 23:12:07 +020018399 This fetch function is not really useful outside of ACLs. Currently only http
18400 basic auth is supported.
18401
18402 ACL derivatives :
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010018403 http_auth_group(<userlist>) : group ...
18404 Returns true when the user extracted from the request and whose password is
18405 valid according to the specified userlist belongs to at least one of the
18406 groups.
Willy Tarreau74ca5042013-06-11 23:12:07 +020018407
Christopher Fauleta4063562019-08-02 11:51:37 +020018408http_auth_pass : string
Willy Tarreauc9c6cdb2020-03-05 16:03:58 +010018409 Returns the user's password found in the authentication data received from
18410 the client, as supplied in the Authorization header. Not checks are
18411 performed by this sample fetch. Only Basic authentication is supported.
Christopher Fauleta4063562019-08-02 11:51:37 +020018412
18413http_auth_type : string
Willy Tarreauc9c6cdb2020-03-05 16:03:58 +010018414 Returns the authentication method found in the authentication data received from
18415 the client, as supplied in the Authorization header. Not checks are
18416 performed by this sample fetch. Only Basic authentication is supported.
Christopher Fauleta4063562019-08-02 11:51:37 +020018417
18418http_auth_user : string
Willy Tarreauc9c6cdb2020-03-05 16:03:58 +010018419 Returns the user name found in the authentication data received from the
18420 client, as supplied in the Authorization header. Not checks are performed by
18421 this sample fetch. Only Basic authentication is supported.
Christopher Fauleta4063562019-08-02 11:51:37 +020018422
Willy Tarreau74ca5042013-06-11 23:12:07 +020018423http_first_req : boolean
Willy Tarreau7f18e522010-10-22 20:04:13 +020018424 Returns true when the request being processed is the first one of the
18425 connection. This can be used to add or remove headers that may be missing
Willy Tarreau74ca5042013-06-11 23:12:07 +020018426 from some requests when a request is not the first one, or to help grouping
18427 requests in the logs.
Willy Tarreau7f18e522010-10-22 20:04:13 +020018428
Willy Tarreau74ca5042013-06-11 23:12:07 +020018429method : integer + string
18430 Returns an integer value corresponding to the method in the HTTP request. For
18431 example, "GET" equals 1 (check sources to establish the matching). Value 9
18432 means "other method" and may be converted to a string extracted from the
18433 stream. This should not be used directly as a sample, this is only meant to
18434 be used from ACLs, which transparently convert methods from patterns to these
18435 integer + string values. Some predefined ACL already check for most common
18436 methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020018437
Willy Tarreau74ca5042013-06-11 23:12:07 +020018438 ACL derivatives :
18439 method : case insensitive method match
Willy Tarreau6a06a402007-07-15 20:15:28 +020018440
Willy Tarreau74ca5042013-06-11 23:12:07 +020018441 Example :
18442 # only accept GET and HEAD requests
18443 acl valid_method method GET HEAD
18444 http-request deny if ! valid_method
Willy Tarreau6a06a402007-07-15 20:15:28 +020018445
Willy Tarreau74ca5042013-06-11 23:12:07 +020018446path : string
18447 This extracts the request's URL path, which starts at the first slash and
18448 ends before the question mark (without the host part). A typical use is with
18449 prefetch-capable caches, and with portals which need to aggregate multiple
18450 information from databases and keep them in caches. Note that with outgoing
18451 caches, it would be wiser to use "url" instead. With ACLs, it's typically
Davor Ocelice9ed2812017-12-25 17:49:28 +010018452 used to match exact file names (e.g. "/login.php"), or directory parts using
Willy Tarreau74ca5042013-06-11 23:12:07 +020018453 the derivative forms. See also the "url" and "base" fetch methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020018454
Willy Tarreau74ca5042013-06-11 23:12:07 +020018455 ACL derivatives :
18456 path : exact string match
18457 path_beg : prefix match
18458 path_dir : subdir match
18459 path_dom : domain match
18460 path_end : suffix match
18461 path_len : length match
18462 path_reg : regex match
18463 path_sub : substring match
Willy Tarreau6a06a402007-07-15 20:15:28 +020018464
Christopher Faulete720c322020-09-02 17:25:18 +020018465pathq : string
18466 This extracts the request's URL path with the query-string, which starts at
18467 the first slash. This sample fetch is pretty handy to always retrieve a
18468 relative URI, excluding the scheme and the authority part, if any. Indeed,
18469 while it is the common representation for an HTTP/1.1 request target, in
18470 HTTP/2, an absolute URI is often used. This sample fetch will return the same
18471 result in both cases.
18472
Willy Tarreau49ad95c2015-01-19 15:06:26 +010018473query : string
18474 This extracts the request's query string, which starts after the first
18475 question mark. If no question mark is present, this fetch returns nothing. If
18476 a question mark is present but nothing follows, it returns an empty string.
18477 This means it's possible to easily know whether a query string is present
Tim Düsterhus4896c442016-11-29 02:15:19 +010018478 using the "found" matching method. This fetch is the complement of "path"
Willy Tarreau49ad95c2015-01-19 15:06:26 +010018479 which stops before the question mark.
18480
Willy Tarreaueb27ec72015-02-20 13:55:29 +010018481req.hdr_names([<delim>]) : string
18482 This builds a string made from the concatenation of all header names as they
18483 appear in the request when the rule is evaluated. The default delimiter is
18484 the comma (',') but it may be overridden as an optional argument <delim>. In
18485 this case, only the first character of <delim> is considered.
18486
Willy Tarreau74ca5042013-06-11 23:12:07 +020018487req.ver : string
18488req_ver : string (deprecated)
18489 Returns the version string from the HTTP request, for example "1.1". This can
18490 be useful for logs, but is mostly there for ACL. Some predefined ACL already
18491 check for versions 1.0 and 1.1.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018492
Willy Tarreau74ca5042013-06-11 23:12:07 +020018493 ACL derivatives :
18494 req_ver : exact string match
Willy Tarreau0e698542011-09-16 08:32:32 +020018495
Christopher Faulete596d182020-05-05 17:46:34 +020018496res.body : binary
18497 This returns the HTTP response's available body as a block of data. Unlike
18498 the request side, there is no directive to wait for the response's body. This
Tim Duesterhus7fb56282021-01-23 17:50:21 +010018499 sample fetch is really useful (and usable) in the health-check context.
18500
18501 It may be used in tcp-check based expect rules.
Christopher Faulete596d182020-05-05 17:46:34 +020018502
18503res.body_len : integer
18504 This returns the length of the HTTP response available body in bytes. Unlike
18505 the request side, there is no directive to wait for the response's body. This
Tim Duesterhus7fb56282021-01-23 17:50:21 +010018506 sample fetch is really useful (and usable) in the health-check context.
18507
18508 It may be used in tcp-check based expect rules.
Christopher Faulete596d182020-05-05 17:46:34 +020018509
18510res.body_size : integer
18511 This returns the advertised length of the HTTP response body in bytes. It
18512 will represent the advertised Content-Length header, or the size of the
18513 available data in case of chunked encoding. Unlike the request side, there is
18514 no directive to wait for the response body. This sample fetch is really
Tim Duesterhus7fb56282021-01-23 17:50:21 +010018515 useful (and usable) in the health-check context.
18516
18517 It may be used in tcp-check based expect rules.
Christopher Faulete596d182020-05-05 17:46:34 +020018518
Remi Tricot-Le Bretonbf971212020-10-27 11:55:57 +010018519res.cache_hit : boolean
18520 Returns the boolean "true" value if the response has been built out of an
18521 HTTP cache entry, otherwise returns boolean "false".
18522
18523res.cache_name : string
18524 Returns a string containing the name of the HTTP cache that was used to
18525 build the HTTP response if res.cache_hit is true, otherwise returns an
18526 empty string.
18527
Willy Tarreau74ca5042013-06-11 23:12:07 +020018528res.comp : boolean
18529 Returns the boolean "true" value if the response has been compressed by
18530 HAProxy, otherwise returns boolean "false". This may be used to add
18531 information in the logs.
Willy Tarreau6a06a402007-07-15 20:15:28 +020018532
Willy Tarreau74ca5042013-06-11 23:12:07 +020018533res.comp_algo : string
18534 Returns a string containing the name of the algorithm used if the response
18535 was compressed by HAProxy, for example : "deflate". This may be used to add
18536 some information in the logs.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018537
Willy Tarreau74ca5042013-06-11 23:12:07 +020018538res.cook([<name>]) : string
18539scook([<name>]) : string (deprecated)
18540 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
18541 header line from the response, and returns its value as string. If no name is
Tim Duesterhus7fb56282021-01-23 17:50:21 +010018542 specified, the first cookie value is returned.
18543
18544 It may be used in tcp-check based expect rules.
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020018545
Willy Tarreau74ca5042013-06-11 23:12:07 +020018546 ACL derivatives :
18547 scook([<name>] : exact string match
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020018548
Willy Tarreau74ca5042013-06-11 23:12:07 +020018549res.cook_cnt([<name>]) : integer
18550scook_cnt([<name>]) : integer (deprecated)
18551 Returns an integer value representing the number of occurrences of the cookie
18552 <name> in the response, or all cookies if <name> is not specified. This is
Tim Duesterhus7fb56282021-01-23 17:50:21 +010018553 mostly useful when combined with ACLs to detect suspicious responses.
18554
18555 It may be used in tcp-check based expect rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018556
Willy Tarreau74ca5042013-06-11 23:12:07 +020018557res.cook_val([<name>]) : integer
18558scook_val([<name>]) : integer (deprecated)
18559 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
18560 header line from the response, and converts its value to an integer which is
Tim Duesterhus7fb56282021-01-23 17:50:21 +010018561 returned. If no name is specified, the first cookie value is returned.
18562
18563 It may be used in tcp-check based expect rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018564
Willy Tarreau74ca5042013-06-11 23:12:07 +020018565res.fhdr([<name>[,<occ>]]) : string
Tim Duesterhus7fb56282021-01-23 17:50:21 +010018566 This fetch works like the req.fhdr() fetch with the difference that it acts
18567 on the headers within an HTTP response.
18568
18569 Like req.fhdr() the res.fhdr() fetch returns full values. If the header is
18570 defined to be a list you should use res.hdr().
18571
18572 This fetch is sometimes useful with headers such as Date or Expires.
18573
18574 It may be used in tcp-check based expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020018575
Willy Tarreau74ca5042013-06-11 23:12:07 +020018576res.fhdr_cnt([<name>]) : integer
Tim Duesterhus7fb56282021-01-23 17:50:21 +010018577 This fetch works like the req.fhdr_cnt() fetch with the difference that it
18578 acts on the headers within an HTTP response.
18579
18580 Like req.fhdr_cnt() the res.fhdr_cnt() fetch acts on full values. If the
18581 header is defined to be a list you should use res.hdr_cnt().
18582
18583 It may be used in tcp-check based expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020018584
Willy Tarreau74ca5042013-06-11 23:12:07 +020018585res.hdr([<name>[,<occ>]]) : string
18586shdr([<name>[,<occ>]]) : string (deprecated)
Tim Duesterhus7fb56282021-01-23 17:50:21 +010018587 This fetch works like the req.hdr() fetch with the difference that it acts
18588 on the headers within an HTTP response.
18589
18590 Like req.hdr() the res.hdr() fetch considers the comma to be a delimeter. If
18591 this is not desired res.fhdr() should be used.
18592
18593 It may be used in tcp-check based expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020018594
Willy Tarreau74ca5042013-06-11 23:12:07 +020018595 ACL derivatives :
18596 shdr([<name>[,<occ>]]) : exact string match
18597 shdr_beg([<name>[,<occ>]]) : prefix match
18598 shdr_dir([<name>[,<occ>]]) : subdir match
18599 shdr_dom([<name>[,<occ>]]) : domain match
18600 shdr_end([<name>[,<occ>]]) : suffix match
18601 shdr_len([<name>[,<occ>]]) : length match
18602 shdr_reg([<name>[,<occ>]]) : regex match
18603 shdr_sub([<name>[,<occ>]]) : substring match
18604
18605res.hdr_cnt([<name>]) : integer
18606shdr_cnt([<name>]) : integer (deprecated)
Tim Duesterhus7fb56282021-01-23 17:50:21 +010018607 This fetch works like the req.hdr_cnt() fetch with the difference that it
18608 acts on the headers within an HTTP response.
18609
18610 Like req.hdr_cnt() the res.hdr_cnt() fetch considers the comma to be a
18611 delimeter. If this is not desired res.fhdr_cnt() should be used.
18612
18613 It may be used in tcp-check based expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020018614
Willy Tarreau74ca5042013-06-11 23:12:07 +020018615res.hdr_ip([<name>[,<occ>]]) : ip
18616shdr_ip([<name>[,<occ>]]) : ip (deprecated)
Tim Duesterhus7fb56282021-01-23 17:50:21 +010018617 This fetch works like the req.hdr_ip() fetch with the difference that it
18618 acts on the headers within an HTTP response.
18619
18620 This can be useful to learn some data into a stick table.
18621
18622 It may be used in tcp-check based expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020018623
Willy Tarreaueb27ec72015-02-20 13:55:29 +010018624res.hdr_names([<delim>]) : string
18625 This builds a string made from the concatenation of all header names as they
18626 appear in the response when the rule is evaluated. The default delimiter is
18627 the comma (',') but it may be overridden as an optional argument <delim>. In
Tim Duesterhus7fb56282021-01-23 17:50:21 +010018628 this case, only the first character of <delim> is considered.
18629
18630 It may be used in tcp-check based expect rules.
Willy Tarreaueb27ec72015-02-20 13:55:29 +010018631
Willy Tarreau74ca5042013-06-11 23:12:07 +020018632res.hdr_val([<name>[,<occ>]]) : integer
18633shdr_val([<name>[,<occ>]]) : integer (deprecated)
Tim Duesterhus7fb56282021-01-23 17:50:21 +010018634 This fetch works like the req.hdr_val() fetch with the difference that it
18635 acts on the headers within an HTTP response.
18636
18637 This can be useful to learn some data into a stick table.
18638
18639 It may be used in tcp-check based expect rules.
Christopher Faulete596d182020-05-05 17:46:34 +020018640
18641res.hdrs : string
18642 Returns the current response headers as string including the last empty line
18643 separating headers from the request body. The last empty line can be used to
18644 detect a truncated header block. This sample fetch is useful for some SPOE
Tim Duesterhus7fb56282021-01-23 17:50:21 +010018645 headers analyzers and for advanced logging.
18646
18647 It may also be used in tcp-check based expect rules.
Christopher Faulete596d182020-05-05 17:46:34 +020018648
18649res.hdrs_bin : binary
18650 Returns the current response headers contained in preparsed binary form. This
18651 is useful for offloading some processing with SPOE. It may be used in
18652 tcp-check based expect rules. Each string is described by a length followed
18653 by the number of bytes indicated in the length. The length is represented
18654 using the variable integer encoding detailed in the SPOE documentation. The
18655 end of the list is marked by a couple of empty header names and values
18656 (length of 0 for both).
18657
18658 *(<str:header-name><str:header-value>)<empty string><empty string>
18659
18660 int: refer to the SPOE documentation for the encoding
18661 str: <int:length><bytes>
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010018662
Willy Tarreau74ca5042013-06-11 23:12:07 +020018663res.ver : string
18664resp_ver : string (deprecated)
18665 Returns the version string from the HTTP response, for example "1.1". This
Tim Duesterhus7fb56282021-01-23 17:50:21 +010018666 can be useful for logs, but is mostly there for ACL.
