blob: 083f1f60055ee4796b4908c21db4a20e19fab851 [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 Faulet05f01882020-09-25 18:40:47 +02007 2020/09/25
Willy Tarreau6a06a402007-07-15 20:15:28 +02008
9
10This document covers the configuration language as implemented in the version
Davor Ocelice9ed2812017-12-25 17:49:28 +010011specified above. It does not provide any hints, examples, or advice. For such
Willy Tarreau0ba27502007-12-24 16:55:16 +010012documentation, please refer to the Reference Manual or the Architecture Manual.
Davor Ocelice9ed2812017-12-25 17:49:28 +010013The summary below is meant to help you find sections by name and navigate
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014through the document.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016Note to documentation contributors :
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017 This document is formatted with 80 columns per line, with even number of
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018 spaces for indentation and without tabs. Please follow these rules strictly
19 so that it remains easily printable everywhere. If a line needs to be
20 printed verbatim and does not fit, please end each line with a backslash
Willy Tarreau62a36c42010-08-17 15:53:10 +020021 ('\') and continue on next line, indented by two characters. It is also
Davor Ocelice9ed2812017-12-25 17:49:28 +010022 sometimes useful to prefix all output lines (logs, console outputs) with 3
23 closing angle brackets ('>>>') in order to emphasize the difference between
24 inputs and outputs when they may be ambiguous. If you add sections,
Willy Tarreau62a36c42010-08-17 15:53:10 +020025 please update the summary below for easier searching.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020026
27
28Summary
29-------
30
311. Quick reminder about HTTP
321.1. The HTTP transaction model
331.2. HTTP request
Davor Ocelice9ed2812017-12-25 17:49:28 +0100341.2.1. The request line
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200351.2.2. The request headers
361.3. HTTP response
Davor Ocelice9ed2812017-12-25 17:49:28 +0100371.3.1. The response line
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200381.3.2. The response headers
39
402. Configuring HAProxy
412.1. Configuration file format
William Lallemandf9873ba2015-05-05 17:37:14 +0200422.2. Quoting and escaping
William Lallemandb2f07452015-05-12 14:27:13 +0200432.3. Environment variables
442.4. Time format
452.5. Examples
Willy Tarreauc57f0e22009-05-10 13:12:33 +020046
473. Global parameters
483.1. Process management and security
493.2. Performance tuning
503.3. Debugging
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +0100513.4. Userlists
Cyril Bontédc4d9032012-04-08 21:57:39 +0200523.5. Peers
Cyril Bonté307ee1e2015-09-28 23:16:06 +0200533.6. Mailers
William Lallemandc9515522019-06-12 16:32:11 +0200543.7. Programs
Christopher Faulet76edc0f2020-01-13 15:52:01 +0100553.8. HTTP-errors
Emeric Brun99c453d2020-05-25 15:01:04 +0200563.9. Rings
Willy Tarreauc57f0e22009-05-10 13:12:33 +020057
584. Proxies
594.1. Proxy keywords matrix
604.2. Alphabetically sorted keywords reference
61
Davor Ocelice9ed2812017-12-25 17:49:28 +0100625. Bind and server options
Willy Tarreau086fbf52012-09-24 20:34:51 +0200635.1. Bind options
645.2. Server and default-server options
Baptiste Assmann1fa66662015-04-14 00:28:47 +0200655.3. Server DNS resolution
665.3.1. Global overview
675.3.2. The resolvers section
Willy Tarreauc57f0e22009-05-10 13:12:33 +020068
Julien Pivotto6ccee412019-11-27 15:49:54 +0100696. Cache
706.1. Limitation
716.2. Setup
726.2.1. Cache section
736.2.2. Proxy section
74
Willy Tarreau74ca5042013-06-11 23:12:07 +0200757. Using ACLs and fetching samples
767.1. ACL basics
777.1.1. Matching booleans
787.1.2. Matching integers
797.1.3. Matching strings
807.1.4. Matching regular expressions (regexes)
817.1.5. Matching arbitrary data blocks
827.1.6. Matching IPv4 and IPv6 addresses
837.2. Using ACLs to form conditions
847.3. Fetching samples
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200857.3.1. Converters
867.3.2. Fetching samples from internal states
877.3.3. Fetching samples at Layer 4
887.3.4. Fetching samples at Layer 5
897.3.5. Fetching samples from buffer contents (Layer 6)
907.3.6. Fetching HTTP samples (Layer 7)
Christopher Faulete596d182020-05-05 17:46:34 +0200917.3.7. Fetching samples for developers
Willy Tarreau74ca5042013-06-11 23:12:07 +0200927.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020093
948. Logging
958.1. Log levels
968.2. Log formats
978.2.1. Default log format
988.2.2. TCP log format
998.2.3. HTTP log format
William Lallemand48940402012-01-30 16:47:22 +01001008.2.4. Custom log format
Willy Tarreau5f51e1a2012-12-03 18:40:10 +01001018.2.5. Error log format
Willy Tarreauc57f0e22009-05-10 13:12:33 +02001028.3. Advanced logging options
1038.3.1. Disabling logging of external tests
1048.3.2. Logging before waiting for the session to terminate
1058.3.3. Raising log level upon errors
1068.3.4. Disabling logging of successful connections
1078.4. Timing events
1088.5. Session state at disconnection
1098.6. Non-printable characters
1108.7. Capturing HTTP cookies
1118.8. Capturing HTTP headers
1128.9. Examples of logs
113
Christopher Fauletc3fe5332016-04-07 15:30:10 +02001149. Supported filters
1159.1. Trace
1169.2. HTTP compression
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +02001179.3. Stream Processing Offload Engine (SPOE)
Christopher Faulet99a17a22018-12-11 09:18:27 +01001189.4. Cache
Christopher Fauletb30b3102019-09-12 23:03:09 +02001199.5. fcgi-app
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200120
Christopher Fauletb30b3102019-09-12 23:03:09 +020012110. FastCGI applications
12210.1. Setup
12310.1.1. Fcgi-app section
12410.1.2. Proxy section
12510.1.3. Example
12610.2. Default parameters
12710.3. Limitations
128
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200129
1301. Quick reminder about HTTP
131----------------------------
132
Davor Ocelice9ed2812017-12-25 17:49:28 +0100133When HAProxy is running in HTTP mode, both the request and the response are
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200134fully analyzed and indexed, thus it becomes possible to build matching criteria
135on almost anything found in the contents.
136
137However, it is important to understand how HTTP requests and responses are
138formed, and how HAProxy decomposes them. It will then become easier to write
139correct rules and to debug existing configurations.
140
141
1421.1. The HTTP transaction model
143-------------------------------
144
145The HTTP protocol is transaction-driven. This means that each request will lead
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100146to one and only one response. Traditionally, a TCP connection is established
Davor Ocelice9ed2812017-12-25 17:49:28 +0100147from the client to the server, a request is sent by the client through the
148connection, the server responds, and the connection is closed. A new request
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200149will involve a new connection :
150
151 [CON1] [REQ1] ... [RESP1] [CLO1] [CON2] [REQ2] ... [RESP2] [CLO2] ...
152
153In this mode, called the "HTTP close" mode, there are as many connection
154establishments as there are HTTP transactions. Since the connection is closed
155by the server after the response, the client does not need to know the content
156length.
157
158Due to the transactional nature of the protocol, it was possible to improve it
159to avoid closing a connection between two subsequent transactions. In this mode
160however, it is mandatory that the server indicates the content length for each
161response so that the client does not wait indefinitely. For this, a special
162header is used: "Content-length". This mode is called the "keep-alive" mode :
163
164 [CON] [REQ1] ... [RESP1] [REQ2] ... [RESP2] [CLO] ...
165
166Its advantages are a reduced latency between transactions, and less processing
167power required on the server side. It is generally better than the close mode,
168but not always because the clients often limit their concurrent connections to
Patrick Mezard9ec2ec42010-06-12 17:02:45 +0200169a smaller value.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200170
Willy Tarreau95c4e142017-11-26 12:18:55 +0100171Another improvement in the communications is the pipelining mode. It still uses
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200172keep-alive, but the client does not wait for the first response to send the
173second request. This is useful for fetching large number of images composing a
174page :
175
176 [CON] [REQ1] [REQ2] ... [RESP1] [RESP2] [CLO] ...
177
178This can obviously have a tremendous benefit on performance because the network
179latency is eliminated between subsequent requests. Many HTTP agents do not
180correctly support pipelining since there is no way to associate a response with
181the corresponding request in HTTP. For this reason, it is mandatory for the
Cyril Bonté78caf842010-03-10 22:41:43 +0100182server to reply in the exact same order as the requests were received.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200183
Willy Tarreau95c4e142017-11-26 12:18:55 +0100184The next improvement is the multiplexed mode, as implemented in HTTP/2. This
185time, each transaction is assigned a single stream identifier, and all streams
186are multiplexed over an existing connection. Many requests can be sent in
187parallel by the client, and responses can arrive in any order since they also
188carry the stream identifier.
189
Willy Tarreau70dffda2014-01-30 03:07:23 +0100190By default HAProxy operates in keep-alive mode with regards to persistent
191connections: for each connection it processes each request and response, and
192leaves the connection idle on both sides between the end of a response and the
Willy Tarreau95c4e142017-11-26 12:18:55 +0100193start of a new request. When it receives HTTP/2 connections from a client, it
194processes all the requests in parallel and leaves the connection idling,
195waiting for new requests, just as if it was a keep-alive HTTP connection.
Patrick Mezard9ec2ec42010-06-12 17:02:45 +0200196
Christopher Faulet315b39c2018-09-21 16:26:19 +0200197HAProxy supports 4 connection modes :
Willy Tarreau70dffda2014-01-30 03:07:23 +0100198 - keep alive : all requests and responses are processed (default)
199 - tunnel : only the first request and response are processed,
Christopher Faulet6c9bbb22019-03-26 21:37:23 +0100200 everything else is forwarded with no analysis (deprecated).
Willy Tarreau70dffda2014-01-30 03:07:23 +0100201 - server close : the server-facing connection is closed after the response.
Christopher Faulet315b39c2018-09-21 16:26:19 +0200202 - close : the connection is actively closed after end of response.
Willy Tarreau70dffda2014-01-30 03:07:23 +0100203
Willy Tarreau95c4e142017-11-26 12:18:55 +0100204
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200205
2061.2. HTTP request
207-----------------
208
209First, let's consider this HTTP request :
210
211 Line Contents
Willy Tarreaud72758d2010-01-12 10:42:19 +0100212 number
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200213 1 GET /serv/login.php?lang=en&profile=2 HTTP/1.1
214 2 Host: www.mydomain.com
215 3 User-agent: my small browser
216 4 Accept: image/jpeg, image/gif
217 5 Accept: image/png
218
219
2201.2.1. The Request line
221-----------------------
222
223Line 1 is the "request line". It is always composed of 3 fields :
224
225 - a METHOD : GET
226 - a URI : /serv/login.php?lang=en&profile=2
227 - a version tag : HTTP/1.1
228
229All of them are delimited by what the standard calls LWS (linear white spaces),
230which are commonly spaces, but can also be tabs or line feeds/carriage returns
231followed by spaces/tabs. The method itself cannot contain any colon (':') and
232is limited to alphabetic letters. All those various combinations make it
233desirable that HAProxy performs the splitting itself rather than leaving it to
234the user to write a complex or inaccurate regular expression.
235
236The URI itself can have several forms :
237
238 - A "relative URI" :
239
240 /serv/login.php?lang=en&profile=2
241
242 It is a complete URL without the host part. This is generally what is
243 received by servers, reverse proxies and transparent proxies.
244
245 - An "absolute URI", also called a "URL" :
246
247 http://192.168.0.12:8080/serv/login.php?lang=en&profile=2
248
249 It is composed of a "scheme" (the protocol name followed by '://'), a host
250 name or address, optionally a colon (':') followed by a port number, then
251 a relative URI beginning at the first slash ('/') after the address part.
252 This is generally what proxies receive, but a server supporting HTTP/1.1
253 must accept this form too.
254
255 - a star ('*') : this form is only accepted in association with the OPTIONS
256 method and is not relayable. It is used to inquiry a next hop's
257 capabilities.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100258
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200259 - an address:port combination : 192.168.0.12:80
260 This is used with the CONNECT method, which is used to establish TCP
261 tunnels through HTTP proxies, generally for HTTPS, but sometimes for
262 other protocols too.
263
264In a relative URI, two sub-parts are identified. The part before the question
265mark is called the "path". It is typically the relative path to static objects
266on the server. The part after the question mark is called the "query string".
267It is mostly used with GET requests sent to dynamic scripts and is very
268specific to the language, framework or application in use.
269
Willy Tarreau95c4e142017-11-26 12:18:55 +0100270HTTP/2 doesn't convey a version information with the request, so the version is
Davor Ocelice9ed2812017-12-25 17:49:28 +0100271assumed to be the same as the one of the underlying protocol (i.e. "HTTP/2").
Willy Tarreau95c4e142017-11-26 12:18:55 +0100272
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200273
2741.2.2. The request headers
275--------------------------
276
277The headers start at the second line. They are composed of a name at the
278beginning of the line, immediately followed by a colon (':'). Traditionally,
279an LWS is added after the colon but that's not required. Then come the values.
280Multiple identical headers may be folded into one single line, delimiting the
281values with commas, provided that their order is respected. This is commonly
282encountered in the "Cookie:" field. A header may span over multiple lines if
283the subsequent lines begin with an LWS. In the example in 1.2, lines 4 and 5
284define a total of 3 values for the "Accept:" header.
285
Davor Ocelice9ed2812017-12-25 17:49:28 +0100286Contrary to a common misconception, header names are not case-sensitive, and
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200287their values are not either if they refer to other header names (such as the
Willy Tarreau95c4e142017-11-26 12:18:55 +0100288"Connection:" header). In HTTP/2, header names are always sent in lower case,
Willy Tarreau253c2512020-07-07 15:55:23 +0200289as can be seen when running in debug mode. Internally, all header names are
290normalized to lower case so that HTTP/1.x and HTTP/2 use the exact same
291representation, and they are sent as-is on the other side. This explains why an
292HTTP/1.x request typed with camel case is delivered in lower case.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200293
294The end of the headers is indicated by the first empty line. People often say
295that it's a double line feed, which is not exact, even if a double line feed
296is one valid form of empty line.
297
298Fortunately, HAProxy takes care of all these complex combinations when indexing
299headers, checking values and counting them, so there is no reason to worry
300about the way they could be written, but it is important not to accuse an
301application of being buggy if it does unusual, valid things.
302
303Important note:
Lukas Tribus23953682017-04-28 13:24:30 +0000304 As suggested by RFC7231, HAProxy normalizes headers by replacing line breaks
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200305 in the middle of headers by LWS in order to join multi-line headers. This
306 is necessary for proper analysis and helps less capable HTTP parsers to work
307 correctly and not to be fooled by such complex constructs.
308
309
3101.3. HTTP response
311------------------
312
313An HTTP response looks very much like an HTTP request. Both are called HTTP
314messages. Let's consider this HTTP response :
315
316 Line Contents
Willy Tarreaud72758d2010-01-12 10:42:19 +0100317 number
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200318 1 HTTP/1.1 200 OK
319 2 Content-length: 350
320 3 Content-Type: text/html
321
Willy Tarreau816b9792009-09-15 21:25:21 +0200322As a special case, HTTP supports so called "Informational responses" as status
323codes 1xx. These messages are special in that they don't convey any part of the
324response, they're just used as sort of a signaling message to ask a client to
Willy Tarreau5843d1a2010-02-01 15:13:32 +0100325continue to post its request for instance. In the case of a status 100 response
326the requested information will be carried by the next non-100 response message
327following the informational one. This implies that multiple responses may be
328sent to a single request, and that this only works when keep-alive is enabled
329(1xx messages are HTTP/1.1 only). HAProxy handles these messages and is able to
330correctly forward and skip them, and only process the next non-100 response. As
331such, these messages are neither logged nor transformed, unless explicitly
332state otherwise. Status 101 messages indicate that the protocol is changing
333over the same connection and that haproxy must switch to tunnel mode, just as
334if a CONNECT had occurred. Then the Upgrade header would contain additional
335information about the type of protocol the connection is switching to.
Willy Tarreau816b9792009-09-15 21:25:21 +0200336
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200337
Davor Ocelice9ed2812017-12-25 17:49:28 +01003381.3.1. The response line
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200339------------------------
340
341Line 1 is the "response line". It is always composed of 3 fields :
342
343 - a version tag : HTTP/1.1
344 - a status code : 200
345 - a reason : OK
346
347The status code is always 3-digit. The first digit indicates a general status :
Davor Ocelice9ed2812017-12-25 17:49:28 +0100348 - 1xx = informational message to be skipped (e.g. 100, 101)
349 - 2xx = OK, content is following (e.g. 200, 206)
350 - 3xx = OK, no content following (e.g. 302, 304)
351 - 4xx = error caused by the client (e.g. 401, 403, 404)
352 - 5xx = error caused by the server (e.g. 500, 502, 503)
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200353
Lukas Tribus23953682017-04-28 13:24:30 +0000354Please refer to RFC7231 for the detailed meaning of all such codes. The
Willy Tarreaud72758d2010-01-12 10:42:19 +0100355"reason" field is just a hint, but is not parsed by clients. Anything can be
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200356found there, but it's a common practice to respect the well-established
357messages. It can be composed of one or multiple words, such as "OK", "Found",
358or "Authentication Required".
359
Davor Ocelice9ed2812017-12-25 17:49:28 +0100360HAProxy may emit the following status codes by itself :
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200361
362 Code When / reason
363 200 access to stats page, and when replying to monitoring requests
364 301 when performing a redirection, depending on the configured code
365 302 when performing a redirection, depending on the configured code
366 303 when performing a redirection, depending on the configured code
Willy Tarreaub67fdc42013-03-29 19:28:11 +0100367 307 when performing a redirection, depending on the configured code
368 308 when performing a redirection, depending on the configured code
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200369 400 for an invalid or too large request
370 401 when an authentication is required to perform the action (when
371 accessing the stats page)
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200372 403 when a request is forbidden by a "http-request deny" rule
Florian Tham9205fea2020-01-08 13:35:30 +0100373 404 when the requested resource could not be found
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200374 408 when the request timeout strikes before the request is complete
Florian Tham272e29b2020-01-08 10:19:05 +0100375 410 when the requested resource is no longer available and will not
376 be available again
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200377 500 when haproxy encounters an unrecoverable internal error, such as a
378 memory allocation failure, which should never happen
379 502 when the server returns an empty, invalid or incomplete response, or
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200380 when an "http-response deny" rule blocks the response.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200381 503 when no server was available to handle the request, or in response to
382 monitoring requests which match the "monitor fail" condition
383 504 when the response timeout strikes before the server responds
384
385The error 4xx and 5xx codes above may be customized (see "errorloc" in section
3864.2).
387
388
3891.3.2. The response headers
390---------------------------
391
392Response headers work exactly like request headers, and as such, HAProxy uses
393the same parsing function for both. Please refer to paragraph 1.2.2 for more
394details.
395
396
3972. Configuring HAProxy
398----------------------
399
4002.1. Configuration file format
401------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +0200402
403HAProxy's configuration process involves 3 major sources of parameters :
404
405 - the arguments from the command-line, which always take precedence
406 - the "global" section, which sets process-wide parameters
407 - the proxies sections which can take form of "defaults", "listen",
408 "frontend" and "backend".
409
Willy Tarreau0ba27502007-12-24 16:55:16 +0100410The configuration file syntax consists in lines beginning with a keyword
411referenced in this manual, optionally followed by one or several parameters
William Lallemandf9873ba2015-05-05 17:37:14 +0200412delimited by spaces.
Willy Tarreau0ba27502007-12-24 16:55:16 +0100413
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200414
William Lallemandf9873ba2015-05-05 17:37:14 +02004152.2. Quoting and escaping
416-------------------------
417
418HAProxy's configuration introduces a quoting and escaping system similar to
419many programming languages. The configuration file supports 3 types: escaping
420with a backslash, weak quoting with double quotes, and strong quoting with
421single quotes.
422
423If spaces have to be entered in strings, then they must be escaped by preceding
424them by a backslash ('\') or by quoting them. Backslashes also have to be
425escaped by doubling or strong quoting them.
426
427Escaping is achieved by preceding a special character by a backslash ('\'):
428
429 \ to mark a space and differentiate it from a delimiter
430 \# to mark a hash and differentiate it from a comment
431 \\ to use a backslash
432 \' to use a single quote and differentiate it from strong quoting
433 \" to use a double quote and differentiate it from weak quoting
434
435Weak quoting is achieved by using double quotes (""). Weak quoting prevents
436the interpretation of:
437
438 space as a parameter separator
439 ' single quote as a strong quoting delimiter
440 # hash as a comment start
441
William Lallemandb2f07452015-05-12 14:27:13 +0200442Weak quoting permits the interpretation of variables, if you want to use a non
443-interpreted dollar within a double quoted string, you should escape it with a
444backslash ("\$"), it does not work outside weak quoting.
445
446Interpretation of escaping and special characters are not prevented by weak
William Lallemandf9873ba2015-05-05 17:37:14 +0200447quoting.
448
449Strong quoting is achieved by using single quotes (''). Inside single quotes,
450nothing is interpreted, it's the efficient way to quote regexes.
451
452Quoted and escaped strings are replaced in memory by their interpreted
453equivalent, it allows you to perform concatenation.
454
455 Example:
456 # those are equivalents:
457 log-format %{+Q}o\ %t\ %s\ %{-Q}r
458 log-format "%{+Q}o %t %s %{-Q}r"
459 log-format '%{+Q}o %t %s %{-Q}r'
460 log-format "%{+Q}o %t"' %s %{-Q}r'
461 log-format "%{+Q}o %t"' %s'\ %{-Q}r
462
463 # those are equivalents:
464 reqrep "^([^\ :]*)\ /static/(.*)" \1\ /\2
465 reqrep "^([^ :]*)\ /static/(.*)" '\1 /\2'
466 reqrep "^([^ :]*)\ /static/(.*)" "\1 /\2"
467 reqrep "^([^ :]*)\ /static/(.*)" "\1\ /\2"
468
469
William Lallemandb2f07452015-05-12 14:27:13 +02004702.3. Environment variables
471--------------------------
472
473HAProxy's configuration supports environment variables. Those variables are
474interpreted only within double quotes. Variables are expanded during the
475configuration parsing. Variable names must be preceded by a dollar ("$") and
476optionally enclosed with braces ("{}") similarly to what is done in Bourne
477shell. Variable names can contain alphanumerical characters or the character
Amaury Denoyellefa41cb62020-10-01 14:32:35 +0200478underscore ("_") but should not start with a digit. If the variable contains a
479list of several values separated by spaces, it can be expanded as individual
480arguments by enclosing the variable with braces and appending the suffix '[*]'
481before the closing brace.
William Lallemandb2f07452015-05-12 14:27:13 +0200482
483 Example:
484
485 bind "fd@${FD_APP1}"
486
487 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
488
489 user "$HAPROXY_USER"
490
William Lallemand4d03e432019-06-14 15:35:37 +0200491Some variables are defined by HAProxy, they can be used in the configuration
492file, or could be inherited by a program (See 3.7. Programs):
William Lallemanddaf4cd22018-04-17 16:46:13 +0200493
William Lallemand4d03e432019-06-14 15:35:37 +0200494* HAPROXY_LOCALPEER: defined at the startup of the process which contains the
495 name of the local peer. (See "-L" in the management guide.)
496
497* HAPROXY_CFGFILES: list of the configuration files loaded by HAProxy,
498 separated by semicolons. Can be useful in the case you specified a
499 directory.
500
501* HAPROXY_MWORKER: In master-worker mode, this variable is set to 1.
502
John Roeslerfb2fce12019-07-10 15:45:51 -0500503* HAPROXY_CLI: configured listeners addresses of the stats socket for every
William Lallemand4d03e432019-06-14 15:35:37 +0200504 processes, separated by semicolons.
505
John Roeslerfb2fce12019-07-10 15:45:51 -0500506* HAPROXY_MASTER_CLI: In master-worker mode, listeners addresses of the master
William Lallemand4d03e432019-06-14 15:35:37 +0200507 CLI, separated by semicolons.
508
509See also "external-check command" for other variables.
William Lallemandb2f07452015-05-12 14:27:13 +0200510
5112.4. Time format
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200512----------------
513
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100514Some parameters involve values representing time, such as timeouts. These
Willy Tarreau0ba27502007-12-24 16:55:16 +0100515values are generally expressed in milliseconds (unless explicitly stated
516otherwise) but may be expressed in any other unit by suffixing the unit to the
517numeric value. It is important to consider this because it will not be repeated
518for every keyword. Supported units are :
519
520 - us : microseconds. 1 microsecond = 1/1000000 second
521 - ms : milliseconds. 1 millisecond = 1/1000 second. This is the default.
522 - s : seconds. 1s = 1000ms
523 - m : minutes. 1m = 60s = 60000ms
524 - h : hours. 1h = 60m = 3600s = 3600000ms
525 - d : days. 1d = 24h = 1440m = 86400s = 86400000ms
526
527
Lukas Tribusaa83a312017-03-21 09:25:09 +00005282.5. Examples
Patrick Mezard35da19c2010-06-12 17:02:47 +0200529-------------
530
531 # Simple configuration for an HTTP proxy listening on port 80 on all
532 # interfaces and forwarding requests to a single backend "servers" with a
533 # single server "server1" listening on 127.0.0.1:8000
534 global
535 daemon
536 maxconn 256
537
538 defaults
539 mode http
540 timeout connect 5000ms
541 timeout client 50000ms
542 timeout server 50000ms
543
544 frontend http-in
545 bind *:80
546 default_backend servers
547
548 backend servers
549 server server1 127.0.0.1:8000 maxconn 32
550
551
552 # The same configuration defined with a single listen block. Shorter but
553 # less expressive, especially in HTTP mode.
554 global
555 daemon
556 maxconn 256
557
558 defaults
559 mode http
560 timeout connect 5000ms
561 timeout client 50000ms
562 timeout server 50000ms
563
564 listen http-in
565 bind *:80
566 server server1 127.0.0.1:8000 maxconn 32
567
568
569Assuming haproxy is in $PATH, test these configurations in a shell with:
570
Willy Tarreauccb289d2010-12-11 20:19:38 +0100571 $ sudo haproxy -f configuration.conf -c
Patrick Mezard35da19c2010-06-12 17:02:47 +0200572
573
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005743. Global parameters
Willy Tarreau6a06a402007-07-15 20:15:28 +0200575--------------------
576
577Parameters in the "global" section are process-wide and often OS-specific. They
578are generally set once for all and do not need being changed once correct. Some
579of them have command-line equivalents.