18667
18668 It may be used in tcp-check based expect rules.
Willy Tarreau0e698542011-09-16 08:32:32 +020018669
Willy Tarreau74ca5042013-06-11 23:12:07 +020018670 ACL derivatives :
18671 resp_ver : exact string match
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010018672
Willy Tarreau74ca5042013-06-11 23:12:07 +020018673set-cookie([<name>]) : string (deprecated)
18674 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
18675 header line from the response and uses the corresponding value to match. This
Willy Tarreau294d0f02015-08-10 19:40:12 +020018676 can be comparable to what "appsession" did with default options, but with
Willy Tarreau74ca5042013-06-11 23:12:07 +020018677 support for multi-peer synchronization and state keeping across restarts.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010018678
Willy Tarreau74ca5042013-06-11 23:12:07 +020018679 This fetch function is deprecated and has been superseded by the "res.cook"
18680 fetch. This keyword will disappear soon.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010018681
Willy Tarreau74ca5042013-06-11 23:12:07 +020018682status : integer
18683 Returns an integer containing the HTTP status code in the HTTP response, for
18684 example, 302. It is mostly used within ACLs and integer ranges, for example,
Tim Duesterhus7fb56282021-01-23 17:50:21 +010018685 to remove any Location header if the response is not a 3xx.
18686
18687 It may be used in tcp-check based expect rules.
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018688
Thierry Fournier0e00dca2016-04-07 15:47:40 +020018689unique-id : string
18690 Returns the unique-id attached to the request. The directive
18691 "unique-id-format" must be set. If it is not set, the unique-id sample fetch
18692 fails. Note that the unique-id is usually used with HTTP requests, however this
18693 sample fetch can be used with other protocols. Obviously, if it is used with
18694 other protocols than HTTP, the unique-id-format directive must not contain
18695 HTTP parts. See: unique-id-format and unique-id-header
18696
Willy Tarreau74ca5042013-06-11 23:12:07 +020018697url : string
18698 This extracts the request's URL as presented in the request. A typical use is
18699 with prefetch-capable caches, and with portals which need to aggregate
18700 multiple information from databases and keep them in caches. With ACLs, using
18701 "path" is preferred over using "url", because clients may send a full URL as
18702 is normally done with proxies. The only real use is to match "*" which does
18703 not match in "path", and for which there is already a predefined ACL. See
18704 also "path" and "base".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018705
Willy Tarreau74ca5042013-06-11 23:12:07 +020018706 ACL derivatives :
18707 url : exact string match
18708 url_beg : prefix match
18709 url_dir : subdir match
18710 url_dom : domain match
18711 url_end : suffix match
18712 url_len : length match
18713 url_reg : regex match
18714 url_sub : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018715
Willy Tarreau74ca5042013-06-11 23:12:07 +020018716url_ip : ip
18717 This extracts the IP address from the request's URL when the host part is
18718 presented as an IP address. Its use is very limited. For instance, a
18719 monitoring system might use this field as an alternative for the source IP in
18720 order to test what path a given source address would follow, or to force an
18721 entry in a table for a given source address. With ACLs it can be used to
18722 restrict access to certain systems through a proxy, for example when combined
18723 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018724
Willy Tarreau74ca5042013-06-11 23:12:07 +020018725url_port : integer
18726 This extracts the port part from the request's URL. Note that if the port is
18727 not specified in the request, port 80 is assumed. With ACLs it can be used to
18728 restrict access to certain systems through a proxy, for example when combined
18729 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018730
Willy Tarreau1ede1da2015-05-07 16:06:18 +020018731urlp([<name>[,<delim>]]) : string
18732url_param([<name>[,<delim>]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020018733 This extracts the first occurrence of the parameter <name> in the query
18734 string, which begins after either '?' or <delim>, and which ends before '&',
Willy Tarreau1ede1da2015-05-07 16:06:18 +020018735 ';' or <delim>. The parameter name is case-sensitive. If no name is given,
18736 any parameter will match, and the first one will be returned. The result is
18737 a string corresponding to the value of the parameter <name> as presented in
18738 the request (no URL decoding is performed). This can be used for session
Willy Tarreau74ca5042013-06-11 23:12:07 +020018739 stickiness based on a client ID, to extract an application cookie passed as a
18740 URL parameter, or in ACLs to apply some checks. Note that the ACL version of
Willy Tarreau1ede1da2015-05-07 16:06:18 +020018741 this fetch iterates over multiple parameters and will iteratively report all
18742 parameters values if no name is given
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018743
Willy Tarreau74ca5042013-06-11 23:12:07 +020018744 ACL derivatives :
18745 urlp(<name>[,<delim>]) : exact string match
18746 urlp_beg(<name>[,<delim>]) : prefix match
18747 urlp_dir(<name>[,<delim>]) : subdir match
18748 urlp_dom(<name>[,<delim>]) : domain match
18749 urlp_end(<name>[,<delim>]) : suffix match
18750 urlp_len(<name>[,<delim>]) : length match
18751 urlp_reg(<name>[,<delim>]) : regex match
18752 urlp_sub(<name>[,<delim>]) : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018753
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018754
Willy Tarreau74ca5042013-06-11 23:12:07 +020018755 Example :
18756 # match http://example.com/foo?PHPSESSIONID=some_id
18757 stick on urlp(PHPSESSIONID)
18758 # match http://example.com/foo;JSESSIONID=some_id
18759 stick on urlp(JSESSIONID,;)
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018760
Jarno Huuskonen676f6222017-03-30 09:19:45 +030018761urlp_val([<name>[,<delim>]]) : integer
Willy Tarreau74ca5042013-06-11 23:12:07 +020018762 See "urlp" above. This one extracts the URL parameter <name> in the request
18763 and converts it to an integer value. This can be used for session stickiness
18764 based on a user ID for example, or with ACLs to match a page number or price.
Willy Tarreaua9fddca2012-07-31 07:51:48 +020018765
Dragan Dosen0070cd52016-06-16 12:19:49 +020018766url32 : integer
18767 This returns a 32-bit hash of the value obtained by concatenating the first
18768 Host header and the whole URL including parameters (not only the path part of
18769 the request, as in the "base32" fetch above). This is useful to track per-URL
18770 activity. A shorter hash is stored, saving a lot of memory. The output type
18771 is an unsigned integer.
18772
18773url32+src : binary
18774 This returns the concatenation of the "url32" fetch and the "src" fetch. The
18775 resulting type is of type binary, with a size of 8 or 20 bytes depending on
18776 the source address family. This can be used to track per-IP, per-URL counters.
18777
Christopher Faulet16032ab2020-04-30 11:30:00 +020018778
Christopher Faulete596d182020-05-05 17:46:34 +0200187797.3.7. Fetching samples for developers
Christopher Fauletd47941d2020-01-08 14:40:19 +010018780---------------------------------------
18781
18782This set of sample fetch methods is reserved to developers and must never be
18783used on a production environment, except on developer demand, for debugging
18784purposes. Moreover, no special care will be taken on backwards compatibility.
18785There is no warranty the following sample fetches will never change, be renamed
18786or simply removed. So be really careful if you should use one of them. To avoid
18787any ambiguity, these sample fetches are placed in the dedicated scope "internal",
18788for instance "internal.strm.is_htx".
18789
18790internal.htx.data : integer
18791 Returns the size in bytes used by data in the HTX message associated to a
18792 channel. The channel is chosen depending on the sample direction.
18793
18794internal.htx.free : integer
18795 Returns the free space (size - used) in bytes in the HTX message associated
18796 to a channel. The channel is chosen depending on the sample direction.
18797
18798internal.htx.free_data : integer
18799 Returns the free space for the data in bytes in the HTX message associated to
18800 a channel. The channel is chosen depending on the sample direction.
18801
18802internal.htx.has_eom : boolean
18803 Returns true if the HTX message associated to a channel contains an
18804 end-of-message block (EOM). Otherwise, it returns false. The channel is
18805 chosen depending on the sample direction.
18806
18807internal.htx.nbblks : integer
18808 Returns the number of blocks present in the HTX message associated to a
18809 channel. The channel is chosen depending on the sample direction.
18810
18811internal.htx.size : integer
18812 Returns the total size in bytes of the HTX message associated to a
18813 channel. The channel is chosen depending on the sample direction.
18814
18815internal.htx.used : integer
18816 Returns the total size used in bytes (data + metadata) in the HTX message
18817 associated to a channel. The channel is chosen depending on the sample
18818 direction.
18819
18820internal.htx_blk.size(<idx>) : integer
18821 Returns the size of the block at the position <idx> in the HTX message
18822 associated to a channel or 0 if it does not exist. The channel is chosen
18823 depending on the sample direction. <idx> may be any positive integer or one
18824 of the special value :
18825 * head : The oldest inserted block
18826 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050018827 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010018828
18829internal.htx_blk.type(<idx>) : string
18830 Returns the type of the block at the position <idx> in the HTX message
18831 associated to a channel or "HTX_BLK_UNUSED" if it does not exist. The channel
18832 is chosen depending on the sample direction. <idx> may be any positive
18833 integer or one of the special value :
18834 * head : The oldest inserted block
18835 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050018836 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010018837
18838internal.htx_blk.data(<idx>) : binary
18839 Returns the value of the DATA block at the position <idx> in the HTX message
18840 associated to a channel or an empty string if it does not exist or if it is
18841 not a DATA block. The channel is chosen depending on the sample direction.
18842 <idx> may be any positive integer or one of the special value :
18843
18844 * head : The oldest inserted block
18845 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050018846 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010018847
18848internal.htx_blk.hdrname(<idx>) : string
18849 Returns the header name of the HEADER block at the position <idx> in the HTX
18850 message associated to a channel or an empty string if it does not exist or if
18851 it is not an HEADER block. The channel is chosen depending on the sample
18852 direction. <idx> may be any positive integer or one of the special value :
18853
18854 * head : The oldest inserted block
18855 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050018856 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010018857
18858internal.htx_blk.hdrval(<idx>) : string
18859 Returns the header value of the HEADER block at the position <idx> in the HTX
18860 message associated to a channel or an empty string if it does not exist or if
18861 it is not an HEADER block. The channel is chosen depending on the sample
18862 direction. <idx> may be any positive integer or one of the special value :
18863
18864 * head : The oldest inserted block
18865 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050018866 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010018867
18868internal.htx_blk.start_line(<idx>) : string
18869 Returns the value of the REQ_SL or RES_SL block at the position <idx> in the
18870 HTX message associated to a channel or an empty string if it does not exist
18871 or if it is not a SL block. The channel is chosen depending on the sample
18872 direction. <idx> may be any positive integer or one of the special value :
18873
18874 * head : The oldest inserted block
18875 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050018876 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010018877
18878internal.strm.is_htx : boolean
18879 Returns true if the current stream is an HTX stream. It means the data in the
18880 channels buffers are stored using the internal HTX representation. Otherwise,
18881 it returns false.
18882
18883
Willy Tarreau74ca5042013-06-11 23:12:07 +0200188847.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018885---------------------
Willy Tarreauced27012008-01-17 20:35:34 +010018886
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018887Some predefined ACLs are hard-coded so that they do not have to be declared in
18888every frontend which needs them. They all have their names in upper case in
Patrick Mézard2382ad62010-05-09 10:43:32 +020018889order to avoid confusion. Their equivalence is provided below.
Willy Tarreauced27012008-01-17 20:35:34 +010018890
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018891ACL name Equivalent to Usage
Christopher Faulet2c9c0c92021-04-01 17:24:04 +020018892---------------+----------------------------------+------------------------------------------------------
18893FALSE always_false never match
18894HTTP req.proto_http match if request protocol is valid HTTP
18895HTTP_1.0 req.ver 1.0 match if HTTP request version is 1.0
18896HTTP_1.1 req.ver 1.1 match if HTTP request version is 1.1
18897HTTP_CONTENT req.hdr_val(content-length) gt 0 match an existing content-length in the HTTP request
18898HTTP_URL_ABS url_reg ^[^/:]*:// match absolute URL with scheme
18899HTTP_URL_SLASH url_beg / match URL beginning with "/"
18900HTTP_URL_STAR url * match URL equal to "*"
18901LOCALHOST src 127.0.0.1/8 match connection from local host
18902METH_CONNECT method CONNECT match HTTP CONNECT method
18903METH_DELETE method DELETE match HTTP DELETE method
18904METH_GET method GET HEAD match HTTP GET or HEAD method
18905METH_HEAD method HEAD match HTTP HEAD method
18906METH_OPTIONS method OPTIONS match HTTP OPTIONS method
18907METH_POST method POST match HTTP POST method
18908METH_PUT method PUT match HTTP PUT method
18909METH_TRACE method TRACE match HTTP TRACE method
18910RDP_COOKIE req.rdp_cookie_cnt gt 0 match presence of an RDP cookie in the request buffer
18911REQ_CONTENT req.len gt 0 match data in the request buffer
18912TRUE always_true always match
18913WAIT_END wait_end wait for end of content analysis
18914---------------+----------------------------------+------------------------------------------------------
Willy Tarreauced27012008-01-17 20:35:34 +010018915
Willy Tarreaub937b7e2010-01-12 15:27:54 +010018916
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200189178. Logging
18918----------
Willy Tarreau844e3c52008-01-11 16:28:18 +010018919
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018920One of HAProxy's strong points certainly lies is its precise logs. It probably
18921provides the finest level of information available for such a product, which is
18922very important for troubleshooting complex environments. Standard information
18923provided in logs include client ports, TCP/HTTP state timers, precise session
18924state at termination and precise termination cause, information about decisions
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010018925to direct traffic to a server, and of course the ability to capture arbitrary
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018926headers.
18927
18928In order to improve administrators reactivity, it offers a great transparency
18929about encountered problems, both internal and external, and it is possible to
18930send logs to different sources at the same time with different level filters :
18931
18932 - global process-level logs (system errors, start/stop, etc..)
18933 - per-instance system and internal errors (lack of resource, bugs, ...)
18934 - per-instance external troubles (servers up/down, max connections)
18935 - per-instance activity (client connections), either at the establishment or
18936 at the termination.
Davor Ocelice9ed2812017-12-25 17:49:28 +010018937 - per-request control of log-level, e.g.
Jim Freeman9e8714b2015-05-26 09:16:34 -060018938 http-request set-log-level silent if sensitive_request
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018939
18940The ability to distribute different levels of logs to different log servers
18941allow several production teams to interact and to fix their problems as soon
18942as possible. For example, the system team might monitor system-wide errors,
18943while the application team might be monitoring the up/down for their servers in
18944real time, and the security team might analyze the activity logs with one hour
18945delay.
18946
18947
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200189488.1. Log levels
18949---------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018950
Simon Hormandf791f52011-05-29 15:01:10 +090018951TCP and HTTP connections can be logged with information such as the date, time,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018952source IP address, destination address, connection duration, response times,
Simon Hormandf791f52011-05-29 15:01:10 +090018953HTTP request, HTTP return code, number of bytes transmitted, conditions
18954in which the session ended, and even exchanged cookies values. For example
18955track a particular user's problems. All messages may be sent to up to two
18956syslog servers. Check the "log" keyword in section 4.2 for more information
18957about log facilities.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018958
18959
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200189608.2. Log formats
18961----------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018962
William Lallemand48940402012-01-30 16:47:22 +010018963HAProxy supports 5 log formats. Several fields are common between these formats
Simon Hormandf791f52011-05-29 15:01:10 +090018964and will be detailed in the following sections. A few of them may vary
18965slightly with the configuration, due to indicators specific to certain
18966options. The supported formats are as follows :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018967
18968 - the default format, which is very basic and very rarely used. It only
18969 provides very basic information about the incoming connection at the moment
18970 it is accepted : source IP:port, destination IP:port, and frontend-name.
18971 This mode will eventually disappear so it will not be described to great
18972 extents.
18973
18974 - the TCP format, which is more advanced. This format is enabled when "option
18975 tcplog" is set on the frontend. HAProxy will then usually wait for the
18976 connection to terminate before logging. This format provides much richer
18977 information, such as timers, connection counts, queue size, etc... This
18978 format is recommended for pure TCP proxies.
18979
18980 - the HTTP format, which is the most advanced for HTTP proxying. This format
18981 is enabled when "option httplog" is set on the frontend. It provides the
18982 same information as the TCP format with some HTTP-specific fields such as
18983 the request, the status code, and captures of headers and cookies. This
18984 format is recommended for HTTP proxies.