580
581The following keywords are supported in the "global" section :
582
583 * Process management and security
Emeric Brunc8e8d122012-10-02 18:42:10 +0200584 - ca-base
Willy Tarreau6a06a402007-07-15 20:15:28 +0200585 - chroot
Emeric Brunc8e8d122012-10-02 18:42:10 +0200586 - crt-base
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200587 - cpu-map
Willy Tarreau6a06a402007-07-15 20:15:28 +0200588 - daemon
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200589 - description
590 - deviceatlas-json-file
591 - deviceatlas-log-level
592 - deviceatlas-separator
593 - deviceatlas-properties-cookie
Simon Horman98637e52014-06-20 12:30:16 +0900594 - external-check
Willy Tarreau6a06a402007-07-15 20:15:28 +0200595 - gid
596 - group
Cyril Bonté203ec5a2017-03-23 22:44:13 +0100597 - hard-stop-after
Christopher Faulet98fbe952019-07-22 16:18:24 +0200598 - h1-case-adjust
599 - h1-case-adjust-file
Willy Tarreaud96f1122019-12-03 07:07:36 +0100600 - insecure-fork-wanted
Willy Tarreaua45a8b52019-12-06 16:31:45 +0100601 - insecure-setuid-wanted
Emmanuel Hocdet70df7bf2019-01-04 11:08:20 +0100602 - issuers-chain-path
Dragan Dosen13cd54c2020-06-18 18:24:05 +0200603 - localpeer
Willy Tarreau6a06a402007-07-15 20:15:28 +0200604 - log
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200605 - log-tag
Joe Williamsdf5b38f2010-12-29 17:05:48 +0100606 - log-send-hostname
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200607 - lua-load
Tim Duesterhusdd74b5f2020-01-12 13:55:40 +0100608 - lua-prepend-path
William Lallemand27edc4b2019-05-07 17:49:33 +0200609 - mworker-max-reloads
Willy Tarreau6a06a402007-07-15 20:15:28 +0200610 - nbproc
Christopher Fauletbe0faa22017-08-29 15:37:10 +0200611 - nbthread
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200612 - node
Willy Tarreau6a06a402007-07-15 20:15:28 +0200613 - pidfile
Willy Tarreau119e50e2020-05-22 13:53:29 +0200614 - pp2-never-send-local
Willy Tarreau1d549722016-02-16 12:41:57 +0100615 - presetenv
616 - resetenv
Willy Tarreau6a06a402007-07-15 20:15:28 +0200617 - uid
618 - ulimit-n
619 - user
Willy Tarreau636848a2019-04-15 19:38:50 +0200620 - set-dumpable
Willy Tarreau1d549722016-02-16 12:41:57 +0100621 - setenv
Willy Tarreaufbee7132007-10-18 13:53:22 +0200622 - stats
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200623 - ssl-default-bind-ciphers
Dirkjan Bussink415150f2018-09-14 11:14:21 +0200624 - ssl-default-bind-ciphersuites
Jerome Magninb203ff62020-04-03 15:28:22 +0200625 - ssl-default-bind-curves
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200626 - ssl-default-bind-options
627 - ssl-default-server-ciphers
Dirkjan Bussink415150f2018-09-14 11:14:21 +0200628 - ssl-default-server-ciphersuites
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200629 - ssl-default-server-options
630 - ssl-dh-param-file
Emeric Brun850efd52014-01-29 12:24:34 +0100631 - ssl-server-verify
Emmanuel Hocdetc3b7e742020-04-22 11:06:19 +0200632 - ssl-skip-self-issued-ca
Willy Tarreauceb24bc2010-11-09 12:46:41 +0100633 - unix-bind
Willy Tarreau1d549722016-02-16 12:41:57 +0100634 - unsetenv
Thomas Holmesdb04f192015-05-18 13:21:39 +0100635 - 51degrees-data-file
636 - 51degrees-property-name-list
Dragan Dosen93b38d92015-06-29 16:43:25 +0200637 - 51degrees-property-separator
Dragan Dosenae6d39a2015-06-29 16:43:27 +0200638 - 51degrees-cache-size
Willy Tarreaub3cc9f22019-04-19 16:03:32 +0200639 - wurfl-data-file
640 - wurfl-information-list
641 - wurfl-information-list-separator
Willy Tarreaub3cc9f22019-04-19 16:03:32 +0200642 - wurfl-cache-size
William Dauchy0fec3ab2019-10-27 20:08:11 +0100643 - strict-limits
Willy Tarreaud72758d2010-01-12 10:42:19 +0100644
Willy Tarreau6a06a402007-07-15 20:15:28 +0200645 * Performance tuning
William Dauchy0a8824f2019-10-27 20:08:09 +0100646 - busy-polling
Willy Tarreau1746eec2014-04-25 10:46:47 +0200647 - max-spread-checks
Willy Tarreau6a06a402007-07-15 20:15:28 +0200648 - maxconn
Willy Tarreau81c25d02011-09-07 15:17:21 +0200649 - maxconnrate
William Lallemandd85f9172012-11-09 17:05:39 +0100650 - maxcomprate
William Lallemand072a2bf2012-11-20 17:01:01 +0100651 - maxcompcpuusage
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100652 - maxpipes
Willy Tarreau93e7c002013-10-07 18:51:07 +0200653 - maxsessrate
Willy Tarreau403edff2012-09-06 11:58:37 +0200654 - maxsslconn
Willy Tarreaue43d5322013-10-07 20:01:52 +0200655 - maxsslrate
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200656 - maxzlibmem
Willy Tarreau6a06a402007-07-15 20:15:28 +0200657 - noepoll
658 - nokqueue
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +0000659 - noevports
Willy Tarreau6a06a402007-07-15 20:15:28 +0200660 - nopoll
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100661 - nosplice
Jarno Huuskonen0e82b922014-04-12 18:22:19 +0300662 - nogetaddrinfo
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +0000663 - noreuseport
Willy Tarreau75c62c22018-11-22 11:02:09 +0100664 - profiling.tasks
Willy Tarreaufe255b72007-10-14 23:09:26 +0200665 - spread-checks
Baptiste Assmann5626f482015-08-23 10:00:10 +0200666 - server-state-base
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +0200667 - server-state-file
Grant Zhang872f9c22017-01-21 01:10:18 +0000668 - ssl-engine
Grant Zhangfa6c7ee2017-01-14 01:42:15 +0000669 - ssl-mode-async
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200670 - tune.buffers.limit
671 - tune.buffers.reserve
Willy Tarreau27a674e2009-08-17 07:23:33 +0200672 - tune.bufsize
Willy Tarreau43961d52010-10-04 20:39:20 +0200673 - tune.chksize
William Lallemandf3747832012-11-09 12:33:10 +0100674 - tune.comp.maxlevel
Willy Tarreaubc52bec2020-06-18 08:58:47 +0200675 - tune.fd.edge-triggered
Willy Tarreaufe20e5b2017-07-27 11:42:14 +0200676 - tune.h2.header-table-size
Willy Tarreaue6baec02017-07-27 11:45:11 +0200677 - tune.h2.initial-window-size
Willy Tarreau5242ef82017-07-27 11:47:28 +0200678 - tune.h2.max-concurrent-streams
Willy Tarreau193b8c62012-11-22 00:17:38 +0100679 - tune.http.cookielen
Stéphane Cottin23e9e932017-05-18 08:58:41 +0200680 - tune.http.logurilen
Willy Tarreauac1932d2011-10-24 19:14:41 +0200681 - tune.http.maxhdr
Willy Tarreau76cc6992020-07-01 18:49:24 +0200682 - tune.idle-pool.shared
Willy Tarreau7e312732014-02-12 16:35:14 +0100683 - tune.idletimer
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100684 - tune.lua.forced-yield
Willy Tarreau32f61e22015-03-18 17:54:59 +0100685 - tune.lua.maxmem
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100686 - tune.lua.session-timeout
687 - tune.lua.task-timeout
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +0200688 - tune.lua.service-timeout
Willy Tarreaua0250ba2008-01-06 11:22:57 +0100689 - tune.maxaccept
690 - tune.maxpollevents
Willy Tarreau27a674e2009-08-17 07:23:33 +0200691 - tune.maxrewrite
Willy Tarreauf3045d22015-04-29 16:24:50 +0200692 - tune.pattern.cache-size
Willy Tarreaubd9a0a72011-10-23 21:14:29 +0200693 - tune.pipesize
Willy Tarreaua8e2d972020-07-01 18:27:16 +0200694 - tune.pool-high-fd-ratio
695 - tune.pool-low-fd-ratio
Willy Tarreaue803de22010-01-21 17:43:04 +0100696 - tune.rcvbuf.client
697 - tune.rcvbuf.server
Willy Tarreaub22fc302015-12-14 12:04:35 +0100698 - tune.recv_enough
Olivier Houchard1599b802018-05-24 18:59:04 +0200699 - tune.runqueue-depth
Willy Tarreaue7723bd2020-06-24 11:11:02 +0200700 - tune.sched.low-latency
Willy Tarreaue803de22010-01-21 17:43:04 +0100701 - tune.sndbuf.client
702 - tune.sndbuf.server
Willy Tarreau6ec58db2012-11-16 16:32:15 +0100703 - tune.ssl.cachesize
William Lallemand7d42ef52020-07-06 11:41:30 +0200704 - tune.ssl.keylog
Willy Tarreaubfd59462013-02-21 07:46:09 +0100705 - tune.ssl.lifetime
Emeric Brun8dc60392014-05-09 13:52:00 +0200706 - tune.ssl.force-private-cache
Willy Tarreaubfd59462013-02-21 07:46:09 +0100707 - tune.ssl.maxrecord
Remi Gacognef46cd6e2014-06-12 14:58:40 +0200708 - tune.ssl.default-dh-param
Christopher Faulet31af49d2015-06-09 17:29:50 +0200709 - tune.ssl.ssl-ctx-cache-size
Thierry FOURNIER5bf77322017-02-25 12:45:22 +0100710 - tune.ssl.capture-cipherlist-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200711 - tune.vars.global-max-size
Christopher Fauletff2613e2016-11-09 11:36:17 +0100712 - tune.vars.proc-max-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200713 - tune.vars.reqres-max-size
714 - tune.vars.sess-max-size
715 - tune.vars.txn-max-size
William Lallemanda509e4c2012-11-07 16:54:34 +0100716 - tune.zlib.memlevel
717 - tune.zlib.windowsize
Willy Tarreaud72758d2010-01-12 10:42:19 +0100718
Willy Tarreau6a06a402007-07-15 20:15:28 +0200719 * Debugging
720 - debug
721 - quiet
Willy Tarreau3eb10b82020-04-15 16:42:39 +0200722 - zero-warning
Willy Tarreau6a06a402007-07-15 20:15:28 +0200723
724
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007253.1. Process management and security
Willy Tarreau6a06a402007-07-15 20:15:28 +0200726------------------------------------
727
Emeric Brunc8e8d122012-10-02 18:42:10 +0200728ca-base <dir>
729 Assigns a default directory to fetch SSL CA certificates and CRLs from when a
Emmanuel Hocdet842e94e2019-12-16 16:39:17 +0100730 relative path is used with "ca-file", "ca-verify-file" or "crl-file"
731 directives. Absolute locations specified in "ca-file", "ca-verify-file" and
732 "crl-file" prevail and ignore "ca-base".
Emeric Brunc8e8d122012-10-02 18:42:10 +0200733
Willy Tarreau6a06a402007-07-15 20:15:28 +0200734chroot <jail dir>
735 Changes current directory to <jail dir> and performs a chroot() there before
736 dropping privileges. This increases the security level in case an unknown
737 vulnerability would be exploited, since it would make it very hard for the
738 attacker to exploit the system. This only works when the process is started
739 with superuser privileges. It is important to ensure that <jail_dir> is both
Davor Ocelice9ed2812017-12-25 17:49:28 +0100740 empty and non-writable to anyone.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100741
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100742cpu-map [auto:]<process-set>[/<thread-set>] <cpu-set>...
743 On Linux 2.6 and above, it is possible to bind a process or a thread to a
744 specific CPU set. This means that the process or the thread will never run on
745 other CPUs. The "cpu-map" directive specifies CPU sets for process or thread
746 sets. The first argument is a process set, eventually followed by a thread
747 set. These sets have the format
748
749 all | odd | even | number[-[number]]
750
751 <number>> must be a number between 1 and 32 or 64, depending on the machine's
Davor Ocelice9ed2812017-12-25 17:49:28 +0100752 word size. Any process IDs above nbproc and any thread IDs above nbthread are
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100753 ignored. It is possible to specify a range with two such number delimited by
754 a dash ('-'). It also is possible to specify all processes at once using
Christopher Faulet1dcb9cb2017-11-22 10:24:40 +0100755 "all", only odd numbers using "odd" or even numbers using "even", just like
756 with the "bind-process" directive. The second and forthcoming arguments are
Davor Ocelice9ed2812017-12-25 17:49:28 +0100757 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 +0100758 range with two such numbers delimited by a dash ('-'). Multiple CPU numbers
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100759 or ranges may be specified, and the processes or threads will be allowed to
Davor Ocelice9ed2812017-12-25 17:49:28 +0100760 bind to all of them. Obviously, multiple "cpu-map" directives may be
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100761 specified. Each "cpu-map" directive will replace the previous ones when they
762 overlap. A thread will be bound on the intersection of its mapping and the
763 one of the process on which it is attached. If the intersection is null, no
764 specific binding will be set for the thread.
Willy Tarreaufc6c0322012-11-16 16:12:27 +0100765
Christopher Fauletff4121f2017-11-22 16:38:49 +0100766 Ranges can be partially defined. The higher bound can be omitted. In such
767 case, it is replaced by the corresponding maximum value, 32 or 64 depending
768 on the machine's word size.
769
Christopher Faulet26028f62017-11-22 15:01:51 +0100770 The prefix "auto:" can be added before the process set to let HAProxy
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100771 automatically bind a process or a thread to a CPU by incrementing
772 process/thread and CPU sets. To be valid, both sets must have the same
773 size. No matter the declaration order of the CPU sets, it will be bound from
774 the lowest to the highest bound. Having a process and a thread range with the
775 "auto:" prefix is not supported. Only one range is supported, the other one
776 must be a fixed number.
Christopher Faulet26028f62017-11-22 15:01:51 +0100777
778 Examples:
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100779 cpu-map 1-4 0-3 # bind processes 1 to 4 on the first 4 CPUs
780
781 cpu-map 1/all 0-3 # bind all threads of the first process on the
782 # first 4 CPUs
783
784 cpu-map 1- 0- # will be replaced by "cpu-map 1-64 0-63"
785 # or "cpu-map 1-32 0-31" depending on the machine's
786 # word size.
787
Christopher Faulet26028f62017-11-22 15:01:51 +0100788 # all these lines bind the process 1 to the cpu 0, the process 2 to cpu 1
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100789 # and so on.
Christopher Faulet26028f62017-11-22 15:01:51 +0100790 cpu-map auto:1-4 0-3
791 cpu-map auto:1-4 0-1 2-3
792 cpu-map auto:1-4 3 2 1 0
793
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100794 # all these lines bind the thread 1 to the cpu 0, the thread 2 to cpu 1
795 # and so on.
796 cpu-map auto:1/1-4 0-3
797 cpu-map auto:1/1-4 0-1 2-3
798 cpu-map auto:1/1-4 3 2 1 0
799
Davor Ocelice9ed2812017-12-25 17:49:28 +0100800 # bind each process to exactly one CPU using all/odd/even keyword
Christopher Faulet26028f62017-11-22 15:01:51 +0100801 cpu-map auto:all 0-63
802 cpu-map auto:even 0-31
803 cpu-map auto:odd 32-63
804
805 # invalid cpu-map because process and CPU sets have different sizes.
806 cpu-map auto:1-4 0 # invalid
807 cpu-map auto:1 0-3 # invalid
808
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100809 # invalid cpu-map because automatic binding is used with a process range
810 # and a thread range.
811 cpu-map auto:all/all 0 # invalid
812 cpu-map auto:all/1-4 0 # invalid
813 cpu-map auto:1-4/all 0 # invalid
814
Emeric Brunc8e8d122012-10-02 18:42:10 +0200815crt-base <dir>
816 Assigns a default directory to fetch SSL certificates from when a relative
William Dauchy238ea3b2020-01-11 13:09:12 +0100817 path is used with "crtfile" or "crt" directives. Absolute locations specified
818 prevail and ignore "crt-base".
Emeric Brunc8e8d122012-10-02 18:42:10 +0200819
Willy Tarreau6a06a402007-07-15 20:15:28 +0200820daemon
821 Makes the process fork into background. This is the recommended mode of
822 operation. It is equivalent to the command line "-D" argument. It can be
Lukas Tribusf46bf952017-11-21 12:39:34 +0100823 disabled by the command line "-db" argument. This option is ignored in
824 systemd mode.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200825
David Carlier8167f302015-06-01 13:50:06 +0200826deviceatlas-json-file <path>
827 Sets the path of the DeviceAtlas JSON data file to be loaded by the API.
Davor Ocelice9ed2812017-12-25 17:49:28 +0100828 The path must be a valid JSON data file and accessible by HAProxy process.
David Carlier8167f302015-06-01 13:50:06 +0200829
830deviceatlas-log-level <value>
Davor Ocelice9ed2812017-12-25 17:49:28 +0100831 Sets the level of information returned by the API. This directive is
David Carlier8167f302015-06-01 13:50:06 +0200832 optional and set to 0 by default if not set.
833
834deviceatlas-separator <char>
835 Sets the character separator for the API properties results. This directive
836 is optional and set to | by default if not set.
837
Cyril Bonté0306c4a2015-10-26 22:37:38 +0100838deviceatlas-properties-cookie <name>
Cyril Bonté307ee1e2015-09-28 23:16:06 +0200839 Sets the client cookie's name used for the detection if the DeviceAtlas
840 Client-side component was used during the request. This directive is optional
841 and set to DAPROPS by default if not set.
David Carlier29b3ca32015-09-25 14:09:21 +0100842
Simon Horman98637e52014-06-20 12:30:16 +0900843external-check
Willy Tarreaud96f1122019-12-03 07:07:36 +0100844 Allows the use of an external agent to perform health checks. This is
845 disabled by default as a security precaution, and even when enabled, checks
Willy Tarreaua45a8b52019-12-06 16:31:45 +0100846 may still fail unless "insecure-fork-wanted" is enabled as well. If the
847 program launched makes use of a setuid executable (it should really not),
848 you may also need to set "insecure-setuid-wanted" in the global section.
849 See "option external-check", and "insecure-fork-wanted", and
850 "insecure-setuid-wanted".
Simon Horman98637e52014-06-20 12:30:16 +0900851
Willy Tarreau6a06a402007-07-15 20:15:28 +0200852gid <number>
853 Changes the process' group ID to <number>. It is recommended that the group
854 ID is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
855 be started with a user belonging to this group, or with superuser privileges.
Michael Schererab012dd2013-01-12 18:35:19 +0100856 Note that if haproxy is started from a user having supplementary groups, it
857 will only be able to drop these groups if started with superuser privileges.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200858 See also "group" and "uid".
Willy Tarreaud72758d2010-01-12 10:42:19 +0100859
Willy Tarreau11770ce2019-12-03 08:29:22 +0100860group <group name>
861 Similar to "gid" but uses the GID of group name <group name> from /etc/group.
862 See also "gid" and "user".
863
Cyril Bonté203ec5a2017-03-23 22:44:13 +0100864hard-stop-after <time>
865 Defines the maximum time allowed to perform a clean soft-stop.
866
867 Arguments :
868 <time> is the maximum time (by default in milliseconds) for which the
869 instance will remain alive when a soft-stop is received via the
870 SIGUSR1 signal.
871
872 This may be used to ensure that the instance will quit even if connections
873 remain opened during a soft-stop (for example with long timeouts for a proxy
874 in tcp mode). It applies both in TCP and HTTP mode.
875
876 Example:
877 global
878 hard-stop-after 30s
879
Christopher Faulet98fbe952019-07-22 16:18:24 +0200880h1-case-adjust <from> <to>
881 Defines the case adjustment to apply, when enabled, to the header name
882 <from>, to change it to <to> before sending it to HTTP/1 clients or
883 servers. <from> must be in lower case, and <from> and <to> must not differ
884 except for their case. It may be repeated if several header names need to be
Ilya Shipitsin8525fd92020-02-29 12:34:59 +0500885 adjusted. Duplicate entries are not allowed. If a lot of header names have to
Christopher Faulet98fbe952019-07-22 16:18:24 +0200886 be adjusted, it might be more convenient to use "h1-case-adjust-file".
887 Please note that no transformation will be applied unless "option
888 h1-case-adjust-bogus-client" or "option h1-case-adjust-bogus-server" is
889 specified in a proxy.
890
891 There is no standard case for header names because, as stated in RFC7230,
892 they are case-insensitive. So applications must handle them in a case-
893 insensitive manner. But some bogus applications violate the standards and
894 erroneously rely on the cases most commonly used by browsers. This problem
895 becomes critical with HTTP/2 because all header names must be exchanged in
896 lower case, and HAProxy follows the same convention. All header names are
897 sent in lower case to clients and servers, regardless of the HTTP version.
898
899 Applications which fail to properly process requests or responses may require
900 to temporarily use such workarounds to adjust header names sent to them for
901 the time it takes the application to be fixed. Please note that an
902 application which requires such workarounds might be vulnerable to content
903 smuggling attacks and must absolutely be fixed.
904
905 Example:
906 global
907 h1-case-adjust content-length Content-Length
908
909 See "h1-case-adjust-file", "option h1-case-adjust-bogus-client" and
910 "option h1-case-adjust-bogus-server".
911
912h1-case-adjust-file <hdrs-file>
913 Defines a file containing a list of key/value pairs used to adjust the case
914 of some header names before sending them to HTTP/1 clients or servers. The
915 file <hdrs-file> must contain 2 header names per line. The first one must be
916 in lower case and both must not differ except for their case. Lines which
917 start with '#' are ignored, just like empty lines. Leading and trailing tabs
918 and spaces are stripped. Duplicate entries are not allowed. Please note that
919 no transformation will be applied unless "option h1-case-adjust-bogus-client"
920 or "option h1-case-adjust-bogus-server" is specified in a proxy.
921
922 If this directive is repeated, only the last one will be processed. It is an
923 alternative to the directive "h1-case-adjust" if a lot of header names need
924 to be adjusted. Please read the risks associated with using this.
925
926 See "h1-case-adjust", "option h1-case-adjust-bogus-client" and
927 "option h1-case-adjust-bogus-server".
928
Willy Tarreaud96f1122019-12-03 07:07:36 +0100929insecure-fork-wanted
930 By default haproxy tries hard to prevent any thread and process creation
931 after it starts. Doing so is particularly important when using Lua files of
932 uncertain origin, and when experimenting with development versions which may
933 still contain bugs whose exploitability is uncertain. And generally speaking
934 it's good hygiene to make sure that no unexpected background activity can be
935 triggered by traffic. But this prevents external checks from working, and may
936 break some very specific Lua scripts which actively rely on the ability to
937 fork. This option is there to disable this protection. Note that it is a bad
938 idea to disable it, as a vulnerability in a library or within haproxy itself
939 will be easier to exploit once disabled. In addition, forking from Lua or
940 anywhere else is not reliable as the forked process may randomly embed a lock
941 set by another thread and never manage to finish an operation. As such it is
942 highly recommended that this option is never used and that any workload
943 requiring such a fork be reconsidered and moved to a safer solution (such as
944 agents instead of external checks). This option supports the "no" prefix to
945 disable it.
946
Willy Tarreaua45a8b52019-12-06 16:31:45 +0100947insecure-setuid-wanted
948 HAProxy doesn't need to call executables at run time (except when using
949 external checks which are strongly recommended against), and is even expected
950 to isolate itself into an empty chroot. As such, there basically is no valid
951 reason to allow a setuid executable to be called without the user being fully
952 aware of the risks. In a situation where haproxy would need to call external
953 checks and/or disable chroot, exploiting a vulnerability in a library or in
954 haproxy itself could lead to the execution of an external program. On Linux
955 it is possible to lock the process so that any setuid bit present on such an
956 executable is ignored. This significantly reduces the risk of privilege
957 escalation in such a situation. This is what haproxy does by default. In case
958 this causes a problem to an external check (for example one which would need
959 the "ping" command), then it is possible to disable this protection by
960 explicitly adding this directive in the global section. If enabled, it is
961 possible to turn it back off by prefixing it with the "no" keyword.
962
Emmanuel Hocdet70df7bf2019-01-04 11:08:20 +0100963issuers-chain-path <dir>
964 Assigns a directory to load certificate chain for issuer completion. All
965 files must be in PEM format. For certificates loaded with "crt" or "crt-list",
966 if certificate chain is not included in PEM (also commonly known as
967 intermediate certificate), haproxy will complete chain if the issuer of the
968 certificate corresponds to the first certificate of the chain loaded with
969 "issuers-chain-path".
970 A "crt" file with PrivateKey+Certificate+IntermediateCA2+IntermediateCA1
971 could be replaced with PrivateKey+Certificate. HAProxy will complete the
972 chain if a file with IntermediateCA2+IntermediateCA1 is present in
973 "issuers-chain-path" directory. All other certificates with the same issuer
974 will share the chain in memory.
975
Dragan Dosen13cd54c2020-06-18 18:24:05 +0200976localpeer <name>
977 Sets the local instance's peer name. It will be ignored if the "-L"
978 command line argument is specified or if used after "peers" section
979 definitions. In such cases, a warning message will be emitted during
980 the configuration parsing.
981
982 This option will also set the HAPROXY_LOCALPEER environment variable.
983 See also "-L" in the management guide and "peers" section below.
984
Frédéric Lécailled690dfa2019-04-25 10:52:17 +0200985log <address> [len <length>] [format <format>] [sample <ranges>:<smp_size>]
986 <facility> [max level [min level]]
Cyril Bonté3e954872018-03-20 23:30:27 +0100987 Adds a global syslog server. Several global servers can be defined. They
Davor Ocelice9ed2812017-12-25 17:49:28 +0100988 will receive logs for starts and exits, as well as all logs from proxies
Robert Tsai81ae1952007-12-05 10:47:29 +0100989 configured with "log global".
990
991 <address> can be one of:
992
Willy Tarreau2769aa02007-12-27 18:26:09 +0100993 - An IPv4 address optionally followed by a colon and a UDP port. If
Robert Tsai81ae1952007-12-05 10:47:29 +0100994 no port is specified, 514 is used by default (the standard syslog
995 port).
996
David du Colombier24bb5f52011-03-17 10:40:23 +0100997 - An IPv6 address followed by a colon and optionally a UDP port. If
998 no port is specified, 514 is used by default (the standard syslog
999 port).
1000
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01001001 - A filesystem path to a datagram UNIX domain socket, keeping in mind
Robert Tsai81ae1952007-12-05 10:47:29 +01001002 considerations for chroot (be sure the path is accessible inside
1003 the chroot) and uid/gid (be sure the path is appropriately
Davor Ocelice9ed2812017-12-25 17:49:28 +01001004 writable).
Robert Tsai81ae1952007-12-05 10:47:29 +01001005
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01001006 - A file descriptor number in the form "fd@<number>", which may point
1007 to a pipe, terminal, or socket. In this case unbuffered logs are used
1008 and one writev() call per log is performed. This is a bit expensive
1009 but acceptable for most workloads. Messages sent this way will not be
1010 truncated but may be dropped, in which case the DroppedLogs counter
1011 will be incremented. The writev() call is atomic even on pipes for
1012 messages up to PIPE_BUF size, which POSIX recommends to be at least
1013 512 and which is 4096 bytes on most modern operating systems. Any
1014 larger message may be interleaved with messages from other processes.
1015 Exceptionally for debugging purposes the file descriptor may also be
1016 directed to a file, but doing so will significantly slow haproxy down
1017 as non-blocking calls will be ignored. Also there will be no way to
1018 purge nor rotate this file without restarting the process. Note that
1019 the configured syslog format is preserved, so the output is suitable
Willy Tarreauc1b06452018-11-12 11:57:56 +01001020 for use with a TCP syslog server. See also the "short" and "raw"
1021 format below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01001022
1023 - "stdout" / "stderr", which are respectively aliases for "fd@1" and
1024 "fd@2", see above.
1025
Willy Tarreauc046d162019-08-30 15:24:59 +02001026 - A ring buffer in the form "ring@<name>", which will correspond to an
1027 in-memory ring buffer accessible over the CLI using the "show events"
1028 command, which will also list existing rings and their sizes. Such
1029 buffers are lost on reload or restart but when used as a complement
1030 this can help troubleshooting by having the logs instantly available.
1031
William Lallemandb2f07452015-05-12 14:27:13 +02001032 You may want to reference some environment variables in the address
1033 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01001034
Willy Tarreau18324f52014-06-27 18:10:07 +02001035 <length> is an optional maximum line length. Log lines larger than this value
1036 will be truncated before being sent. The reason is that syslog
1037 servers act differently on log line length. All servers support the
1038 default value of 1024, but some servers simply drop larger lines
1039 while others do log them. If a server supports long lines, it may
1040 make sense to set this value here in order to avoid truncating long
1041 lines. Similarly, if a server drops long lines, it is preferable to
1042 truncate them before sending them. Accepted values are 80 to 65535
1043 inclusive. The default value of 1024 is generally fine for all
1044 standard usages. Some specific cases of long captures or
Davor Ocelice9ed2812017-12-25 17:49:28 +01001045 JSON-formatted logs may require larger values. You may also need to
1046 increase "tune.http.logurilen" if your request URIs are truncated.
Willy Tarreau18324f52014-06-27 18:10:07 +02001047
Dragan Dosen7ad31542015-09-28 17:16:47 +02001048 <format> is the log format used when generating syslog messages. It may be
1049 one of the following :
1050
1051 rfc3164 The RFC3164 syslog message format. This is the default.
1052 (https://tools.ietf.org/html/rfc3164)
1053
1054 rfc5424 The RFC5424 syslog message format.
1055 (https://tools.ietf.org/html/rfc5424)
1056
Emeric Brun54648852020-07-06 15:54:06 +02001057 priority A message containing only a level plus syslog facility between
1058 angle brackets such as '<63>', followed by the text. The PID,
1059 date, time, process name and system name are omitted. This is
1060 designed to be used with a local log server.
1061
Willy Tarreaue8746a02018-11-12 08:45:00 +01001062 short A message containing only a level between angle brackets such as
1063 '<3>', followed by the text. The PID, date, time, process name
1064 and system name are omitted. This is designed to be used with a
1065 local log server. This format is compatible with what the systemd
1066 logger consumes.
1067
Emeric Brun54648852020-07-06 15:54:06 +02001068 timed A message containing only a level between angle brackets such as
1069 '<3>', followed by ISO date and by the text. The PID, process
1070 name and system name are omitted. This is designed to be
1071 used with a local log server.
1072
1073 iso A message containing only the ISO date, followed by the text.
1074 The PID, process name and system name are omitted. This is
1075 designed to be used with a local log server.
1076
Willy Tarreauc1b06452018-11-12 11:57:56 +01001077 raw A message containing only the text. The level, PID, date, time,
1078 process name and system name are omitted. This is designed to be
1079 used in containers or during development, where the severity only
1080 depends on the file descriptor used (stdout/stderr).
1081
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02001082 <ranges> A list of comma-separated ranges to identify the logs to sample.
1083 This is used to balance the load of the logs to send to the log
1084 server. The limits of the ranges cannot be null. They are numbered
1085 from 1. The size or period (in number of logs) of the sample must be
1086 set with <sample_size> parameter.
1087
1088 <sample_size>
1089 The size of the sample in number of logs to consider when balancing
1090 their logging loads. It is used to balance the load of the logs to
1091 send to the syslog server. This size must be greater or equal to the
1092 maximum of the high limits of the ranges.
1093 (see also <ranges> parameter).
1094
Robert Tsai81ae1952007-12-05 10:47:29 +01001095 <facility> must be one of the 24 standard syslog facilities :
Willy Tarreau6a06a402007-07-15 20:15:28 +02001096
Willy Tarreaue8746a02018-11-12 08:45:00 +01001097 kern user mail daemon auth syslog lpr news
1098 uucp cron auth2 ftp ntp audit alert cron2
1099 local0 local1 local2 local3 local4 local5 local6 local7
1100
Willy Tarreauc1b06452018-11-12 11:57:56 +01001101 Note that the facility is ignored for the "short" and "raw"
1102 formats, but still required as a positional field. It is
1103 recommended to use "daemon" in this case to make it clear that
1104 it's only supposed to be used locally.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001105
1106 An optional level can be specified to filter outgoing messages. By default,
Willy Tarreauf7edefa2009-05-10 17:20:05 +02001107 all messages are sent. If a maximum level is specified, only messages with a
1108 severity at least as important as this level will be sent. An optional minimum
1109 level can be specified. If it is set, logs emitted with a more severe level
1110 than this one will be capped to this level. This is used to avoid sending
1111 "emerg" messages on all terminals on some default syslog configurations.
1112 Eight levels are known :
Willy Tarreau6a06a402007-07-15 20:15:28 +02001113
Cyril Bontédc4d9032012-04-08 21:57:39 +02001114 emerg alert crit err warning notice info debug
Willy Tarreau6a06a402007-07-15 20:15:28 +02001115
Joe Williamsdf5b38f2010-12-29 17:05:48 +01001116log-send-hostname [<string>]
1117 Sets the hostname field in the syslog header. If optional "string" parameter
1118 is set the header is set to the string contents, otherwise uses the hostname
1119 of the system. Generally used if one is not relaying logs through an
1120 intermediate syslog server or for simply customizing the hostname printed in
1121 the logs.
1122
Kevinm48936af2010-12-22 16:08:21 +00001123log-tag <string>
1124 Sets the tag field in the syslog header to this string. It defaults to the
1125 program name as launched from the command line, which usually is "haproxy".
1126 Sometimes it can be useful to differentiate between multiple processes
Willy Tarreau094af4e2015-01-07 15:03:42 +01001127 running on the same host. See also the per-proxy "log-tag" directive.
Kevinm48936af2010-12-22 16:08:21 +00001128
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001129lua-load <file>
1130 This global directive loads and executes a Lua file. This directive can be
1131 used multiple times.
1132
Tim Duesterhusdd74b5f2020-01-12 13:55:40 +01001133lua-prepend-path <string> [<type>]
1134 Prepends the given string followed by a semicolon to Lua's package.<type>
1135 variable.
1136 <type> must either be "path" or "cpath". If <type> is not given it defaults
1137 to "path".
1138
1139 Lua's paths are semicolon delimited lists of patterns that specify how the
1140 `require` function attempts to find the source file of a library. Question
1141 marks (?) within a pattern will be replaced by module name. The path is
1142 evaluated left to right. This implies that paths that are prepended later
1143 will be checked earlier.
1144
1145 As an example by specifying the following path:
1146
1147 lua-prepend-path /usr/share/haproxy-lua/?/init.lua
1148 lua-prepend-path /usr/share/haproxy-lua/?.lua
1149
1150 When `require "example"` is being called Lua will first attempt to load the
1151 /usr/share/haproxy-lua/example.lua script, if that does not exist the
1152 /usr/share/haproxy-lua/example/init.lua will be attempted and the default
1153 paths if that does not exist either.
1154
1155 See https://www.lua.org/pil/8.1.html for the details within the Lua
1156 documentation.
1157
William Lallemand4cfede82017-11-24 22:02:34 +01001158master-worker [no-exit-on-failure]
William Lallemande202b1e2017-06-01 17:38:56 +02001159 Master-worker mode. It is equivalent to the command line "-W" argument.
1160 This mode will launch a "master" which will monitor the "workers". Using
1161 this mode, you can reload HAProxy directly by sending a SIGUSR2 signal to
Davor Ocelice9ed2812017-12-25 17:49:28 +01001162 the master. The master-worker mode is compatible either with the foreground
William Lallemande202b1e2017-06-01 17:38:56 +02001163 or daemon mode. It is recommended to use this mode with multiprocess and
1164 systemd.
William Lallemand4cfede82017-11-24 22:02:34 +01001165 By default, if a worker exits with a bad return code, in the case of a
1166 segfault for example, all workers will be killed, and the master will leave.
1167 It is convenient to combine this behavior with Restart=on-failure in a
1168 systemd unit file in order to relaunch the whole process. If you don't want
1169 this behavior, you must use the keyword "no-exit-on-failure".
William Lallemande202b1e2017-06-01 17:38:56 +02001170
William Lallemand4cfede82017-11-24 22:02:34 +01001171 See also "-W" in the management guide.
William Lallemande202b1e2017-06-01 17:38:56 +02001172
William Lallemand27edc4b2019-05-07 17:49:33 +02001173mworker-max-reloads <number>
1174 In master-worker mode, this option limits the number of time a worker can
John Roeslerfb2fce12019-07-10 15:45:51 -05001175 survive to a reload. If the worker did not leave after a reload, once its
William Lallemand27edc4b2019-05-07 17:49:33 +02001176 number of reloads is greater than this number, the worker will receive a
1177 SIGTERM. This option helps to keep under control the number of workers.
1178 See also "show proc" in the Management Guide.
1179
Willy Tarreau6a06a402007-07-15 20:15:28 +02001180nbproc <number>
1181 Creates <number> processes when going daemon. This requires the "daemon"
1182 mode. By default, only one process is created, which is the recommended mode
1183 of operation. For systems limited to small sets of file descriptors per
Willy Tarreau149ab772019-01-26 14:27:06 +01001184 process, it may be needed to fork multiple daemons. When set to a value
1185 larger than 1, threads are automatically disabled. USING MULTIPLE PROCESSES
Willy Tarreau1f672a82019-01-26 14:20:55 +01001186 IS HARDER TO DEBUG AND IS REALLY DISCOURAGED. See also "daemon" and
1187 "nbthread".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001188
Christopher Fauletbe0faa22017-08-29 15:37:10 +02001189nbthread <number>
1190 This setting is only available when support for threads was built in. It
Willy Tarreau26f6ae12019-02-02 12:56:15 +01001191 makes haproxy run on <number> threads. This is exclusive with "nbproc". While
1192 "nbproc" historically used to be the only way to use multiple processors, it
1193 also involved a number of shortcomings related to the lack of synchronization
1194 between processes (health-checks, peers, stick-tables, stats, ...) which do
1195 not affect threads. As such, any modern configuration is strongly encouraged
Willy Tarreau149ab772019-01-26 14:27:06 +01001196 to migrate away from "nbproc" to "nbthread". "nbthread" also works when
1197 HAProxy is started in foreground. On some platforms supporting CPU affinity,
1198 when nbproc is not used, the default "nbthread" value is automatically set to
1199 the number of CPUs the process is bound to upon startup. This means that the
1200 thread count can easily be adjusted from the calling process using commands
1201 like "taskset" or "cpuset". Otherwise, this value defaults to 1. The default
1202 value is reported in the output of "haproxy -vv". See also "nbproc".
Christopher Fauletbe0faa22017-08-29 15:37:10 +02001203
Willy Tarreau6a06a402007-07-15 20:15:28 +02001204pidfile <pidfile>
MIZUTA Takeshic32f3942020-08-26 13:46:19 +09001205 Writes PIDs of all daemons into file <pidfile> when daemon mode or writes PID
1206 of master process into file <pidfile> when master-worker mode. This option is
1207 equivalent to the "-p" command line argument. The file must be accessible to
1208 the user starting the process. See also "daemon" and "master-worker".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001209
Willy Tarreau119e50e2020-05-22 13:53:29 +02001210pp2-never-send-local
1211 A bug in the PROXY protocol v2 implementation was present in HAProxy up to
1212 version 2.1, causing it to emit a PROXY command instead of a LOCAL command
1213 for health checks. This is particularly minor but confuses some servers'
1214 logs. Sadly, the bug was discovered very late and revealed that some servers
1215 which possibly only tested their PROXY protocol implementation against
1216 HAProxy fail to properly handle the LOCAL command, and permanently remain in
1217 the "down" state when HAProxy checks them. When this happens, it is possible
1218 to enable this global option to revert to the older (bogus) behavior for the
1219 time it takes to contact the affected components' vendors and get them fixed.
1220 This option is disabled by default and acts on all servers having the
1221 "send-proxy-v2" statement.
1222
Willy Tarreau1d549722016-02-16 12:41:57 +01001223presetenv <name> <value>
1224 Sets environment variable <name> to value <value>. If the variable exists, it
1225 is NOT overwritten. The changes immediately take effect so that the next line
1226 in the configuration file sees the new value. See also "setenv", "resetenv",
1227 and "unsetenv".
1228
1229resetenv [<name> ...]
1230 Removes all environment variables except the ones specified in argument. It
1231 allows to use a clean controlled environment before setting new values with
1232 setenv or unsetenv. Please note that some internal functions may make use of
1233 some environment variables, such as time manipulation functions, but also
1234 OpenSSL or even external checks. This must be used with extreme care and only
1235 after complete validation. The changes immediately take effect so that the
1236 next line in the configuration file sees the new environment. See also
1237 "setenv", "presetenv", and "unsetenv".
1238
Christopher Fauletff4121f2017-11-22 16:38:49 +01001239stats bind-process [ all | odd | even | <process_num>[-[process_num>]] ] ...
Willy Tarreau35b7b162012-10-22 23:17:18 +02001240 Limits the stats socket to a certain set of processes numbers. By default the
1241 stats socket is bound to all processes, causing a warning to be emitted when
1242 nbproc is greater than 1 because there is no way to select the target process
1243 when connecting. However, by using this setting, it becomes possible to pin
1244 the stats socket to a specific set of processes, typically the first one. The
1245 warning will automatically be disabled when this setting is used, whatever
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001246 the number of processes used. The maximum process ID depends on the machine's
Christopher Fauletff4121f2017-11-22 16:38:49 +01001247 word size (32 or 64). Ranges can be partially defined. The higher bound can
1248 be omitted. In such case, it is replaced by the corresponding maximum
1249 value. A better option consists in using the "process" setting of the "stats
1250 socket" line to force the process on each line.
Willy Tarreau35b7b162012-10-22 23:17:18 +02001251
Baptiste Assmann5626f482015-08-23 10:00:10 +02001252server-state-base <directory>
1253 Specifies the directory prefix to be prepended in front of all servers state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001254 file names which do not start with a '/'. See also "server-state-file",
1255 "load-server-state-from-file" and "server-state-file-name".
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +02001256
1257server-state-file <file>
1258 Specifies the path to the file containing state of servers. If the path starts
1259 with a slash ('/'), it is considered absolute, otherwise it is considered
1260 relative to the directory specified using "server-state-base" (if set) or to
1261 the current directory. Before reloading HAProxy, it is possible to save the
1262 servers' current state using the stats command "show servers state". The
1263 output of this command must be written in the file pointed by <file>. When
1264 starting up, before handling traffic, HAProxy will read, load and apply state
1265 for each server found in the file and available in its current running
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001266 configuration. See also "server-state-base" and "show servers state",
1267 "load-server-state-from-file" and "server-state-file-name"
Baptiste Assmann5626f482015-08-23 10:00:10 +02001268
Willy Tarreau1d549722016-02-16 12:41:57 +01001269setenv <name> <value>
1270 Sets environment variable <name> to value <value>. If the variable exists, it
1271 is overwritten. The changes immediately take effect so that the next line in
1272 the configuration file sees the new value. See also "presetenv", "resetenv",
1273 and "unsetenv".
1274
Willy Tarreau636848a2019-04-15 19:38:50 +02001275set-dumpable
1276 This option is better left disabled by default and enabled only upon a
William Dauchyec730982019-10-27 20:08:10 +01001277 developer's request. If it has been enabled, it may still be forcibly
1278 disabled by prefixing it with the "no" keyword. It has no impact on
1279 performance nor stability but will try hard to re-enable core dumps that were
1280 possibly disabled by file size limitations (ulimit -f), core size limitations
1281 (ulimit -c), or "dumpability" of a process after changing its UID/GID (such
1282 as /proc/sys/fs/suid_dumpable on Linux). Core dumps might still be limited by
1283 the current directory's permissions (check what directory the file is started
1284 from), the chroot directory's permission (it may be needed to temporarily
1285 disable the chroot directive or to move it to a dedicated writable location),
1286 or any other system-specific constraint. For example, some Linux flavours are
1287 notorious for replacing the default core file with a path to an executable
1288 not even installed on the system (check /proc/sys/kernel/core_pattern). Often,
1289 simply writing "core", "core.%p" or "/var/log/core/core.%p" addresses the
1290 issue. When trying to enable this option waiting for a rare issue to
1291 re-appear, it's often a good idea to first try to obtain such a dump by
1292 issuing, for example, "kill -11" to the haproxy process and verify that it
1293 leaves a core where expected when dying.
Willy Tarreau636848a2019-04-15 19:38:50 +02001294
Willy Tarreau610f04b2014-02-13 11:36:41 +01001295ssl-default-bind-ciphers <ciphers>
1296 This setting is only available when support for OpenSSL was built in. It sets
1297 the default string describing the list of cipher algorithms ("cipher suite")
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +00001298 that are negotiated during the SSL/TLS handshake up to TLSv1.2 for all
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001299 "bind" lines which do not explicitly define theirs. The format of the string
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001300 is defined in "man 1 ciphers" from OpenSSL man pages. For background
1301 information and recommendations see e.g.
1302 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
1303 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
1304 cipher configuration, please check the "ssl-default-bind-ciphersuites" keyword.
1305 Please check the "bind" keyword for more information.
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001306
1307ssl-default-bind-ciphersuites <ciphersuites>
1308 This setting is only available when support for OpenSSL was built in and
1309 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default string
1310 describing the list of cipher algorithms ("cipher suite") that are negotiated
1311 during the TLSv1.3 handshake for all "bind" lines which do not explicitly define
1312 theirs. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001313 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites". For
1314 cipher configuration for TLSv1.2 and earlier, please check the
1315 "ssl-default-bind-ciphers" keyword. Please check the "bind" keyword for more
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001316 information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001317
Jerome Magninb203ff62020-04-03 15:28:22 +02001318ssl-default-bind-curves <curves>
1319 This setting is only available when support for OpenSSL was built in. It sets
1320 the default string describing the list of elliptic curves algorithms ("curve
1321 suite") that are negotiated during the SSL/TLS handshake with ECDHE. The format
1322 of the string is a colon-delimited list of curve name.
1323 Please check the "bind" keyword for more information.
1324
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001325ssl-default-bind-options [<option>]...
1326 This setting is only available when support for OpenSSL was built in. It sets
1327 default ssl-options to force on all "bind" lines. Please check the "bind"
1328 keyword to see available options.
1329
1330 Example:
1331 global
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +02001332 ssl-default-bind-options ssl-min-ver TLSv1.0 no-tls-tickets
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001333
Willy Tarreau610f04b2014-02-13 11:36:41 +01001334ssl-default-server-ciphers <ciphers>
1335 This setting is only available when support for OpenSSL was built in. It
1336 sets the default string describing the list of cipher algorithms that are
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +00001337 negotiated during the SSL/TLS handshake up to TLSv1.2 with the server,
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001338 for all "server" lines which do not explicitly define theirs. The format of
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001339 the string is defined in "man 1 ciphers" from OpenSSL man pages. For background
1340 information and recommendations see e.g.
1341 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
1342 (https://mozilla.github.io/server-side-tls/ssl-config-generator/).
1343 For TLSv1.3 cipher configuration, please check the
1344 "ssl-default-server-ciphersuites" keyword. Please check the "server" keyword
1345 for more information.
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001346
1347ssl-default-server-ciphersuites <ciphersuites>
1348 This setting is only available when support for OpenSSL was built in and
1349 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default
1350 string describing the list of cipher algorithms that are negotiated during
1351 the TLSv1.3 handshake with the server, for all "server" lines which do not
1352 explicitly define theirs. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001353 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites". For
1354 cipher configuration for TLSv1.2 and earlier, please check the
1355 "ssl-default-server-ciphers" keyword. Please check the "server" keyword for
1356 more information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001357
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001358ssl-default-server-options [<option>]...
1359 This setting is only available when support for OpenSSL was built in. It sets
1360 default ssl-options to force on all "server" lines. Please check the "server"
1361 keyword to see available options.
1362
Remi Gacogne47783ef2015-05-29 15:53:22 +02001363ssl-dh-param-file <file>
1364 This setting is only available when support for OpenSSL was built in. It sets
1365 the default DH parameters that are used during the SSL/TLS handshake when
1366 ephemeral Diffie-Hellman (DHE) key exchange is used, for all "bind" lines
Davor Ocelice9ed2812017-12-25 17:49:28 +01001367 which do not explicitly define theirs. It will be overridden by custom DH
Remi Gacogne47783ef2015-05-29 15:53:22 +02001368 parameters found in a bind certificate file if any. If custom DH parameters
Cyril Bonté307ee1e2015-09-28 23:16:06 +02001369 are not specified either by using ssl-dh-param-file or by setting them
1370 directly in the certificate file, pre-generated DH parameters of the size
1371 specified by tune.ssl.default-dh-param will be used. Custom parameters are
1372 known to be more secure and therefore their use is recommended.
Remi Gacogne47783ef2015-05-29 15:53:22 +02001373 Custom DH parameters may be generated by using the OpenSSL command
1374 "openssl dhparam <size>", where size should be at least 2048, as 1024-bit DH
1375 parameters should not be considered secure anymore.
1376
William Lallemand4c5adbf2020-02-24 14:23:22 +01001377ssl-load-extra-files <none|all|bundle|sctl|ocsp|issuer|key>*
William Lallemand3af48e72020-02-03 17:15:52 +01001378 This setting alters the way HAProxy will look for unspecified files during
Jerome Magnin587be9c2020-09-07 11:55:57 +02001379 the loading of the SSL certificates associated to "bind" lines. It does not
1380 apply to certificates used for client authentication on "server" lines.
William Lallemand3af48e72020-02-03 17:15:52 +01001381
1382 By default, HAProxy discovers automatically a lot of files not specified in
1383 the configuration, and you may want to disable this behavior if you want to
1384 optimize the startup time.
1385
1386 "none": Only load the files specified in the configuration. Don't try to load
1387 a certificate bundle if the file does not exist. In the case of a directory,
1388 it won't try to bundle the certificates if they have the same basename.
1389
1390 "all": This is the default behavior, it will try to load everything,
William Lallemand4c5adbf2020-02-24 14:23:22 +01001391 bundles, sctl, ocsp, issuer, key.
William Lallemand3af48e72020-02-03 17:15:52 +01001392
1393 "bundle": When a file specified in the configuration does not exist, HAProxy
William Lallemandf9ff3ec2020-10-02 17:57:44 +02001394 will try to load a "cert bundle".
1395
1396 Starting from HAProxy 2.3, the bundles are not loaded in the same OpenSSL
1397 certificate store, instead it will loads each certificate in a separate
1398 store which is equivalent to declaring multiple "crt". OpenSSL 1.1.1 is
1399 required to achieve this. Which means that bundles are now used only for
1400 backward compatibility and are not mandatory anymore to do an hybrid RSA/ECC
1401 bind configuration..
1402
1403 To associate these PEM files into a "cert bundle" that is recognized by
1404 haproxy, they must be named in the following way: All PEM files that are to
1405 be bundled must have the same base name, with a suffix indicating the key
1406 type. Currently, three suffixes are supported: rsa, dsa and ecdsa. For
1407 example, if www.example.com has two PEM files, an RSA file and an ECDSA
1408 file, they must be named: "example.pem.rsa" and "example.pem.ecdsa". The
1409 first part of the filename is arbitrary; only the suffix matters. To load
1410 this bundle into haproxy, specify the base name only:
1411
1412 Example : bind :8443 ssl crt example.pem
1413
1414 Note that the suffix is not given to haproxy; this tells haproxy to look for
1415 a cert bundle.
1416
1417 HAProxy will load all PEM files in the bundle as if they were configured
1418 separately in several "crt".
1419
1420 The bundle loading does not have an impact anymore on the directory loading
1421 since files are loading separately.
1422
1423 On the CLI, bundles are seen as separate files, and the bundle extension is
1424 required to commit them.
1425
William Dauchy57dd6f12020-10-06 15:22:37 +02001426 OCSP files (.ocsp), issuer files (.issuer), Certificate Transparency (.sctl)
William Lallemandf9ff3ec2020-10-02 17:57:44 +02001427 as well as private keys (.key) are supported with multi-cert bundling.
William Lallemand3af48e72020-02-03 17:15:52 +01001428
1429 "sctl": Try to load "<basename>.sctl" for each crt keyword.
1430
1431 "ocsp": Try to load "<basename>.ocsp" for each crt keyword.
1432
1433 "issuer": Try to load "<basename>.issuer" if the issuer of the OCSP file is
1434 not provided in the PEM file.
1435
William Lallemand4c5adbf2020-02-24 14:23:22 +01001436 "key": If the private key was not provided by the PEM file, try to load a
1437 file "<basename>.key" containing a private key.
1438
William Lallemand3af48e72020-02-03 17:15:52 +01001439 The default behavior is "all".
1440
1441 Example:
1442 ssl-load-extra-files bundle sctl
1443 ssl-load-extra-files sctl ocsp issuer
1444 ssl-load-extra-files none
1445
1446 See also: "crt", section 5.1 about bind options.
1447
Emeric Brun850efd52014-01-29 12:24:34 +01001448ssl-server-verify [none|required]
1449 The default behavior for SSL verify on servers side. If specified to 'none',
1450 servers certificates are not verified. The default is 'required' except if
1451 forced using cmdline option '-dV'.
1452
Emmanuel Hocdetc3b7e742020-04-22 11:06:19 +02001453ssl-skip-self-issued-ca
Daniel Corbett67a82712020-07-06 23:01:19 -04001454 Self issued CA, aka x509 root CA, is the anchor for chain validation: as a
Emmanuel Hocdetc3b7e742020-04-22 11:06:19 +02001455 server is useless to send it, client must have it. Standard configuration
1456 need to not include such CA in PEM file. This option allows you to keep such
1457 CA in PEM file without sending it to the client. Use case is to provide
1458 issuer for ocsp without the need for '.issuer' file and be able to share it
1459 with 'issuers-chain-path'. This concerns all certificates without intermediate
1460 certificates. It's useless for BoringSSL, .issuer is ignored because ocsp
William Lallemand9a1d8392020-08-10 17:28:23 +02001461 bits does not need it. Requires at least OpenSSL 1.0.2.
Emmanuel Hocdetc3b7e742020-04-22 11:06:19 +02001462
Willy Tarreauabb175f2012-09-24 12:43:26 +02001463stats socket [<address:port>|<path>] [param*]
1464 Binds a UNIX socket to <path> or a TCPv4/v6 address to <address:port>.
1465 Connections to this socket will return various statistics outputs and even
1466 allow some commands to be issued to change some runtime settings. Please
Willy Tarreau1af20c72017-06-23 16:01:14 +02001467 consult section 9.3 "Unix Socket commands" of Management Guide for more
Kevin Decherf949c7202015-10-13 23:26:44 +02001468 details.
Willy Tarreau6162db22009-10-10 17:13:00 +02001469
Willy Tarreauabb175f2012-09-24 12:43:26 +02001470 All parameters supported by "bind" lines are supported, for instance to
1471 restrict access to some users or their access rights. Please consult
1472 section 5.1 for more information.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001473
1474stats timeout <timeout, in milliseconds>
1475 The default timeout on the stats socket is set to 10 seconds. It is possible
1476 to change this value with "stats timeout". The value must be passed in
Willy Tarreaubefdff12007-12-02 22:27:38 +01001477 milliseconds, or be suffixed by a time unit among { us, ms, s, m, h, d }.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001478
1479stats maxconn <connections>
1480 By default, the stats socket is limited to 10 concurrent connections. It is
1481 possible to change this value with "stats maxconn".
1482
Willy Tarreau6a06a402007-07-15 20:15:28 +02001483uid <number>
1484 Changes the process' user ID to <number>. It is recommended that the user ID
1485 is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
1486 be started with superuser privileges in order to be able to switch to another
1487 one. See also "gid" and "user".
1488
1489ulimit-n <number>
1490 Sets the maximum number of per-process file-descriptors to <number>. By
1491 default, it is automatically computed, so it is recommended not to use this
1492 option.
1493
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001494unix-bind [ prefix <prefix> ] [ mode <mode> ] [ user <user> ] [ uid <uid> ]
1495 [ group <group> ] [ gid <gid> ]
1496
1497 Fixes common settings to UNIX listening sockets declared in "bind" statements.
1498 This is mainly used to simplify declaration of those UNIX sockets and reduce
1499 the risk of errors, since those settings are most commonly required but are
1500 also process-specific. The <prefix> setting can be used to force all socket
1501 path to be relative to that directory. This might be needed to access another
1502 component's chroot. Note that those paths are resolved before haproxy chroots
1503 itself, so they are absolute. The <mode>, <user>, <uid>, <group> and <gid>
1504 all have the same meaning as their homonyms used by the "bind" statement. If
1505 both are specified, the "bind" statement has priority, meaning that the
1506 "unix-bind" settings may be seen as process-wide default settings.
1507
Willy Tarreau1d549722016-02-16 12:41:57 +01001508unsetenv [<name> ...]
1509 Removes environment variables specified in arguments. This can be useful to
1510 hide some sensitive information that are occasionally inherited from the
1511 user's environment during some operations. Variables which did not exist are
1512 silently ignored so that after the operation, it is certain that none of
1513 these variables remain. The changes immediately take effect so that the next
1514 line in the configuration file will not see these variables. See also
1515 "setenv", "presetenv", and "resetenv".
1516
Willy Tarreau6a06a402007-07-15 20:15:28 +02001517user <user name>
1518 Similar to "uid" but uses the UID of user name <user name> from /etc/passwd.
1519 See also "uid" and "group".
1520
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02001521node <name>
1522 Only letters, digits, hyphen and underscore are allowed, like in DNS names.
1523
1524 This statement is useful in HA configurations where two or more processes or
1525 servers share the same IP address. By setting a different node-name on all
1526 nodes, it becomes easy to immediately spot what server is handling the
1527 traffic.
1528
1529description <text>
1530 Add a text that describes the instance.
1531
1532 Please note that it is required to escape certain characters (# for example)
1533 and this text is inserted into a html page so you should avoid using
1534 "<" and ">" characters.
1535
Thomas Holmesdb04f192015-05-18 13:21:39 +0100153651degrees-data-file <file path>
1537 The path of the 51Degrees data file to provide device detection services. The
Davor Ocelice9ed2812017-12-25 17:49:28 +01001538 file should be unzipped and accessible by HAProxy with relevant permissions.
Thomas Holmesdb04f192015-05-18 13:21:39 +01001539
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001540 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001541 compiled with USE_51DEGREES.
1542
Ben Shillitof25e8e52016-12-02 14:25:37 +0000154351degrees-property-name-list [<string> ...]
Thomas Holmesdb04f192015-05-18 13:21:39 +01001544 A list of 51Degrees property names to be load from the dataset. A full list
1545 of names is available on the 51Degrees website:
1546 https://51degrees.com/resources/property-dictionary
1547
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001548 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001549 compiled with USE_51DEGREES.
1550
Dragan Dosen93b38d92015-06-29 16:43:25 +0200155151degrees-property-separator <char>
Thomas Holmesdb04f192015-05-18 13:21:39 +01001552 A char that will be appended to every property value in a response header
1553 containing 51Degrees results. If not set that will be set as ','.
1554
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001555 Please note that this option is only available when haproxy has been
1556 compiled with USE_51DEGREES.
1557
155851degrees-cache-size <number>
1559 Sets the size of the 51Degrees converter cache to <number> entries. This
1560 is an LRU cache which reminds previous device detections and their results.
1561 By default, this cache is disabled.
1562
1563 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001564 compiled with USE_51DEGREES.
1565
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001566wurfl-data-file <file path>
1567 The path of the WURFL data file to provide device detection services. The
1568 file should be accessible by HAProxy with relevant permissions.
1569
1570 Please note that this option is only available when haproxy has been compiled
1571 with USE_WURFL=1.
1572
1573wurfl-information-list [<capability>]*
1574 A space-delimited list of WURFL capabilities, virtual capabilities, property
1575 names we plan to use in injected headers. A full list of capability and
1576 virtual capability names is available on the Scientiamobile website :
1577
1578 https://www.scientiamobile.com/wurflCapability
1579
1580 Valid WURFL properties are:
1581 - wurfl_id Contains the device ID of the matched device.
1582
1583 - wurfl_root_id Contains the device root ID of the matched
1584 device.
1585
1586 - wurfl_isdevroot Tells if the matched device is a root device.
1587 Possible values are "TRUE" or "FALSE".
1588
1589 - wurfl_useragent The original useragent coming with this
1590 particular web request.
1591
1592 - wurfl_api_version Contains a string representing the currently
1593 used Libwurfl API version.
1594
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001595 - wurfl_info A string containing information on the parsed
1596 wurfl.xml and its full path.
1597
1598 - wurfl_last_load_time Contains the UNIX timestamp of the last time
1599 WURFL has been loaded successfully.
1600
1601 - wurfl_normalized_useragent The normalized useragent.
1602
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001603 Please note that this option is only available when haproxy has been compiled
1604 with USE_WURFL=1.
1605
1606wurfl-information-list-separator <char>
1607 A char that will be used to separate values in a response header containing
1608 WURFL results. If not set that a comma (',') will be used by default.
1609
1610 Please note that this option is only available when haproxy has been compiled
1611 with USE_WURFL=1.
1612
1613wurfl-patch-file [<file path>]
1614 A list of WURFL patch file paths. Note that patches are loaded during startup
1615 thus before the chroot.
1616
1617 Please note that this option is only available when haproxy has been compiled
1618 with USE_WURFL=1.
1619
paulborilebad132c2019-04-18 11:57:04 +02001620wurfl-cache-size <size>
1621 Sets the WURFL Useragent cache size. For faster lookups, already processed user
1622 agents are kept in a LRU cache :
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001623 - "0" : no cache is used.
paulborilebad132c2019-04-18 11:57:04 +02001624 - <size> : size of lru cache in elements.
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001625
1626 Please note that this option is only available when haproxy has been compiled
1627 with USE_WURFL=1.
1628
William Dauchy0fec3ab2019-10-27 20:08:11 +01001629strict-limits
William Dauchya5194602020-03-28 19:29:58 +01001630 Makes process fail at startup when a setrlimit fails. Haproxy tries to set the
1631 best setrlimit according to what has been calculated. If it fails, it will
1632 emit a warning. This option is here to guarantee an explicit failure of
1633 haproxy when those limits fail. It is enabled by default. It may still be
1634 forcibly disabled by prefixing it with the "no" keyword.
William Dauchy0fec3ab2019-10-27 20:08:11 +01001635
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016363.2. Performance tuning
Willy Tarreau6a06a402007-07-15 20:15:28 +02001637-----------------------
1638
Willy Tarreaubeb859a2018-11-22 18:07:59 +01001639busy-polling
1640 In some situations, especially when dealing with low latency on processors
1641 supporting a variable frequency or when running inside virtual machines, each
1642 time the process waits for an I/O using the poller, the processor goes back
1643 to sleep or is offered to another VM for a long time, and it causes
1644 excessively high latencies. This option provides a solution preventing the
1645 processor from sleeping by always using a null timeout on the pollers. This
1646 results in a significant latency reduction (30 to 100 microseconds observed)
1647 at the expense of a risk to overheat the processor. It may even be used with
1648 threads, in which case improperly bound threads may heavily conflict,
1649 resulting in a worse performance and high values for the CPU stolen fields
1650 in "show info" output, indicating which threads are misconfigured. It is
1651 important not to let the process run on the same processor as the network
1652 interrupts when this option is used. It is also better to avoid using it on
1653 multiple CPU threads sharing the same core. This option is disabled by
1654 default. If it has been enabled, it may still be forcibly disabled by
1655 prefixing it with the "no" keyword. It is ignored by the "select" and
1656 "poll" pollers.
1657
William Dauchy3894d972019-12-28 15:36:02 +01001658 This option is automatically disabled on old processes in the context of
1659 seamless reload; it avoids too much cpu conflicts when multiple processes
1660 stay around for some time waiting for the end of their current connections.
1661
Willy Tarreau1746eec2014-04-25 10:46:47 +02001662max-spread-checks <delay in milliseconds>
1663 By default, haproxy tries to spread the start of health checks across the
1664 smallest health check interval of all the servers in a farm. The principle is
1665 to avoid hammering services running on the same server. But when using large
1666 check intervals (10 seconds or more), the last servers in the farm take some
1667 time before starting to be tested, which can be a problem. This parameter is
1668 used to enforce an upper bound on delay between the first and the last check,
1669 even if the servers' check intervals are larger. When servers run with
1670 shorter intervals, their intervals will be respected though.
1671
Willy Tarreau6a06a402007-07-15 20:15:28 +02001672maxconn <number>
1673 Sets the maximum per-process number of concurrent connections to <number>. It
1674 is equivalent to the command-line argument "-n". Proxies will stop accepting
1675 connections when this limit is reached. The "ulimit-n" parameter is
Willy Tarreau8274e102014-06-19 15:31:25 +02001676 automatically adjusted according to this value. See also "ulimit-n". Note:
1677 the "select" poller cannot reliably use more than 1024 file descriptors on
1678 some platforms. If your platform only supports select and reports "select
1679 FAILED" on startup, you need to reduce maxconn until it works (slightly
Willy Tarreaub28f3442019-03-04 08:13:43 +01001680 below 500 in general). If this value is not set, it will automatically be
1681 calculated based on the current file descriptors limit reported by the
1682 "ulimit -n" command, possibly reduced to a lower value if a memory limit
1683 is enforced, based on the buffer size, memory allocated to compression, SSL
1684 cache size, and use or not of SSL and the associated maxsslconn (which can
1685 also be automatic).
Willy Tarreau6a06a402007-07-15 20:15:28 +02001686
Willy Tarreau81c25d02011-09-07 15:17:21 +02001687maxconnrate <number>
1688 Sets the maximum per-process number of connections per second to <number>.
1689 Proxies will stop accepting connections when this limit is reached. It can be
1690 used to limit the global capacity regardless of each frontend capacity. It is
1691 important to note that this can only be used as a service protection measure,
1692 as there will not necessarily be a fair share between frontends when the
1693 limit is reached, so it's a good idea to also limit each frontend to some
1694 value close to its expected share. Also, lowering tune.maxaccept can improve
1695 fairness.