18985
Emeric Brun3a058f32009-06-30 18:26:00 +020018986 - the CLF HTTP format, which is equivalent to the HTTP format, but with the
18987 fields arranged in the same order as the CLF format. In this mode, all
18988 timers, captures, flags, etc... appear one per field after the end of the
18989 common fields, in the same order they appear in the standard HTTP format.
18990
William Lallemand48940402012-01-30 16:47:22 +010018991 - the custom log format, allows you to make your own log line.
18992
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018993Next sections will go deeper into details for each of these formats. Format
18994specification will be performed on a "field" basis. Unless stated otherwise, a
18995field is a portion of text delimited by any number of spaces. Since syslog
18996servers are susceptible of inserting fields at the beginning of a line, it is
18997always assumed that the first field is the one containing the process name and
18998identifier.
18999
19000Note : Since log lines may be quite long, the log examples in sections below
19001 might be broken into multiple lines. The example log lines will be
19002 prefixed with 3 closing angle brackets ('>>>') and each time a log is
19003 broken into multiple lines, each non-final line will end with a
19004 backslash ('\') and the next line will start indented by two characters.
19005
19006
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200190078.2.1. Default log format
19008-------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019009
19010This format is used when no specific option is set. The log is emitted as soon
19011as the connection is accepted. One should note that this currently is the only
19012format which logs the request's destination IP and ports.
19013
19014 Example :
19015 listen www
19016 mode http
19017 log global
19018 server srv1 127.0.0.1:8000
19019
19020 >>> Feb 6 12:12:09 localhost \
19021 haproxy[14385]: Connect from 10.0.1.2:33312 to 10.0.3.31:8012 \
19022 (www/HTTP)
19023
19024 Field Format Extract from the example above
19025 1 process_name '[' pid ']:' haproxy[14385]:
19026 2 'Connect from' Connect from
19027 3 source_ip ':' source_port 10.0.1.2:33312
19028 4 'to' to
19029 5 destination_ip ':' destination_port 10.0.3.31:8012
19030 6 '(' frontend_name '/' mode ')' (www/HTTP)
19031
19032Detailed fields description :
19033 - "source_ip" is the IP address of the client which initiated the connection.
19034 - "source_port" is the TCP port of the client which initiated the connection.
19035 - "destination_ip" is the IP address the client connected to.
19036 - "destination_port" is the TCP port the client connected to.
19037 - "frontend_name" is the name of the frontend (or listener) which received
19038 and processed the connection.
19039 - "mode is the mode the frontend is operating (TCP or HTTP).
19040
Willy Tarreauceb24bc2010-11-09 12:46:41 +010019041In case of a UNIX socket, the source and destination addresses are marked as
19042"unix:" and the ports reflect the internal ID of the socket which accepted the
19043connection (the same ID as reported in the stats).
19044
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019045It is advised not to use this deprecated format for newer installations as it
19046will eventually disappear.
19047
19048
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200190498.2.2. TCP log format
19050---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019051
19052The TCP format is used when "option tcplog" is specified in the frontend, and
19053is the recommended format for pure TCP proxies. It provides a lot of precious
19054information for troubleshooting. Since this format includes timers and byte
19055counts, the log is normally emitted at the end of the session. It can be
19056emitted earlier if "option logasap" is specified, which makes sense in most
19057environments with long sessions such as remote terminals. Sessions which match
19058the "monitor" rules are never logged. It is also possible not to emit logs for
19059sessions for which no data were exchanged between the client and the server, by
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020019060specifying "option dontlognull" in the frontend. Successful connections will
19061not be logged if "option dontlog-normal" is specified in the frontend. A few
19062fields may slightly vary depending on some configuration options, those are
19063marked with a star ('*') after the field name below.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019064
19065 Example :
19066 frontend fnt
19067 mode tcp
19068 option tcplog
19069 log global
19070 default_backend bck
19071
19072 backend bck
19073 server srv1 127.0.0.1:8000
19074
19075 >>> Feb 6 12:12:56 localhost \
19076 haproxy[14387]: 10.0.1.2:33313 [06/Feb/2009:12:12:51.443] fnt \
19077 bck/srv1 0/0/5007 212 -- 0/0/0/0/3 0/0
19078
19079 Field Format Extract from the example above
19080 1 process_name '[' pid ']:' haproxy[14387]:
19081 2 client_ip ':' client_port 10.0.1.2:33313
19082 3 '[' accept_date ']' [06/Feb/2009:12:12:51.443]
19083 4 frontend_name fnt
19084 5 backend_name '/' server_name bck/srv1
19085 6 Tw '/' Tc '/' Tt* 0/0/5007
19086 7 bytes_read* 212
19087 8 termination_state --
19088 9 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 0/0/0/0/3
19089 10 srv_queue '/' backend_queue 0/0
19090
19091Detailed fields description :
19092 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010019093 connection to haproxy. If the connection was accepted on a UNIX socket
19094 instead, the IP address would be replaced with the word "unix". Note that
19095 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010019096 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010019097 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010019098 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019099
19100 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010019101 If the connection was accepted on a UNIX socket instead, the port would be
19102 replaced with the ID of the accepting socket, which is also reported in the
19103 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019104
19105 - "accept_date" is the exact date when the connection was received by haproxy
19106 (which might be very slightly different from the date observed on the
19107 network if there was some queuing in the system's backlog). This is usually
Willy Tarreau590a0512018-09-05 11:56:48 +020019108 the same date which may appear in any upstream firewall's log. When used in
19109 HTTP mode, the accept_date field will be reset to the first moment the
19110 connection is ready to receive a new request (end of previous response for
19111 HTTP/1, immediately after previous request for HTTP/2).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019112
19113 - "frontend_name" is the name of the frontend (or listener) which received
19114 and processed the connection.
19115
19116 - "backend_name" is the name of the backend (or listener) which was selected
19117 to manage the connection to the server. This will be the same as the
19118 frontend if no switching rule has been applied, which is common for TCP
19119 applications.
19120
19121 - "server_name" is the name of the last server to which the connection was
19122 sent, which might differ from the first one if there were connection errors
19123 and a redispatch occurred. Note that this server belongs to the backend
19124 which processed the request. If the connection was aborted before reaching
19125 a server, "<NOSRV>" is indicated instead of a server name.
19126
19127 - "Tw" is the total time in milliseconds spent waiting in the various queues.
19128 It can be "-1" if the connection was aborted before reaching the queue.
19129 See "Timers" below for more details.
19130
19131 - "Tc" is the total time in milliseconds spent waiting for the connection to
19132 establish to the final server, including retries. It can be "-1" if the
19133 connection was aborted before a connection could be established. See
19134 "Timers" below for more details.
19135
19136 - "Tt" is the total time in milliseconds elapsed between the accept and the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030019137 last close. It covers all possible processing. There is one exception, if
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019138 "option logasap" was specified, then the time counting stops at the moment
19139 the log is emitted. In this case, a '+' sign is prepended before the value,
19140 indicating that the final one will be larger. See "Timers" below for more
19141 details.
19142
19143 - "bytes_read" is the total number of bytes transmitted from the server to
19144 the client when the log is emitted. If "option logasap" is specified, the
19145 this value will be prefixed with a '+' sign indicating that the final one
19146 may be larger. Please note that this value is a 64-bit counter, so log
19147 analysis tools must be able to handle it without overflowing.
19148
19149 - "termination_state" is the condition the session was in when the session
19150 ended. This indicates the session state, which side caused the end of
19151 session to happen, and for what reason (timeout, error, ...). The normal
19152 flags should be "--", indicating the session was closed by either end with
19153 no data remaining in buffers. See below "Session state at disconnection"
19154 for more details.
19155
19156 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040019157 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019158 limits have been reached. For instance, if actconn is close to 512 when
19159 multiple connection errors occur, chances are high that the system limits
19160 the process to use a maximum of 1024 file descriptors and that all of them
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019161 are used. See section 3 "Global parameters" to find how to tune the system.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019162
19163 - "feconn" is the total number of concurrent connections on the frontend when
19164 the session was logged. It is useful to estimate the amount of resource
19165 required to sustain high loads, and to detect when the frontend's "maxconn"
19166 has been reached. Most often when this value increases by huge jumps, it is
19167 because there is congestion on the backend servers, but sometimes it can be
19168 caused by a denial of service attack.
19169
19170 - "beconn" is the total number of concurrent connections handled by the
19171 backend when the session was logged. It includes the total number of
19172 concurrent connections active on servers as well as the number of
19173 connections pending in queues. It is useful to estimate the amount of
19174 additional servers needed to support high loads for a given application.
19175 Most often when this value increases by huge jumps, it is because there is
19176 congestion on the backend servers, but sometimes it can be caused by a
19177 denial of service attack.
19178
19179 - "srv_conn" is the total number of concurrent connections still active on
19180 the server when the session was logged. It can never exceed the server's
19181 configured "maxconn" parameter. If this value is very often close or equal
19182 to the server's "maxconn", it means that traffic regulation is involved a
19183 lot, meaning that either the server's maxconn value is too low, or that
19184 there aren't enough servers to process the load with an optimal response
19185 time. When only one of the server's "srv_conn" is high, it usually means
19186 that this server has some trouble causing the connections to take longer to
19187 be processed than on other servers.
19188
19189 - "retries" is the number of connection retries experienced by this session
19190 when trying to connect to the server. It must normally be zero, unless a
19191 server is being stopped at the same moment the connection was attempted.
19192 Frequent retries generally indicate either a network problem between
19193 haproxy and the server, or a misconfigured system backlog on the server
19194 preventing new connections from being queued. This field may optionally be
19195 prefixed with a '+' sign, indicating that the session has experienced a
19196 redispatch after the maximal retry count has been reached on the initial
19197 server. In this case, the server name appearing in the log is the one the
19198 connection was redispatched to, and not the first one, though both may
19199 sometimes be the same in case of hashing for instance. So as a general rule
19200 of thumb, when a '+' is present in front of the retry count, this count
19201 should not be attributed to the logged server.
19202
19203 - "srv_queue" is the total number of requests which were processed before
19204 this one in the server queue. It is zero when the request has not gone
19205 through the server queue. It makes it possible to estimate the approximate
19206 server's response time by dividing the time spent in queue by the number of
19207 requests in the queue. It is worth noting that if a session experiences a
19208 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010019209 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019210 backend queue unless a redispatch occurs.
19211
19212 - "backend_queue" is the total number of requests which were processed before
19213 this one in the backend's global queue. It is zero when the request has not
19214 gone through the global queue. It makes it possible to estimate the average
19215 queue length, which easily translates into a number of missing servers when
19216 divided by a server's "maxconn" parameter. It is worth noting that if a
19217 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010019218 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019219 through both the server queue and the backend queue unless a redispatch
19220 occurs.
19221
19222
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200192238.2.3. HTTP log format
19224----------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019225
19226The HTTP format is the most complete and the best suited for HTTP proxies. It
19227is enabled by when "option httplog" is specified in the frontend. It provides
19228the same level of information as the TCP format with additional features which
19229are specific to the HTTP protocol. Just like the TCP format, the log is usually
19230emitted at the end of the session, unless "option logasap" is specified, which
19231generally only makes sense for download sites. A session which matches the
19232"monitor" rules will never logged. It is also possible not to log sessions for
19233which no data were sent by the client by specifying "option dontlognull" in the
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020019234frontend. Successful connections will not be logged if "option dontlog-normal"
19235is specified in the frontend.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019236
19237Most fields are shared with the TCP log, some being different. A few fields may
19238slightly vary depending on some configuration options. Those ones are marked
19239with a star ('*') after the field name below.
19240
19241 Example :
19242 frontend http-in
19243 mode http
19244 option httplog
19245 log global
19246 default_backend bck
19247
19248 backend static
19249 server srv1 127.0.0.1:8000
19250
19251 >>> Feb 6 12:14:14 localhost \
19252 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
19253 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 +010019254 {} "GET /index.html HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019255
19256 Field Format Extract from the example above
19257 1 process_name '[' pid ']:' haproxy[14389]:
19258 2 client_ip ':' client_port 10.0.1.2:33317
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019259 3 '[' request_date ']' [06/Feb/2009:12:14:14.655]
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019260 4 frontend_name http-in
19261 5 backend_name '/' server_name static/srv1
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019262 6 TR '/' Tw '/' Tc '/' Tr '/' Ta* 10/0/30/69/109
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019263 7 status_code 200
19264 8 bytes_read* 2750
19265 9 captured_request_cookie -
19266 10 captured_response_cookie -
19267 11 termination_state ----
19268 12 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 1/1/1/1/0
19269 13 srv_queue '/' backend_queue 0/0
19270 14 '{' captured_request_headers* '}' {haproxy.1wt.eu}
19271 15 '{' captured_response_headers* '}' {}
19272 16 '"' http_request '"' "GET /index.html HTTP/1.1"
Willy Tarreaud72758d2010-01-12 10:42:19 +010019273
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019274Detailed fields description :
19275 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010019276 connection to haproxy. If the connection was accepted on a UNIX socket
19277 instead, the IP address would be replaced with the word "unix". Note that
19278 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010019279 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010019280 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010019281 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019282
19283 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010019284 If the connection was accepted on a UNIX socket instead, the port would be
19285 replaced with the ID of the accepting socket, which is also reported in the
19286 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019287
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019288 - "request_date" is the exact date when the first byte of the HTTP request
19289 was received by haproxy (log field %tr).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019290
19291 - "frontend_name" is the name of the frontend (or listener) which received
19292 and processed the connection.
19293
19294 - "backend_name" is the name of the backend (or listener) which was selected
19295 to manage the connection to the server. This will be the same as the
19296 frontend if no switching rule has been applied.
19297
19298 - "server_name" is the name of the last server to which the connection was
19299 sent, which might differ from the first one if there were connection errors
19300 and a redispatch occurred. Note that this server belongs to the backend
19301 which processed the request. If the request was aborted before reaching a
19302 server, "<NOSRV>" is indicated instead of a server name. If the request was
19303 intercepted by the stats subsystem, "<STATS>" is indicated instead.
19304
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019305 - "TR" is the total time in milliseconds spent waiting for a full HTTP
19306 request from the client (not counting body) after the first byte was
19307 received. It can be "-1" if the connection was aborted before a complete
John Roeslerfb2fce12019-07-10 15:45:51 -050019308 request could be received or a bad request was received. It should
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019309 always be very small because a request generally fits in one single packet.
19310 Large times here generally indicate network issues between the client and
Willy Tarreau590a0512018-09-05 11:56:48 +020019311 haproxy or requests being typed by hand. See section 8.4 "Timing Events"
19312 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019313
19314 - "Tw" is the total time in milliseconds spent waiting in the various queues.
19315 It can be "-1" if the connection was aborted before reaching the queue.
Willy Tarreau590a0512018-09-05 11:56:48 +020019316 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019317
19318 - "Tc" is the total time in milliseconds spent waiting for the connection to
19319 establish to the final server, including retries. It can be "-1" if the
Willy Tarreau590a0512018-09-05 11:56:48 +020019320 request was aborted before a connection could be established. See section
19321 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019322
19323 - "Tr" is the total time in milliseconds spent waiting for the server to send
19324 a full HTTP response, not counting data. It can be "-1" if the request was
19325 aborted before a complete response could be received. It generally matches
19326 the server's processing time for the request, though it may be altered by
19327 the amount of data sent by the client to the server. Large times here on
Willy Tarreau590a0512018-09-05 11:56:48 +020019328 "GET" requests generally indicate an overloaded server. See section 8.4
19329 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019330
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019331 - "Ta" is the time the request remained active in haproxy, which is the total
19332 time in milliseconds elapsed between the first byte of the request was
19333 received and the last byte of response was sent. It covers all possible
19334 processing except the handshake (see Th) and idle time (see Ti). There is
19335 one exception, if "option logasap" was specified, then the time counting
19336 stops at the moment the log is emitted. In this case, a '+' sign is
19337 prepended before the value, indicating that the final one will be larger.