1696
William Lallemandd85f9172012-11-09 17:05:39 +01001697maxcomprate <number>
1698 Sets the maximum per-process input compression rate to <number> kilobytes
Davor Ocelice9ed2812017-12-25 17:49:28 +01001699 per second. For each session, if the maximum is reached, the compression
William Lallemandd85f9172012-11-09 17:05:39 +01001700 level will be decreased during the session. If the maximum is reached at the
1701 beginning of a session, the session will not compress at all. If the maximum
1702 is not reached, the compression level will be increased up to
Davor Ocelice9ed2812017-12-25 17:49:28 +01001703 tune.comp.maxlevel. A value of zero means there is no limit, this is the
William Lallemandd85f9172012-11-09 17:05:39 +01001704 default value.
1705
William Lallemand072a2bf2012-11-20 17:01:01 +01001706maxcompcpuusage <number>
1707 Sets the maximum CPU usage HAProxy can reach before stopping the compression
1708 for new requests or decreasing the compression level of current requests.
1709 It works like 'maxcomprate' but measures CPU usage instead of incoming data
1710 bandwidth. The value is expressed in percent of the CPU used by haproxy. In
1711 case of multiple processes (nbproc > 1), each process manages its individual
1712 usage. A value of 100 disable the limit. The default value is 100. Setting
1713 a lower value will prevent the compression work from slowing the whole
1714 process down and from introducing high latencies.
1715
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001716maxpipes <number>
1717 Sets the maximum per-process number of pipes to <number>. Currently, pipes
1718 are only used by kernel-based tcp splicing. Since a pipe contains two file
1719 descriptors, the "ulimit-n" value will be increased accordingly. The default
1720 value is maxconn/4, which seems to be more than enough for most heavy usages.
1721 The splice code dynamically allocates and releases pipes, and can fall back
1722 to standard copy, so setting this value too low may only impact performance.
1723
Willy Tarreau93e7c002013-10-07 18:51:07 +02001724maxsessrate <number>
1725 Sets the maximum per-process number of sessions per second to <number>.
1726 Proxies will stop accepting connections when this limit is reached. It can be
1727 used to limit the global capacity regardless of each frontend capacity. It is
1728 important to note that this can only be used as a service protection measure,
1729 as there will not necessarily be a fair share between frontends when the
1730 limit is reached, so it's a good idea to also limit each frontend to some
1731 value close to its expected share. Also, lowering tune.maxaccept can improve
1732 fairness.
1733
Willy Tarreau403edff2012-09-06 11:58:37 +02001734maxsslconn <number>
1735 Sets the maximum per-process number of concurrent SSL connections to
1736 <number>. By default there is no SSL-specific limit, which means that the
1737 global maxconn setting will apply to all connections. Setting this limit
1738 avoids having openssl use too much memory and crash when malloc returns NULL
1739 (since it unfortunately does not reliably check for such conditions). Note
1740 that the limit applies both to incoming and outgoing connections, so one
1741 connection which is deciphered then ciphered accounts for 2 SSL connections.
Willy Tarreaud0256482015-01-15 21:45:22 +01001742 If this value is not set, but a memory limit is enforced, this value will be
1743 automatically computed based on the memory limit, maxconn, the buffer size,
1744 memory allocated to compression, SSL cache size, and use of SSL in either
1745 frontends, backends or both. If neither maxconn nor maxsslconn are specified
1746 when there is a memory limit, haproxy will automatically adjust these values
1747 so that 100% of the connections can be made over SSL with no risk, and will
1748 consider the sides where it is enabled (frontend, backend, both).
Willy Tarreau403edff2012-09-06 11:58:37 +02001749
Willy Tarreaue43d5322013-10-07 20:01:52 +02001750maxsslrate <number>
1751 Sets the maximum per-process number of SSL sessions per second to <number>.
1752 SSL listeners will stop accepting connections when this limit is reached. It
1753 can be used to limit the global SSL CPU usage regardless of each frontend
1754 capacity. It is important to note that this can only be used as a service
1755 protection measure, as there will not necessarily be a fair share between
1756 frontends when the limit is reached, so it's a good idea to also limit each
1757 frontend to some value close to its expected share. It is also important to
1758 note that the sessions are accounted before they enter the SSL stack and not
1759 after, which also protects the stack against bad handshakes. Also, lowering
1760 tune.maxaccept can improve fairness.
1761
William Lallemand9d5f5482012-11-07 16:12:57 +01001762maxzlibmem <number>
1763 Sets the maximum amount of RAM in megabytes per process usable by the zlib.
1764 When the maximum amount is reached, future sessions will not compress as long
1765 as RAM is unavailable. When sets to 0, there is no limit.
William Lallemande3a7d992012-11-20 11:25:20 +01001766 The default value is 0. The value is available in bytes on the UNIX socket
1767 with "show info" on the line "MaxZlibMemUsage", the memory used by zlib is
1768 "ZlibMemUsage" in bytes.
1769
Willy Tarreau6a06a402007-07-15 20:15:28 +02001770noepoll
1771 Disables the use of the "epoll" event polling system on Linux. It is
1772 equivalent to the command-line argument "-de". The next polling system
Willy Tarreaue9f49e72012-11-11 17:42:00 +01001773 used will generally be "poll". See also "nopoll".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001774
1775nokqueue
1776 Disables the use of the "kqueue" event polling system on BSD. It is
1777 equivalent to the command-line argument "-dk". The next polling system
1778 used will generally be "poll". See also "nopoll".
1779
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +00001780noevports
1781 Disables the use of the event ports event polling system on SunOS systems
1782 derived from Solaris 10 and later. It is equivalent to the command-line
1783 argument "-dv". The next polling system used will generally be "poll". See
1784 also "nopoll".
1785
Willy Tarreau6a06a402007-07-15 20:15:28 +02001786nopoll
1787 Disables the use of the "poll" event polling system. It is equivalent to the
1788 command-line argument "-dp". The next polling system used will be "select".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001789 It should never be needed to disable "poll" since it's available on all
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +00001790 platforms supported by HAProxy. See also "nokqueue", "noepoll" and
1791 "noevports".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001792
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001793nosplice
1794 Disables the use of kernel tcp splicing between sockets on Linux. It is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001795 equivalent to the command line argument "-dS". Data will then be copied
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001796 using conventional and more portable recv/send calls. Kernel tcp splicing is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001797 limited to some very recent instances of kernel 2.6. Most versions between
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001798 2.6.25 and 2.6.28 are buggy and will forward corrupted data, so they must not
1799 be used. This option makes it easier to globally disable kernel splicing in
1800 case of doubt. See also "option splice-auto", "option splice-request" and
1801 "option splice-response".
1802
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001803nogetaddrinfo
1804 Disables the use of getaddrinfo(3) for name resolving. It is equivalent to
1805 the command line argument "-dG". Deprecated gethostbyname(3) will be used.
1806
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +00001807noreuseport
1808 Disables the use of SO_REUSEPORT - see socket(7). It is equivalent to the
1809 command line argument "-dR".
1810
Willy Tarreaud2d33482019-04-25 17:09:07 +02001811profiling.tasks { auto | on | off }
1812 Enables ('on') or disables ('off') per-task CPU profiling. When set to 'auto'
1813 the profiling automatically turns on a thread when it starts to suffer from
1814 an average latency of 1000 microseconds or higher as reported in the
1815 "avg_loop_us" activity field, and automatically turns off when the latency
John Roeslerfb2fce12019-07-10 15:45:51 -05001816 returns below 990 microseconds (this value is an average over the last 1024
Willy Tarreaud2d33482019-04-25 17:09:07 +02001817 loops so it does not vary quickly and tends to significantly smooth short
1818 spikes). It may also spontaneously trigger from time to time on overloaded
1819 systems, containers, or virtual machines, or when the system swaps (which
1820 must absolutely never happen on a load balancer).
1821
1822 CPU profiling per task can be very convenient to report where the time is
1823 spent and which requests have what effect on which other request. Enabling
1824 it will typically affect the overall's performance by less than 1%, thus it
1825 is recommended to leave it to the default 'auto' value so that it only
1826 operates when a problem is identified. This feature requires a system
Willy Tarreau75c62c22018-11-22 11:02:09 +01001827 supporting the clock_gettime(2) syscall with clock identifiers
1828 CLOCK_MONOTONIC and CLOCK_THREAD_CPUTIME_ID, otherwise the reported time will
1829 be zero. This option may be changed at run time using "set profiling" on the
1830 CLI.
1831
Willy Tarreaufe255b72007-10-14 23:09:26 +02001832spread-checks <0..50, in percent>
Simon Hormand60d6912013-11-25 10:46:36 +09001833 Sometimes it is desirable to avoid sending agent and health checks to
1834 servers at exact intervals, for instance when many logical servers are
1835 located on the same physical server. With the help of this parameter, it
1836 becomes possible to add some randomness in the check interval between 0
1837 and +/- 50%. A value between 2 and 5 seems to show good results. The
1838 default value remains at 0.
Willy Tarreaufe255b72007-10-14 23:09:26 +02001839
Davor Ocelice9ed2812017-12-25 17:49:28 +01001840ssl-engine <name> [algo <comma-separated list of algorithms>]
Grant Zhang872f9c22017-01-21 01:10:18 +00001841 Sets the OpenSSL engine to <name>. List of valid values for <name> may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01001842 obtained using the command "openssl engine". This statement may be used
Grant Zhang872f9c22017-01-21 01:10:18 +00001843 multiple times, it will simply enable multiple crypto engines. Referencing an
1844 unsupported engine will prevent haproxy from starting. Note that many engines
1845 will lead to lower HTTPS performance than pure software with recent
1846 processors. The optional command "algo" sets the default algorithms an ENGINE
1847 will supply using the OPENSSL function ENGINE_set_default_string(). A value
Davor Ocelice9ed2812017-12-25 17:49:28 +01001848 of "ALL" uses the engine for all cryptographic operations. If no list of
1849 algo is specified then the value of "ALL" is used. A comma-separated list
Grant Zhang872f9c22017-01-21 01:10:18 +00001850 of different algorithms may be specified, including: RSA, DSA, DH, EC, RAND,
1851 CIPHERS, DIGESTS, PKEY, PKEY_CRYPTO, PKEY_ASN1. This is the same format that
1852 openssl configuration file uses:
1853 https://www.openssl.org/docs/man1.0.2/apps/config.html
1854
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00001855ssl-mode-async
1856 Adds SSL_MODE_ASYNC mode to the SSL context. This enables asynchronous TLS
Emeric Brun3854e012017-05-17 20:42:48 +02001857 I/O operations if asynchronous capable SSL engines are used. The current
Emeric Brunb5e42a82017-06-06 12:35:14 +00001858 implementation supports a maximum of 32 engines. The Openssl ASYNC API
1859 doesn't support moving read/write buffers and is not compliant with
1860 haproxy's buffer management. So the asynchronous mode is disabled on
John Roeslerfb2fce12019-07-10 15:45:51 -05001861 read/write operations (it is only enabled during initial and renegotiation
Emeric Brunb5e42a82017-06-06 12:35:14 +00001862 handshakes).
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00001863
Willy Tarreau33cb0652014-12-23 22:52:37 +01001864tune.buffers.limit <number>
1865 Sets a hard limit on the number of buffers which may be allocated per process.
1866 The default value is zero which means unlimited. The minimum non-zero value
1867 will always be greater than "tune.buffers.reserve" and should ideally always
1868 be about twice as large. Forcing this value can be particularly useful to
1869 limit the amount of memory a process may take, while retaining a sane
Davor Ocelice9ed2812017-12-25 17:49:28 +01001870 behavior. When this limit is reached, sessions which need a buffer wait for
Willy Tarreau33cb0652014-12-23 22:52:37 +01001871 another one to be released by another session. Since buffers are dynamically
1872 allocated and released, the waiting time is very short and not perceptible
1873 provided that limits remain reasonable. In fact sometimes reducing the limit
1874 may even increase performance by increasing the CPU cache's efficiency. Tests
1875 have shown good results on average HTTP traffic with a limit to 1/10 of the
1876 expected global maxconn setting, which also significantly reduces memory
1877 usage. The memory savings come from the fact that a number of connections
1878 will not allocate 2*tune.bufsize. It is best not to touch this value unless
1879 advised to do so by an haproxy core developer.
1880
Willy Tarreau1058ae72014-12-23 22:40:40 +01001881tune.buffers.reserve <number>
1882 Sets the number of buffers which are pre-allocated and reserved for use only
1883 during memory shortage conditions resulting in failed memory allocations. The
1884 minimum value is 2 and is also the default. There is no reason a user would
1885 want to change this value, it's mostly aimed at haproxy core developers.
1886
Willy Tarreau27a674e2009-08-17 07:23:33 +02001887tune.bufsize <number>
1888 Sets the buffer size to this size (in bytes). Lower values allow more
1889 sessions to coexist in the same amount of RAM, and higher values allow some
1890 applications with very large cookies to work. The default value is 16384 and
1891 can be changed at build time. It is strongly recommended not to change this
1892 from the default value, as very low values will break some services such as
1893 statistics, and values larger than default size will increase memory usage,
1894 possibly causing the system to run out of memory. At least the global maxconn
Willy Tarreau45a66cc2017-11-24 11:28:00 +01001895 parameter should be decreased by the same factor as this one is increased. In
1896 addition, use of HTTP/2 mandates that this value must be 16384 or more. If an
1897 HTTP request is larger than (tune.bufsize - tune.maxrewrite), haproxy will
Dmitry Sivachenkof6f4f7b2012-10-21 18:10:25 +04001898 return HTTP 400 (Bad Request) error. Similarly if an HTTP response is larger
Willy Tarreauc77d3642018-12-12 06:19:42 +01001899 than this size, haproxy will return HTTP 502 (Bad Gateway). Note that the
1900 value set using this parameter will automatically be rounded up to the next
1901 multiple of 8 on 32-bit machines and 16 on 64-bit machines.
Willy Tarreau27a674e2009-08-17 07:23:33 +02001902
Willy Tarreau43961d52010-10-04 20:39:20 +02001903tune.chksize <number>
1904 Sets the check buffer size to this size (in bytes). Higher values may help
1905 find string or regex patterns in very large pages, though doing so may imply
1906 more memory and CPU usage. The default value is 16384 and can be changed at
1907 build time. It is not recommended to change this value, but to use better
1908 checks whenever possible.
1909
William Lallemandf3747832012-11-09 12:33:10 +01001910tune.comp.maxlevel <number>
1911 Sets the maximum compression level. The compression level affects CPU
1912 usage during compression. This value affects CPU usage during compression.
1913 Each session using compression initializes the compression algorithm with
1914 this value. The default value is 1.
1915
Willy Tarreauc299e1e2019-02-27 11:35:12 +01001916tune.fail-alloc
1917 If compiled with DEBUG_FAIL_ALLOC, gives the percentage of chances an
1918 allocation attempt fails. Must be between 0 (no failure) and 100 (no
1919 success). This is useful to debug and make sure memory failures are handled
1920 gracefully.
1921
Willy Tarreaubc52bec2020-06-18 08:58:47 +02001922tune.fd.edge-triggered { on | off } [ EXPERIMENTAL ]
1923 Enables ('on') or disables ('off') the edge-triggered polling mode for FDs
1924 that support it. This is currently only support with epoll. It may noticeably
1925 reduce the number of epoll_ctl() calls and slightly improve performance in
1926 certain scenarios. This is still experimental, it may result in frozen
1927 connections if bugs are still present, and is disabled by default.
1928
Willy Tarreaufe20e5b2017-07-27 11:42:14 +02001929tune.h2.header-table-size <number>
1930 Sets the HTTP/2 dynamic header table size. It defaults to 4096 bytes and
1931 cannot be larger than 65536 bytes. A larger value may help certain clients
1932 send more compact requests, depending on their capabilities. This amount of
1933 memory is consumed for each HTTP/2 connection. It is recommended not to
1934 change it.
1935
Willy Tarreaue6baec02017-07-27 11:45:11 +02001936tune.h2.initial-window-size <number>
1937 Sets the HTTP/2 initial window size, which is the number of bytes the client
Davor Ocelice9ed2812017-12-25 17:49:28 +01001938 can upload before waiting for an acknowledgment from haproxy. This setting
1939 only affects payload contents (i.e. the body of POST requests), not headers.
Willy Tarreaue6baec02017-07-27 11:45:11 +02001940 The default value is 65535, which roughly allows up to 5 Mbps of upload
1941 bandwidth per client over a network showing a 100 ms ping time, or 500 Mbps
1942 over a 1-ms local network. It can make sense to increase this value to allow
1943 faster uploads, or to reduce it to increase fairness when dealing with many
1944 clients. It doesn't affect resource usage.
1945
Willy Tarreau5242ef82017-07-27 11:47:28 +02001946tune.h2.max-concurrent-streams <number>
1947 Sets the HTTP/2 maximum number of concurrent streams per connection (ie the
1948 number of outstanding requests on a single connection). The default value is
1949 100. A larger one may slightly improve page load time for complex sites when
1950 visited over high latency networks, but increases the amount of resources a
1951 single client may allocate. A value of zero disables the limit so a single
1952 client may create as many streams as allocatable by haproxy. It is highly
1953 recommended not to change this value.
1954
Willy Tarreaua24b35c2019-02-21 13:24:36 +01001955tune.h2.max-frame-size <number>
1956 Sets the HTTP/2 maximum frame size that haproxy announces it is willing to
1957 receive to its peers. The default value is the largest between 16384 and the
1958 buffer size (tune.bufsize). In any case, haproxy will not announce support
1959 for frame sizes larger than buffers. The main purpose of this setting is to
1960 allow to limit the maximum frame size setting when using large buffers. Too
1961 large frame sizes might have performance impact or cause some peers to
1962 misbehave. It is highly recommended not to change this value.
1963
Willy Tarreau193b8c62012-11-22 00:17:38 +01001964tune.http.cookielen <number>
1965 Sets the maximum length of captured cookies. This is the maximum value that
1966 the "capture cookie xxx len yyy" will be allowed to take, and any upper value
1967 will automatically be truncated to this one. It is important not to set too
1968 high a value because all cookie captures still allocate this size whatever
1969 their configured value (they share a same pool). This value is per request
1970 per response, so the memory allocated is twice this value per connection.
1971 When not specified, the limit is set to 63 characters. It is recommended not
1972 to change this value.
1973
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001974tune.http.logurilen <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001975 Sets the maximum length of request URI in logs. This prevents truncating long
1976 request URIs with valuable query strings in log lines. This is not related
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001977 to syslog limits. If you increase this limit, you may also increase the
Davor Ocelice9ed2812017-12-25 17:49:28 +01001978 'log ... len yyy' parameter. Your syslog daemon may also need specific
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001979 configuration directives too.
1980 The default value is 1024.
1981
Willy Tarreauac1932d2011-10-24 19:14:41 +02001982tune.http.maxhdr <number>
1983 Sets the maximum number of headers in a request. When a request comes with a
1984 number of headers greater than this value (including the first line), it is
1985 rejected with a "400 Bad Request" status code. Similarly, too large responses
1986 are blocked with "502 Bad Gateway". The default value is 101, which is enough
1987 for all usages, considering that the widely deployed Apache server uses the
1988 same limit. It can be useful to push this limit further to temporarily allow
Christopher Faulet50174f32017-06-21 16:31:35 +02001989 a buggy application to work by the time it gets fixed. The accepted range is
1990 1..32767. Keep in mind that each new header consumes 32bits of memory for
1991 each session, so don't push this limit too high.
Willy Tarreauac1932d2011-10-24 19:14:41 +02001992
Willy Tarreau76cc6992020-07-01 18:49:24 +02001993tune.idle-pool.shared { on | off }
1994 Enables ('on') or disables ('off') sharing of idle connection pools between
1995 threads for a same server. The default is to share them between threads in
1996 order to minimize the number of persistent connections to a server, and to
1997 optimize the connection reuse rate. But to help with debugging or when
1998 suspecting a bug in HAProxy around connection reuse, it can be convenient to
1999 forcefully disable this idle pool sharing between multiple threads, and force
2000 this option to "off". The default is on.
2001
Willy Tarreau7e312732014-02-12 16:35:14 +01002002tune.idletimer <timeout>
2003 Sets the duration after which haproxy will consider that an empty buffer is
2004 probably associated with an idle stream. This is used to optimally adjust
2005 some packet sizes while forwarding large and small data alternatively. The
2006 decision to use splice() or to send large buffers in SSL is modulated by this
2007 parameter. The value is in milliseconds between 0 and 65535. A value of zero
2008 means that haproxy will not try to detect idle streams. The default is 1000,
Davor Ocelice9ed2812017-12-25 17:49:28 +01002009 which seems to correctly detect end user pauses (e.g. read a page before
John Roeslerfb2fce12019-07-10 15:45:51 -05002010 clicking). There should be no reason for changing this value. Please check
Willy Tarreau7e312732014-02-12 16:35:14 +01002011 tune.ssl.maxrecord below.
2012
Willy Tarreau7ac908b2019-02-27 12:02:18 +01002013tune.listener.multi-queue { on | off }
2014 Enables ('on') or disables ('off') the listener's multi-queue accept which
2015 spreads the incoming traffic to all threads a "bind" line is allowed to run
2016 on instead of taking them for itself. This provides a smoother traffic
2017 distribution and scales much better, especially in environments where threads
2018 may be unevenly loaded due to external activity (network interrupts colliding
2019 with one thread for example). This option is enabled by default, but it may
2020 be forcefully disabled for troubleshooting or for situations where it is
2021 estimated that the operating system already provides a good enough
2022 distribution and connections are extremely short-lived.
2023
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002024tune.lua.forced-yield <number>
2025 This directive forces the Lua engine to execute a yield each <number> of
Tim Düsterhus4896c442016-11-29 02:15:19 +01002026 instructions executed. This permits interrupting a long script and allows the
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002027 HAProxy scheduler to process other tasks like accepting connections or
2028 forwarding traffic. The default value is 10000 instructions. If HAProxy often
Davor Ocelice9ed2812017-12-25 17:49:28 +01002029 executes some Lua code but more responsiveness is required, this value can be
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002030 lowered. If the Lua code is quite long and its result is absolutely required
2031 to process the data, the <number> can be increased.
2032
Willy Tarreau32f61e22015-03-18 17:54:59 +01002033tune.lua.maxmem
2034 Sets the maximum amount of RAM in megabytes per process usable by Lua. By
2035 default it is zero which means unlimited. It is important to set a limit to
2036 ensure that a bug in a script will not result in the system running out of
2037 memory.
2038
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002039tune.lua.session-timeout <timeout>
2040 This is the execution timeout for the Lua sessions. This is useful for
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02002041 preventing infinite loops or spending too much time in Lua. This timeout
2042 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01002043 not taken in account. The default timeout is 4s.
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002044
2045tune.lua.task-timeout <timeout>
2046 Purpose is the same as "tune.lua.session-timeout", but this timeout is
2047 dedicated to the tasks. By default, this timeout isn't set because a task may
2048 remain alive during of the lifetime of HAProxy. For example, a task used to
2049 check servers.
2050
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02002051tune.lua.service-timeout <timeout>
2052 This is the execution timeout for the Lua services. This is useful for
2053 preventing infinite loops or spending too much time in Lua. This timeout
2054 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01002055 not taken in account. The default timeout is 4s.
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02002056
Willy Tarreaua0250ba2008-01-06 11:22:57 +01002057tune.maxaccept <number>
Willy Tarreau16a21472012-11-19 12:39:59 +01002058 Sets the maximum number of consecutive connections a process may accept in a
2059 row before switching to other work. In single process mode, higher numbers
2060 give better performance at high connection rates. However in multi-process
2061 modes, keeping a bit of fairness between processes generally is better to
2062 increase performance. This value applies individually to each listener, so
2063 that the number of processes a listener is bound to is taken into account.
2064 This value defaults to 64. In multi-process mode, it is divided by twice
2065 the number of processes the listener is bound to. Setting this value to -1
2066 completely disables the limitation. It should normally not be needed to tweak
2067 this value.
Willy Tarreaua0250ba2008-01-06 11:22:57 +01002068
2069tune.maxpollevents <number>
2070 Sets the maximum amount of events that can be processed at once in a call to
2071 the polling system. The default value is adapted to the operating system. It
2072 has been noticed that reducing it below 200 tends to slightly decrease
2073 latency at the expense of network bandwidth, and increasing it above 200
2074 tends to trade latency for slightly increased bandwidth.
2075
Willy Tarreau27a674e2009-08-17 07:23:33 +02002076tune.maxrewrite <number>
2077 Sets the reserved buffer space to this size in bytes. The reserved space is
2078 used for header rewriting or appending. The first reads on sockets will never
2079 fill more than bufsize-maxrewrite. Historically it has defaulted to half of
2080 bufsize, though that does not make much sense since there are rarely large
2081 numbers of headers to add. Setting it too high prevents processing of large
2082 requests or responses. Setting it too low prevents addition of new headers
2083 to already large requests or to POST requests. It is generally wise to set it
2084 to about 1024. It is automatically readjusted to half of bufsize if it is
2085 larger than that. This means you don't have to worry about it when changing
2086 bufsize.
2087
Willy Tarreauf3045d22015-04-29 16:24:50 +02002088tune.pattern.cache-size <number>
2089 Sets the size of the pattern lookup cache to <number> entries. This is an LRU
2090 cache which reminds previous lookups and their results. It is used by ACLs
2091 and maps on slow pattern lookups, namely the ones using the "sub", "reg",
2092 "dir", "dom", "end", "bin" match methods as well as the case-insensitive
2093 strings. It applies to pattern expressions which means that it will be able
2094 to memorize the result of a lookup among all the patterns specified on a
2095 configuration line (including all those loaded from files). It automatically
2096 invalidates entries which are updated using HTTP actions or on the CLI. The
2097 default cache size is set to 10000 entries, which limits its footprint to
Willy Tarreau403bfbb2019-10-23 06:59:31 +02002098 about 5 MB per process/thread on 32-bit systems and 8 MB per process/thread
2099 on 64-bit systems, as caches are thread/process local. There is a very low
Willy Tarreauf3045d22015-04-29 16:24:50 +02002100 risk of collision in this cache, which is in the order of the size of the
2101 cache divided by 2^64. Typically, at 10000 requests per second with the
2102 default cache size of 10000 entries, there's 1% chance that a brute force
2103 attack could cause a single collision after 60 years, or 0.1% after 6 years.
2104 This is considered much lower than the risk of a memory corruption caused by
2105 aging components. If this is not acceptable, the cache can be disabled by
2106 setting this parameter to 0.
2107
Willy Tarreaubd9a0a72011-10-23 21:14:29 +02002108tune.pipesize <number>
2109 Sets the kernel pipe buffer size to this size (in bytes). By default, pipes
2110 are the default size for the system. But sometimes when using TCP splicing,
2111 it can improve performance to increase pipe sizes, especially if it is
2112 suspected that pipes are not filled and that many calls to splice() are
2113 performed. This has an impact on the kernel's memory footprint, so this must
2114 not be changed if impacts are not understood.
2115
Olivier Houchard88698d92019-04-16 19:07:22 +02002116tune.pool-high-fd-ratio <number>
2117 This setting sets the max number of file descriptors (in percentage) used by
2118 haproxy globally against the maximum number of file descriptors haproxy can
2119 use before we start killing idle connections when we can't reuse a connection
2120 and we have to create a new one. The default is 25 (one quarter of the file
2121 descriptor will mean that roughly half of the maximum front connections can
2122 keep an idle connection behind, anything beyond this probably doesn't make
John Roeslerfb2fce12019-07-10 15:45:51 -05002123 much sense in the general case when targeting connection reuse).
Olivier Houchard88698d92019-04-16 19:07:22 +02002124
Willy Tarreau83ca3052020-07-01 18:30:16 +02002125tune.pool-low-fd-ratio <number>
2126 This setting sets the max number of file descriptors (in percentage) used by
2127 haproxy globally against the maximum number of file descriptors haproxy can
2128 use before we stop putting connection into the idle pool for reuse. The
2129 default is 20.
2130
Willy Tarreaue803de22010-01-21 17:43:04 +01002131tune.rcvbuf.client <number>
2132tune.rcvbuf.server <number>
2133 Forces the kernel socket receive buffer size on the client or the server side
2134 to the specified value in bytes. This value applies to all TCP/HTTP frontends
2135 and backends. It should normally never be set, and the default size (0) lets
John Roeslerfb2fce12019-07-10 15:45:51 -05002136 the kernel auto-tune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01002137 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01002138 order to save kernel memory by preventing it from buffering too large amounts
2139 of received data. Lower values will significantly increase CPU usage though.
2140
Willy Tarreaub22fc302015-12-14 12:04:35 +01002141tune.recv_enough <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01002142 HAProxy uses some hints to detect that a short read indicates the end of the
Willy Tarreaub22fc302015-12-14 12:04:35 +01002143 socket buffers. One of them is that a read returns more than <recv_enough>
2144 bytes, which defaults to 10136 (7 segments of 1448 each). This default value
2145 may be changed by this setting to better deal with workloads involving lots
2146 of short messages such as telnet or SSH sessions.
2147
Olivier Houchard1599b802018-05-24 18:59:04 +02002148tune.runqueue-depth <number>
John Roeslerfb2fce12019-07-10 15:45:51 -05002149 Sets the maximum amount of task that can be processed at once when running
Olivier Houchard1599b802018-05-24 18:59:04 +02002150 tasks. The default value is 200. Increasing it may incur latency when
Willy Tarreaue7723bd2020-06-24 11:11:02 +02002151 dealing with I/Os, making it too small can incur extra overhead. When
2152 experimenting with much larger values, it may be useful to also enable
2153 tune.sched.low-latency to limit the maximum latency to the lowest possible.
2154
2155tune.sched.low-latency { on | off }
2156 Enables ('on') or disables ('off') the low-latency task scheduler. By default
2157 haproxy processes tasks from several classes one class at a time as this is
2158 the most efficient. But when running with large values of tune.runqueue-depth
2159 this can have a measurable effect on request or connection latency. When this
2160 low-latency setting is enabled, tasks of lower priority classes will always
2161 be executed before other ones if they exist. This will permit to lower the
2162 maximum latency experienced by new requests or connections in the middle of
2163 massive traffic, at the expense of a higher impact on this large traffic.
2164 For regular usage it is better to leave this off. The default value is off.
Olivier Houchard1599b802018-05-24 18:59:04 +02002165
Willy Tarreaue803de22010-01-21 17:43:04 +01002166tune.sndbuf.client <number>
2167tune.sndbuf.server <number>
2168 Forces the kernel socket send buffer size on the client or the server side to
2169 the specified value in bytes. This value applies to all TCP/HTTP frontends
2170 and backends. It should normally never be set, and the default size (0) lets
John Roeslerfb2fce12019-07-10 15:45:51 -05002171 the kernel auto-tune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01002172 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01002173 order to save kernel memory by preventing it from buffering too large amounts
2174 of received data. Lower values will significantly increase CPU usage though.
2175 Another use case is to prevent write timeouts with extremely slow clients due
2176 to the kernel waiting for a large part of the buffer to be read before
2177 notifying haproxy again.
2178
Willy Tarreau6ec58db2012-11-16 16:32:15 +01002179tune.ssl.cachesize <number>
Emeric Brunaf9619d2012-11-28 18:47:52 +01002180 Sets the size of the global SSL session cache, in a number of blocks. A block
2181 is large enough to contain an encoded session without peer certificate.