Willy Tarreau590a0512018-09-05 11:56:48 +020019338 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019339
19340 - "status_code" is the HTTP status code returned to the client. This status
19341 is generally set by the server, but it might also be set by haproxy when
19342 the server cannot be reached or when its response is blocked by haproxy.
19343
19344 - "bytes_read" is the total number of bytes transmitted to the client when
19345 the log is emitted. This does include HTTP headers. If "option logasap" is
John Roeslerfb2fce12019-07-10 15:45:51 -050019346 specified, this value will be prefixed with a '+' sign indicating that
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019347 the final one may be larger. Please note that this value is a 64-bit
19348 counter, so log analysis tools must be able to handle it without
19349 overflowing.
19350
19351 - "captured_request_cookie" is an optional "name=value" entry indicating that
19352 the client had this cookie in the request. The cookie name and its maximum
19353 length are defined by the "capture cookie" statement in the frontend
19354 configuration. The field is a single dash ('-') when the option is not
19355 set. Only one cookie may be captured, it is generally used to track session
19356 ID exchanges between a client and a server to detect session crossing
19357 between clients due to application bugs. For more details, please consult
19358 the section "Capturing HTTP headers and cookies" below.
19359
19360 - "captured_response_cookie" is an optional "name=value" entry indicating
19361 that the server has returned a cookie with its response. The cookie name
19362 and its maximum length are defined by the "capture cookie" statement in the
19363 frontend configuration. The field is a single dash ('-') when the option is
19364 not set. Only one cookie may be captured, it is generally used to track
19365 session ID exchanges between a client and a server to detect session
19366 crossing between clients due to application bugs. For more details, please
19367 consult the section "Capturing HTTP headers and cookies" below.
19368
19369 - "termination_state" is the condition the session was in when the session
19370 ended. This indicates the session state, which side caused the end of
19371 session to happen, for what reason (timeout, error, ...), just like in TCP
19372 logs, and information about persistence operations on cookies in the last
19373 two characters. The normal flags should begin with "--", indicating the
19374 session was closed by either end with no data remaining in buffers. See
19375 below "Session state at disconnection" for more details.
19376
19377 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040019378 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019379 limits have been reached. For instance, if actconn is close to 512 or 1024
19380 when multiple connection errors occur, chances are high that the system
19381 limits the process to use a maximum of 1024 file descriptors and that all
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019382 of them are used. See section 3 "Global parameters" to find how to tune the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019383 system.
19384
19385 - "feconn" is the total number of concurrent connections on the frontend when
19386 the session was logged. It is useful to estimate the amount of resource
19387 required to sustain high loads, and to detect when the frontend's "maxconn"
19388 has been reached. Most often when this value increases by huge jumps, it is
19389 because there is congestion on the backend servers, but sometimes it can be
19390 caused by a denial of service attack.
19391
19392 - "beconn" is the total number of concurrent connections handled by the
19393 backend when the session was logged. It includes the total number of
19394 concurrent connections active on servers as well as the number of
19395 connections pending in queues. It is useful to estimate the amount of
19396 additional servers needed to support high loads for a given application.
19397 Most often when this value increases by huge jumps, it is because there is
19398 congestion on the backend servers, but sometimes it can be caused by a
19399 denial of service attack.
19400
19401 - "srv_conn" is the total number of concurrent connections still active on
19402 the server when the session was logged. It can never exceed the server's
19403 configured "maxconn" parameter. If this value is very often close or equal
19404 to the server's "maxconn", it means that traffic regulation is involved a
19405 lot, meaning that either the server's maxconn value is too low, or that
19406 there aren't enough servers to process the load with an optimal response
19407 time. When only one of the server's "srv_conn" is high, it usually means
19408 that this server has some trouble causing the requests to take longer to be
19409 processed than on other servers.
19410
19411 - "retries" is the number of connection retries experienced by this session
19412 when trying to connect to the server. It must normally be zero, unless a
19413 server is being stopped at the same moment the connection was attempted.
19414 Frequent retries generally indicate either a network problem between
19415 haproxy and the server, or a misconfigured system backlog on the server
19416 preventing new connections from being queued. This field may optionally be
19417 prefixed with a '+' sign, indicating that the session has experienced a
19418 redispatch after the maximal retry count has been reached on the initial
19419 server. In this case, the server name appearing in the log is the one the
19420 connection was redispatched to, and not the first one, though both may
19421 sometimes be the same in case of hashing for instance. So as a general rule
19422 of thumb, when a '+' is present in front of the retry count, this count
19423 should not be attributed to the logged server.
19424
19425 - "srv_queue" is the total number of requests which were processed before
19426 this one in the server queue. It is zero when the request has not gone
19427 through the server queue. It makes it possible to estimate the approximate
19428 server's response time by dividing the time spent in queue by the number of
19429 requests in the queue. It is worth noting that if a session experiences a
19430 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010019431 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019432 backend queue unless a redispatch occurs.
19433
19434 - "backend_queue" is the total number of requests which were processed before
19435 this one in the backend's global queue. It is zero when the request has not
19436 gone through the global queue. It makes it possible to estimate the average
19437 queue length, which easily translates into a number of missing servers when
19438 divided by a server's "maxconn" parameter. It is worth noting that if a
19439 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010019440 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019441 through both the server queue and the backend queue unless a redispatch
19442 occurs.
19443
19444 - "captured_request_headers" is a list of headers captured in the request due
19445 to the presence of the "capture request header" statement in the frontend.
19446 Multiple headers can be captured, they will be delimited by a vertical bar
19447 ('|'). When no capture is enabled, the braces do not appear, causing a
19448 shift of remaining fields. It is important to note that this field may
19449 contain spaces, and that using it requires a smarter log parser than when
19450 it's not used. Please consult the section "Capturing HTTP headers and
19451 cookies" below for more details.
19452
19453 - "captured_response_headers" is a list of headers captured in the response
19454 due to the presence of the "capture response header" statement in the
19455 frontend. Multiple headers can be captured, they will be delimited by a
19456 vertical bar ('|'). When no capture is enabled, the braces do not appear,
19457 causing a shift of remaining fields. It is important to note that this
19458 field may contain spaces, and that using it requires a smarter log parser
19459 than when it's not used. Please consult the section "Capturing HTTP headers
19460 and cookies" below for more details.
19461
19462 - "http_request" is the complete HTTP request line, including the method,
19463 request and HTTP version string. Non-printable characters are encoded (see
19464 below the section "Non-printable characters"). This is always the last
19465 field, and it is always delimited by quotes and is the only one which can
19466 contain quotes. If new fields are added to the log format, they will be
19467 added before this field. This field might be truncated if the request is
19468 huge and does not fit in the standard syslog buffer (1024 characters). This
19469 is the reason why this field must always remain the last one.
19470
19471
Cyril Bontédc4d9032012-04-08 21:57:39 +0200194728.2.4. Custom log format
19473------------------------
William Lallemand48940402012-01-30 16:47:22 +010019474
Willy Tarreau2beef582012-12-20 17:22:52 +010019475The directive log-format allows you to customize the logs in http mode and tcp
William Lallemandbddd4fd2012-02-27 11:23:10 +010019476mode. It takes a string as argument.
William Lallemand48940402012-01-30 16:47:22 +010019477
Davor Ocelice9ed2812017-12-25 17:49:28 +010019478HAProxy understands some log format variables. % precedes log format variables.
William Lallemand48940402012-01-30 16:47:22 +010019479Variables can take arguments using braces ('{}'), and multiple arguments are
19480separated by commas within the braces. Flags may be added or removed by
19481prefixing them with a '+' or '-' sign.
19482
19483Special variable "%o" may be used to propagate its flags to all other
19484variables on the same format string. This is particularly handy with quoted
Dragan Dosen835b9212016-02-12 13:23:03 +010019485("Q") and escaped ("E") string formats.
William Lallemand48940402012-01-30 16:47:22 +010019486
Willy Tarreauc8368452012-12-21 00:09:23 +010019487If a variable is named between square brackets ('[' .. ']') then it is used
Willy Tarreaube722a22014-06-13 16:31:59 +020019488as a sample expression rule (see section 7.3). This it useful to add some
Willy Tarreauc8368452012-12-21 00:09:23 +010019489less common information such as the client's SSL certificate's DN, or to log
19490the key that would be used to store an entry into a stick table.
19491
Dragan Dosen1e3b16f2020-06-23 18:16:44 +020019492Note: spaces must be escaped. In configuration directives "log-format",
19493"log-format-sd" and "unique-id-format", spaces are considered as
19494delimiters and are merged. In order to emit a verbatim '%', it must be
19495preceded by another '%' resulting in '%%'.
William Lallemand48940402012-01-30 16:47:22 +010019496
Dragan Dosen835b9212016-02-12 13:23:03 +010019497Note: when using the RFC5424 syslog message format, the characters '"',
19498'\' and ']' inside PARAM-VALUE should be escaped with '\' as prefix (see
19499https://tools.ietf.org/html/rfc5424#section-6.3.3 for more details). In
19500such cases, the use of the flag "E" should be considered.
19501
William Lallemand48940402012-01-30 16:47:22 +010019502Flags are :
19503 * Q: quote a string
Jamie Gloudonaaa21002012-08-25 00:18:33 -040019504 * X: hexadecimal representation (IPs, Ports, %Ts, %rt, %pid)
Dragan Dosen835b9212016-02-12 13:23:03 +010019505 * E: escape characters '"', '\' and ']' in a string with '\' as prefix
19506 (intended purpose is for the RFC5424 structured-data log formats)
William Lallemand48940402012-01-30 16:47:22 +010019507
19508 Example:
19509
19510 log-format %T\ %t\ Some\ Text
19511 log-format %{+Q}o\ %t\ %s\ %{-Q}r
19512
Dragan Dosen835b9212016-02-12 13:23:03 +010019513 log-format-sd %{+Q,+E}o\ [exampleSDID@1234\ header=%[capture.req.hdr(0)]]
19514
William Lallemand48940402012-01-30 16:47:22 +010019515At the moment, the default HTTP format is defined this way :
19516
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019517 log-format "%ci:%cp [%tr] %ft %b/%s %TR/%Tw/%Tc/%Tr/%Ta %ST %B %CC \
19518 %CS %tsc %ac/%fc/%bc/%sc/%rc %sq/%bq %hr %hs %{+Q}r"
William Lallemand48940402012-01-30 16:47:22 +010019519
William Lallemandbddd4fd2012-02-27 11:23:10 +010019520the default CLF format is defined this way :
William Lallemand48940402012-01-30 16:47:22 +010019521
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019522 log-format "%{+Q}o %{-Q}ci - - [%trg] %r %ST %B \"\" \"\" %cp \
19523 %ms %ft %b %s %TR %Tw %Tc %Tr %Ta %tsc %ac %fc \
19524 %bc %sc %rc %sq %bq %CC %CS %hrl %hsl"
William Lallemand48940402012-01-30 16:47:22 +010019525
William Lallemandbddd4fd2012-02-27 11:23:10 +010019526and the default TCP format is defined this way :
19527
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019528 log-format "%ci:%cp [%t] %ft %b/%s %Tw/%Tc/%Tt %B %ts \
19529 %ac/%fc/%bc/%sc/%rc %sq/%bq"
William Lallemandbddd4fd2012-02-27 11:23:10 +010019530
William Lallemand48940402012-01-30 16:47:22 +010019531Please refer to the table below for currently defined variables :
19532
William Lallemandbddd4fd2012-02-27 11:23:10 +010019533 +---+------+-----------------------------------------------+-------------+
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020019534 | R | var | field name (8.2.2 and 8.2.3 for description) | type |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019535 +---+------+-----------------------------------------------+-------------+
19536 | | %o | special variable, apply flags on all next var | |
19537 +---+------+-----------------------------------------------+-------------+
Willy Tarreau2beef582012-12-20 17:22:52 +010019538 | | %B | bytes_read (from server to client) | numeric |
19539 | H | %CC | captured_request_cookie | string |
19540 | H | %CS | captured_response_cookie | string |
William Lallemand5f232402012-04-05 18:02:55 +020019541 | | %H | hostname | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000019542 | H | %HM | HTTP method (ex: POST) | string |
Maciej Zdebf9f0f942020-11-26 10:45:52 +000019543 | H | %HP | HTTP request URI without query string | string |
Andrew Hayworthe63ac872015-07-31 16:14:16 +000019544 | H | %HQ | HTTP request URI query string (ex: ?bar=baz) | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000019545 | H | %HU | HTTP request URI (ex: /foo?bar=baz) | string |
19546 | H | %HV | HTTP version (ex: HTTP/1.0) | string |
William Lallemanda73203e2012-03-12 12:48:57 +010019547 | | %ID | unique-id | string |
Willy Tarreau4bf99632014-06-13 12:21:40 +020019548 | | %ST | status_code | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020019549 | | %T | gmt_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019550 | | %Ta | Active time of the request (from TR to end) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019551 | | %Tc | Tc | numeric |
Willy Tarreau27b639d2016-05-17 17:55:27 +020019552 | | %Td | Td = Tt - (Tq + Tw + Tc + Tr) | numeric |
Yuxans Yao4e25b012012-10-19 10:36:09 +080019553 | | %Tl | local_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019554 | | %Th | connection handshake time (SSL, PROXY proto) | numeric |
19555 | H | %Ti | idle time before the HTTP request | numeric |
19556 | H | %Tq | Th + Ti + TR | numeric |
19557 | H | %TR | time to receive the full request from 1st byte| numeric |
19558 | H | %Tr | Tr (response time) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020019559 | | %Ts | timestamp | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019560 | | %Tt | Tt | numeric |
Damien Claisse57c8eb92020-04-28 12:09:19 +000019561 | | %Tu | Tu | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019562 | | %Tw | Tw | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010019563 | | %U | bytes_uploaded (from client to server) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019564 | | %ac | actconn | numeric |
19565 | | %b | backend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010019566 | | %bc | beconn (backend concurrent connections) | numeric |
19567 | | %bi | backend_source_ip (connecting address) | IP |
19568 | | %bp | backend_source_port (connecting address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019569 | | %bq | backend_queue | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010019570 | | %ci | client_ip (accepted address) | IP |
19571 | | %cp | client_port (accepted address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019572 | | %f | frontend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010019573 | | %fc | feconn (frontend concurrent connections) | numeric |
19574 | | %fi | frontend_ip (accepting address) | IP |
19575 | | %fp | frontend_port (accepting address) | numeric |
Willy Tarreau773d65f2012-10-12 14:56:11 +020019576 | | %ft | frontend_name_transport ('~' suffix for SSL) | string |
Willy Tarreau7346acb2014-08-28 15:03:15 +020019577 | | %lc | frontend_log_counter | numeric |
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020019578 | | %hr | captured_request_headers default style | string |
19579 | | %hrl | captured_request_headers CLF style | string list |
19580 | | %hs | captured_response_headers default style | string |
19581 | | %hsl | captured_response_headers CLF style | string list |
Willy Tarreau812c88e2015-08-09 10:56:35 +020019582 | | %ms | accept date milliseconds (left-padded with 0) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020019583 | | %pid | PID | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020019584 | H | %r | http_request | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019585 | | %rc | retries | numeric |
Willy Tarreau1f0da242014-01-25 11:01:50 +010019586 | | %rt | request_counter (HTTP req or TCP session) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019587 | | %s | server_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010019588 | | %sc | srv_conn (server concurrent connections) | numeric |
19589 | | %si | server_IP (target address) | IP |
19590 | | %sp | server_port (target address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019591 | | %sq | srv_queue | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020019592 | S | %sslc| ssl_ciphers (ex: AES-SHA) | string |
19593 | S | %sslv| ssl_version (ex: TLSv1) | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010019594 | | %t | date_time (with millisecond resolution) | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019595 | H | %tr | date_time of HTTP request | date |
19596 | H | %trg | gmt_date_time of start of HTTP request | date |
Jens Bissinger15c64ff2018-08-23 14:11:27 +020019597 | H | %trl | local_date_time of start of HTTP request | date |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019598 | | %ts | termination_state | string |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020019599 | H | %tsc | termination_state with cookie status | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019600 +---+------+-----------------------------------------------+-------------+
William Lallemand48940402012-01-30 16:47:22 +010019601
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020019602 R = Restrictions : H = mode http only ; S = SSL only
William Lallemand48940402012-01-30 16:47:22 +010019603
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010019604
196058.2.5. Error log format
19606-----------------------
19607
19608When an incoming connection fails due to an SSL handshake or an invalid PROXY
19609protocol header, haproxy will log the event using a shorter, fixed line format.