2182 An encoded session with peer certificate is stored in multiple blocks
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002183 depending on the size of the peer certificate. A block uses approximately
Emeric Brunaf9619d2012-11-28 18:47:52 +01002184 200 bytes of memory. The default value may be forced at build time, otherwise
Davor Ocelice9ed2812017-12-25 17:49:28 +01002185 defaults to 20000. When the cache is full, the most idle entries are purged
Emeric Brunaf9619d2012-11-28 18:47:52 +01002186 and reassigned. Higher values reduce the occurrence of such a purge, hence
2187 the number of CPU-intensive SSL handshakes by ensuring that all users keep
2188 their session as long as possible. All entries are pre-allocated upon startup
Emeric Brun22890a12012-12-28 14:41:32 +01002189 and are shared between all processes if "nbproc" is greater than 1. Setting
2190 this value to 0 disables the SSL session cache.
Willy Tarreau6ec58db2012-11-16 16:32:15 +01002191
Emeric Brun8dc60392014-05-09 13:52:00 +02002192tune.ssl.force-private-cache
Lukas Tribus27935782018-10-01 02:00:16 +02002193 This option disables SSL session cache sharing between all processes. It
Emeric Brun8dc60392014-05-09 13:52:00 +02002194 should normally not be used since it will force many renegotiations due to
2195 clients hitting a random process. But it may be required on some operating
2196 systems where none of the SSL cache synchronization method may be used. In
2197 this case, adding a first layer of hash-based load balancing before the SSL
2198 layer might limit the impact of the lack of session sharing.
2199
William Lallemand7d42ef52020-07-06 11:41:30 +02002200tune.ssl.keylog { on | off }
2201 This option activates the logging of the TLS keys. It should be used with
2202 care as it will consume more memory per SSL session and could decrease
2203 performances. This is disabled by default.
2204
2205 These sample fetches should be used to generate the SSLKEYLOGFILE that is
2206 required to decipher traffic with wireshark.
2207
2208 https://developer.mozilla.org/en-US/docs/Mozilla/Projects/NSS/Key_Log_Format
2209
2210 The SSLKEYLOG is a series of lines which are formatted this way:
2211
2212 <Label> <space> <ClientRandom> <space> <Secret>
2213
2214 The ClientRandom is provided by the %[ssl_fc_client_random,hex] sample
2215 fetch, the secret and the Label could be find in the array below. You need
2216 to generate a SSLKEYLOGFILE with all the labels in this array.
2217
2218 The following sample fetches are hexadecimal strings and does not need to be
2219 converted.
2220
2221 SSLKEYLOGFILE Label | Sample fetches for the Secrets
2222 --------------------------------|-----------------------------------------
2223 CLIENT_EARLY_TRAFFIC_SECRET | %[ssl_fc_client_early_traffic_secret]
2224 CLIENT_HANDSHAKE_TRAFFIC_SECRET | %[ssl_fc_client_handshake_traffic_secret]
2225 SERVER_HANDSHAKE_TRAFFIC_SECRET | %[ssl_fc_server_handshake_traffic_secret]
2226 CLIENT_TRAFFIC_SECRET_0 | %[ssl_fc_client_traffic_secret_0]
2227 SERVER_TRAFFIC_SECRET_0 | %[ssl_fc_server_traffic_secret_0]
William Lallemandd742b6c2020-07-07 10:14:56 +02002228 EXPORTER_SECRET | %[ssl_fc_exporter_secret]
2229 EARLY_EXPORTER_SECRET | %[ssl_fc_early_exporter_secret]
William Lallemand7d42ef52020-07-06 11:41:30 +02002230
2231 This is only available with OpenSSL 1.1.1, and useful with TLS1.3 session.
2232
2233 If you want to generate the content of a SSLKEYLOGFILE with TLS < 1.3, you
2234 only need this line:
2235
2236 "CLIENT_RANDOM %[ssl_fc_client_random,hex] %[ssl_fc_session_key,hex]"
2237
Emeric Brun4f65bff2012-11-16 15:11:00 +01002238tune.ssl.lifetime <timeout>
2239 Sets how long a cached SSL session may remain valid. This time is expressed
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002240 in seconds and defaults to 300 (5 min). It is important to understand that it
Emeric Brun4f65bff2012-11-16 15:11:00 +01002241 does not guarantee that sessions will last that long, because if the cache is
2242 full, the longest idle sessions will be purged despite their configured
2243 lifetime. The real usefulness of this setting is to prevent sessions from
2244 being used for too long.
2245
Willy Tarreaubfd59462013-02-21 07:46:09 +01002246tune.ssl.maxrecord <number>
2247 Sets the maximum amount of bytes passed to SSL_write() at a time. Default
2248 value 0 means there is no limit. Over SSL/TLS, the client can decipher the
2249 data only once it has received a full record. With large records, it means
2250 that clients might have to download up to 16kB of data before starting to
2251 process them. Limiting the value can improve page load times on browsers
2252 located over high latency or low bandwidth networks. It is suggested to find
2253 optimal values which fit into 1 or 2 TCP segments (generally 1448 bytes over
2254 Ethernet with TCP timestamps enabled, or 1460 when timestamps are disabled),
2255 keeping in mind that SSL/TLS add some overhead. Typical values of 1419 and
2256 2859 gave good results during tests. Use "strace -e trace=write" to find the
Davor Ocelice9ed2812017-12-25 17:49:28 +01002257 best value. HAProxy will automatically switch to this setting after an idle
Willy Tarreau7e312732014-02-12 16:35:14 +01002258 stream has been detected (see tune.idletimer above).
Willy Tarreaubfd59462013-02-21 07:46:09 +01002259
Remi Gacognef46cd6e2014-06-12 14:58:40 +02002260tune.ssl.default-dh-param <number>
2261 Sets the maximum size of the Diffie-Hellman parameters used for generating
2262 the ephemeral/temporary Diffie-Hellman key in case of DHE key exchange. The
2263 final size will try to match the size of the server's RSA (or DSA) key (e.g,
2264 a 2048 bits temporary DH key for a 2048 bits RSA key), but will not exceed
Willy Tarreau3ba77d22020-05-08 09:31:18 +02002265 this maximum value. Default value if 2048. Only 1024 or higher values are
Remi Gacognef46cd6e2014-06-12 14:58:40 +02002266 allowed. Higher values will increase the CPU load, and values greater than
2267 1024 bits are not supported by Java 7 and earlier clients. This value is not
Remi Gacogne47783ef2015-05-29 15:53:22 +02002268 used if static Diffie-Hellman parameters are supplied either directly
2269 in the certificate file or by using the ssl-dh-param-file parameter.
Remi Gacognef46cd6e2014-06-12 14:58:40 +02002270
Christopher Faulet31af49d2015-06-09 17:29:50 +02002271tune.ssl.ssl-ctx-cache-size <number>
2272 Sets the size of the cache used to store generated certificates to <number>
2273 entries. This is a LRU cache. Because generating a SSL certificate
2274 dynamically is expensive, they are cached. The default cache size is set to
2275 1000 entries.
2276
Thierry FOURNIER5bf77322017-02-25 12:45:22 +01002277tune.ssl.capture-cipherlist-size <number>
2278 Sets the maximum size of the buffer used for capturing client-hello cipher
2279 list. If the value is 0 (default value) the capture is disabled, otherwise
2280 a buffer is allocated for each SSL/TLS connection.
2281
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002282tune.vars.global-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01002283tune.vars.proc-max-size <size>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002284tune.vars.reqres-max-size <size>
2285tune.vars.sess-max-size <size>
2286tune.vars.txn-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01002287 These five tunes help to manage the maximum amount of memory used by the
2288 variables system. "global" limits the overall amount of memory available for
2289 all scopes. "proc" limits the memory for the process scope, "sess" limits the
2290 memory for the session scope, "txn" for the transaction scope, and "reqres"
2291 limits the memory for each request or response processing.
2292 Memory accounting is hierarchical, meaning more coarse grained limits include
2293 the finer grained ones: "proc" includes "sess", "sess" includes "txn", and
2294 "txn" includes "reqres".
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002295
Daniel Schneller0b547052016-03-21 20:46:57 +01002296 For example, when "tune.vars.sess-max-size" is limited to 100,
2297 "tune.vars.txn-max-size" and "tune.vars.reqres-max-size" cannot exceed
2298 100 either. If we create a variable "txn.var" that contains 100 bytes,
2299 all available space is consumed.
2300 Notice that exceeding the limits at runtime will not result in an error
2301 message, but values might be cut off or corrupted. So make sure to accurately
2302 plan for the amount of space needed to store all your variables.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002303
William Lallemanda509e4c2012-11-07 16:54:34 +01002304tune.zlib.memlevel <number>
2305 Sets the memLevel parameter in zlib initialization for each session. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002306 defines how much memory should be allocated for the internal compression
William Lallemanda509e4c2012-11-07 16:54:34 +01002307 state. A value of 1 uses minimum memory but is slow and reduces compression
Davor Ocelice9ed2812017-12-25 17:49:28 +01002308 ratio, a value of 9 uses maximum memory for optimal speed. Can be a value
William Lallemanda509e4c2012-11-07 16:54:34 +01002309 between 1 and 9. The default value is 8.
2310
2311tune.zlib.windowsize <number>
2312 Sets the window size (the size of the history buffer) as a parameter of the
2313 zlib initialization for each session. Larger values of this parameter result
Davor Ocelice9ed2812017-12-25 17:49:28 +01002314 in better compression at the expense of memory usage. Can be a value between
2315 8 and 15. The default value is 15.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002316
Willy Tarreauc57f0e22009-05-10 13:12:33 +020023173.3. Debugging
2318--------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02002319
Willy Tarreau1b857852020-02-25 11:27:22 +01002320debug (deprecated)
Willy Tarreau6a06a402007-07-15 20:15:28 +02002321 Enables debug mode which dumps to stdout all exchanges, and disables forking
2322 into background. It is the equivalent of the command-line argument "-d". It
2323 should never be used in a production configuration since it may prevent full
2324 system startup.
2325
2326quiet
2327 Do not display any message during startup. It is equivalent to the command-
2328 line argument "-q".
2329
Willy Tarreau3eb10b82020-04-15 16:42:39 +02002330zero-warning
2331 When this option is set, haproxy will refuse to start if any warning was
2332 emitted while processing the configuration. It is highly recommended to set
2333 this option on configurations that are not changed often, as it helps detect
2334 subtle mistakes and keep the configuration clean and forward-compatible. Note
2335 that "haproxy -c" will also report errors in such a case. This option is
2336 equivalent to command line argument "-dW".
2337
Emeric Brunf099e792010-09-27 12:05:28 +02002338
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010023393.4. Userlists
2340--------------
2341It is possible to control access to frontend/backend/listen sections or to
2342http stats by allowing only authenticated and authorized users. To do this,
2343it is required to create at least one userlist and to define users.
2344
2345userlist <listname>
Cyril Bonté78caf842010-03-10 22:41:43 +01002346 Creates new userlist with name <listname>. Many independent userlists can be
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002347 used to store authentication & authorization data for independent customers.
2348
2349group <groupname> [users <user>,<user>,(...)]
Cyril Bonté78caf842010-03-10 22:41:43 +01002350 Adds group <groupname> to the current userlist. It is also possible to
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002351 attach users to this group by using a comma separated list of names
2352 proceeded by "users" keyword.
2353
Cyril Bontéf0c60612010-02-06 14:44:47 +01002354user <username> [password|insecure-password <password>]
2355 [groups <group>,<group>,(...)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002356 Adds user <username> to the current userlist. Both secure (encrypted) and
2357 insecure (unencrypted) passwords can be used. Encrypted passwords are
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01002358 evaluated using the crypt(3) function, so depending on the system's
2359 capabilities, different algorithms are supported. For example, modern Glibc
2360 based Linux systems support MD5, SHA-256, SHA-512, and, of course, the
2361 classic DES-based method of encrypting passwords.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002362
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01002363 Attention: Be aware that using encrypted passwords might cause significantly
2364 increased CPU usage, depending on the number of requests, and the algorithm
2365 used. For any of the hashed variants, the password for each request must
2366 be processed through the chosen algorithm, before it can be compared to the
2367 value specified in the config file. Most current algorithms are deliberately
2368 designed to be expensive to compute to achieve resistance against brute
2369 force attacks. They do not simply salt/hash the clear text password once,
2370 but thousands of times. This can quickly become a major factor in haproxy's
2371 overall CPU consumption!
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002372
2373 Example:
Cyril Bontéf0c60612010-02-06 14:44:47 +01002374 userlist L1
2375 group G1 users tiger,scott
2376 group G2 users xdb,scott
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002377
Cyril Bontéf0c60612010-02-06 14:44:47 +01002378 user tiger password $6$k6y3o.eP$JlKBx9za9667qe4(...)xHSwRv6J.C0/D7cV91
2379 user scott insecure-password elgato
2380 user xdb insecure-password hello
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002381
Cyril Bontéf0c60612010-02-06 14:44:47 +01002382 userlist L2
2383 group G1
2384 group G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002385
Cyril Bontéf0c60612010-02-06 14:44:47 +01002386 user tiger password $6$k6y3o.eP$JlKBx(...)xHSwRv6J.C0/D7cV91 groups G1
2387 user scott insecure-password elgato groups G1,G2
2388 user xdb insecure-password hello groups G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002389
2390 Please note that both lists are functionally identical.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002391
Emeric Brunf099e792010-09-27 12:05:28 +02002392
23933.5. Peers
Cyril Bontédc4d9032012-04-08 21:57:39 +02002394----------
Emeric Brun94900952015-06-11 18:25:54 +02002395It is possible to propagate entries of any data-types in stick-tables between
2396several haproxy instances over TCP connections in a multi-master fashion. Each
2397instance pushes its local updates and insertions to remote peers. The pushed
2398values overwrite remote ones without aggregation. Interrupted exchanges are
2399automatically detected and recovered from the last known point.
2400In addition, during a soft restart, the old process connects to the new one
2401using such a TCP connection to push all its entries before the new process
2402tries to connect to other peers. That ensures very fast replication during a
2403reload, it typically takes a fraction of a second even for large tables.
2404Note that Server IDs are used to identify servers remotely, so it is important
2405that configurations look similar or at least that the same IDs are forced on
2406each server on all participants.
Emeric Brunf099e792010-09-27 12:05:28 +02002407
2408peers <peersect>
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002409 Creates a new peer list with name <peersect>. It is an independent section,
Emeric Brunf099e792010-09-27 12:05:28 +02002410 which is referenced by one or more stick-tables.
2411
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002412bind [<address>]:<port_range> [, ...] [param*]
2413 Defines the binding parameters of the local peer of this "peers" section.
2414 Such lines are not supported with "peer" line in the same "peers" section.
2415
Willy Tarreau77e4bd12015-05-01 20:02:17 +02002416disabled
2417 Disables a peers section. It disables both listening and any synchronization
2418 related to this section. This is provided to disable synchronization of stick
2419 tables without having to comment out all "peers" references.
2420
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002421default-bind [param*]
2422 Defines the binding parameters for the local peer, excepted its address.
2423
2424default-server [param*]
2425 Change default options for a server in a "peers" section.
2426
2427 Arguments:
2428 <param*> is a list of parameters for this server. The "default-server"
2429 keyword accepts an important number of options and has a complete
2430 section dedicated to it. Please refer to section 5 for more
2431 details.
2432
2433
2434 See also: "server" and section 5 about server options
2435
Willy Tarreau77e4bd12015-05-01 20:02:17 +02002436enable
2437 This re-enables a disabled peers section which was previously disabled.
2438
Frédéric Lécailleb6f759b2019-11-05 09:57:45 +01002439log <address> [len <length>] [format <format>] [sample <ranges>:<smp_size>]
2440 <facility> [<level> [<minlevel>]]
2441 "peers" sections support the same "log" keyword as for the proxies to
2442 log information about the "peers" listener. See "log" option for proxies for
2443 more details.
2444
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002445peer <peername> <ip>:<port> [param*]
Emeric Brunf099e792010-09-27 12:05:28 +02002446 Defines a peer inside a peers section.
2447 If <peername> is set to the local peer name (by default hostname, or forced
Dragan Dosen13cd54c2020-06-18 18:24:05 +02002448 using "-L" command line option or "localpeer" global configuration setting),
2449 haproxy will listen for incoming remote peer connection on <ip>:<port>.
2450 Otherwise, <ip>:<port> defines where to connect to in order to join the
2451 remote peer, and <peername> is used at the protocol level to identify and
2452 validate the remote peer on the server side.
Emeric Brunf099e792010-09-27 12:05:28 +02002453
2454 During a soft restart, local peer <ip>:<port> is used by the old instance to
2455 connect the new one and initiate a complete replication (teaching process).
2456
2457 It is strongly recommended to have the exact same peers declaration on all
Dragan Dosen13cd54c2020-06-18 18:24:05 +02002458 peers and to only rely on the "-L" command line argument or the "localpeer"
2459 global configuration setting to change the local peer name. This makes it
2460 easier to maintain coherent configuration files across all peers.
Emeric Brunf099e792010-09-27 12:05:28 +02002461
William Lallemandb2f07452015-05-12 14:27:13 +02002462 You may want to reference some environment variables in the address
2463 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01002464
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002465 Note: "peer" keyword may transparently be replaced by "server" keyword (see
2466 "server" keyword explanation below).
2467
2468server <peername> [<ip>:<port>] [param*]
Michael Prokop4438c602019-05-24 10:25:45 +02002469 As previously mentioned, "peer" keyword may be replaced by "server" keyword
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002470 with a support for all "server" parameters found in 5.2 paragraph.
2471 If the underlying peer is local, <ip>:<port> parameters must not be present.
2472 These parameters must be provided on a "bind" line (see "bind" keyword
2473 of this "peers" section).
2474 Some of these parameters are irrelevant for "peers" sections.
2475
2476
Cyril Bontédc4d9032012-04-08 21:57:39 +02002477 Example:
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002478 # The old way.
Emeric Brunf099e792010-09-27 12:05:28 +02002479 peers mypeers
Willy Tarreauf7b30a92010-12-06 22:59:17 +01002480 peer haproxy1 192.168.0.1:1024
2481 peer haproxy2 192.168.0.2:1024
2482 peer haproxy3 10.2.0.1:1024
Emeric Brunf099e792010-09-27 12:05:28 +02002483
2484 backend mybackend
2485 mode tcp
2486 balance roundrobin
2487 stick-table type ip size 20k peers mypeers
2488 stick on src
2489
Willy Tarreauf7b30a92010-12-06 22:59:17 +01002490 server srv1 192.168.0.30:80
2491 server srv2 192.168.0.31:80
Emeric Brunf099e792010-09-27 12:05:28 +02002492
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002493 Example:
2494 peers mypeers
2495 bind 127.0.0.11:10001 ssl crt mycerts/pem
2496 default-server ssl verify none
2497 server hostA 127.0.0.10:10000
2498 server hostB #local peer
Emeric Brunf099e792010-09-27 12:05:28 +02002499
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01002500
2501table <tablename> type {ip | integer | string [len <length>] | binary [len <length>]}
2502 size <size> [expire <expire>] [nopurge] [store <data_type>]*
2503
2504 Configure a stickiness table for the current section. This line is parsed
2505 exactly the same way as the "stick-table" keyword in others section, except
John Roeslerfb2fce12019-07-10 15:45:51 -05002506 for the "peers" argument which is not required here and with an additional
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01002507 mandatory first parameter to designate the stick-table. Contrary to others
2508 sections, there may be several "table" lines in "peers" sections (see also
2509 "stick-table" keyword).
2510
2511 Also be aware of the fact that "peers" sections have their own stick-table
2512 namespaces to avoid collisions between stick-table names identical in
2513 different "peers" section. This is internally handled prepending the "peers"
2514 sections names to the name of the stick-tables followed by a '/' character.
2515 If somewhere else in the configuration file you have to refer to such
2516 stick-tables declared in "peers" sections you must use the prefixed version
2517 of the stick-table name as follows:
2518
2519 peers mypeers
2520 peer A ...
2521 peer B ...
2522 table t1 ...
2523
2524 frontend fe1
2525 tcp-request content track-sc0 src table mypeers/t1
2526
2527 This is also this prefixed version of the stick-table names which must be
2528 used to refer to stick-tables through the CLI.
2529
2530 About "peers" protocol, as only "peers" belonging to the same section may
2531 communicate with each others, there is no need to do such a distinction.
2532 Several "peers" sections may declare stick-tables with the same name.
2533 This is shorter version of the stick-table name which is sent over the network.
2534 There is only a '/' character as prefix to avoid stick-table name collisions between
2535 stick-tables declared as backends and stick-table declared in "peers" sections
2536 as follows in this weird but supported configuration:
2537
2538 peers mypeers
2539 peer A ...
2540 peer B ...
2541 table t1 type string size 10m store gpc0
2542
2543 backend t1
2544 stick-table type string size 10m store gpc0 peers mypeers
2545
Daniel Corbett67a82712020-07-06 23:01:19 -04002546 Here "t1" table declared in "mypeers" section has "mypeers/t1" as global name.
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01002547 "t1" table declared as a backend as "t1" as global name. But at peer protocol
2548 level the former table is named "/t1", the latter is again named "t1".
2549
Simon Horman51a1cf62015-02-03 13:00:44 +090025503.6. Mailers
2551------------
2552It is possible to send email alerts when the state of servers changes.
2553If configured email alerts are sent to each mailer that is configured
2554in a mailers section. Email is sent to mailers using SMTP.
2555
Pieter Baauw386a1272015-08-16 15:26:24 +02002556mailers <mailersect>
Simon Horman51a1cf62015-02-03 13:00:44 +09002557 Creates a new mailer list with the name <mailersect>. It is an
2558 independent section which is referenced by one or more proxies.
2559
2560mailer <mailername> <ip>:<port>
2561 Defines a mailer inside a mailers section.
2562
2563 Example:
2564 mailers mymailers
2565 mailer smtp1 192.168.0.1:587
2566 mailer smtp2 192.168.0.2:587
2567
2568 backend mybackend
2569 mode tcp
2570 balance roundrobin
2571
2572 email-alert mailers mymailers
2573 email-alert from test1@horms.org
2574 email-alert to test2@horms.org
2575
2576 server srv1 192.168.0.30:80
2577 server srv2 192.168.0.31:80
2578
Pieter Baauw235fcfc2016-02-13 15:33:40 +01002579timeout mail <time>
2580 Defines the time available for a mail/connection to be made and send to
2581 the mail-server. If not defined the default value is 10 seconds. To allow
2582 for at least two SYN-ACK packets to be send during initial TCP handshake it
2583 is advised to keep this value above 4 seconds.
2584
2585 Example:
2586 mailers mymailers
2587 timeout mail 20s
2588 mailer smtp1 192.168.0.1:587
Simon Horman51a1cf62015-02-03 13:00:44 +09002589
William Lallemandc9515522019-06-12 16:32:11 +020025903.7. Programs
2591-------------
2592In master-worker mode, it is possible to launch external binaries with the
2593master, these processes are called programs. These programs are launched and
2594managed the same way as the workers.
2595
2596During a reload of HAProxy, those processes are dealing with the same
2597sequence as a worker:
2598
2599 - the master is re-executed
2600 - the master sends a SIGUSR1 signal to the program
2601 - if "option start-on-reload" is not disabled, the master launches a new
2602 instance of the program
2603
2604During a stop, or restart, a SIGTERM is sent to the programs.
2605
2606program <name>
2607 This is a new program section, this section will create an instance <name>
2608 which is visible in "show proc" on the master CLI. (See "9.4. Master CLI" in
2609 the management guide).
2610
2611command <command> [arguments*]
2612 Define the command to start with optional arguments. The command is looked
2613 up in the current PATH if it does not include an absolute path. This is a
2614 mandatory option of the program section. Arguments containing spaces must
2615 be enclosed in quotes or double quotes or be prefixed by a backslash.
2616
Andrew Heberle97236962019-07-12 11:50:26 +08002617user <user name>
2618 Changes the executed command user ID to the <user name> from /etc/passwd.
2619 See also "group".
2620
2621group <group name>
2622 Changes the executed command group ID to the <group name> from /etc/group.
2623 See also "user".
2624
William Lallemandc9515522019-06-12 16:32:11 +02002625option start-on-reload
2626no option start-on-reload
2627 Start (or not) a new instance of the program upon a reload of the master.
2628 The default is to start a new instance. This option may only be used in a
2629 program section.
2630
2631
Christopher Faulet76edc0f2020-01-13 15:52:01 +010026323.8. HTTP-errors
2633----------------
2634
2635It is possible to globally declare several groups of HTTP errors, to be
2636imported afterwards in any proxy section. Same group may be referenced at
2637several places and can be fully or partially imported.
2638
2639http-errors <name>
2640 Create a new http-errors group with the name <name>. It is an independent
2641 section that may be referenced by one or more proxies using its name.
2642
2643errorfile <code> <file>
2644 Associate a file contents to an HTTP error code
2645
2646 Arguments :
2647 <code> is the HTTP status code. Currently, HAProxy is capable of
Christopher Faulet612f2ea2020-05-27 09:57:28 +02002648 generating codes 200, 400, 401, 403, 404, 405, 407, 408, 410,
2649 425, 429, 500, 502, 503, and 504.
Christopher Faulet76edc0f2020-01-13 15:52:01 +01002650
2651 <file> designates a file containing the full HTTP response. It is
2652 recommended to follow the common practice of appending ".http" to
2653 the filename so that people do not confuse the response with HTML
2654 error pages, and to use absolute paths, since files are read
2655 before any chroot is performed.
2656
2657 Please referrers to "errorfile" keyword in section 4 for details.
2658
2659 Example:
2660 http-errors website-1
2661 errorfile 400 /etc/haproxy/errorfiles/site1/400.http
2662 errorfile 404 /etc/haproxy/errorfiles/site1/404.http
2663 errorfile 408 /dev/null # work around Chrome pre-connect bug
2664
2665 http-errors website-2
2666 errorfile 400 /etc/haproxy/errorfiles/site2/400.http
2667 errorfile 404 /etc/haproxy/errorfiles/site2/404.http
2668 errorfile 408 /dev/null # work around Chrome pre-connect bug
2669
Emeric Brun99c453d2020-05-25 15:01:04 +020026703.9. Rings
2671----------
2672
2673It is possible to globally declare ring-buffers, to be used as target for log
2674servers or traces.
2675
2676ring <ringname>
2677 Creates a new ring-buffer with name <ringname>.
2678
2679description <text>
Daniel Corbett67a82712020-07-06 23:01:19 -04002680 The description is an optional description string of the ring. It will
Emeric Brun99c453d2020-05-25 15:01:04 +02002681 appear on CLI. By default, <name> is reused to fill this field.
2682
2683format <format>
2684 Format used to store events into the ring buffer.
2685
2686 Arguments:
2687 <format> is the log format used when generating syslog messages. It may be
2688 one of the following :
2689
2690 iso A message containing only the ISO date, followed by the text.
2691 The PID, process name and system name are omitted. This is
2692 designed to be used with a local log server.
2693
2694 raw A message containing only the text. The level, PID, date, time,
2695 process name and system name are omitted. This is designed to be
2696 used in containers or during development, where the severity
2697 only depends on the file descriptor used (stdout/stderr). This
2698 is the default.
2699
2700 rfc3164 The RFC3164 syslog message format. This is the default.
2701 (https://tools.ietf.org/html/rfc3164)
2702
2703 rfc5424 The RFC5424 syslog message format.
2704 (https://tools.ietf.org/html/rfc5424)
2705
2706 short A message containing only a level between angle brackets such as
2707 '<3>', followed by the text. The PID, date, time, process name
2708 and system name are omitted. This is designed to be used with a
2709 local log server. This format is compatible with what the systemd
2710 logger consumes.
2711
Emeric Brun54648852020-07-06 15:54:06 +02002712 priority A message containing only a level plus syslog facility between angle
2713 brackets such as '<63>', followed by the text. The PID, date, time,
2714 process name and system name are omitted. This is designed to be used
2715 with a local log server.
2716
Emeric Brun99c453d2020-05-25 15:01:04 +02002717 timed A message containing only a level between angle brackets such as
2718 '<3>', followed by ISO date and by the text. The PID, process
2719 name and system name are omitted. This is designed to be
2720 used with a local log server.
2721
2722maxlen <length>
2723 The maximum length of an event message stored into the ring,
2724 including formatted header. If an event message is longer than
2725 <length>, it will be truncated to this length.
2726
Emeric Brun494c5052020-05-28 11:13:15 +02002727server <name> <address> [param*]
2728 Used to configure a syslog tcp server to forward messages from ring buffer.
2729 This supports for all "server" parameters found in 5.2 paragraph. Some of
2730 these parameters are irrelevant for "ring" sections. Important point: there
2731 is little reason to add more than one server to a ring, because all servers
2732 will receive the exact same copy of the ring contents, and as such the ring
2733 will progress at the speed of the slowest server. If one server does not
2734 respond, it will prevent old messages from being purged and may block new
2735 messages from being inserted into the ring. The proper way to send messages
2736 to multiple servers is to use one distinct ring per log server, not to
Emeric Brun97556472020-05-30 01:42:45 +02002737 attach multiple servers to the same ring. Note that specific server directive
2738 "log-proto" is used to set the protocol used to send messages.
Emeric Brun494c5052020-05-28 11:13:15 +02002739
Emeric Brun99c453d2020-05-25 15:01:04 +02002740size <size>
2741 This is the optional size in bytes for the ring-buffer. Default value is
2742 set to BUFSIZE.
2743
Emeric Brun494c5052020-05-28 11:13:15 +02002744timeout connect <timeout>
2745 Set the maximum time to wait for a connection attempt to a server to succeed.
2746
2747 Arguments :
2748 <timeout> is the timeout value specified in milliseconds by default, but
2749 can be in any other unit if the number is suffixed by the unit,
2750 as explained at the top of this document.
2751
2752timeout server <timeout>
2753 Set the maximum time for pending data staying into output buffer.
2754
2755 Arguments :
2756 <timeout> is the timeout value specified in milliseconds by default, but
2757 can be in any other unit if the number is suffixed by the unit,
2758 as explained at the top of this document.
2759
Emeric Brun99c453d2020-05-25 15:01:04 +02002760 Example:
2761 global
2762 log ring@myring local7
2763
2764 ring myring
2765 description "My local buffer"
2766 format rfc3164
2767 maxlen 1200
2768 size 32764
Emeric Brun494c5052020-05-28 11:13:15 +02002769 timeout connect 5s
2770 timeout server 10s
Emeric Brun97556472020-05-30 01:42:45 +02002771 server mysyslogsrv 127.0.0.1:6514 log-proto octet-count
Emeric Brun99c453d2020-05-25 15:01:04 +02002772
Emeric Brun12941c82020-07-07 14:19:42 +020027733.10. Log forwarding
2774-------------------
2775
2776It is possible to declare one or multiple log forwarding section,
2777haproxy will forward all received log messages to a log servers list.
2778
2779log-forward <name>
2780 Creates a new log forwarder proxy identified as <name>.
2781
Emeric Bruncbb7bf72020-10-05 14:39:35 +02002782backlog <conns>
2783 Give hints to the system about the approximate listen backlog desired size
2784 on connections accept.
2785
2786bind <addr> [param*]
2787 Used to configure a stream log listener to receive messages to forward.
Emeric Brunda46c1c2020-10-08 08:39:02 +02002788 This supports the "bind" parameters found in 5.1 paragraph including
2789 those about ssl but some statements such as "alpn" may be irrelevant for
2790 syslog protocol over TCP.