19610By default, logs are emitted at the LOG_INFO level, unless the option
19611"log-separate-errors" is set in the backend, in which case the LOG_ERR level
Davor Ocelice9ed2812017-12-25 17:49:28 +010019612will be used. Connections on which no data are exchanged (e.g. probes) are not
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010019613logged if the "dontlognull" option is set.
19614
19615The format looks like this :
19616
19617 >>> Dec 3 18:27:14 localhost \
19618 haproxy[6103]: 127.0.0.1:56059 [03/Dec/2012:17:35:10.380] frt/f1: \
19619 Connection error during SSL handshake
19620
19621 Field Format Extract from the example above
19622 1 process_name '[' pid ']:' haproxy[6103]:
19623 2 client_ip ':' client_port 127.0.0.1:56059
19624 3 '[' accept_date ']' [03/Dec/2012:17:35:10.380]
19625 4 frontend_name "/" bind_name ":" frt/f1:
19626 5 message Connection error during SSL handshake
19627
19628These fields just provide minimal information to help debugging connection
19629failures.
19630
19631
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200196328.3. Advanced logging options
19633-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019634
19635Some advanced logging options are often looked for but are not easy to find out
19636just by looking at the various options. Here is an entry point for the few
19637options which can enable better logging. Please refer to the keywords reference
19638for more information about their usage.
19639
19640
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200196418.3.1. Disabling logging of external tests
19642------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019643
19644It is quite common to have some monitoring tools perform health checks on
19645haproxy. Sometimes it will be a layer 3 load-balancer such as LVS or any
19646commercial load-balancer, and sometimes it will simply be a more complete
19647monitoring system such as Nagios. When the tests are very frequent, users often
19648ask how to disable logging for those checks. There are three possibilities :
19649
19650 - if connections come from everywhere and are just TCP probes, it is often
19651 desired to simply disable logging of connections without data exchange, by
19652 setting "option dontlognull" in the frontend. It also disables logging of
19653 port scans, which may or may not be desired.
19654
Willy Tarreau9e9919d2020-10-14 15:55:23 +020019655 - it is possible to use the "http-request set-log-level silent" action using
19656 a variety of conditions (source networks, paths, user-agents, etc).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019657
19658 - if the tests are performed on a known URI, use "monitor-uri" to declare
19659 this URI as dedicated to monitoring. Any host sending this request will
19660 only get the result of a health-check, and the request will not be logged.
19661
19662
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200196638.3.2. Logging before waiting for the session to terminate
19664----------------------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019665
19666The problem with logging at end of connection is that you have no clue about
19667what is happening during very long sessions, such as remote terminal sessions
19668or large file downloads. This problem can be worked around by specifying
Davor Ocelice9ed2812017-12-25 17:49:28 +010019669"option logasap" in the frontend. HAProxy will then log as soon as possible,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019670just before data transfer begins. This means that in case of TCP, it will still
19671log the connection status to the server, and in case of HTTP, it will log just
19672after processing the server headers. In this case, the number of bytes reported
19673is the number of header bytes sent to the client. In order to avoid confusion
19674with normal logs, the total time field and the number of bytes are prefixed
19675with a '+' sign which means that real numbers are certainly larger.
19676
19677
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200196788.3.3. Raising log level upon errors
19679------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020019680
19681Sometimes it is more convenient to separate normal traffic from errors logs,
19682for instance in order to ease error monitoring from log files. When the option
19683"log-separate-errors" is used, connections which experience errors, timeouts,
19684retries, redispatches or HTTP status codes 5xx will see their syslog level
19685raised from "info" to "err". This will help a syslog daemon store the log in
19686a separate file. It is very important to keep the errors in the normal traffic
19687file too, so that log ordering is not altered. You should also be careful if
19688you already have configured your syslog daemon to store all logs higher than
19689"notice" in an "admin" file, because the "err" level is higher than "notice".
19690
19691
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200196928.3.4. Disabling logging of successful connections
19693--------------------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020019694
19695Although this may sound strange at first, some large sites have to deal with
19696multiple thousands of logs per second and are experiencing difficulties keeping
19697them intact for a long time or detecting errors within them. If the option
19698"dontlog-normal" is set on the frontend, all normal connections will not be
19699logged. In this regard, a normal connection is defined as one without any
19700error, timeout, retry nor redispatch. In HTTP, the status code is checked too,
19701and a response with a status 5xx is not considered normal and will be logged
19702too. Of course, doing is is really discouraged as it will remove most of the
19703useful information from the logs. Do this only if you have no other
19704alternative.
19705
19706
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200197078.4. Timing events
19708------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019709
19710Timers provide a great help in troubleshooting network problems. All values are
19711reported in milliseconds (ms). These timers should be used in conjunction with
19712the session termination flags. In TCP mode with "option tcplog" set on the
19713frontend, 3 control points are reported under the form "Tw/Tc/Tt", and in HTTP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019714mode, 5 control points are reported under the form "TR/Tw/Tc/Tr/Ta". In
19715addition, three other measures are provided, "Th", "Ti", and "Tq".
19716
Guillaume de Lafondf27cddc2016-12-23 17:32:43 +010019717Timings events in HTTP mode:
19718
19719 first request 2nd request
19720 |<-------------------------------->|<-------------- ...
19721 t tr t tr ...
19722 ---|----|----|----|----|----|----|----|----|--
19723 : Th Ti TR Tw Tc Tr Td : Ti ...
19724 :<---- Tq ---->: :
19725 :<-------------- Tt -------------->:
Damien Claisse57c8eb92020-04-28 12:09:19 +000019726 :<-- -----Tu--------------->:
Guillaume de Lafondf27cddc2016-12-23 17:32:43 +010019727 :<--------- Ta --------->:
19728
19729Timings events in TCP mode:
19730
19731 TCP session
19732 |<----------------->|
19733 t t
19734 ---|----|----|----|----|---
19735 | Th Tw Tc Td |
19736 |<------ Tt ------->|
19737
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019738 - Th: total time to accept tcp connection and execute handshakes for low level
Davor Ocelice9ed2812017-12-25 17:49:28 +010019739 protocols. Currently, these protocols are proxy-protocol and SSL. This may
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019740 only happen once during the whole connection's lifetime. A large time here
19741 may indicate that the client only pre-established the connection without
19742 speaking, that it is experiencing network issues preventing it from
Davor Ocelice9ed2812017-12-25 17:49:28 +010019743 completing a handshake in a reasonable time (e.g. MTU issues), or that an
Willy Tarreau590a0512018-09-05 11:56:48 +020019744 SSL handshake was very expensive to compute. Please note that this time is
19745 reported only before the first request, so it is safe to average it over
19746 all request to calculate the amortized value. The second and subsequent
19747 request will always report zero here.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019748
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019749 - Ti: is the idle time before the HTTP request (HTTP mode only). This timer
19750 counts between the end of the handshakes and the first byte of the HTTP
19751 request. When dealing with a second request in keep-alive mode, it starts
Willy Tarreau590a0512018-09-05 11:56:48 +020019752 to count after the end of the transmission the previous response. When a
19753 multiplexed protocol such as HTTP/2 is used, it starts to count immediately
19754 after the previous request. Some browsers pre-establish connections to a
19755 server in order to reduce the latency of a future request, and keep them
19756 pending until they need it. This delay will be reported as the idle time. A
19757 value of -1 indicates that nothing was received on the connection.
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019758
19759 - TR: total time to get the client request (HTTP mode only). It's the time
19760 elapsed between the first bytes received and the moment the proxy received
19761 the empty line marking the end of the HTTP headers. The value "-1"
19762 indicates that the end of headers has never been seen. This happens when
19763 the client closes prematurely or times out. This time is usually very short
19764 since most requests fit in a single packet. A large time may indicate a
19765 request typed by hand during a test.
19766
19767 - Tq: total time to get the client request from the accept date or since the
19768 emission of the last byte of the previous response (HTTP mode only). It's
Davor Ocelice9ed2812017-12-25 17:49:28 +010019769 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 +020019770 returns -1 as well. This timer used to be very useful before the arrival of
19771 HTTP keep-alive and browsers' pre-connect feature. It's recommended to drop
19772 it in favor of TR nowadays, as the idle time adds a lot of noise to the
19773 reports.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019774
19775 - Tw: total time spent in the queues waiting for a connection slot. It
19776 accounts for backend queue as well as the server queues, and depends on the
19777 queue size, and the time needed for the server to complete previous
19778 requests. The value "-1" means that the request was killed before reaching
19779 the queue, which is generally what happens with invalid or denied requests.
19780
19781 - Tc: total time to establish the TCP connection to the server. It's the time
19782 elapsed between the moment the proxy sent the connection request, and the
19783 moment it was acknowledged by the server, or between the TCP SYN packet and
19784 the matching SYN/ACK packet in return. The value "-1" means that the
19785 connection never established.
19786
19787 - Tr: server response time (HTTP mode only). It's the time elapsed between
19788 the moment the TCP connection was established to the server and the moment
19789 the server sent its complete response headers. It purely shows its request
19790 processing time, without the network overhead due to the data transmission.
19791 It is worth noting that when the client has data to send to the server, for
19792 instance during a POST request, the time already runs, and this can distort
19793 apparent response time. For this reason, it's generally wise not to trust
19794 too much this field for POST requests initiated from clients behind an
19795 untrusted network. A value of "-1" here means that the last the response
19796 header (empty line) was never seen, most likely because the server timeout
19797 stroke before the server managed to process the request.
19798
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019799 - Ta: total active time for the HTTP request, between the moment the proxy
19800 received the first byte of the request header and the emission of the last
19801 byte of the response body. The exception is when the "logasap" option is
19802 specified. In this case, it only equals (TR+Tw+Tc+Tr), and is prefixed with
19803 a '+' sign. From this field, we can deduce "Td", the data transmission time,
19804 by subtracting other timers when valid :
19805
19806 Td = Ta - (TR + Tw + Tc + Tr)
19807
19808 Timers with "-1" values have to be excluded from this equation. Note that
19809 "Ta" can never be negative.
19810
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019811 - Tt: total session duration time, between the moment the proxy accepted it
19812 and the moment both ends were closed. The exception is when the "logasap"
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019813 option is specified. In this case, it only equals (Th+Ti+TR+Tw+Tc+Tr), and
19814 is prefixed with a '+' sign. From this field, we can deduce "Td", the data
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030019815 transmission time, by subtracting other timers when valid :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019816
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019817 Td = Tt - (Th + Ti + TR + Tw + Tc + Tr)
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019818
19819 Timers with "-1" values have to be excluded from this equation. In TCP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019820 mode, "Ti", "Tq" and "Tr" have to be excluded too. Note that "Tt" can never
19821 be negative and that for HTTP, Tt is simply equal to (Th+Ti+Ta).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019822
Damien Claisse57c8eb92020-04-28 12:09:19 +000019823 - Tu: total estimated time as seen from client, between the moment the proxy
19824 accepted it and the moment both ends were closed, without idle time.
19825 This is useful to roughly measure end-to-end time as a user would see it,
19826 without idle time pollution from keep-alive time between requests. This
19827 timer in only an estimation of time seen by user as it assumes network
19828 latency is the same in both directions. The exception is when the "logasap"
19829 option is specified. In this case, it only equals (Th+TR+Tw+Tc+Tr), and is
19830 prefixed with a '+' sign.
19831
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019832These timers provide precious indications on trouble causes. Since the TCP
19833protocol defines retransmit delays of 3, 6, 12... seconds, we know for sure
19834that timers close to multiples of 3s are nearly always related to lost packets
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019835due to network problems (wires, negotiation, congestion). Moreover, if "Ta" or
19836"Tt" is close to a timeout value specified in the configuration, it often means
19837that a session has been aborted on timeout.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019838
19839Most common cases :
19840
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019841 - If "Th" or "Ti" are close to 3000, a packet has probably been lost between
19842 the client and the proxy. This is very rare on local networks but might
19843 happen when clients are on far remote networks and send large requests. It
19844 may happen that values larger than usual appear here without any network
19845 cause. Sometimes, during an attack or just after a resource starvation has
19846 ended, haproxy may accept thousands of connections in a few milliseconds.
19847 The time spent accepting these connections will inevitably slightly delay
19848 processing of other connections, and it can happen that request times in the
19849 order of a few tens of milliseconds are measured after a few thousands of
19850 new connections have been accepted at once. Using one of the keep-alive
19851 modes may display larger idle times since "Ti" measures the time spent
Patrick Mezard105faca2010-06-12 17:02:46 +020019852 waiting for additional requests.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019853
19854 - If "Tc" is close to 3000, a packet has probably been lost between the
19855 server and the proxy during the server connection phase. This value should
19856 always be very low, such as 1 ms on local networks and less than a few tens
19857 of ms on remote networks.
19858
Willy Tarreau55165fe2009-05-10 12:02:55 +020019859 - If "Tr" is nearly always lower than 3000 except some rare values which seem
19860 to be the average majored by 3000, there are probably some packets lost
19861 between the proxy and the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019862
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019863 - If "Ta" is large even for small byte counts, it generally is because
19864 neither the client nor the server decides to close the connection while
19865 haproxy is running in tunnel mode and both have agreed on a keep-alive
19866 connection mode. In order to solve this issue, it will be needed to specify
19867 one of the HTTP options to manipulate keep-alive or close options on either
19868 the frontend or the backend. Having the smallest possible 'Ta' or 'Tt' is
19869 important when connection regulation is used with the "maxconn" option on
19870 the servers, since no new connection will be sent to the server until
19871 another one is released.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019872
19873Other noticeable HTTP log cases ('xx' means any value to be ignored) :
19874
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019875 TR/Tw/Tc/Tr/+Ta The "option logasap" is present on the frontend and the log
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019876 was emitted before the data phase. All the timers are valid
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019877 except "Ta" which is shorter than reality.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019878
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019879 -1/xx/xx/xx/Ta The client was not able to send a complete request in time
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019880 or it aborted too early. Check the session termination flags
19881 then "timeout http-request" and "timeout client" settings.
19882
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019883 TR/-1/xx/xx/Ta It was not possible to process the request, maybe because
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019884 servers were out of order, because the request was invalid
19885 or forbidden by ACL rules. Check the session termination
19886 flags.
19887
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019888 TR/Tw/-1/xx/Ta The connection could not establish on the server. Either it
19889 actively refused it or it timed out after Ta-(TR+Tw) ms.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019890 Check the session termination flags, then check the
19891 "timeout connect" setting. Note that the tarpit action might
19892 return similar-looking patterns, with "Tw" equal to the time
19893 the client connection was maintained open.
19894
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019895 TR/Tw/Tc/-1/Ta The server has accepted the connection but did not return
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030019896 a complete response in time, or it closed its connection
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019897 unexpectedly after Ta-(TR+Tw+Tc) ms. Check the session
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019898 termination flags, then check the "timeout server" setting.
19899
19900
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200199018.5. Session state at disconnection
19902-----------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019903
19904TCP and HTTP logs provide a session termination indicator in the
19905"termination_state" field, just before the number of active connections. It is
199062-characters long in TCP mode, and is extended to 4 characters in HTTP mode,
19907each of which has a special meaning :
19908
19909 - On the first character, a code reporting the first event which caused the
19910 session to terminate :
19911
19912 C : the TCP session was unexpectedly aborted by the client.
19913
19914 S : the TCP session was unexpectedly aborted by the server, or the
19915 server explicitly refused it.