2791 Those listeners support both "Octet Counting" and "Non-Transparent-Framing"
Emeric Bruncbb7bf72020-10-05 14:39:35 +02002792 modes as defined in rfc-6587.
2793
Willy Tarreau76aaa7f2020-09-16 15:07:22 +02002794dgram-bind <addr> [param*]
Emeric Bruncbb7bf72020-10-05 14:39:35 +02002795 Used to configure a datagram log listener to receive messages to forward.
2796 Addresses must be in IPv4 or IPv6 form,followed by a port. This supports
2797 for some of the "bind" parameters found in 5.1 paragraph among which
2798 "interface", "namespace" or "transparent", the other ones being
Willy Tarreau26ff5da2020-09-16 15:22:19 +02002799 silently ignored as irrelevant for UDP/syslog case.
Emeric Brun12941c82020-07-07 14:19:42 +02002800
2801log global
2802log <address> [len <length>] [format <format>] [sample <ranges>:<smp_size>]
2803 <facility> [<level> [<minlevel>]]
2804 Used to configure target log servers. See more details on proxies
2805 documentation.
2806 If no format specified, haproxy tries to keep the incoming log format.
2807 Configured facility is ignored, except if incoming message does not
2808 present a facility but one is mandatory on the outgoing format.
2809 If there is no timestamp available in the input format, but the field
2810 exists in output format, haproxy will use the local date.
2811
2812 Example:
2813 global
2814 log stderr format iso local7
2815
2816 ring myring
2817 description "My local buffer"
2818 format rfc5424
2819 maxlen 1200
2820 size 32764
2821 timeout connect 5s
2822 timeout server 10s
2823 # syslog tcp server
2824 server mysyslogsrv 127.0.0.1:514 log-proto octet-count
2825
2826 log-forward sylog-loadb
Emeric Bruncbb7bf72020-10-05 14:39:35 +02002827 dgram-bind 127.0.0.1:1514
2828 bind 127.0.0.1:1514
Emeric Brun12941c82020-07-07 14:19:42 +02002829 # all messages on stderr
2830 log global
2831 # all messages on local tcp syslog server
2832 log ring@myring local0
2833 # load balance messages on 4 udp syslog servers
2834 log 127.0.0.1:10001 sample 1:4 local0
2835 log 127.0.0.1:10002 sample 2:4 local0
2836 log 127.0.0.1:10003 sample 3:4 local0
2837 log 127.0.0.1:10004 sample 4:4 local0
Christopher Faulet76edc0f2020-01-13 15:52:01 +01002838
Emeric Bruncbb7bf72020-10-05 14:39:35 +02002839maxconn <conns>
2840 Fix the maximum number of concurrent connections on a log forwarder.
2841 10 is the default.
2842
2843timeout client <timeout>
2844 Set the maximum inactivity time on the client side.
2845
Willy Tarreauc57f0e22009-05-10 13:12:33 +020028464. Proxies
Willy Tarreau6a06a402007-07-15 20:15:28 +02002847----------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002848
Willy Tarreau6a06a402007-07-15 20:15:28 +02002849Proxy configuration can be located in a set of sections :
William Lallemand6e62fb62015-04-28 16:55:23 +02002850 - defaults [<name>]
Willy Tarreau6a06a402007-07-15 20:15:28 +02002851 - frontend <name>
2852 - backend <name>
2853 - listen <name>
2854
2855A "defaults" section sets default parameters for all other sections following
2856its declaration. Those default parameters are reset by the next "defaults"
2857section. See below for the list of parameters which can be set in a "defaults"
Willy Tarreau0ba27502007-12-24 16:55:16 +01002858section. The name is optional but its use is encouraged for better readability.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002859
2860A "frontend" section describes a set of listening sockets accepting client
2861connections.
2862
2863A "backend" section describes a set of servers to which the proxy will connect
2864to forward incoming connections.
2865
2866A "listen" section defines a complete proxy with its frontend and backend
2867parts combined in one section. It is generally useful for TCP-only traffic.
2868
Willy Tarreau0ba27502007-12-24 16:55:16 +01002869All proxy names must be formed from upper and lower case letters, digits,
2870'-' (dash), '_' (underscore) , '.' (dot) and ':' (colon). ACL names are
2871case-sensitive, which means that "www" and "WWW" are two different proxies.
2872
2873Historically, all proxy names could overlap, it just caused troubles in the
2874logs. Since the introduction of content switching, it is mandatory that two
2875proxies with overlapping capabilities (frontend/backend) have different names.
2876However, it is still permitted that a frontend and a backend share the same
2877name, as this configuration seems to be commonly encountered.
2878
2879Right now, two major proxy modes are supported : "tcp", also known as layer 4,
2880and "http", also known as layer 7. In layer 4 mode, HAProxy simply forwards
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002881bidirectional traffic between two sides. In layer 7 mode, HAProxy analyzes the
Willy Tarreau0ba27502007-12-24 16:55:16 +01002882protocol, and can interact with it by allowing, blocking, switching, adding,
2883modifying, or removing arbitrary contents in requests or responses, based on
2884arbitrary criteria.
2885
Willy Tarreau70dffda2014-01-30 03:07:23 +01002886In HTTP mode, the processing applied to requests and responses flowing over
2887a connection depends in the combination of the frontend's HTTP options and
Julien Pivotto21ad3152019-12-10 13:11:17 +01002888the backend's. HAProxy supports 3 connection modes :
Willy Tarreau70dffda2014-01-30 03:07:23 +01002889
2890 - KAL : keep alive ("option http-keep-alive") which is the default mode : all
2891 requests and responses are processed, and connections remain open but idle
2892 between responses and new requests.
2893
Willy Tarreau70dffda2014-01-30 03:07:23 +01002894 - SCL: server close ("option http-server-close") : the server-facing
2895 connection is closed after the end of the response is received, but the
2896 client-facing connection remains open.
2897
Christopher Faulet315b39c2018-09-21 16:26:19 +02002898 - CLO: close ("option httpclose"): the connection is closed after the end of
2899 the response and "Connection: close" appended in both directions.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002900
2901The effective mode that will be applied to a connection passing through a
2902frontend and a backend can be determined by both proxy modes according to the
2903following matrix, but in short, the modes are symmetric, keep-alive is the
Christopher Faulet315b39c2018-09-21 16:26:19 +02002904weakest option and close is the strongest.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002905
Christopher Faulet315b39c2018-09-21 16:26:19 +02002906 Backend mode
Willy Tarreau70dffda2014-01-30 03:07:23 +01002907
Christopher Faulet315b39c2018-09-21 16:26:19 +02002908 | KAL | SCL | CLO
2909 ----+-----+-----+----
2910 KAL | KAL | SCL | CLO
2911 ----+-----+-----+----
Christopher Faulet315b39c2018-09-21 16:26:19 +02002912 mode SCL | SCL | SCL | CLO
2913 ----+-----+-----+----
2914 CLO | CLO | CLO | CLO
Willy Tarreau70dffda2014-01-30 03:07:23 +01002915
Willy Tarreau0ba27502007-12-24 16:55:16 +01002916
Willy Tarreau70dffda2014-01-30 03:07:23 +01002917
Willy Tarreauc57f0e22009-05-10 13:12:33 +020029184.1. Proxy keywords matrix
2919--------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002920
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002921The following list of keywords is supported. Most of them may only be used in a
2922limited set of section types. Some of them are marked as "deprecated" because
2923they are inherited from an old syntax which may be confusing or functionally
2924limited, and there are new recommended keywords to replace them. Keywords
Davor Ocelice9ed2812017-12-25 17:49:28 +01002925marked with "(*)" can be optionally inverted using the "no" prefix, e.g. "no
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002926option contstats". This makes sense when the option has been enabled by default
Willy Tarreau3842f002009-06-14 11:39:52 +02002927and must be disabled for a specific instance. Such options may also be prefixed
2928with "default" in order to restore default settings regardless of what has been
2929specified in a previous "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002930
Willy Tarreau6a06a402007-07-15 20:15:28 +02002931
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002932 keyword defaults frontend listen backend
2933------------------------------------+----------+----------+---------+---------
2934acl - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002935backlog X X X -
2936balance X - X X
2937bind - X X -
2938bind-process X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002939capture cookie - X X -
2940capture request header - X X -
2941capture response header - X X -
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09002942clitcpka-cnt X X X -
2943clitcpka-idle X X X -
2944clitcpka-intvl X X X -
William Lallemand82fe75c2012-10-23 10:25:10 +02002945compression X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002946cookie X - X X
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02002947declare capture - X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002948default-server X - X X
2949default_backend X X X -
2950description - X X X
2951disabled X X X X
2952dispatch - - X X
Simon Horman51a1cf62015-02-03 13:00:44 +09002953email-alert from X X X X
Simon Horman64e34162015-02-06 11:11:57 +09002954email-alert level X X X X
Simon Horman51a1cf62015-02-03 13:00:44 +09002955email-alert mailers X X X X
2956email-alert myhostname X X X X
2957email-alert to X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002958enabled X X X X
2959errorfile X X X X
Christopher Faulet76edc0f2020-01-13 15:52:01 +01002960errorfiles X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002961errorloc X X X X
2962errorloc302 X X X X
2963-- keyword -------------------------- defaults - frontend - listen -- backend -
2964errorloc303 X X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01002965force-persist - - X X
Christopher Fauletc3fe5332016-04-07 15:30:10 +02002966filter - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002967fullconn X - X X
2968grace X X X X
2969hash-type X - X X
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01002970http-after-response - X X X
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02002971http-check comment X - X X
Christopher Faulete5870d82020-04-15 11:32:03 +02002972http-check connect X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002973http-check disable-on-404 X - X X
Christopher Faulete5870d82020-04-15 11:32:03 +02002974http-check expect X - X X
Peter Gervai8912ae62020-06-11 18:26:36 +02002975http-check send X - X X
Willy Tarreau7ab6aff2010-10-12 06:30:16 +02002976http-check send-state X - X X
Christopher Faulete5870d82020-04-15 11:32:03 +02002977http-check set-var X - X X
2978http-check unset-var X - X X
Christopher Faulet3b967c12020-05-15 15:47:44 +02002979http-error X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002980http-request - X X X
Willy Tarreaue365c0b2013-06-11 16:06:12 +02002981http-response - X X X
Willy Tarreau30631952015-08-06 15:05:24 +02002982http-reuse X - X X
Baptiste Assmann2c42ef52013-10-09 21:57:02 +02002983http-send-name-header - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002984id - X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01002985ignore-persist - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02002986load-server-state-from-file X - X X
William Lallemand0f99e342011-10-12 17:50:54 +02002987log (*) X X X X
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01002988log-format X X X -
Dragan Dosen7ad31542015-09-28 17:16:47 +02002989log-format-sd X X X -
Willy Tarreau094af4e2015-01-07 15:03:42 +01002990log-tag X X X X
Willy Tarreauc35362a2014-04-25 13:58:37 +02002991max-keep-alive-queue X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002992maxconn X X X -
2993mode X X X X
2994monitor fail - X X -
2995monitor-net X X X -
2996monitor-uri X X X -
2997option abortonclose (*) X - X X
2998option accept-invalid-http-request (*) X X X -
2999option accept-invalid-http-response (*) X - X X
3000option allbackups (*) X - X X
3001option checkcache (*) X - X X
3002option clitcpka (*) X X X -
3003option contstats (*) X X X -
Christopher Faulet89aed322020-06-02 17:33:56 +02003004option disable-h2-upgrade (*) X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003005option dontlog-normal (*) X X X -
3006option dontlognull (*) X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003007-- keyword -------------------------- defaults - frontend - listen -- backend -
3008option forwardfor X X X X
Christopher Faulet98fbe952019-07-22 16:18:24 +02003009option h1-case-adjust-bogus-client (*) X X X -
3010option h1-case-adjust-bogus-server (*) X - X X
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02003011option http-buffer-request (*) X X X X
Willy Tarreau82649f92015-05-01 22:40:51 +02003012option http-ignore-probes (*) X X X -
Willy Tarreau16bfb022010-01-16 19:48:41 +01003013option http-keep-alive (*) X X X X
Willy Tarreau96e31212011-05-30 18:10:30 +02003014option http-no-delay (*) X X X X
Christopher Faulet98db9762018-09-21 10:25:19 +02003015option http-pretend-keepalive (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003016option http-server-close (*) X X X X
3017option http-use-proxy-header (*) X X X -
3018option httpchk X - X X
3019option httpclose (*) X X X X
Freddy Spierenburge88b7732019-03-25 14:35:17 +01003020option httplog X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003021option http_proxy (*) X X X X
Jamie Gloudon801a0a32012-08-25 00:18:33 -04003022option independent-streams (*) X X X X
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02003023option ldap-check X - X X
Simon Horman98637e52014-06-20 12:30:16 +09003024option external-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003025option log-health-checks (*) X - X X
3026option log-separate-errors (*) X X X -
3027option logasap (*) X X X -
3028option mysql-check X - X X
3029option nolinger (*) X X X X
3030option originalto X X X X
3031option persist (*) X - X X
Baptiste Assmann809e22a2015-10-12 20:22:55 +02003032option pgsql-check X - X X
3033option prefer-last-server (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003034option redispatch (*) X - X X
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02003035option redis-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003036option smtpchk X - X X
3037option socket-stats (*) X X X -
3038option splice-auto (*) X X X X
3039option splice-request (*) X X X X
3040option splice-response (*) X X X X
Christopher Fauletba7bc162016-11-07 21:07:38 +01003041option spop-check - - - X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003042option srvtcpka (*) X - X X
3043option ssl-hello-chk X - X X
3044-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreaued179852013-12-16 01:07:00 +01003045option tcp-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003046option tcp-smart-accept (*) X X X -
3047option tcp-smart-connect (*) X - X X
3048option tcpka X X X X
3049option tcplog X X X X
3050option transparent (*) X - X X
Simon Horman98637e52014-06-20 12:30:16 +09003051external-check command X - X X
3052external-check path X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003053persist rdp-cookie X - X X
3054rate-limit sessions X X X -
3055redirect - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003056-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003057retries X - X X
Olivier Houcharda254a372019-04-05 15:30:12 +02003058retry-on X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003059server - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02003060server-state-file-name X - X X
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02003061server-template - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003062source X - X X
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09003063srvtcpka-cnt X - X X
3064srvtcpka-idle X - X X
3065srvtcpka-intvl X - X X
Baptiste Assmann5a549212015-10-12 20:30:24 +02003066stats admin - X X X
3067stats auth X X X X
3068stats enable X X X X
3069stats hide-version X X X X
3070stats http-request - X X X
3071stats realm X X X X
3072stats refresh X X X X
3073stats scope X X X X
3074stats show-desc X X X X
3075stats show-legends X X X X
3076stats show-node X X X X
3077stats uri X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003078-- keyword -------------------------- defaults - frontend - listen -- backend -
3079stick match - - X X
3080stick on - - X X
3081stick store-request - - X X
Willy Tarreaud8dc99f2011-07-01 11:33:25 +02003082stick store-response - - X X
Adam Spiers68af3c12017-04-06 16:31:39 +01003083stick-table - X X X
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02003084tcp-check comment X - X X
Christopher Faulet404f9192020-04-09 23:13:54 +02003085tcp-check connect X - X X
3086tcp-check expect X - X X
3087tcp-check send X - X X
Christopher Fauletb50b3e62020-05-05 18:43:43 +02003088tcp-check send-lf X - X X
Christopher Faulet404f9192020-04-09 23:13:54 +02003089tcp-check send-binary X - X X
Christopher Fauletb50b3e62020-05-05 18:43:43 +02003090tcp-check send-binary-lf X - X X
Christopher Faulet404f9192020-04-09 23:13:54 +02003091tcp-check set-var X - X X
3092tcp-check unset-var X - X X
Willy Tarreaue9656522010-08-17 15:40:09 +02003093tcp-request connection - X X -
3094tcp-request content - X X X
Willy Tarreaua56235c2010-09-14 11:31:36 +02003095tcp-request inspect-delay - X X X
Willy Tarreau4f614292016-10-21 17:49:36 +02003096tcp-request session - X X -
Emeric Brun0a3b67f2010-09-24 15:34:53 +02003097tcp-response content - - X X
3098tcp-response inspect-delay - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003099timeout check X - X X
3100timeout client X X X -
Willy Tarreau05cdd962014-05-10 14:30:07 +02003101timeout client-fin X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003102timeout connect X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003103timeout http-keep-alive X X X X
3104timeout http-request X X X X
3105timeout queue X - X X
3106timeout server X - X X
Willy Tarreau05cdd962014-05-10 14:30:07 +02003107timeout server-fin X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003108timeout tarpit X X X X
Willy Tarreauce887fd2012-05-12 12:50:00 +02003109timeout tunnel X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003110transparent (deprecated) X - X X
William Lallemanda73203e2012-03-12 12:48:57 +01003111unique-id-format X X X -
3112unique-id-header X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003113use_backend - X X -
Christopher Fauletb30b3102019-09-12 23:03:09 +02003114use-fcgi-app - - X X
Willy Tarreau4a5cade2012-04-05 21:09:48 +02003115use-server - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003116------------------------------------+----------+----------+---------+---------
3117 keyword defaults frontend listen backend
Willy Tarreau6a06a402007-07-15 20:15:28 +02003118
Willy Tarreau0ba27502007-12-24 16:55:16 +01003119
Willy Tarreauc57f0e22009-05-10 13:12:33 +020031204.2. Alphabetically sorted keywords reference
3121---------------------------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01003122
3123This section provides a description of each keyword and its usage.
3124
3125
3126acl <aclname> <criterion> [flags] [operator] <value> ...
3127 Declare or complete an access list.
3128 May be used in sections : defaults | frontend | listen | backend
3129 no | yes | yes | yes
3130 Example:
3131 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
3132 acl invalid_src src_port 0:1023
3133 acl local_dst hdr(host) -i localhost
3134
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003135 See section 7 about ACL usage.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003136
3137
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01003138backlog <conns>
3139 Give hints to the system about the approximate listen backlog desired size
3140 May be used in sections : defaults | frontend | listen | backend
3141 yes | yes | yes | no
3142 Arguments :
3143 <conns> is the number of pending connections. Depending on the operating
3144 system, it may represent the number of already acknowledged
Cyril Bontédc4d9032012-04-08 21:57:39 +02003145 connections, of non-acknowledged ones, or both.
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01003146
3147 In order to protect against SYN flood attacks, one solution is to increase
3148 the system's SYN backlog size. Depending on the system, sometimes it is just
3149 tunable via a system parameter, sometimes it is not adjustable at all, and
3150 sometimes the system relies on hints given by the application at the time of
3151 the listen() syscall. By default, HAProxy passes the frontend's maxconn value
3152 to the listen() syscall. On systems which can make use of this value, it can
3153 sometimes be useful to be able to specify a different value, hence this
3154 backlog parameter.
3155
3156 On Linux 2.4, the parameter is ignored by the system. On Linux 2.6, it is
3157 used as a hint and the system accepts up to the smallest greater power of
3158 two, and never more than some limits (usually 32768).
3159
3160 See also : "maxconn" and the target operating system's tuning guide.
3161
3162
Willy Tarreau0ba27502007-12-24 16:55:16 +01003163balance <algorithm> [ <arguments> ]
Willy Tarreau226071e2014-04-10 11:55:45 +02003164balance url_param <param> [check_post]
Willy Tarreau0ba27502007-12-24 16:55:16 +01003165 Define the load balancing algorithm to be used in a backend.
3166 May be used in sections : defaults | frontend | listen | backend
3167 yes | no | yes | yes
3168 Arguments :
3169 <algorithm> is the algorithm used to select a server when doing load
3170 balancing. This only applies when no persistence information
3171 is available, or when a connection is redispatched to another
3172 server. <algorithm> may be one of the following :
3173
3174 roundrobin Each server is used in turns, according to their weights.
3175 This is the smoothest and fairest algorithm when the server's
3176 processing time remains equally distributed. This algorithm
3177 is dynamic, which means that server weights may be adjusted
Willy Tarreau9757a382009-10-03 12:56:50 +02003178 on the fly for slow starts for instance. It is limited by
Godbacha34bdc02013-07-22 07:44:53 +08003179 design to 4095 active servers per backend. Note that in some
Willy Tarreau9757a382009-10-03 12:56:50 +02003180 large farms, when a server becomes up after having been down
3181 for a very short time, it may sometimes take a few hundreds
3182 requests for it to be re-integrated into the farm and start
3183 receiving traffic. This is normal, though very rare. It is
3184 indicated here in case you would have the chance to observe
3185 it, so that you don't worry.
3186
3187 static-rr Each server is used in turns, according to their weights.
3188 This algorithm is as similar to roundrobin except that it is
3189 static, which means that changing a server's weight on the
3190 fly will have no effect. On the other hand, it has no design
3191 limitation on the number of servers, and when a server goes
3192 up, it is always immediately reintroduced into the farm, once
3193 the full map is recomputed. It also uses slightly less CPU to
3194 run (around -1%).
Willy Tarreau0ba27502007-12-24 16:55:16 +01003195
Willy Tarreau2d2a7f82008-03-17 12:07:56 +01003196 leastconn The server with the lowest number of connections receives the
3197 connection. Round-robin is performed within groups of servers
3198 of the same load to ensure that all servers will be used. Use
3199 of this algorithm is recommended where very long sessions are
3200 expected, such as LDAP, SQL, TSE, etc... but is not very well
3201 suited for protocols using short sessions such as HTTP. This
3202 algorithm is dynamic, which means that server weights may be
3203 adjusted on the fly for slow starts for instance.
3204
Willy Tarreauf09c6602012-02-13 17:12:08 +01003205 first The first server with available connection slots receives the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03003206 connection. The servers are chosen from the lowest numeric
Willy Tarreauf09c6602012-02-13 17:12:08 +01003207 identifier to the highest (see server parameter "id"), which
3208 defaults to the server's position in the farm. Once a server
Willy Tarreau64559c52012-04-07 09:08:45 +02003209 reaches its maxconn value, the next server is used. It does
Willy Tarreauf09c6602012-02-13 17:12:08 +01003210 not make sense to use this algorithm without setting maxconn.
3211 The purpose of this algorithm is to always use the smallest
3212 number of servers so that extra servers can be powered off
3213 during non-intensive hours. This algorithm ignores the server
3214 weight, and brings more benefit to long session such as RDP
Willy Tarreau64559c52012-04-07 09:08:45 +02003215 or IMAP than HTTP, though it can be useful there too. In
3216 order to use this algorithm efficiently, it is recommended
3217 that a cloud controller regularly checks server usage to turn
3218 them off when unused, and regularly checks backend queue to
3219 turn new servers on when the queue inflates. Alternatively,
3220 using "http-check send-state" may inform servers on the load.
Willy Tarreauf09c6602012-02-13 17:12:08 +01003221
Willy Tarreau0ba27502007-12-24 16:55:16 +01003222 source The source IP address is hashed and divided by the total
3223 weight of the running servers to designate which server will
3224 receive the request. This ensures that the same client IP
3225 address will always reach the same server as long as no
3226 server goes down or up. If the hash result changes due to the
3227 number of running servers changing, many clients will be
3228 directed to a different server. This algorithm is generally
3229 used in TCP mode where no cookie may be inserted. It may also
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003230 be used on the Internet to provide a best-effort stickiness
Willy Tarreau0ba27502007-12-24 16:55:16 +01003231 to clients which refuse session cookies. This algorithm is
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003232 static by default, which means that changing a server's
3233 weight on the fly will have no effect, but this can be
3234 changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003235
Oskar Stolc8dc41842012-05-19 10:19:54 +01003236 uri This algorithm hashes either the left part of the URI (before
3237 the question mark) or the whole URI (if the "whole" parameter
3238 is present) and divides the hash value by the total weight of
3239 the running servers. The result designates which server will
3240 receive the request. This ensures that the same URI will
3241 always be directed to the same server as long as no server
3242 goes up or down. This is used with proxy caches and
3243 anti-virus proxies in order to maximize the cache hit rate.
3244 Note that this algorithm may only be used in an HTTP backend.
3245 This algorithm is static by default, which means that
3246 changing a server's weight on the fly will have no effect,
3247 but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003248
Oskar Stolc8dc41842012-05-19 10:19:54 +01003249 This algorithm supports two optional parameters "len" and
Marek Majkowski9c30fc12008-04-27 23:25:55 +02003250 "depth", both followed by a positive integer number. These
3251 options may be helpful when it is needed to balance servers
3252 based on the beginning of the URI only. The "len" parameter
3253 indicates that the algorithm should only consider that many
3254 characters at the beginning of the URI to compute the hash.
3255 Note that having "len" set to 1 rarely makes sense since most
3256 URIs start with a leading "/".
3257
3258 The "depth" parameter indicates the maximum directory depth
3259 to be used to compute the hash. One level is counted for each
3260 slash in the request. If both parameters are specified, the
3261 evaluation stops when either is reached.
3262
Willy Tarreau57a37412020-09-23 08:56:29 +02003263 A "path-only" parameter indicates that the hashing key starts
3264 at the first '/' of the path. This can be used to ignore the
3265 authority part of absolute URIs, and to make sure that HTTP/1
3266 and HTTP/2 URIs will provide the same hash.
3267
Willy Tarreau0ba27502007-12-24 16:55:16 +01003268 url_param The URL parameter specified in argument will be looked up in
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003269 the query string of each HTTP GET request.
3270
3271 If the modifier "check_post" is used, then an HTTP POST
Cyril Bontédc4d9032012-04-08 21:57:39 +02003272 request entity will be searched for the parameter argument,
3273 when it is not found in a query string after a question mark
Willy Tarreau226071e2014-04-10 11:55:45 +02003274 ('?') in the URL. The message body will only start to be
3275 analyzed once either the advertised amount of data has been
3276 received or the request buffer is full. In the unlikely event
3277 that chunked encoding is used, only the first chunk is
Cyril Bontédc4d9032012-04-08 21:57:39 +02003278 scanned. Parameter values separated by a chunk boundary, may
Willy Tarreau226071e2014-04-10 11:55:45 +02003279 be randomly balanced if at all. This keyword used to support
3280 an optional <max_wait> parameter which is now ignored.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003281
3282 If the parameter is found followed by an equal sign ('=') and
3283 a value, then the value is hashed and divided by the total
3284 weight of the running servers. The result designates which
3285 server will receive the request.
3286
3287 This is used to track user identifiers in requests and ensure
3288 that a same user ID will always be sent to the same server as
3289 long as no server goes up or down. If no value is found or if
3290 the parameter is not found, then a round robin algorithm is
3291 applied. Note that this algorithm may only be used in an HTTP
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003292 backend. This algorithm is static by default, which means
3293 that changing a server's weight on the fly will have no
3294 effect, but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003295
Cyril Bontédc4d9032012-04-08 21:57:39 +02003296 hdr(<name>) The HTTP header <name> will be looked up in each HTTP
3297 request. Just as with the equivalent ACL 'hdr()' function,
3298 the header name in parenthesis is not case sensitive. If the
3299 header is absent or if it does not contain any value, the
3300 roundrobin algorithm is applied instead.
Benoitaffb4812009-03-25 13:02:10 +01003301
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003302 An optional 'use_domain_only' parameter is available, for
Benoitaffb4812009-03-25 13:02:10 +01003303 reducing the hash algorithm to the main domain part with some
3304 specific headers such as 'Host'. For instance, in the Host
3305 value "haproxy.1wt.eu", only "1wt" will be considered.
3306
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003307 This algorithm is static by default, which means that
3308 changing a server's weight on the fly will have no effect,
3309 but this can be changed using "hash-type".
3310
Willy Tarreau21c741a2019-01-14 18:14:27 +01003311 random
3312 random(<draws>)
3313 A random number will be used as the key for the consistent
Willy Tarreau760e81d2018-05-03 07:20:40 +02003314 hashing function. This means that the servers' weights are
3315 respected, dynamic weight changes immediately take effect, as
3316 well as new server additions. Random load balancing can be
3317 useful with large farms or when servers are frequently added
Willy Tarreau21c741a2019-01-14 18:14:27 +01003318 or removed as it may avoid the hammering effect that could
3319 result from roundrobin or leastconn in this situation. The
3320 hash-balance-factor directive can be used to further improve
3321 fairness of the load balancing, especially in situations
3322 where servers show highly variable response times. When an
3323 argument <draws> is present, it must be an integer value one
3324 or greater, indicating the number of draws before selecting
3325 the least loaded of these servers. It was indeed demonstrated
3326 that picking the least loaded of two servers is enough to
3327 significantly improve the fairness of the algorithm, by
3328 always avoiding to pick the most loaded server within a farm
3329 and getting rid of any bias that could be induced by the
3330 unfair distribution of the consistent list. Higher values N
3331 will take away N-1 of the highest loaded servers at the
3332 expense of performance. With very high values, the algorithm
3333 will converge towards the leastconn's result but much slower.
3334 The default value is 2, which generally shows very good
3335 distribution and performance. This algorithm is also known as
3336 the Power of Two Random Choices and is described here :
3337 http://www.eecs.harvard.edu/~michaelm/postscripts/handbook2001.pdf
Willy Tarreau760e81d2018-05-03 07:20:40 +02003338
Emeric Brun736aa232009-06-30 17:56:00 +02003339 rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02003340 rdp-cookie(<name>)
Emeric Brun736aa232009-06-30 17:56:00 +02003341 The RDP cookie <name> (or "mstshash" if omitted) will be
3342 looked up and hashed for each incoming TCP request. Just as
3343 with the equivalent ACL 'req_rdp_cookie()' function, the name
3344 is not case-sensitive. This mechanism is useful as a degraded
3345 persistence mode, as it makes it possible to always send the
3346 same user (or the same session ID) to the same server. If the
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003347 cookie is not found, the normal roundrobin algorithm is
Emeric Brun736aa232009-06-30 17:56:00 +02003348 used instead.
3349
3350 Note that for this to work, the frontend must ensure that an
3351 RDP cookie is already present in the request buffer. For this
3352 you must use 'tcp-request content accept' rule combined with
3353 a 'req_rdp_cookie_cnt' ACL.
3354
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003355 This algorithm is static by default, which means that
3356 changing a server's weight on the fly will have no effect,
3357 but this can be changed using "hash-type".
3358
Cyril Bontédc4d9032012-04-08 21:57:39 +02003359 See also the rdp_cookie pattern fetch function.
Simon Hormanab814e02011-06-24 14:50:20 +09003360
Willy Tarreau0ba27502007-12-24 16:55:16 +01003361 <arguments> is an optional list of arguments which may be needed by some
Marek Majkowski9c30fc12008-04-27 23:25:55 +02003362 algorithms. Right now, only "url_param" and "uri" support an
3363 optional argument.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003364
Willy Tarreau3cd9af22009-03-15 14:06:41 +01003365 The load balancing algorithm of a backend is set to roundrobin when no other
3366 algorithm, mode nor option have been set. The algorithm may only be set once
3367 for each backend.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003368
Lukas Tribus80512b12018-10-27 20:07:40 +02003369 With authentication schemes that require the same connection like NTLM, URI
John Roeslerfb2fce12019-07-10 15:45:51 -05003370 based algorithms must not be used, as they would cause subsequent requests
Lukas Tribus80512b12018-10-27 20:07:40 +02003371 to be routed to different backend servers, breaking the invalid assumptions
3372 NTLM relies on.
3373
Willy Tarreau0ba27502007-12-24 16:55:16 +01003374 Examples :
3375 balance roundrobin
3376 balance url_param userid
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003377 balance url_param session_id check_post 64
Benoitaffb4812009-03-25 13:02:10 +01003378 balance hdr(User-Agent)
3379 balance hdr(host)
3380 balance hdr(Host) use_domain_only
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003381
3382 Note: the following caveats and limitations on using the "check_post"
3383 extension with "url_param" must be considered :
3384
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003385 - all POST requests are eligible for consideration, because there is no way
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003386 to determine if the parameters will be found in the body or entity which
3387 may contain binary data. Therefore another method may be required to
3388 restrict consideration of POST requests that have no URL parameters in
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02003389 the body. (see acl http_end)
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003390
3391 - using a <max_wait> value larger than the request buffer size does not
3392 make sense and is useless. The buffer size is set at build time, and
3393 defaults to 16 kB.
3394
3395 - Content-Encoding is not supported, the parameter search will probably
3396 fail; and load balancing will fall back to Round Robin.
3397
3398 - Expect: 100-continue is not supported, load balancing will fall back to
3399 Round Robin.
3400
Lukas Tribus23953682017-04-28 13:24:30 +00003401 - Transfer-Encoding (RFC7230 3.3.1) is only supported in the first chunk.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003402 If the entire parameter value is not present in the first chunk, the
3403 selection of server is undefined (actually, defined by how little
3404 actually appeared in the first chunk).
3405
3406 - This feature does not support generation of a 100, 411 or 501 response.
3407
3408 - In some cases, requesting "check_post" MAY attempt to scan the entire
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003409 contents of a message body. Scanning normally terminates when linear
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003410 white space or control characters are found, indicating the end of what
3411 might be a URL parameter list. This is probably not a concern with SGML
3412 type message bodies.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003413
Willy Tarreau294d0f02015-08-10 19:40:12 +02003414 See also : "dispatch", "cookie", "transparent", "hash-type" and "http_proxy".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003415
3416
Willy Tarreaub6205fd2012-09-24 12:27:33 +02003417bind [<address>]:<port_range> [, ...] [param*]
3418bind /<path> [, ...] [param*]
Willy Tarreau0ba27502007-12-24 16:55:16 +01003419 Define one or several listening addresses and/or ports in a frontend.
3420 May be used in sections : defaults | frontend | listen | backend
3421 no | yes | yes | no
3422 Arguments :
Willy Tarreaub1e52e82008-01-13 14:49:51 +01003423 <address> is optional and can be a host name, an IPv4 address, an IPv6
3424 address, or '*'. It designates the address the frontend will
3425 listen on. If unset, all IPv4 addresses of the system will be
3426 listened on. The same will apply for '*' or the system's
David du Colombier9c938da2011-03-17 10:40:27 +01003427 special address "0.0.0.0". The IPv6 equivalent is '::'.
Willy Tarreau24709282013-03-10 21:32:12 +01003428 Optionally, an address family prefix may be used before the
3429 address to force the family regardless of the address format,
3430 which can be useful to specify a path to a unix socket with
3431 no slash ('/'). Currently supported prefixes are :
3432 - 'ipv4@' -> address is always IPv4
3433 - 'ipv6@' -> address is always IPv6
Emeric Brun3835c0d2020-07-07 09:46:09 +02003434 - 'udp@' -> address is resolved as IPv4 or IPv6 and
Emeric Brun12941c82020-07-07 14:19:42 +02003435 protocol UDP is used. Currently those listeners are
3436 supported only in log-forward sections.
Emeric Brun3835c0d2020-07-07 09:46:09 +02003437 - 'udp4@' -> address is always IPv4 and protocol UDP
Emeric Brun12941c82020-07-07 14:19:42 +02003438 is used. Currently those listeners are supported
3439 only in log-forward sections.
Emeric Brun3835c0d2020-07-07 09:46:09 +02003440 - 'udp6@' -> address is always IPv6 and protocol UDP
Emeric Brun12941c82020-07-07 14:19:42 +02003441 is used. Currently those listeners are supported
3442 only in log-forward sections.
Willy Tarreau24709282013-03-10 21:32:12 +01003443 - 'unix@' -> address is a path to a local unix socket
Willy Tarreau70f72e02014-07-08 00:37:50 +02003444 - 'abns@' -> address is in abstract namespace (Linux only).
3445 Note: since abstract sockets are not "rebindable", they
3446 do not cope well with multi-process mode during
3447 soft-restart, so it is better to avoid them if
3448 nbproc is greater than 1. The effect is that if the
3449 new process fails to start, only one of the old ones
3450 will be able to rebind to the socket.
Willy Tarreau40aa0702013-03-10 23:51:38 +01003451 - 'fd@<n>' -> use file descriptor <n> inherited from the
3452 parent. The fd must be bound and may or may not already
3453 be listening.
William Lallemand2fe7dd02018-09-11 16:51:29 +02003454 - 'sockpair@<n>'-> like fd@ but you must use the fd of a
3455 connected unix socket or of a socketpair. The bind waits
3456 to receive a FD over the unix socket and uses it as if it
3457 was the FD of an accept(). Should be used carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02003458 You may want to reference some environment variables in the
3459 address parameter, see section 2.3 about environment
3460 variables.
Willy Tarreaub1e52e82008-01-13 14:49:51 +01003461
Willy Tarreauc5011ca2010-03-22 11:53:56 +01003462 <port_range> is either a unique TCP port, or a port range for which the
3463 proxy will accept connections for the IP address specified
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003464 above. The port is mandatory for TCP listeners. Note that in
3465 the case of an IPv6 address, the port is always the number
3466 after the last colon (':'). A range can either be :
Willy Tarreauc5011ca2010-03-22 11:53:56 +01003467 - a numerical port (ex: '80')
3468 - a dash-delimited ports range explicitly stating the lower
3469 and upper bounds (ex: '2000-2100') which are included in
3470 the range.
3471
3472 Particular care must be taken against port ranges, because
3473 every <address:port> couple consumes one socket (= a file
3474 descriptor), so it's easy to consume lots of descriptors
3475 with a simple range, and to run out of sockets. Also, each
3476 <address:port> couple must be used only once among all
3477 instances running on a same system. Please note that binding
3478 to ports lower than 1024 generally require particular
Jamie Gloudon801a0a32012-08-25 00:18:33 -04003479 privileges to start the program, which are independent of
Willy Tarreauc5011ca2010-03-22 11:53:56 +01003480 the 'uid' parameter.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003481
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003482 <path> is a UNIX socket path beginning with a slash ('/'). This is
Davor Ocelice9ed2812017-12-25 17:49:28 +01003483 alternative to the TCP listening port. HAProxy will then
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003484 receive UNIX connections on the socket located at this place.
3485 The path must begin with a slash and by default is absolute.
3486 It can be relative to the prefix defined by "unix-bind" in
3487 the global section. Note that the total length of the prefix
3488 followed by the socket path cannot exceed some system limits
3489 for UNIX sockets, which commonly are set to 107 characters.
3490
Willy Tarreaub6205fd2012-09-24 12:27:33 +02003491 <param*> is a list of parameters common to all sockets declared on the
3492 same line. These numerous parameters depend on OS and build
3493 options and have a complete section dedicated to them. Please
3494 refer to section 5 to for more details.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02003495
Willy Tarreau0ba27502007-12-24 16:55:16 +01003496 It is possible to specify a list of address:port combinations delimited by
3497 commas. The frontend will then listen on all of these addresses. There is no
3498 fixed limit to the number of addresses and ports which can be listened on in
3499 a frontend, as well as there is no limit to the number of "bind" statements
3500 in a frontend.
3501
3502 Example :
3503 listen http_proxy
3504 bind :80,:443
3505 bind 10.0.0.1:10080,10.0.0.1:10443
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003506 bind /var/run/ssl-frontend.sock user root mode 600 accept-proxy
Willy Tarreau0ba27502007-12-24 16:55:16 +01003507
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02003508 listen http_https_proxy
3509 bind :80
Cyril Bonté0d44fc62012-10-09 22:45:33 +02003510 bind :443 ssl crt /etc/haproxy/site.pem
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02003511
Willy Tarreau24709282013-03-10 21:32:12 +01003512 listen http_https_proxy_explicit
3513 bind ipv6@:80
3514 bind ipv4@public_ssl:443 ssl crt /etc/haproxy/site.pem
3515 bind unix@ssl-frontend.sock user root mode 600 accept-proxy
3516
Willy Tarreaudad36a32013-03-11 01:20:04 +01003517 listen external_bind_app1
William Lallemandb2f07452015-05-12 14:27:13 +02003518 bind "fd@${FD_APP1}"
Willy Tarreaudad36a32013-03-11 01:20:04 +01003519
Willy Tarreau55dcaf62015-09-27 15:03:15 +02003520 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
3521 sun_path length is used for the address length. Some other programs
3522 such as socat use the string length only by default. Pass the option
3523 ",unix-tightsocklen=0" to any abstract socket definition in socat to
3524 make it compatible with HAProxy's.
3525
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003526 See also : "source", "option forwardfor", "unix-bind" and the PROXY protocol
Willy Tarreaub6205fd2012-09-24 12:27:33 +02003527 documentation, and section 5 about bind options.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003528
3529
Christopher Fauletff4121f2017-11-22 16:38:49 +01003530bind-process [ all | odd | even | <process_num>[-[<process_num>]] ] ...
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003531 Limit visibility of an instance to a certain set of processes numbers.
3532 May be used in sections : defaults | frontend | listen | backend
3533 yes | yes | yes | yes
3534 Arguments :
3535 all All process will see this instance. This is the default. It
3536 may be used to override a default value.
3537
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003538 odd This instance will be enabled on processes 1,3,5,...63. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003539 option may be combined with other numbers.
3540
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003541 even This instance will be enabled on processes 2,4,6,...64. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003542 option may be combined with other numbers. Do not use it
3543 with less than 2 processes otherwise some instances might be
3544 missing from all processes.
3545
Christopher Fauletff4121f2017-11-22 16:38:49 +01003546 process_num The instance will be enabled on this process number or range,
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003547 whose values must all be between 1 and 32 or 64 depending on
Christopher Fauletff4121f2017-11-22 16:38:49 +01003548 the machine's word size. Ranges can be partially defined. The
3549 higher bound can be omitted. In such case, it is replaced by
3550 the corresponding maximum value. If a proxy is bound to
3551 process numbers greater than the configured global.nbproc, it
3552 will either be forced to process #1 if a single process was
Willy Tarreau102df612014-05-07 23:56:38 +02003553 specified, or to all processes otherwise.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003554
3555 This keyword limits binding of certain instances to certain processes. This
3556 is useful in order not to have too many processes listening to the same
3557 ports. For instance, on a dual-core machine, it might make sense to set
3558 'nbproc 2' in the global section, then distributes the listeners among 'odd'
3559 and 'even' instances.
3560
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003561 At the moment, it is not possible to reference more than 32 or 64 processes
3562 using this keyword, but this should be more than enough for most setups.
3563 Please note that 'all' really means all processes regardless of the machine's
3564 word size, and is not limited to the first 32 or 64.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003565
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02003566 Each "bind" line may further be limited to a subset of the proxy's processes,
3567 please consult the "process" bind keyword in section 5.1.
3568
Willy Tarreaub369a042014-09-16 13:21:03 +02003569 When a frontend has no explicit "bind-process" line, it tries to bind to all
3570 the processes referenced by its "bind" lines. That means that frontends can
3571 easily adapt to their listeners' processes.
3572
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003573 If some backends are referenced by frontends bound to other processes, the
3574 backend automatically inherits the frontend's processes.
3575
3576 Example :
3577 listen app_ip1
3578 bind 10.0.0.1:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02003579 bind-process odd
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003580
3581 listen app_ip2
3582 bind 10.0.0.2:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02003583 bind-process even
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003584
3585 listen management
3586 bind 10.0.0.3:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02003587 bind-process 1 2 3 4
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003588
Willy Tarreau110ecc12012-11-15 17:50:01 +01003589 listen management
3590 bind 10.0.0.4:80
3591 bind-process 1-4
3592
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02003593 See also : "nbproc" in global section, and "process" in section 5.1.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003594
3595
Willy Tarreau0ba27502007-12-24 16:55:16 +01003596capture cookie <name> len <length>
3597 Capture and log a cookie in the request and in the response.
3598 May be used in sections : defaults | frontend | listen | backend
3599 no | yes | yes | no
3600 Arguments :
3601 <name> is the beginning of the name of the cookie to capture. In order
3602 to match the exact name, simply suffix the name with an equal
3603 sign ('='). The full name will appear in the logs, which is
3604 useful with application servers which adjust both the cookie name
Davor Ocelice9ed2812017-12-25 17:49:28 +01003605 and value (e.g. ASPSESSIONXXX).
Willy Tarreau0ba27502007-12-24 16:55:16 +01003606
3607 <length> is the maximum number of characters to report in the logs, which
3608 include the cookie name, the equal sign and the value, all in the
3609 standard "name=value" form. The string will be truncated on the
3610 right if it exceeds <length>.
3611
3612 Only the first cookie is captured. Both the "cookie" request headers and the
3613 "set-cookie" response headers are monitored. This is particularly useful to
3614 check for application bugs causing session crossing or stealing between
3615 users, because generally the user's cookies can only change on a login page.
3616
3617 When the cookie was not presented by the client, the associated log column
3618 will report "-". When a request does not cause a cookie to be assigned by the
3619 server, a "-" is reported in the response column.
3620
3621 The capture is performed in the frontend only because it is necessary that
3622 the log format does not change for a given frontend depending on the
3623 backends. This may change in the future. Note that there can be only one
Willy Tarreau193b8c62012-11-22 00:17:38 +01003624 "capture cookie" statement in a frontend. The maximum capture length is set
3625 by the global "tune.http.cookielen" setting and defaults to 63 characters. It
3626 is not possible to specify a capture in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003627
3628 Example:
3629 capture cookie ASPSESSION len 32
3630
3631 See also : "capture request header", "capture response header" as well as
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003632 section 8 about logging.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003633
3634
3635capture request header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01003636 Capture and log the last occurrence of the specified request header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003637 May be used in sections : defaults | frontend | listen | backend
3638 no | yes | yes | no
3639 Arguments :
3640 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003641 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01003642 appear in the requests, with the first letter of each word in
3643 upper case. The header name will not appear in the logs, only the
3644 value is reported, but the position in the logs is respected.
3645
3646 <length> is the maximum number of characters to extract from the value and
3647 report in the logs. The string will be truncated on the right if
3648 it exceeds <length>.
3649
Willy Tarreau4460d032012-11-21 23:37:37 +01003650 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01003651 value will be added to the logs between braces ('{}'). If multiple headers
3652 are captured, they will be delimited by a vertical bar ('|') and will appear
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003653 in the same order they were declared in the configuration. Non-existent
3654 headers will be logged just as an empty string. Common uses for request
3655 header captures include the "Host" field in virtual hosting environments, the
3656 "Content-length" when uploads are supported, "User-agent" to quickly
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003657 differentiate between real users and robots, and "X-Forwarded-For" in proxied
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003658 environments to find where the request came from.
3659
3660 Note that when capturing headers such as "User-agent", some spaces may be
3661 logged, making the log analysis more difficult. Thus be careful about what
3662 you log if you know your log parser is not smart enough to rely on the
3663 braces.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003664
Willy Tarreau0900abb2012-11-22 00:21:46 +01003665 There is no limit to the number of captured request headers nor to their
3666 length, though it is wise to keep them low to limit memory usage per session.
3667 In order to keep log format consistent for a same frontend, header captures
3668 can only be declared in a frontend. It is not possible to specify a capture
3669 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003670
3671 Example:
3672 capture request header Host len 15
3673 capture request header X-Forwarded-For len 15
Cyril Bontéd1b0f7c2015-10-26 22:37:39 +01003674 capture request header Referer len 15
Willy Tarreau0ba27502007-12-24 16:55:16 +01003675
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003676 See also : "capture cookie", "capture response header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01003677 about logging.
3678
3679
3680capture response header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01003681 Capture and log the last occurrence of the specified response header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003682 May be used in sections : defaults | frontend | listen | backend
3683 no | yes | yes | no
3684 Arguments :
3685 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003686 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01003687 appear in the response, with the first letter of each word in
3688 upper case. The header name will not appear in the logs, only the
3689 value is reported, but the position in the logs is respected.
3690
3691 <length> is the maximum number of characters to extract from the value and
3692 report in the logs. The string will be truncated on the right if
3693 it exceeds <length>.
3694
Willy Tarreau4460d032012-11-21 23:37:37 +01003695 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01003696 result will be added to the logs between braces ('{}') after the captured
3697 request headers. If multiple headers are captured, they will be delimited by
3698 a vertical bar ('|') and will appear in the same order they were declared in
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003699 the configuration. Non-existent headers will be logged just as an empty
3700 string. Common uses for response header captures include the "Content-length"
3701 header which indicates how many bytes are expected to be returned, the
3702 "Location" header to track redirections.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003703
Willy Tarreau0900abb2012-11-22 00:21:46 +01003704 There is no limit to the number of captured response headers nor to their
3705 length, though it is wise to keep them low to limit memory usage per session.
3706 In order to keep log format consistent for a same frontend, header captures
3707 can only be declared in a frontend. It is not possible to specify a capture
3708 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003709
3710 Example:
3711 capture response header Content-length len 9
3712 capture response header Location len 15
3713
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003714 See also : "capture cookie", "capture request header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01003715 about logging.
3716
3717
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09003718clitcpka-cnt <count>
3719 Sets the maximum number of keepalive probes TCP should send before dropping
3720 the connection on the client side.
3721 May be used in sections : defaults | frontend | listen | backend
3722 yes | yes | yes | no
3723 Arguments :
3724 <count> is the maximum number of keepalive probes.
3725
3726 This keyword corresponds to the socket option TCP_KEEPCNT. If this keyword
3727 is not specified, system-wide TCP parameter (tcp_keepalive_probes) is used.
Willy Tarreau52543212020-07-09 05:58:51 +02003728 The availability of this setting depends on the operating system. It is
3729 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09003730
3731 See also : "option clitcpka", "clitcpka-idle", "clitcpka-intvl".
3732
3733
3734clitcpka-idle <timeout>
3735 Sets the time the connection needs to remain idle before TCP starts sending
3736 keepalive probes, if enabled the sending of TCP keepalive packets on the
3737 client side.
3738 May be used in sections : defaults | frontend | listen | backend
3739 yes | yes | yes | no
3740 Arguments :
3741 <timeout> is the time the connection needs to remain idle before TCP starts
3742 sending keepalive probes. It is specified in seconds by default,
3743 but can be in any other unit if the number is suffixed by the
3744 unit, as explained at the top of this document.
3745
3746 This keyword corresponds to the socket option TCP_KEEPIDLE. If this keyword
3747 is not specified, system-wide TCP parameter (tcp_keepalive_time) is used.
Willy Tarreau52543212020-07-09 05:58:51 +02003748 The availability of this setting depends on the operating system. It is
3749 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09003750
3751 See also : "option clitcpka", "clitcpka-cnt", "clitcpka-intvl".
3752
3753
3754clitcpka-intvl <timeout>
3755 Sets the time between individual keepalive probes on the client side.
3756 May be used in sections : defaults | frontend | listen | backend
3757 yes | yes | yes | no
3758 Arguments :
3759 <timeout> is the time between individual keepalive probes. It is specified
3760 in seconds by default, but can be in any other unit if the number
3761 is suffixed by the unit, as explained at the top of this
3762 document.
3763
3764 This keyword corresponds to the socket option TCP_KEEPINTVL. If this keyword
3765 is not specified, system-wide TCP parameter (tcp_keepalive_intvl) is used.
Willy Tarreau52543212020-07-09 05:58:51 +02003766 The availability of this setting depends on the operating system. It is
3767 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09003768
3769 See also : "option clitcpka", "clitcpka-cnt", "clitcpka-idle".
3770
3771
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003772compression algo <algorithm> ...
3773compression type <mime type> ...
Willy Tarreau70737d12012-10-27 00:34:28 +02003774compression offload
William Lallemand82fe75c2012-10-23 10:25:10 +02003775 Enable HTTP compression.
3776 May be used in sections : defaults | frontend | listen | backend
3777 yes | yes | yes | yes
3778 Arguments :
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003779 algo is followed by the list of supported compression algorithms.
3780 type is followed by the list of MIME types that will be compressed.
3781 offload makes haproxy work as a compression offloader only (see notes).
3782
3783 The currently supported algorithms are :
Willy Tarreauc91840a2015-03-28 17:00:39 +01003784 identity this is mostly for debugging, and it was useful for developing
3785 the compression feature. Identity does not apply any change on
3786 data.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003787
Willy Tarreauc91840a2015-03-28 17:00:39 +01003788 gzip applies gzip compression. This setting is only available when
Baptiste Assmannf085d632015-12-21 17:57:32 +01003789 support for zlib or libslz was built in.
Willy Tarreauc91840a2015-03-28 17:00:39 +01003790
3791 deflate same as "gzip", but with deflate algorithm and zlib format.
3792 Note that this algorithm has ambiguous support on many
3793 browsers and no support at all from recent ones. It is
3794 strongly recommended not to use it for anything else than
3795 experimentation. This setting is only available when support
Baptiste Assmannf085d632015-12-21 17:57:32 +01003796 for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003797
Willy Tarreauc91840a2015-03-28 17:00:39 +01003798 raw-deflate same as "deflate" without the zlib wrapper, and used as an
3799 alternative when the browser wants "deflate". All major
3800 browsers understand it and despite violating the standards,
3801 it is known to work better than "deflate", at least on MSIE
3802 and some versions of Safari. Do not use it in conjunction
3803 with "deflate", use either one or the other since both react
3804 to the same Accept-Encoding token. This setting is only
Baptiste Assmannf085d632015-12-21 17:57:32 +01003805 available when support for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003806
Dmitry Sivachenko87c208b2012-11-22 20:03:26 +04003807 Compression will be activated depending on the Accept-Encoding request
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003808 header. With identity, it does not take care of that header.
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04003809 If backend servers support HTTP compression, these directives
3810 will be no-op: haproxy will see the compressed response and will not
3811 compress again. If backend servers do not support HTTP compression and
3812 there is Accept-Encoding header in request, haproxy will compress the
3813 matching response.
Willy Tarreau70737d12012-10-27 00:34:28 +02003814
3815 The "offload" setting makes haproxy remove the Accept-Encoding header to
3816 prevent backend servers from compressing responses. It is strongly
3817 recommended not to do this because this means that all the compression work
3818 will be done on the single point where haproxy is located. However in some
3819 deployment scenarios, haproxy may be installed in front of a buggy gateway
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04003820 with broken HTTP compression implementation which can't be turned off.
3821 In that case haproxy can be used to prevent that gateway from emitting
3822 invalid payloads. In this case, simply removing the header in the
3823 configuration does not work because it applies before the header is parsed,
3824 so that prevents haproxy from compressing. The "offload" setting should
Willy Tarreauffea9fd2014-07-12 16:37:02 +02003825 then be used for such scenarios. Note: for now, the "offload" setting is
3826 ignored when set in a defaults section.
William Lallemand82fe75c2012-10-23 10:25:10 +02003827
William Lallemand05097442012-11-20 12:14:28 +01003828 Compression is disabled when:
Baptiste Assmann650d53d2013-01-05 15:44:44 +01003829 * the request does not advertise a supported compression algorithm in the
3830 "Accept-Encoding" header
3831 * the response message is not HTTP/1.1
Tim Duesterhusbb48c9a2019-01-30 23:46:04 +01003832 * HTTP status code is not one of 200, 201, 202, or 203
Baptiste Assmann650d53d2013-01-05 15:44:44 +01003833 * response contain neither a "Content-Length" header nor a
3834 "Transfer-Encoding" whose last value is "chunked"
3835 * response contains a "Content-Type" header whose first value starts with
3836 "multipart"
3837 * the response contains the "no-transform" value in the "Cache-control"
3838 header
3839 * User-Agent matches "Mozilla/4" unless it is MSIE 6 with XP SP2, or MSIE 7
3840 and later
3841 * The response contains a "Content-Encoding" header, indicating that the
3842 response is already compressed (see compression offload)
Tim Duesterhusbb48c9a2019-01-30 23:46:04 +01003843 * The response contains an invalid "ETag" header or multiple ETag headers
William Lallemand05097442012-11-20 12:14:28 +01003844
Tim Duesterhusb229f012019-01-29 16:38:56 +01003845 Note: The compression does not emit the Warning header.
William Lallemand05097442012-11-20 12:14:28 +01003846
William Lallemand82fe75c2012-10-23 10:25:10 +02003847 Examples :
3848 compression algo gzip
3849 compression type text/html text/plain
Willy Tarreau0ba27502007-12-24 16:55:16 +01003850
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003851
Willy Tarreau55165fe2009-05-10 12:02:55 +02003852cookie <name> [ rewrite | insert | prefix ] [ indirect ] [ nocache ]
Willy Tarreau4992dd22012-05-31 21:02:17 +02003853 [ postonly ] [ preserve ] [ httponly ] [ secure ]
3854 [ domain <domain> ]* [ maxidle <idle> ] [ maxlife <life> ]
Christopher Faulet2f533902020-01-21 11:06:48 +01003855 [ dynamic ] [ attr <value> ]*
Willy Tarreau0ba27502007-12-24 16:55:16 +01003856 Enable cookie-based persistence in a backend.
3857 May be used in sections : defaults | frontend | listen | backend
3858 yes | no | yes | yes
3859 Arguments :
3860 <name> is the name of the cookie which will be monitored, modified or
3861 inserted in order to bring persistence. This cookie is sent to
3862 the client via a "Set-Cookie" header in the response, and is
3863 brought back by the client in a "Cookie" header in all requests.
3864 Special care should be taken to choose a name which does not
3865 conflict with any likely application cookie. Also, if the same
Davor Ocelice9ed2812017-12-25 17:49:28 +01003866 backends are subject to be used by the same clients (e.g.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003867 HTTP/HTTPS), care should be taken to use different cookie names
3868 between all backends if persistence between them is not desired.
3869
3870 rewrite This keyword indicates that the cookie will be provided by the
3871 server and that haproxy will have to modify its value to set the
3872 server's identifier in it. This mode is handy when the management
3873 of complex combinations of "Set-cookie" and "Cache-control"
3874 headers is left to the application. The application can then
3875 decide whether or not it is appropriate to emit a persistence
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003876 cookie. Since all responses should be monitored, this mode
3877 doesn't work in HTTP tunnel mode. Unless the application
Davor Ocelice9ed2812017-12-25 17:49:28 +01003878 behavior is very complex and/or broken, it is advised not to
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003879 start with this mode for new deployments. This keyword is
3880 incompatible with "insert" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003881
3882 insert This keyword indicates that the persistence cookie will have to
Willy Tarreaua79094d2010-08-31 22:54:15 +02003883 be inserted by haproxy in server responses if the client did not
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003884
Willy Tarreaua79094d2010-08-31 22:54:15 +02003885 already have a cookie that would have permitted it to access this
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003886 server. When used without the "preserve" option, if the server
Michael Prokop4438c602019-05-24 10:25:45 +02003887 emits a cookie with the same name, it will be removed before
Davor Ocelice9ed2812017-12-25 17:49:28 +01003888 processing. For this reason, this mode can be used to upgrade
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003889 existing configurations running in the "rewrite" mode. The cookie
3890 will only be a session cookie and will not be stored on the
3891 client's disk. By default, unless the "indirect" option is added,
3892 the server will see the cookies emitted by the client. Due to
3893 caching effects, it is generally wise to add the "nocache" or
3894 "postonly" keywords (see below). The "insert" keyword is not
3895 compatible with "rewrite" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003896
3897 prefix This keyword indicates that instead of relying on a dedicated
3898 cookie for the persistence, an existing one will be completed.
3899 This may be needed in some specific environments where the client
3900 does not support more than one single cookie and the application
3901 already needs it. In this case, whenever the server sets a cookie
3902 named <name>, it will be prefixed with the server's identifier
3903 and a delimiter. The prefix will be removed from all client
3904 requests so that the server still finds the cookie it emitted.
3905 Since all requests and responses are subject to being modified,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003906 this mode doesn't work with tunnel mode. The "prefix" keyword is
Willy Tarreau37229df2011-10-17 12:24:55 +02003907 not compatible with "rewrite" and "insert". Note: it is highly
3908 recommended not to use "indirect" with "prefix", otherwise server
3909 cookie updates would not be sent to clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003910
Willy Tarreaua79094d2010-08-31 22:54:15 +02003911 indirect When this option is specified, no cookie will be emitted to a
3912 client which already has a valid one for the server which has
3913 processed the request. If the server sets such a cookie itself,
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003914 it will be removed, unless the "preserve" option is also set. In
3915 "insert" mode, this will additionally remove cookies from the
3916 requests transmitted to the server, making the persistence
3917 mechanism totally transparent from an application point of view.
Willy Tarreau37229df2011-10-17 12:24:55 +02003918 Note: it is highly recommended not to use "indirect" with
3919 "prefix", otherwise server cookie updates would not be sent to
3920 clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003921
3922 nocache This option is recommended in conjunction with the insert mode
3923 when there is a cache between the client and HAProxy, as it
3924 ensures that a cacheable response will be tagged non-cacheable if
3925 a cookie needs to be inserted. This is important because if all
3926 persistence cookies are added on a cacheable home page for
3927 instance, then all customers will then fetch the page from an
3928 outer cache and will all share the same persistence cookie,
3929 leading to one server receiving much more traffic than others.
3930 See also the "insert" and "postonly" options.
3931
3932 postonly This option ensures that cookie insertion will only be performed
3933 on responses to POST requests. It is an alternative to the
3934 "nocache" option, because POST responses are not cacheable, so
3935 this ensures that the persistence cookie will never get cached.
3936 Since most sites do not need any sort of persistence before the
3937 first POST which generally is a login request, this is a very
3938 efficient method to optimize caching without risking to find a
3939 persistence cookie in the cache.
3940 See also the "insert" and "nocache" options.
3941
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003942 preserve This option may only be used with "insert" and/or "indirect". It
3943 allows the server to emit the persistence cookie itself. In this
3944 case, if a cookie is found in the response, haproxy will leave it
3945 untouched. This is useful in order to end persistence after a
3946 logout request for instance. For this, the server just has to
Davor Ocelice9ed2812017-12-25 17:49:28 +01003947 emit a cookie with an invalid value (e.g. empty) or with a date in
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003948 the past. By combining this mechanism with the "disable-on-404"
3949 check option, it is possible to perform a completely graceful
3950 shutdown because users will definitely leave the server after
3951 they logout.
3952
Willy Tarreau4992dd22012-05-31 21:02:17 +02003953 httponly This option tells haproxy to add an "HttpOnly" cookie attribute
3954 when a cookie is inserted. This attribute is used so that a
3955 user agent doesn't share the cookie with non-HTTP components.