19916
19917 P : the session was prematurely aborted by the proxy, because of a
19918 connection limit enforcement, because a DENY filter was matched,
19919 because of a security check which detected and blocked a dangerous
19920 error in server response which might have caused information leak
Davor Ocelice9ed2812017-12-25 17:49:28 +010019921 (e.g. cacheable cookie).
Willy Tarreau570f2212013-06-10 16:42:09 +020019922
19923 L : the session was locally processed by haproxy and was not passed to
19924 a server. This is what happens for stats and redirects.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019925
19926 R : a resource on the proxy has been exhausted (memory, sockets, source
19927 ports, ...). Usually, this appears during the connection phase, and
19928 system logs should contain a copy of the precise error. If this
19929 happens, it must be considered as a very serious anomaly which
19930 should be fixed as soon as possible by any means.
19931
19932 I : an internal error was identified by the proxy during a self-check.
19933 This should NEVER happen, and you are encouraged to report any log
19934 containing this, because this would almost certainly be a bug. It
19935 would be wise to preventively restart the process after such an
19936 event too, in case it would be caused by memory corruption.
19937
Simon Horman752dc4a2011-06-21 14:34:59 +090019938 D : the session was killed by haproxy because the server was detected
19939 as down and was configured to kill all connections when going down.
19940
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070019941 U : the session was killed by haproxy on this backup server because an
19942 active server was detected as up and was configured to kill all
19943 backup connections when going up.
19944
Willy Tarreaua2a64e92011-09-07 23:01:56 +020019945 K : the session was actively killed by an admin operating on haproxy.
19946
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019947 c : the client-side timeout expired while waiting for the client to
19948 send or receive data.
19949
19950 s : the server-side timeout expired while waiting for the server to
19951 send or receive data.
19952
19953 - : normal session completion, both the client and the server closed
19954 with nothing left in the buffers.
19955
19956 - on the second character, the TCP or HTTP session state when it was closed :
19957
Willy Tarreauf7b30a92010-12-06 22:59:17 +010019958 R : the proxy was waiting for a complete, valid REQUEST from the client
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019959 (HTTP mode only). Nothing was sent to any server.
19960
19961 Q : the proxy was waiting in the QUEUE for a connection slot. This can
19962 only happen when servers have a 'maxconn' parameter set. It can
19963 also happen in the global queue after a redispatch consecutive to
19964 a failed attempt to connect to a dying server. If no redispatch is
19965 reported, then no connection attempt was made to any server.
19966
19967 C : the proxy was waiting for the CONNECTION to establish on the
19968 server. The server might at most have noticed a connection attempt.
19969
19970 H : the proxy was waiting for complete, valid response HEADERS from the
19971 server (HTTP only).
19972
19973 D : the session was in the DATA phase.
19974
19975 L : the proxy was still transmitting LAST data to the client while the
19976 server had already finished. This one is very rare as it can only
19977 happen when the client dies while receiving the last packets.
19978
19979 T : the request was tarpitted. It has been held open with the client
19980 during the whole "timeout tarpit" duration or until the client
19981 closed, both of which will be reported in the "Tw" timer.
19982
19983 - : normal session completion after end of data transfer.
19984
19985 - the third character tells whether the persistence cookie was provided by
19986 the client (only in HTTP mode) :
19987
19988 N : the client provided NO cookie. This is usually the case for new
19989 visitors, so counting the number of occurrences of this flag in the
19990 logs generally indicate a valid trend for the site frequentation.
19991
19992 I : the client provided an INVALID cookie matching no known server.
19993 This might be caused by a recent configuration change, mixed
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +020019994 cookies between HTTP/HTTPS sites, persistence conditionally
19995 ignored, or an attack.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019996
19997 D : the client provided a cookie designating a server which was DOWN,
19998 so either "option persist" was used and the client was sent to
19999 this server, or it was not set and the client was redispatched to
20000 another server.
20001
Willy Tarreau996a92c2010-10-13 19:30:47 +020020002 V : the client provided a VALID cookie, and was sent to the associated
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020003 server.
20004
Willy Tarreau996a92c2010-10-13 19:30:47 +020020005 E : the client provided a valid cookie, but with a last date which was
20006 older than what is allowed by the "maxidle" cookie parameter, so
20007 the cookie is consider EXPIRED and is ignored. The request will be
20008 redispatched just as if there was no cookie.
20009
20010 O : the client provided a valid cookie, but with a first date which was
20011 older than what is allowed by the "maxlife" cookie parameter, so
20012 the cookie is consider too OLD and is ignored. The request will be
20013 redispatched just as if there was no cookie.
20014
Willy Tarreauc89ccb62012-04-05 21:18:22 +020020015 U : a cookie was present but was not used to select the server because
20016 some other server selection mechanism was used instead (typically a
20017 "use-server" rule).
20018
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020019 - : does not apply (no cookie set in configuration).
20020
20021 - the last character reports what operations were performed on the persistence
20022 cookie returned by the server (only in HTTP mode) :
20023
20024 N : NO cookie was provided by the server, and none was inserted either.
20025
20026 I : no cookie was provided by the server, and the proxy INSERTED one.
20027 Note that in "cookie insert" mode, if the server provides a cookie,
20028 it will still be overwritten and reported as "I" here.
20029
Willy Tarreau996a92c2010-10-13 19:30:47 +020020030 U : the proxy UPDATED the last date in the cookie that was presented by
20031 the client. This can only happen in insert mode with "maxidle". It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030020032 happens every time there is activity at a different date than the
Willy Tarreau996a92c2010-10-13 19:30:47 +020020033 date indicated in the cookie. If any other change happens, such as
20034 a redispatch, then the cookie will be marked as inserted instead.
20035
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020036 P : a cookie was PROVIDED by the server and transmitted as-is.
20037
20038 R : the cookie provided by the server was REWRITTEN by the proxy, which
20039 happens in "cookie rewrite" or "cookie prefix" modes.
20040
20041 D : the cookie provided by the server was DELETED by the proxy.
20042
20043 - : does not apply (no cookie set in configuration).
20044
Willy Tarreau996a92c2010-10-13 19:30:47 +020020045The combination of the two first flags gives a lot of information about what
20046was happening when the session terminated, and why it did terminate. It can be
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020047helpful to detect server saturation, network troubles, local system resource
20048starvation, attacks, etc...
20049
20050The most common termination flags combinations are indicated below. They are
20051alphabetically sorted, with the lowercase set just after the upper case for
20052easier finding and understanding.
20053
20054 Flags Reason
20055
20056 -- Normal termination.
20057
20058 CC The client aborted before the connection could be established to the
20059 server. This can happen when haproxy tries to connect to a recently
20060 dead (or unchecked) server, and the client aborts while haproxy is
20061 waiting for the server to respond or for "timeout connect" to expire.
20062
20063 CD The client unexpectedly aborted during data transfer. This can be
20064 caused by a browser crash, by an intermediate equipment between the
20065 client and haproxy which decided to actively break the connection,
20066 by network routing issues between the client and haproxy, or by a
20067 keep-alive session between the server and the client terminated first
20068 by the client.
Willy Tarreaud72758d2010-01-12 10:42:19 +010020069
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020070 cD The client did not send nor acknowledge any data for as long as the
20071 "timeout client" delay. This is often caused by network failures on
Cyril Bontédc4d9032012-04-08 21:57:39 +020020072 the client side, or the client simply leaving the net uncleanly.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020073
20074 CH The client aborted while waiting for the server to start responding.
20075 It might be the server taking too long to respond or the client
20076 clicking the 'Stop' button too fast.
20077
20078 cH The "timeout client" stroke while waiting for client data during a
20079 POST request. This is sometimes caused by too large TCP MSS values
20080 for PPPoE networks which cannot transport full-sized packets. It can
20081 also happen when client timeout is smaller than server timeout and
20082 the server takes too long to respond.
20083
20084 CQ The client aborted while its session was queued, waiting for a server
20085 with enough empty slots to accept it. It might be that either all the
20086 servers were saturated or that the assigned server was taking too
20087 long a time to respond.
20088
20089 CR The client aborted before sending a full HTTP request. Most likely
20090 the request was typed by hand using a telnet client, and aborted
20091 too early. The HTTP status code is likely a 400 here. Sometimes this
20092 might also be caused by an IDS killing the connection between haproxy
Willy Tarreau0f228a02015-05-01 15:37:53 +020020093 and the client. "option http-ignore-probes" can be used to ignore
20094 connections without any data transfer.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020095
20096 cR The "timeout http-request" stroke before the client sent a full HTTP
20097 request. This is sometimes caused by too large TCP MSS values on the
20098 client side for PPPoE networks which cannot transport full-sized
20099 packets, or by clients sending requests by hand and not typing fast
20100 enough, or forgetting to enter the empty line at the end of the
Willy Tarreau2705a612014-05-23 17:38:34 +020020101 request. The HTTP status code is likely a 408 here. Note: recently,
Willy Tarreau0f228a02015-05-01 15:37:53 +020020102 some browsers started to implement a "pre-connect" feature consisting
20103 in speculatively connecting to some recently visited web sites just
20104 in case the user would like to visit them. This results in many
20105 connections being established to web sites, which end up in 408
20106 Request Timeout if the timeout strikes first, or 400 Bad Request when
20107 the browser decides to close them first. These ones pollute the log
20108 and feed the error counters. Some versions of some browsers have even
20109 been reported to display the error code. It is possible to work
Davor Ocelice9ed2812017-12-25 17:49:28 +010020110 around the undesirable effects of this behavior by adding "option
Willy Tarreau0f228a02015-05-01 15:37:53 +020020111 http-ignore-probes" in the frontend, resulting in connections with
20112 zero data transfer to be totally ignored. This will definitely hide
20113 the errors of people experiencing connectivity issues though.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020114
20115 CT The client aborted while its session was tarpitted. It is important to
20116 check if this happens on valid requests, in order to be sure that no
Willy Tarreau55165fe2009-05-10 12:02:55 +020020117 wrong tarpit rules have been written. If a lot of them happen, it
20118 might make sense to lower the "timeout tarpit" value to something
20119 closer to the average reported "Tw" timer, in order not to consume
20120 resources for just a few attackers.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020121
Willy Tarreau570f2212013-06-10 16:42:09 +020020122 LR The request was intercepted and locally handled by haproxy. Generally
20123 it means that this was a redirect or a stats request.
20124
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010020125 SC The server or an equipment between it and haproxy explicitly refused
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020126 the TCP connection (the proxy received a TCP RST or an ICMP message
20127 in return). Under some circumstances, it can also be the network
Davor Ocelice9ed2812017-12-25 17:49:28 +010020128 stack telling the proxy that the server is unreachable (e.g. no route,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020129 or no ARP response on local network). When this happens in HTTP mode,
20130 the status code is likely a 502 or 503 here.
20131
20132 sC The "timeout connect" stroke before a connection to the server could
20133 complete. When this happens in HTTP mode, the status code is likely a
20134 503 or 504 here.
20135
20136 SD The connection to the server died with an error during the data
20137 transfer. This usually means that haproxy has received an RST from
20138 the server or an ICMP message from an intermediate equipment while
20139 exchanging data with the server. This can be caused by a server crash
20140 or by a network issue on an intermediate equipment.
20141
20142 sD The server did not send nor acknowledge any data for as long as the
20143 "timeout server" setting during the data phase. This is often caused
Davor Ocelice9ed2812017-12-25 17:49:28 +010020144 by too short timeouts on L4 equipment before the server (firewalls,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020145 load-balancers, ...), as well as keep-alive sessions maintained
20146 between the client and the server expiring first on haproxy.
20147
20148 SH The server aborted before sending its full HTTP response headers, or
20149 it crashed while processing the request. Since a server aborting at
20150 this moment is very rare, it would be wise to inspect its logs to
20151 control whether it crashed and why. The logged request may indicate a
20152 small set of faulty requests, demonstrating bugs in the application.
20153 Sometimes this might also be caused by an IDS killing the connection
20154 between haproxy and the server.
20155
20156 sH The "timeout server" stroke before the server could return its
20157 response headers. This is the most common anomaly, indicating too
20158 long transactions, probably caused by server or database saturation.
20159 The immediate workaround consists in increasing the "timeout server"
20160 setting, but it is important to keep in mind that the user experience
20161 will suffer from these long response times. The only long term
20162 solution is to fix the application.
20163
20164 sQ The session spent too much time in queue and has been expired. See
20165 the "timeout queue" and "timeout connect" settings to find out how to
20166 fix this if it happens too often. If it often happens massively in
20167 short periods, it may indicate general problems on the affected
20168 servers due to I/O or database congestion, or saturation caused by
20169 external attacks.
20170
20171 PC The proxy refused to establish a connection to the server because the
Thayne McCombsdab4ba62021-01-07 21:24:41 -070020172 process's socket limit has been reached while attempting to connect.
Cyril Bontédc4d9032012-04-08 21:57:39 +020020173 The global "maxconn" parameter may be increased in the configuration
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020174 so that it does not happen anymore. This status is very rare and
20175 might happen when the global "ulimit-n" parameter is forced by hand.
20176
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010020177 PD The proxy blocked an incorrectly formatted chunked encoded message in
20178 a request or a response, after the server has emitted its headers. In
20179 most cases, this will indicate an invalid message from the server to
Davor Ocelice9ed2812017-12-25 17:49:28 +010020180 the client. HAProxy supports chunk sizes of up to 2GB - 1 (2147483647
Willy Tarreauf3a3e132013-08-31 08:16:26 +020020181 bytes). Any larger size will be considered as an error.
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010020182
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020183 PH The proxy blocked the server's response, because it was invalid,
20184 incomplete, dangerous (cache control), or matched a security filter.
20185 In any case, an HTTP 502 error is sent to the client. One possible
20186 cause for this error is an invalid syntax in an HTTP header name
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010020187 containing unauthorized characters. It is also possible but quite
20188 rare, that the proxy blocked a chunked-encoding request from the
20189 client due to an invalid syntax, before the server responded. In this
20190 case, an HTTP 400 error is sent to the client and reported in the
20191 logs.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020192
20193 PR The proxy blocked the client's HTTP request, either because of an
20194 invalid HTTP syntax, in which case it returned an HTTP 400 error to
20195 the client, or because a deny filter matched, in which case it
20196 returned an HTTP 403 error.
20197
20198 PT The proxy blocked the client's request and has tarpitted its
20199 connection before returning it a 500 server error. Nothing was sent
20200 to the server. The connection was maintained open for as long as
20201 reported by the "Tw" timer field.
20202
20203 RC A local resource has been exhausted (memory, sockets, source ports)
20204 preventing the connection to the server from establishing. The error
20205 logs will tell precisely what was missing. This is very rare and can
20206 only be solved by proper system tuning.
20207
Willy Tarreau996a92c2010-10-13 19:30:47 +020020208The combination of the two last flags gives a lot of information about how
20209persistence was handled by the client, the server and by haproxy. This is very
20210important to troubleshoot disconnections, when users complain they have to
20211re-authenticate. The commonly encountered flags are :
20212
20213 -- Persistence cookie is not enabled.
20214
20215 NN No cookie was provided by the client, none was inserted in the
20216 response. For instance, this can be in insert mode with "postonly"
20217 set on a GET request.
20218
20219 II A cookie designating an invalid server was provided by the client,
20220 a valid one was inserted in the response. This typically happens when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040020221 a "server" entry is removed from the configuration, since its cookie
Willy Tarreau996a92c2010-10-13 19:30:47 +020020222 value can be presented by a client when no other server knows it.
20223
20224 NI No cookie was provided by the client, one was inserted in the
20225 response. This typically happens for first requests from every user
20226 in "insert" mode, which makes it an easy way to count real users.
20227
20228 VN A cookie was provided by the client, none was inserted in the
20229 response. This happens for most responses for which the client has
20230 already got a cookie.
20231
20232 VU A cookie was provided by the client, with a last visit date which is
20233 not completely up-to-date, so an updated cookie was provided in
20234 response. This can also happen if there was no date at all, or if
20235 there was a date but the "maxidle" parameter was not set, so that the
20236 cookie can be switched to unlimited time.
20237
20238 EI A cookie was provided by the client, with a last visit date which is
20239 too old for the "maxidle" parameter, so the cookie was ignored and a
20240 new cookie was inserted in the response.
20241
20242 OI A cookie was provided by the client, with a first visit date which is
20243 too old for the "maxlife" parameter, so the cookie was ignored and a
20244 new cookie was inserted in the response.
20245
20246 DI The server designated by the cookie was down, a new server was
20247 selected and a new cookie was emitted in the response.
20248
20249 VI The server designated by the cookie was not marked dead but could not
20250 be reached. A redispatch happened and selected another one, which was
20251 then advertised in the response.
20252
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020253
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200202548.6. Non-printable characters
20255-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020256
20257In order not to cause trouble to log analysis tools or terminals during log
20258consulting, non-printable characters are not sent as-is into log files, but are
20259converted to the two-digits hexadecimal representation of their ASCII code,
20260prefixed by the character '#'. The only characters that can be logged without
20261being escaped are comprised between 32 and 126 (inclusive). Obviously, the
20262escape character '#' itself is also encoded to avoid any ambiguity ("#23"). It
20263is the same for the character '"' which becomes "#22", as well as '{', '|' and
20264'}' when logging headers.
20265
20266Note that the space character (' ') is not encoded in headers, which can cause
20267issues for tools relying on space count to locate fields. A typical header
20268containing spaces is "User-Agent".
20269
20270Last, it has been observed that some syslog daemons such as syslog-ng escape
20271the quote ('"') with a backslash ('\'). The reverse operation can safely be
20272performed since no quote may appear anywhere else in the logs.
20273
20274
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200202758.7. Capturing HTTP cookies
20276---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020277
20278Cookie capture simplifies the tracking a complete user session. This can be
20279achieved using the "capture cookie" statement in the frontend. Please refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020280section 4.2 for more details. Only one cookie can be captured, and the same
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020281cookie will simultaneously be checked in the request ("Cookie:" header) and in
20282the response ("Set-Cookie:" header). The respective values will be reported in
20283the HTTP logs at the "captured_request_cookie" and "captured_response_cookie"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020284locations (see section 8.2.3 about HTTP log format). When either cookie is
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020285not seen, a dash ('-') replaces the value. This way, it's easy to detect when a
20286user switches to a new session for example, because the server will reassign it
20287a new cookie. It is also possible to detect if a server unexpectedly sets a
20288wrong cookie to a client, leading to session crossing.
20289
20290 Examples :
20291 # capture the first cookie whose name starts with "ASPSESSION"
20292 capture cookie ASPSESSION len 32
20293
20294 # capture the first cookie whose name is exactly "vgnvisitor"
20295 capture cookie vgnvisitor= len 32
20296
20297
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200202988.8. Capturing HTTP headers
20299---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020300
20301Header captures are useful to track unique request identifiers set by an upper
20302proxy, virtual host names, user-agents, POST content-length, referrers, etc. In
20303the response, one can search for information about the response length, how the
20304server asked the cache to behave, or an object location during a redirection.
20305
20306Header captures are performed using the "capture request header" and "capture
20307response header" statements in the frontend. Please consult their definition in
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020308section 4.2 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020309
20310It is possible to include both request headers and response headers at the same
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010020311time. Non-existent headers are logged as empty strings, and if one header
20312appears more than once, only its last occurrence will be logged. Request headers
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020313are grouped within braces '{' and '}' in the same order as they were declared,
20314and delimited with a vertical bar '|' without any space. Response headers
20315follow the same representation, but are displayed after a space following the
20316request headers block. These blocks are displayed just before the HTTP request
20317in the logs.
20318
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020020319As a special case, it is possible to specify an HTTP header capture in a TCP
20320frontend. The purpose is to enable logging of headers which will be parsed in
20321an HTTP backend if the request is then switched to this HTTP backend.
20322
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020323 Example :
20324 # This instance chains to the outgoing proxy
20325 listen proxy-out
20326 mode http
20327 option httplog
20328 option logasap
20329 log global
20330 server cache1 192.168.1.1:3128
20331
20332 # log the name of the virtual server
20333 capture request header Host len 20
20334
20335 # log the amount of data uploaded during a POST
20336 capture request header Content-Length len 10
20337
20338 # log the beginning of the referrer
20339 capture request header Referer len 20
20340
20341 # server name (useful for outgoing proxies only)
20342 capture response header Server len 20
20343
20344 # logging the content-length is useful with "option logasap"
20345 capture response header Content-Length len 10
20346
Davor Ocelice9ed2812017-12-25 17:49:28 +010020347 # log the expected cache behavior on the response
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020348 capture response header Cache-Control len 8
20349
20350 # the Via header will report the next proxy's name
20351 capture response header Via len 20
20352
20353 # log the URL location during a redirection
20354 capture response header Location len 20
20355
20356 >>> Aug 9 20:26:09 localhost \
20357 haproxy[2022]: 127.0.0.1:34014 [09/Aug/2004:20:26:09] proxy-out \
20358 proxy-out/cache1 0/0/0/162/+162 200 +350 - - ---- 0/0/0/0/0 0/0 \
20359 {fr.adserver.yahoo.co||http://fr.f416.mail.} {|864|private||} \
20360 "GET http://fr.adserver.yahoo.com/"
20361
20362 >>> Aug 9 20:30:46 localhost \
20363 haproxy[2022]: 127.0.0.1:34020 [09/Aug/2004:20:30:46] proxy-out \
20364 proxy-out/cache1 0/0/0/182/+182 200 +279 - - ---- 0/0/0/0/0 0/0 \
20365 {w.ods.org||} {Formilux/0.1.8|3495|||} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010020366 "GET http://trafic.1wt.eu/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020367
20368 >>> Aug 9 20:30:46 localhost \
20369 haproxy[2022]: 127.0.0.1:34028 [09/Aug/2004:20:30:46] proxy-out \
20370 proxy-out/cache1 0/0/2/126/+128 301 +223 - - ---- 0/0/0/0/0 0/0 \
20371 {www.sytadin.equipement.gouv.fr||http://trafic.1wt.eu/} \
20372 {Apache|230|||http://www.sytadin.} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010020373 "GET http://www.sytadin.equipement.gouv.fr/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020374
20375
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200203768.9. Examples of logs
20377---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020378
20379These are real-world examples of logs accompanied with an explanation. Some of
20380them have been made up by hand. The syslog part has been removed for better
20381reading. Their sole purpose is to explain how to decipher them.
20382
20383 >>> haproxy[674]: 127.0.0.1:33318 [15/Oct/2003:08:31:57.130] px-http \
20384 px-http/srv1 6559/0/7/147/6723 200 243 - - ---- 5/3/3/1/0 0/0 \
20385 "HEAD / HTTP/1.0"
20386
20387 => long request (6.5s) entered by hand through 'telnet'. The server replied
20388 in 147 ms, and the session ended normally ('----')
20389
20390 >>> haproxy[674]: 127.0.0.1:33319 [15/Oct/2003:08:31:57.149] px-http \
20391 px-http/srv1 6559/1230/7/147/6870 200 243 - - ---- 324/239/239/99/0 \
20392 0/9 "HEAD / HTTP/1.0"
20393
20394 => Idem, but the request was queued in the global queue behind 9 other
20395 requests, and waited there for 1230 ms.
20396
20397 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.654] px-http \
20398 px-http/srv1 9/0/7/14/+30 200 +243 - - ---- 3/3/3/1/0 0/0 \
20399 "GET /image.iso HTTP/1.0"
20400
20401 => request for a long data transfer. The "logasap" option was specified, so
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010020402 the log was produced just before transferring data. The server replied in
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020403 14 ms, 243 bytes of headers were sent to the client, and total time from
20404 accept to first data byte is 30 ms.
20405
20406 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.925] px-http \
20407 px-http/srv1 9/0/7/14/30 502 243 - - PH-- 3/2/2/0/0 0/0 \
20408 "GET /cgi-bin/bug.cgi? HTTP/1.0"
20409
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020020410 => the proxy blocked a server response either because of an "http-response
20411 deny" rule, or because the response was improperly formatted and not
20412 HTTP-compliant, or because it blocked sensitive information which risked
20413 being cached. In this case, the response is replaced with a "502 bad
20414 gateway". The flags ("PH--") tell us that it was haproxy who decided to
20415 return the 502 and not the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020416
20417 >>> haproxy[18113]: 127.0.0.1:34548 [15/Oct/2003:15:18:55.798] px-http \
Willy Tarreaud72758d2010-01-12 10:42:19 +010020418 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 +010020419
20420 => the client never completed its request and aborted itself ("C---") after
20421 8.5s, while the proxy was waiting for the request headers ("-R--").
20422 Nothing was sent to any server.
20423
20424 >>> haproxy[18113]: 127.0.0.1:34549 [15/Oct/2003:15:19:06.103] px-http \
20425 px-http/<NOSRV> -1/-1/-1/-1/50001 408 0 - - cR-- 2/2/2/0/0 0/0 ""
20426
20427 => The client never completed its request, which was aborted by the
20428 time-out ("c---") after 50s, while the proxy was waiting for the request
Davor Ocelice9ed2812017-12-25 17:49:28 +010020429 headers ("-R--"). Nothing was sent to any server, but the proxy could
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020430 send a 408 return code to the client.
20431
20432 >>> haproxy[18989]: 127.0.0.1:34550 [15/Oct/2003:15:24:28.312] px-tcp \
20433 px-tcp/srv1 0/0/5007 0 cD 0/0/0/0/0 0/0
20434
20435 => This log was produced with "option tcplog". The client timed out after
20436 5 seconds ("c----").
20437
20438 >>> haproxy[18989]: 10.0.0.1:34552 [15/Oct/2003:15:26:31.462] px-http \
20439 px-http/srv1 3183/-1/-1/-1/11215 503 0 - - SC-- 205/202/202/115/3 \
Willy Tarreaud72758d2010-01-12 10:42:19 +010020440 0/0 "HEAD / HTTP/1.0"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020441
20442 => The request took 3s to complete (probably a network problem), and the
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020443 connection to the server failed ('SC--') after 4 attempts of 2 seconds
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020444 (config says 'retries 3'), and no redispatch (otherwise we would have
20445 seen "/+3"). Status code 503 was returned to the client. There were 115
20446 connections on this server, 202 connections on this proxy, and 205 on
20447 the global process. It is possible that the server refused the
20448 connection because of too many already established.
Willy Tarreau844e3c52008-01-11 16:28:18 +010020449
Willy Tarreau52b2d222011-09-07 23:48:48 +020020450
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200204519. Supported filters
20452--------------------
20453
20454Here are listed officially supported filters with the list of parameters they
20455accept. Depending on compile options, some of these filters might be
20456unavailable. The list of available filters is reported in haproxy -vv.
20457
20458See also : "filter"
20459
204609.1. Trace
20461----------
20462
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010020463filter trace [name <name>] [random-parsing] [random-forwarding] [hexdump]
Christopher Fauletc3fe5332016-04-07 15:30:10 +020020464
20465 Arguments:
20466 <name> is an arbitrary name that will be reported in
20467 messages. If no name is provided, "TRACE" is used.
20468
20469 <random-parsing> enables the random parsing of data exchanged between
20470 the client and the server. By default, this filter
20471 parses all available data. With this parameter, it
20472 only parses a random amount of the available data.
20473
Davor Ocelice9ed2812017-12-25 17:49:28 +010020474 <random-forwarding> enables the random forwarding of parsed data. By
Christopher Fauletc3fe5332016-04-07 15:30:10 +020020475 default, this filter forwards all previously parsed
20476 data. With this parameter, it only forwards a random
20477 amount of the parsed data.
20478
Davor Ocelice9ed2812017-12-25 17:49:28 +010020479 <hexdump> dumps all forwarded data to the server and the client.
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010020480
Christopher Fauletc3fe5332016-04-07 15:30:10 +020020481This filter can be used as a base to develop new filters. It defines all
20482callbacks and print a message on the standard error stream (stderr) with useful
20483information for all of them. It may be useful to debug the activity of other
20484filters or, quite simply, HAProxy's activity.
20485
20486Using <random-parsing> and/or <random-forwarding> parameters is a good way to
20487tests the behavior of a filter that parses data exchanged between a client and
20488a server by adding some latencies in the processing.
20489
20490
204919.2. HTTP compression
20492---------------------
20493
20494filter compression
20495
20496The HTTP compression has been moved in a filter in HAProxy 1.7. "compression"
20497keyword must still be used to enable and configure the HTTP compression. And
Christopher Fauletb30b3102019-09-12 23:03:09 +020020498when no other filter is used, it is enough. When used with the cache or the
20499fcgi-app enabled, it is also enough. In this case, the compression is always
20500done after the response is stored in the cache. But it is mandatory to
20501explicitly use a filter line to enable the HTTP compression when at least one
20502filter other than the cache or the fcgi-app is used for the same
20503listener/frontend/backend. This is important to know the filters evaluation
20504order.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020020505
Christopher Fauletb30b3102019-09-12 23:03:09 +020020506See also : "compression", section 9.4 about the cache filter and section 9.5
20507 about the fcgi-app filter.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020020508
20509
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +0200205109.3. Stream Processing Offload Engine (SPOE)
20511--------------------------------------------
20512
20513filter spoe [engine <name>] config <file>
20514
20515 Arguments :
20516
20517 <name> is the engine name that will be used to find the right scope in
20518 the configuration file. If not provided, all the file will be
20519 parsed.
20520
20521 <file> is the path of the engine configuration file. This file can
20522 contain configuration of several engines. In this case, each
20523 part must be placed in its own scope.
20524
20525The Stream Processing Offload Engine (SPOE) is a filter communicating with
20526external components. It allows the offload of some specifics processing on the
Davor Ocelice9ed2812017-12-25 17:49:28 +010020527streams in tiered applications. These external components and information
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020020528exchanged with them are configured in dedicated files, for the main part. It
20529also requires dedicated backends, defined in HAProxy configuration.
20530
20531SPOE communicates with external components using an in-house binary protocol,
20532the Stream Processing Offload Protocol (SPOP).
20533
Tim Düsterhus4896c442016-11-29 02:15:19 +010020534For all information about the SPOE configuration and the SPOP specification, see
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020020535"doc/SPOE.txt".
20536
Christopher Faulet99a17a22018-12-11 09:18:27 +0100205379.4. Cache
20538----------
20539
20540filter cache <name>
20541
20542 Arguments :
20543
20544 <name> is name of the cache section this filter will use.
20545
20546The cache uses a filter to store cacheable responses. The HTTP rules
20547"cache-store" and "cache-use" must be used to define how and when to use a
John Roeslerfb2fce12019-07-10 15:45:51 -050020548cache. By default the corresponding filter is implicitly defined. And when no
Christopher Fauletb30b3102019-09-12 23:03:09 +020020549other filters than fcgi-app or compression are used, it is enough. In such
20550case, the compression filter is always evaluated after the cache filter. But it
20551is mandatory to explicitly use a filter line to use a cache when at least one
20552filter other than the compression or the fcgi-app is used for the same
Christopher Faulet27d93c32018-12-15 22:32:02 +010020553listener/frontend/backend. This is important to know the filters evaluation
20554order.
Christopher Faulet99a17a22018-12-11 09:18:27 +010020555
Christopher Fauletb30b3102019-09-12 23:03:09 +020020556See also : section 9.2 about the compression filter, section 9.5 about the
20557 fcgi-app filter and section 6 about cache.
20558
20559
205609.5. Fcgi-app
20561-------------
20562
Daniel Corbett67a82712020-07-06 23:01:19 -040020563filter fcgi-app <name>
Christopher Fauletb30b3102019-09-12 23:03:09 +020020564
20565 Arguments :
20566
20567 <name> is name of the fcgi-app section this filter will use.
20568
20569The FastCGI application uses a filter to evaluate all custom parameters on the
20570request path, and to process the headers on the response path. the <name> must
20571reference an existing fcgi-app section. The directive "use-fcgi-app" should be
20572used to define the application to use. By default the corresponding filter is
20573implicitly defined. And when no other filters than cache or compression are
20574used, it is enough. But it is mandatory to explicitly use a filter line to a
20575fcgi-app when at least one filter other than the compression or the cache is
20576used for the same backend. This is important to know the filters evaluation
20577order.
20578
20579See also: "use-fcgi-app", section 9.2 about the compression filter, section 9.4
20580 about the cache filter and section 10 about FastCGI application.
20581
20582
2058310. FastCGI applications
20584-------------------------
20585
20586HAProxy is able to send HTTP requests to Responder FastCGI applications. This
20587feature was added in HAProxy 2.1. To do so, servers must be configured to use
20588the FastCGI protocol (using the keyword "proto fcgi" on the server line) and a
20589FastCGI application must be configured and used by the backend managing these
20590servers (using the keyword "use-fcgi-app" into the proxy section). Several
20591FastCGI applications may be defined, but only one can be used at a time by a
20592backend.
20593
20594HAProxy implements all features of the FastCGI specification for Responder
20595application. Especially it is able to multiplex several requests on a simple
20596connection.
20597
2059810.1. Setup
20599-----------
20600
2060110.1.1. Fcgi-app section
20602--------------------------
20603
20604fcgi-app <name>
20605 Declare a FastCGI application named <name>. To be valid, at least the
20606 document root must be defined.
20607
20608acl <aclname> <criterion> [flags] [operator] <value> ...
20609 Declare or complete an access list.
20610
20611 See "acl" keyword in section 4.2 and section 7 about ACL usage for
20612 details. ACLs defined for a FastCGI application are private. They cannot be
20613 used by any other application or by any proxy. In the same way, ACLs defined
20614 in any other section are not usable by a FastCGI application. However,
20615 Pre-defined ACLs are available.
20616
20617docroot <path>
20618 Define the document root on the remote host. <path> will be used to build
20619 the default value of FastCGI parameters SCRIPT_FILENAME and
20620 PATH_TRANSLATED. It is a mandatory setting.
20621
20622index <script-name>
20623 Define the script name that will be appended after an URI that ends with a
20624 slash ("/") to set the default value of the FastCGI parameter SCRIPT_NAME. It
20625 is an optional setting.
20626
20627 Example :
20628 index index.php
20629
20630log-stderr global
20631log-stderr <address> [len <length>] [format <format>]
Jan Wagnerf2f5c4e2020-12-17 22:22:32 +010020632 [sample <ranges>:<sample_size>] <facility> [<level> [<minlevel>]]
Christopher Fauletb30b3102019-09-12 23:03:09 +020020633 Enable logging of STDERR messages reported by the FastCGI application.
20634
20635 See "log" keyword in section 4.2 for details. It is an optional setting. By
20636 default STDERR messages are ignored.
20637
20638pass-header <name> [ { if | unless } <condition> ]
20639 Specify the name of a request header which will be passed to the FastCGI
20640 application. It may optionally be followed by an ACL-based condition, in
20641 which case it will only be evaluated if the condition is true.
20642
20643 Most request headers are already available to the FastCGI application,
20644 prefixed with "HTTP_". Thus, this directive is only required to pass headers
20645 that are purposefully omitted. Currently, the headers "Authorization",
20646 "Proxy-Authorization" and hop-by-hop headers are omitted.
20647
20648 Note that the headers "Content-type" and "Content-length" are never passed to
20649 the FastCGI application because they are already converted into parameters.
20650
20651path-info <regex>
Christopher Faulet28cb3662020-02-14 14:47:37 +010020652 Define a regular expression to extract the script-name and the path-info from
Christopher Faulet6c57f2d2020-02-14 16:55:52 +010020653 the URL-decoded path. Thus, <regex> may have two captures: the first one to
20654 capture the script name and the second one to capture the path-info. The
20655 first one is mandatory, the second one is optional. This way, it is possible
20656 to extract the script-name from the path ignoring the path-info. It is an
20657 optional setting. If it is not defined, no matching is performed on the
20658 path. and the FastCGI parameters PATH_INFO and PATH_TRANSLATED are not
20659 filled.
Christopher Faulet28cb3662020-02-14 14:47:37 +010020660
20661 For security reason, when this regular expression is defined, the newline and
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050020662 the null characters are forbidden from the path, once URL-decoded. The reason
Christopher Faulet28cb3662020-02-14 14:47:37 +010020663 to such limitation is because otherwise the matching always fails (due to a
20664 limitation one the way regular expression are executed in HAProxy). So if one
20665 of these two characters is found in the URL-decoded path, an error is
20666 returned to the client. The principle of least astonishment is applied here.
Christopher Fauletb30b3102019-09-12 23:03:09 +020020667
20668 Example :
Christopher Faulet6c57f2d2020-02-14 16:55:52 +010020669 path-info ^(/.+\.php)(/.*)?$ # both script-name and path-info may be set
20670 path-info ^(/.+\.php) # the path-info is ignored
Christopher Fauletb30b3102019-09-12 23:03:09 +020020671
20672option get-values
20673no option get-values
20674 Enable or disable the retrieve of variables about connection management.
20675
Daniel Corbett67a82712020-07-06 23:01:19 -040020676 HAProxy is able to send the record FCGI_GET_VALUES on connection
Christopher Fauletb30b3102019-09-12 23:03:09 +020020677 establishment to retrieve the value for following variables:
20678
20679 * FCGI_MAX_REQS The maximum number of concurrent requests this
20680 application will accept.
20681
William Lallemand93e548e2019-09-30 13:54:02 +020020682 * FCGI_MPXS_CONNS "0" if this application does not multiplex connections,
20683 "1" otherwise.
Christopher Fauletb30b3102019-09-12 23:03:09 +020020684
20685 Some FastCGI applications does not support this feature. Some others close
Ilya Shipitsin11057a32020-06-21 21:18:27 +050020686 the connection immediately after sending their response. So, by default, this
Christopher Fauletb30b3102019-09-12 23:03:09 +020020687 option is disabled.
20688
20689 Note that the maximum number of concurrent requests accepted by a FastCGI
20690 application is a connection variable. It only limits the number of streams
20691 per connection. If the global load must be limited on the application, the
20692 server parameters "maxconn" and "pool-max-conn" must be set. In addition, if
20693 an application does not support connection multiplexing, the maximum number
20694 of concurrent requests is automatically set to 1.
20695
20696option keep-conn
20697no option keep-conn
20698 Instruct the FastCGI application to keep the connection open or not after
20699 sending a response.
20700
20701 If disabled, the FastCGI application closes the connection after responding
20702 to this request. By default, this option is enabled.
20703
20704option max-reqs <reqs>
20705 Define the maximum number of concurrent requests this application will
20706 accept.
20707
20708 This option may be overwritten if the variable FCGI_MAX_REQS is retrieved
20709 during connection establishment. Furthermore, if the application does not
20710 support connection multiplexing, this option will be ignored. By default set
20711 to 1.
20712
20713option mpxs-conns
20714no option mpxs-conns
20715 Enable or disable the support of connection multiplexing.
20716
20717 This option may be overwritten if the variable FCGI_MPXS_CONNS is retrieved
20718 during connection establishment. It is disabled by default.
20719
20720set-param <name> <fmt> [ { if | unless } <condition> ]
20721 Set a FastCGI parameter that should be passed to this application. Its
20722 value, defined by <fmt> must follows the log-format rules (see section 8.2.4
20723 "Custom Log format"). It may optionally be followed by an ACL-based
20724 condition, in which case it will only be evaluated if the condition is true.
20725
20726 With this directive, it is possible to overwrite the value of default FastCGI
20727 parameters. If the value is evaluated to an empty string, the rule is
20728 ignored. These directives are evaluated in their declaration order.
20729
20730 Example :
20731 # PHP only, required if PHP was built with --enable-force-cgi-redirect
20732 set-param REDIRECT_STATUS 200
20733
20734 set-param PHP_AUTH_DIGEST %[req.hdr(Authorization)]
20735
20736
2073710.1.2. Proxy section
20738---------------------
20739
20740use-fcgi-app <name>
20741 Define the FastCGI application to use for the backend.
20742
20743 Arguments :
20744 <name> is the name of the FastCGI application to use.
20745
20746 This keyword is only available for HTTP proxies with the backend capability
20747 and with at least one FastCGI server. However, FastCGI servers can be mixed
20748 with HTTP servers. But except there is a good reason to do so, it is not
20749 recommended (see section 10.3 about the limitations for details). Only one
20750 application may be defined at a time per backend.
20751
20752 Note that, once a FastCGI application is referenced for a backend, depending
20753 on the configuration some processing may be done even if the request is not
20754 sent to a FastCGI server. Rules to set parameters or pass headers to an
20755 application are evaluated.
20756
20757
2075810.1.3. Example
20759---------------
20760
20761 frontend front-http
20762 mode http
20763 bind *:80
20764 bind *:
20765
20766 use_backend back-dynamic if { path_reg ^/.+\.php(/.*)?$ }
20767 default_backend back-static
20768
20769 backend back-static
20770 mode http
20771 server www A.B.C.D:80
20772
20773 backend back-dynamic
20774 mode http
20775 use-fcgi-app php-fpm
20776 server php-fpm A.B.C.D:9000 proto fcgi
20777
20778 fcgi-app php-fpm
20779 log-stderr global
20780 option keep-conn
20781
20782 docroot /var/www/my-app
20783 index index.php
20784 path-info ^(/.+\.php)(/.*)?$
20785
20786
2078710.2. Default parameters
20788------------------------
20789
20790A Responder FastCGI application has the same purpose as a CGI/1.1 program. In
20791the CGI/1.1 specification (RFC3875), several variables must be passed to the
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050020792script. So HAProxy set them and some others commonly used by FastCGI
Christopher Fauletb30b3102019-09-12 23:03:09 +020020793applications. All these variables may be overwritten, with caution though.
20794
20795 +-------------------+-----------------------------------------------------+
20796 | AUTH_TYPE | Identifies the mechanism, if any, used by HAProxy |
20797 | | to authenticate the user. Concretely, only the |
20798 | | BASIC authentication mechanism is supported. |
20799 | | |
20800 +-------------------+-----------------------------------------------------+
20801 | CONTENT_LENGTH | Contains the size of the message-body attached to |
20802 | | the request. It means only requests with a known |
20803 | | size are considered as valid and sent to the |
20804 | | application. |
20805 | | |
20806 +-------------------+-----------------------------------------------------+
20807 | CONTENT_TYPE | Contains the type of the message-body attached to |
20808 | | the request. It may not be set. |
20809 | | |
20810 +-------------------+-----------------------------------------------------+
20811 | DOCUMENT_ROOT | Contains the document root on the remote host under |
20812 | | which the script should be executed, as defined in |
20813 | | the application's configuration. |
20814 | | |
20815 +-------------------+-----------------------------------------------------+
20816 | GATEWAY_INTERFACE | Contains the dialect of CGI being used by HAProxy |
20817 | | to communicate with the FastCGI application. |
20818 | | Concretely, it is set to "CGI/1.1". |
20819 | | |
20820 +-------------------+-----------------------------------------------------+
20821 | PATH_INFO | Contains the portion of the URI path hierarchy |
20822 | | following the part that identifies the script |
20823 | | itself. To be set, the directive "path-info" must |
20824 | | be defined. |
20825 | | |
20826 +-------------------+-----------------------------------------------------+
20827 | PATH_TRANSLATED | If PATH_INFO is set, it is its translated version. |
20828 | | It is the concatenation of DOCUMENT_ROOT and |
20829 | | PATH_INFO. If PATH_INFO is not set, this parameters |
20830 | | is not set too. |
20831 | | |
20832 +-------------------+-----------------------------------------------------+
20833 | QUERY_STRING | Contains the request's query string. It may not be |
20834 | | set. |
20835 | | |
20836 +-------------------+-----------------------------------------------------+
20837 | REMOTE_ADDR | Contains the network address of the client sending |
20838 | | the request. |
20839 | | |
20840 +-------------------+-----------------------------------------------------+
20841 | REMOTE_USER | Contains the user identification string supplied by |
20842 | | client as part of user authentication. |
20843 | | |
20844 +-------------------+-----------------------------------------------------+
20845 | REQUEST_METHOD | Contains the method which should be used by the |
20846 | | script to process the request. |
20847 | | |
20848 +-------------------+-----------------------------------------------------+
20849 | REQUEST_URI | Contains the request's URI. |
20850 | | |
20851 +-------------------+-----------------------------------------------------+
20852 | SCRIPT_FILENAME | Contains the absolute pathname of the script. it is |
20853 | | the concatenation of DOCUMENT_ROOT and SCRIPT_NAME. |
20854 | | |
20855 +-------------------+-----------------------------------------------------+
20856 | SCRIPT_NAME | Contains the name of the script. If the directive |
20857 | | "path-info" is defined, it is the first part of the |
20858 | | URI path hierarchy, ending with the script name. |
20859 | | Otherwise, it is the entire URI path. |
20860 | | |
20861 +-------------------+-----------------------------------------------------+
20862 | SERVER_NAME | Contains the name of the server host to which the |
20863 | | client request is directed. It is the value of the |
20864 | | header "Host", if defined. Otherwise, the |
20865 | | destination address of the connection on the client |
20866 | | side. |
20867 | | |
20868 +-------------------+-----------------------------------------------------+
20869 | SERVER_PORT | Contains the destination TCP port of the connection |
20870 | | on the client side, which is the port the client |
20871 | | connected to. |
20872 | | |
20873 +-------------------+-----------------------------------------------------+
20874 | SERVER_PROTOCOL | Contains the request's protocol. |
20875 | | |
20876 +-------------------+-----------------------------------------------------+
Christopher Faulet7e0d0d82021-06-11 13:34:42 +020020877 | SERVER_SOFTWARE | Contains the string "HAProxy" followed by the |
20878 | | current HAProxy version. |
20879 | | |
20880 +-------------------+-----------------------------------------------------+
Christopher Fauletb30b3102019-09-12 23:03:09 +020020881 | HTTPS | Set to a non-empty value ("on") if the script was |
20882 | | queried through the HTTPS protocol. |
20883 | | |
20884 +-------------------+-----------------------------------------------------+
20885
20886
2088710.3. Limitations
20888------------------
20889
20890The current implementation have some limitations. The first one is about the
20891way some request headers are hidden to the FastCGI applications. This happens
20892during the headers analysis, on the backend side, before the connection
20893establishment. At this stage, HAProxy know the backend is using a FastCGI
20894application but it don't know if the request will be routed to a FastCGI server
20895or not. But to hide request headers, it simply removes them from the HTX
20896message. So, if the request is finally routed to an HTTP server, it never see
20897these headers. For this reason, it is not recommended to mix FastCGI servers
20898and HTTP servers under the same backend.
20899
20900Similarly, the rules "set-param" and "pass-header" are evaluated during the
20901request headers analysis. So the evaluation is always performed, even if the
20902requests is finally forwarded to an HTTP server.
20903
20904About the rules "set-param", when a rule is applied, a pseudo header is added
20905into the HTX message. So, the same way than for HTTP header rewrites, it may
20906fail if the buffer is full. The rules "set-param" will compete with
20907"http-request" ones.
20908
20909Finally, all FastCGI params and HTTP headers are sent into a unique record
20910FCGI_PARAM. Encoding of this record must be done in one pass, otherwise a
20911processing error is returned. It means the record FCGI_PARAM, once encoded,
20912must not exceeds the size of a buffer. However, there is no reserve to respect
20913here.
William Lallemand86d0df02017-11-24 21:36:45 +010020914
Willy Tarreau0ba27502007-12-24 16:55:16 +010020915/*
20916 * Local variables:
20917 * fill-column: 79
20918 * End:
20919 */