3956 Please check RFC6265 for more information on this attribute.
3957
3958 secure This option tells haproxy to add a "Secure" cookie attribute when
3959 a cookie is inserted. This attribute is used so that a user agent
3960 never emits this cookie over non-secure channels, which means
3961 that a cookie learned with this flag will be presented only over
3962 SSL/TLS connections. Please check RFC6265 for more information on
3963 this attribute.
3964
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02003965 domain This option allows to specify the domain at which a cookie is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003966 inserted. It requires exactly one parameter: a valid domain
Willy Tarreau68a897b2009-12-03 23:28:34 +01003967 name. If the domain begins with a dot, the browser is allowed to
3968 use it for any host ending with that name. It is also possible to
3969 specify several domain names by invoking this option multiple
3970 times. Some browsers might have small limits on the number of
3971 domains, so be careful when doing that. For the record, sending
3972 10 domains to MSIE 6 or Firefox 2 works as expected.
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02003973
Willy Tarreau996a92c2010-10-13 19:30:47 +02003974 maxidle This option allows inserted cookies to be ignored after some idle
3975 time. It only works with insert-mode cookies. When a cookie is
3976 sent to the client, the date this cookie was emitted is sent too.
3977 Upon further presentations of this cookie, if the date is older
3978 than the delay indicated by the parameter (in seconds), it will
3979 be ignored. Otherwise, it will be refreshed if needed when the
3980 response is sent to the client. This is particularly useful to
3981 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01003982 too long on the same server (e.g. after a farm size change). When
Willy Tarreau996a92c2010-10-13 19:30:47 +02003983 this option is set and a cookie has no date, it is always
3984 accepted, but gets refreshed in the response. This maintains the
3985 ability for admins to access their sites. Cookies that have a
3986 date in the future further than 24 hours are ignored. Doing so
3987 lets admins fix timezone issues without risking kicking users off
3988 the site.
3989
3990 maxlife This option allows inserted cookies to be ignored after some life
3991 time, whether they're in use or not. It only works with insert
3992 mode cookies. When a cookie is first sent to the client, the date
3993 this cookie was emitted is sent too. Upon further presentations
3994 of this cookie, if the date is older than the delay indicated by
3995 the parameter (in seconds), it will be ignored. If the cookie in
3996 the request has no date, it is accepted and a date will be set.
3997 Cookies that have a date in the future further than 24 hours are
3998 ignored. Doing so lets admins fix timezone issues without risking
3999 kicking users off the site. Contrary to maxidle, this value is
4000 not refreshed, only the first visit date counts. Both maxidle and
4001 maxlife may be used at the time. This is particularly useful to
4002 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01004003 too long on the same server (e.g. after a farm size change). This
Willy Tarreau996a92c2010-10-13 19:30:47 +02004004 is stronger than the maxidle method in that it forces a
4005 redispatch after some absolute delay.
4006
Olivier Houchard4e694042017-03-14 20:01:29 +01004007 dynamic Activate dynamic cookies. When used, a session cookie is
4008 dynamically created for each server, based on the IP and port
4009 of the server, and a secret key, specified in the
4010 "dynamic-cookie-key" backend directive.
4011 The cookie will be regenerated each time the IP address change,
4012 and is only generated for IPv4/IPv6.
4013
Christopher Faulet2f533902020-01-21 11:06:48 +01004014 attr This option tells haproxy to add an extra attribute when a
4015 cookie is inserted. The attribute value can contain any
4016 characters except control ones or ";". This option may be
4017 repeated.
4018
Willy Tarreau0ba27502007-12-24 16:55:16 +01004019 There can be only one persistence cookie per HTTP backend, and it can be
4020 declared in a defaults section. The value of the cookie will be the value
4021 indicated after the "cookie" keyword in a "server" statement. If no cookie
4022 is declared for a given server, the cookie is not set.
Willy Tarreau6a06a402007-07-15 20:15:28 +02004023
Willy Tarreau0ba27502007-12-24 16:55:16 +01004024 Examples :
4025 cookie JSESSIONID prefix
4026 cookie SRV insert indirect nocache
4027 cookie SRV insert postonly indirect
Willy Tarreau996a92c2010-10-13 19:30:47 +02004028 cookie SRV insert indirect nocache maxidle 30m maxlife 8h
Willy Tarreau0ba27502007-12-24 16:55:16 +01004029
Willy Tarreau294d0f02015-08-10 19:40:12 +02004030 See also : "balance source", "capture cookie", "server" and "ignore-persist".
Willy Tarreau0ba27502007-12-24 16:55:16 +01004031
Willy Tarreau983e01e2010-01-11 18:42:06 +01004032
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02004033declare capture [ request | response ] len <length>
4034 Declares a capture slot.
4035 May be used in sections : defaults | frontend | listen | backend
4036 no | yes | yes | no
4037 Arguments:
4038 <length> is the length allowed for the capture.
4039
4040 This declaration is only available in the frontend or listen section, but the
4041 reserved slot can be used in the backends. The "request" keyword allocates a
4042 capture slot for use in the request, and "response" allocates a capture slot
4043 for use in the response.
4044
4045 See also: "capture-req", "capture-res" (sample converters),
Baptiste Assmann5ac425c2015-10-21 23:13:46 +02004046 "capture.req.hdr", "capture.res.hdr" (sample fetches),
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02004047 "http-request capture" and "http-response capture".
4048
4049
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01004050default-server [param*]
4051 Change default options for a server in a backend
4052 May be used in sections : defaults | frontend | listen | backend
4053 yes | no | yes | yes
4054 Arguments:
Willy Tarreau983e01e2010-01-11 18:42:06 +01004055 <param*> is a list of parameters for this server. The "default-server"
4056 keyword accepts an important number of options and has a complete
4057 section dedicated to it. Please refer to section 5 for more
4058 details.
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01004059
Willy Tarreau983e01e2010-01-11 18:42:06 +01004060 Example :
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01004061 default-server inter 1000 weight 13
4062
4063 See also: "server" and section 5 about server options
Willy Tarreau0ba27502007-12-24 16:55:16 +01004064
Willy Tarreau983e01e2010-01-11 18:42:06 +01004065
Willy Tarreau0ba27502007-12-24 16:55:16 +01004066default_backend <backend>
4067 Specify the backend to use when no "use_backend" rule has been matched.
4068 May be used in sections : defaults | frontend | listen | backend
4069 yes | yes | yes | no
4070 Arguments :
4071 <backend> is the name of the backend to use.
4072
4073 When doing content-switching between frontend and backends using the
4074 "use_backend" keyword, it is often useful to indicate which backend will be
4075 used when no rule has matched. It generally is the dynamic backend which
4076 will catch all undetermined requests.
4077
Willy Tarreau0ba27502007-12-24 16:55:16 +01004078 Example :
4079
4080 use_backend dynamic if url_dyn
4081 use_backend static if url_css url_img extension_img
4082 default_backend dynamic
4083
Willy Tarreau98d04852015-05-26 12:18:29 +02004084 See also : "use_backend"
Willy Tarreau2769aa02007-12-27 18:26:09 +01004085
Willy Tarreau0ba27502007-12-24 16:55:16 +01004086
Baptiste Assmann27f51342013-10-09 06:51:49 +02004087description <string>
4088 Describe a listen, frontend or backend.
4089 May be used in sections : defaults | frontend | listen | backend
4090 no | yes | yes | yes
4091 Arguments : string
4092
4093 Allows to add a sentence to describe the related object in the HAProxy HTML
4094 stats page. The description will be printed on the right of the object name
4095 it describes.
4096 No need to backslash spaces in the <string> arguments.
4097
4098
Willy Tarreau0ba27502007-12-24 16:55:16 +01004099disabled
4100 Disable a proxy, frontend or backend.
4101 May be used in sections : defaults | frontend | listen | backend
4102 yes | yes | yes | yes
4103 Arguments : none
4104
4105 The "disabled" keyword is used to disable an instance, mainly in order to
4106 liberate a listening port or to temporarily disable a service. The instance
4107 will still be created and its configuration will be checked, but it will be
4108 created in the "stopped" state and will appear as such in the statistics. It
4109 will not receive any traffic nor will it send any health-checks or logs. It
4110 is possible to disable many instances at once by adding the "disabled"
4111 keyword in a "defaults" section.
4112
4113 See also : "enabled"
4114
4115
Willy Tarreau5ce94572010-06-07 14:35:41 +02004116dispatch <address>:<port>
4117 Set a default server address
4118 May be used in sections : defaults | frontend | listen | backend
4119 no | no | yes | yes
Cyril Bonté108cf6e2012-04-21 23:30:29 +02004120 Arguments :
Willy Tarreau5ce94572010-06-07 14:35:41 +02004121
4122 <address> is the IPv4 address of the default server. Alternatively, a
4123 resolvable hostname is supported, but this name will be resolved
4124 during start-up.
4125
4126 <ports> is a mandatory port specification. All connections will be sent
4127 to this port, and it is not permitted to use port offsets as is
4128 possible with normal servers.
4129
Willy Tarreau787aed52011-04-15 06:45:37 +02004130 The "dispatch" keyword designates a default server for use when no other
Willy Tarreau5ce94572010-06-07 14:35:41 +02004131 server can take the connection. In the past it was used to forward non
4132 persistent connections to an auxiliary load balancer. Due to its simple
4133 syntax, it has also been used for simple TCP relays. It is recommended not to
4134 use it for more clarity, and to use the "server" directive instead.
4135
4136 See also : "server"
4137
Olivier Houchard4e694042017-03-14 20:01:29 +01004138
4139dynamic-cookie-key <string>
4140 Set the dynamic cookie secret key for a backend.
4141 May be used in sections : defaults | frontend | listen | backend
4142 yes | no | yes | yes
4143 Arguments : The secret key to be used.
4144
4145 When dynamic cookies are enabled (see the "dynamic" directive for cookie),
Davor Ocelice9ed2812017-12-25 17:49:28 +01004146 a dynamic cookie is created for each server (unless one is explicitly
Olivier Houchard4e694042017-03-14 20:01:29 +01004147 specified on the "server" line), using a hash of the IP address of the
4148 server, the TCP port, and the secret key.
Davor Ocelice9ed2812017-12-25 17:49:28 +01004149 That way, we can ensure session persistence across multiple load-balancers,
Olivier Houchard4e694042017-03-14 20:01:29 +01004150 even if servers are dynamically added or removed.
Willy Tarreau5ce94572010-06-07 14:35:41 +02004151
Willy Tarreau0ba27502007-12-24 16:55:16 +01004152enabled
4153 Enable a proxy, frontend or backend.
4154 May be used in sections : defaults | frontend | listen | backend
4155 yes | yes | yes | yes
4156 Arguments : none
4157
4158 The "enabled" keyword is used to explicitly enable an instance, when the
4159 defaults has been set to "disabled". This is very rarely used.
4160
4161 See also : "disabled"
4162
4163
4164errorfile <code> <file>
4165 Return a file contents instead of errors generated by HAProxy
4166 May be used in sections : defaults | frontend | listen | backend
4167 yes | yes | yes | yes
4168 Arguments :
4169 <code> is the HTTP status code. Currently, HAProxy is capable of
Christopher Faulet612f2ea2020-05-27 09:57:28 +02004170 generating codes 200, 400, 401, 403, 404, 405, 407, 408, 410,
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02004171 413, 425, 429, 500, 502, 503, and 504.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004172
4173 <file> designates a file containing the full HTTP response. It is
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01004174 recommended to follow the common practice of appending ".http" to
Willy Tarreau0ba27502007-12-24 16:55:16 +01004175 the filename so that people do not confuse the response with HTML
Willy Tarreau59140a22009-02-22 12:02:30 +01004176 error pages, and to use absolute paths, since files are read
4177 before any chroot is performed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004178
4179 It is important to understand that this keyword is not meant to rewrite
4180 errors returned by the server, but errors detected and returned by HAProxy.
4181 This is why the list of supported errors is limited to a small set.
4182
Willy Tarreauae94d4d2011-05-11 16:28:49 +02004183 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
4184
Christopher Faulet70170672020-05-18 17:42:48 +02004185 The files are parsed when HAProxy starts and must be valid according to the
4186 HTTP specification. They should not exceed the configured buffer size
4187 (BUFSIZE), which generally is 16 kB, otherwise an internal error will be
4188 returned. It is also wise not to put any reference to local contents
4189 (e.g. images) in order to avoid loops between the client and HAProxy when all
4190 servers are down, causing an error to be returned instead of an
4191 image. Finally, The response cannot exceed (tune.bufsize - tune.maxrewrite)
4192 so that "http-after-response" rules still have room to operate (see
4193 "tune.maxrewrite").
Willy Tarreau59140a22009-02-22 12:02:30 +01004194
Willy Tarreau0ba27502007-12-24 16:55:16 +01004195 The files are read at the same time as the configuration and kept in memory.
4196 For this reason, the errors continue to be returned even when the process is
4197 chrooted, and no file change is considered while the process is running. A
Willy Tarreauc27debf2008-01-06 08:57:02 +01004198 simple method for developing those files consists in associating them to the
Willy Tarreau0ba27502007-12-24 16:55:16 +01004199 403 status code and interrogating a blocked URL.
4200
Christopher Faulet3b967c12020-05-15 15:47:44 +02004201 See also : "http-error", "errorloc", "errorloc302", "errorloc303"
Willy Tarreau0ba27502007-12-24 16:55:16 +01004202
Willy Tarreau59140a22009-02-22 12:02:30 +01004203 Example :
4204 errorfile 400 /etc/haproxy/errorfiles/400badreq.http
Willy Tarreau989222a2016-01-15 10:26:26 +01004205 errorfile 408 /dev/null # work around Chrome pre-connect bug
Willy Tarreau59140a22009-02-22 12:02:30 +01004206 errorfile 403 /etc/haproxy/errorfiles/403forbid.http
4207 errorfile 503 /etc/haproxy/errorfiles/503sorry.http
4208
Willy Tarreau2769aa02007-12-27 18:26:09 +01004209
Christopher Faulet76edc0f2020-01-13 15:52:01 +01004210errorfiles <name> [<code> ...]
4211 Import, fully or partially, the error files defined in the <name> http-errors
4212 section.
4213 May be used in sections : defaults | frontend | listen | backend
4214 yes | yes | yes | yes
4215 Arguments :
4216 <name> is the name of an existing http-errors section.
4217
4218 <code> is a HTTP status code. Several status code may be listed.
Christopher Faulet612f2ea2020-05-27 09:57:28 +02004219 Currently, HAProxy is capable of generating codes 200, 400, 401,
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02004220 403, 404, 405, 407, 408, 410, 413, 425, 429, 500, 502, 503, and 504.
Christopher Faulet76edc0f2020-01-13 15:52:01 +01004221
4222 Errors defined in the http-errors section with the name <name> are imported
4223 in the current proxy. If no status code is specified, all error files of the
4224 http-errors section are imported. Otherwise, only error files associated to
4225 the listed status code are imported. Those error files override the already
4226 defined custom errors for the proxy. And they may be overridden by following
Daniel Corbett67a82712020-07-06 23:01:19 -04004227 ones. Functionally, it is exactly the same as declaring all error files by
Christopher Faulet76edc0f2020-01-13 15:52:01 +01004228 hand using "errorfile" directives.
4229
Christopher Faulet3b967c12020-05-15 15:47:44 +02004230 See also : "http-error", "errorfile", "errorloc", "errorloc302" ,
4231 "errorloc303" and section 3.8 about http-errors.
Christopher Faulet76edc0f2020-01-13 15:52:01 +01004232
4233 Example :
4234 errorfiles generic
4235 errorfiles site-1 403 404
4236
4237
Willy Tarreau2769aa02007-12-27 18:26:09 +01004238errorloc <code> <url>
4239errorloc302 <code> <url>
4240 Return an HTTP redirection to a URL instead of errors generated by HAProxy
4241 May be used in sections : defaults | frontend | listen | backend
4242 yes | yes | yes | yes
4243 Arguments :
4244 <code> is the HTTP status code. Currently, HAProxy is capable of
Christopher Faulet612f2ea2020-05-27 09:57:28 +02004245 generating codes 200, 400, 401, 403, 404, 405, 407, 408, 410,
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02004246 413, 425, 429, 500, 502, 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01004247
4248 <url> it is the exact contents of the "Location" header. It may contain
4249 either a relative URI to an error page hosted on the same site,
4250 or an absolute URI designating an error page on another site.
4251 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01004252 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01004253
4254 It is important to understand that this keyword is not meant to rewrite
4255 errors returned by the server, but errors detected and returned by HAProxy.
4256 This is why the list of supported errors is limited to a small set.
4257
Willy Tarreauae94d4d2011-05-11 16:28:49 +02004258 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
4259
Willy Tarreau2769aa02007-12-27 18:26:09 +01004260 Note that both keyword return the HTTP 302 status code, which tells the
4261 client to fetch the designated URL using the same HTTP method. This can be
4262 quite problematic in case of non-GET methods such as POST, because the URL
4263 sent to the client might not be allowed for something other than GET. To
Willy Tarreau989222a2016-01-15 10:26:26 +01004264 work around this problem, please use "errorloc303" which send the HTTP 303
Willy Tarreau2769aa02007-12-27 18:26:09 +01004265 status code, indicating to the client that the URL must be fetched with a GET
4266 request.
4267
Christopher Faulet3b967c12020-05-15 15:47:44 +02004268 See also : "http-error", "errorfile", "errorloc303"
Willy Tarreau2769aa02007-12-27 18:26:09 +01004269
4270
4271errorloc303 <code> <url>
4272 Return an HTTP redirection to a URL instead of errors generated by HAProxy
4273 May be used in sections : defaults | frontend | listen | backend
4274 yes | yes | yes | yes
4275 Arguments :
4276 <code> is the HTTP status code. Currently, HAProxy is capable of
Christopher Faulet612f2ea2020-05-27 09:57:28 +02004277 generating codes 200, 400, 401, 403, 404, 405, 407, 408, 410,
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02004278 413, 425, 429, 500, 502, 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01004279
4280 <url> it is the exact contents of the "Location" header. It may contain
4281 either a relative URI to an error page hosted on the same site,
4282 or an absolute URI designating an error page on another site.
4283 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01004284 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01004285
4286 It is important to understand that this keyword is not meant to rewrite
4287 errors returned by the server, but errors detected and returned by HAProxy.
4288 This is why the list of supported errors is limited to a small set.
4289
Willy Tarreauae94d4d2011-05-11 16:28:49 +02004290 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
4291
Willy Tarreau2769aa02007-12-27 18:26:09 +01004292 Note that both keyword return the HTTP 303 status code, which tells the
4293 client to fetch the designated URL using the same HTTP GET method. This
4294 solves the usual problems associated with "errorloc" and the 302 code. It is
4295 possible that some very old browsers designed before HTTP/1.1 do not support
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01004296 it, but no such problem has been reported till now.
Willy Tarreau2769aa02007-12-27 18:26:09 +01004297
Christopher Faulet3b967c12020-05-15 15:47:44 +02004298 See also : "http-error", "errorfile", "errorloc", "errorloc302"
Willy Tarreau2769aa02007-12-27 18:26:09 +01004299
4300
Simon Horman51a1cf62015-02-03 13:00:44 +09004301email-alert from <emailaddr>
4302 Declare the from email address to be used in both the envelope and header
Davor Ocelice9ed2812017-12-25 17:49:28 +01004303 of email alerts. This is the address that email alerts are sent from.
Simon Horman51a1cf62015-02-03 13:00:44 +09004304 May be used in sections: defaults | frontend | listen | backend
4305 yes | yes | yes | yes
4306
4307 Arguments :
4308
4309 <emailaddr> is the from email address to use when sending email alerts
4310
4311 Also requires "email-alert mailers" and "email-alert to" to be set
4312 and if so sending email alerts is enabled for the proxy.
4313
Simon Horman64e34162015-02-06 11:11:57 +09004314 See also : "email-alert level", "email-alert mailers",
Cyril Bonté307ee1e2015-09-28 23:16:06 +02004315 "email-alert myhostname", "email-alert to", section 3.6 about
4316 mailers.
Simon Horman64e34162015-02-06 11:11:57 +09004317
4318
4319email-alert level <level>
4320 Declare the maximum log level of messages for which email alerts will be
4321 sent. This acts as a filter on the sending of email alerts.
4322 May be used in sections: defaults | frontend | listen | backend
4323 yes | yes | yes | yes
4324
4325 Arguments :
4326
4327 <level> One of the 8 syslog levels:
4328 emerg alert crit err warning notice info debug
4329 The above syslog levels are ordered from lowest to highest.
4330
4331 By default level is alert
4332
4333 Also requires "email-alert from", "email-alert mailers" and
4334 "email-alert to" to be set and if so sending email alerts is enabled
4335 for the proxy.
4336
Simon Horman1421e212015-04-30 13:10:35 +09004337 Alerts are sent when :
4338
4339 * An un-paused server is marked as down and <level> is alert or lower
4340 * A paused server is marked as down and <level> is notice or lower
4341 * A server is marked as up or enters the drain state and <level>
4342 is notice or lower
4343 * "option log-health-checks" is enabled, <level> is info or lower,
4344 and a health check status update occurs
4345
Simon Horman64e34162015-02-06 11:11:57 +09004346 See also : "email-alert from", "email-alert mailers",
4347 "email-alert myhostname", "email-alert to",
Simon Horman51a1cf62015-02-03 13:00:44 +09004348 section 3.6 about mailers.
4349
4350
4351email-alert mailers <mailersect>
4352 Declare the mailers to be used when sending email alerts
4353 May be used in sections: defaults | frontend | listen | backend
4354 yes | yes | yes | yes
4355
4356 Arguments :
4357
4358 <mailersect> is the name of the mailers section to send email alerts.
4359
4360 Also requires "email-alert from" and "email-alert to" to be set
4361 and if so sending email alerts is enabled for the proxy.
4362
Simon Horman64e34162015-02-06 11:11:57 +09004363 See also : "email-alert from", "email-alert level", "email-alert myhostname",
4364 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09004365
4366
4367email-alert myhostname <hostname>
4368 Declare the to hostname address to be used when communicating with
4369 mailers.
4370 May be used in sections: defaults | frontend | listen | backend
4371 yes | yes | yes | yes
4372
4373 Arguments :
4374
Baptiste Assmann738bad92015-12-21 15:27:53 +01004375 <hostname> is the hostname to use when communicating with mailers
Simon Horman51a1cf62015-02-03 13:00:44 +09004376
4377 By default the systems hostname is used.
4378
4379 Also requires "email-alert from", "email-alert mailers" and
4380 "email-alert to" to be set and if so sending email alerts is enabled
4381 for the proxy.
4382
Simon Horman64e34162015-02-06 11:11:57 +09004383 See also : "email-alert from", "email-alert level", "email-alert mailers",
4384 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09004385
4386
4387email-alert to <emailaddr>
Davor Ocelice9ed2812017-12-25 17:49:28 +01004388 Declare both the recipient address in the envelope and to address in the
Simon Horman51a1cf62015-02-03 13:00:44 +09004389 header of email alerts. This is the address that email alerts are sent to.
4390 May be used in sections: defaults | frontend | listen | backend
4391 yes | yes | yes | yes
4392
4393 Arguments :
4394
4395 <emailaddr> is the to email address to use when sending email alerts
4396
4397 Also requires "email-alert mailers" and "email-alert to" to be set
4398 and if so sending email alerts is enabled for the proxy.
4399
Simon Horman64e34162015-02-06 11:11:57 +09004400 See also : "email-alert from", "email-alert level", "email-alert mailers",
Simon Horman51a1cf62015-02-03 13:00:44 +09004401 "email-alert myhostname", section 3.6 about mailers.
4402
4403
Willy Tarreau4de91492010-01-22 19:10:05 +01004404force-persist { if | unless } <condition>
4405 Declare a condition to force persistence on down servers
4406 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01004407 no | no | yes | yes
Willy Tarreau4de91492010-01-22 19:10:05 +01004408
4409 By default, requests are not dispatched to down servers. It is possible to
4410 force this using "option persist", but it is unconditional and redispatches
4411 to a valid server if "option redispatch" is set. That leaves with very little
4412 possibilities to force some requests to reach a server which is artificially
4413 marked down for maintenance operations.
4414
4415 The "force-persist" statement allows one to declare various ACL-based
4416 conditions which, when met, will cause a request to ignore the down status of
4417 a server and still try to connect to it. That makes it possible to start a
4418 server, still replying an error to the health checks, and run a specially
4419 configured browser to test the service. Among the handy methods, one could
4420 use a specific source IP address, or a specific cookie. The cookie also has
4421 the advantage that it can easily be added/removed on the browser from a test
4422 page. Once the service is validated, it is then possible to open the service
4423 to the world by returning a valid response to health checks.
4424
4425 The forced persistence is enabled when an "if" condition is met, or unless an
4426 "unless" condition is met. The final redispatch is always disabled when this
4427 is used.
4428
Cyril Bonté0d4bf012010-04-25 23:21:46 +02004429 See also : "option redispatch", "ignore-persist", "persist",
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +02004430 and section 7 about ACL usage.
Willy Tarreau4de91492010-01-22 19:10:05 +01004431
Christopher Fauletc3fe5332016-04-07 15:30:10 +02004432
4433filter <name> [param*]
4434 Add the filter <name> in the filter list attached to the proxy.
4435 May be used in sections : defaults | frontend | listen | backend
4436 no | yes | yes | yes
4437 Arguments :
4438 <name> is the name of the filter. Officially supported filters are
4439 referenced in section 9.
4440
Tim Düsterhus4896c442016-11-29 02:15:19 +01004441 <param*> is a list of parameters accepted by the filter <name>. The
Christopher Fauletc3fe5332016-04-07 15:30:10 +02004442 parsing of these parameters are the responsibility of the
Tim Düsterhus4896c442016-11-29 02:15:19 +01004443 filter. Please refer to the documentation of the corresponding
4444 filter (section 9) for all details on the supported parameters.
Christopher Fauletc3fe5332016-04-07 15:30:10 +02004445
4446 Multiple occurrences of the filter line can be used for the same proxy. The
4447 same filter can be referenced many times if needed.
4448
4449 Example:
4450 listen
4451 bind *:80
4452
4453 filter trace name BEFORE-HTTP-COMP
4454 filter compression
4455 filter trace name AFTER-HTTP-COMP
4456
4457 compression algo gzip
4458 compression offload
4459
4460 server srv1 192.168.0.1:80
4461
4462 See also : section 9.
4463
Willy Tarreau4de91492010-01-22 19:10:05 +01004464
Willy Tarreau2769aa02007-12-27 18:26:09 +01004465fullconn <conns>
4466 Specify at what backend load the servers will reach their maxconn
4467 May be used in sections : defaults | frontend | listen | backend
4468 yes | no | yes | yes
4469 Arguments :
4470 <conns> is the number of connections on the backend which will make the
4471 servers use the maximal number of connections.
4472
Willy Tarreau198a7442008-01-17 12:05:32 +01004473 When a server has a "maxconn" parameter specified, it means that its number
Willy Tarreau2769aa02007-12-27 18:26:09 +01004474 of concurrent connections will never go higher. Additionally, if it has a
Willy Tarreau198a7442008-01-17 12:05:32 +01004475 "minconn" parameter, it indicates a dynamic limit following the backend's
Willy Tarreau2769aa02007-12-27 18:26:09 +01004476 load. The server will then always accept at least <minconn> connections,
4477 never more than <maxconn>, and the limit will be on the ramp between both
4478 values when the backend has less than <conns> concurrent connections. This
4479 makes it possible to limit the load on the servers during normal loads, but
4480 push it further for important loads without overloading the servers during
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004481 exceptional loads.
Willy Tarreau2769aa02007-12-27 18:26:09 +01004482
Willy Tarreaufbb78422011-06-05 15:38:35 +02004483 Since it's hard to get this value right, haproxy automatically sets it to
4484 10% of the sum of the maxconns of all frontends that may branch to this
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01004485 backend (based on "use_backend" and "default_backend" rules). That way it's
4486 safe to leave it unset. However, "use_backend" involving dynamic names are
4487 not counted since there is no way to know if they could match or not.
Willy Tarreaufbb78422011-06-05 15:38:35 +02004488
Willy Tarreau2769aa02007-12-27 18:26:09 +01004489 Example :
4490 # The servers will accept between 100 and 1000 concurrent connections each
4491 # and the maximum of 1000 will be reached when the backend reaches 10000
4492 # connections.
4493 backend dynamic
4494 fullconn 10000
4495 server srv1 dyn1:80 minconn 100 maxconn 1000
4496 server srv2 dyn2:80 minconn 100 maxconn 1000
4497
4498 See also : "maxconn", "server"
4499
4500
4501grace <time>
4502 Maintain a proxy operational for some time after a soft stop
4503 May be used in sections : defaults | frontend | listen | backend
Cyril Bonté99ed3272010-01-24 23:29:44 +01004504 yes | yes | yes | yes
Willy Tarreau2769aa02007-12-27 18:26:09 +01004505 Arguments :
4506 <time> is the time (by default in milliseconds) for which the instance
4507 will remain operational with the frontend sockets still listening
4508 when a soft-stop is received via the SIGUSR1 signal.
4509
4510 This may be used to ensure that the services disappear in a certain order.
4511 This was designed so that frontends which are dedicated to monitoring by an
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004512 external equipment fail immediately while other ones remain up for the time
Willy Tarreau2769aa02007-12-27 18:26:09 +01004513 needed by the equipment to detect the failure.
4514
4515 Note that currently, there is very little benefit in using this parameter,
4516 and it may in fact complicate the soft-reconfiguration process more than
4517 simplify it.
4518
Willy Tarreau0ba27502007-12-24 16:55:16 +01004519
Andrew Rodland17be45e2016-10-25 17:04:12 -04004520hash-balance-factor <factor>
4521 Specify the balancing factor for bounded-load consistent hashing
4522 May be used in sections : defaults | frontend | listen | backend
4523 yes | no | no | yes
4524 Arguments :
4525 <factor> is the control for the maximum number of concurrent requests to
4526 send to a server, expressed as a percentage of the average number
Frédéric Lécaille93d33162019-03-06 09:35:59 +01004527 of concurrent requests across all of the active servers.
Andrew Rodland17be45e2016-10-25 17:04:12 -04004528
4529 Specifying a "hash-balance-factor" for a server with "hash-type consistent"
4530 enables an algorithm that prevents any one server from getting too many
4531 requests at once, even if some hash buckets receive many more requests than
4532 others. Setting <factor> to 0 (the default) disables the feature. Otherwise,
4533 <factor> is a percentage greater than 100. For example, if <factor> is 150,
4534 then no server will be allowed to have a load more than 1.5 times the average.
4535 If server weights are used, they will be respected.
4536
4537 If the first-choice server is disqualified, the algorithm will choose another
4538 server based on the request hash, until a server with additional capacity is
4539 found. A higher <factor> allows more imbalance between the servers, while a
4540 lower <factor> means that more servers will be checked on average, affecting
4541 performance. Reasonable values are from 125 to 200.
4542
Willy Tarreau760e81d2018-05-03 07:20:40 +02004543 This setting is also used by "balance random" which internally relies on the
4544 consistent hashing mechanism.
4545
Andrew Rodland17be45e2016-10-25 17:04:12 -04004546 See also : "balance" and "hash-type".
4547
4548
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004549hash-type <method> <function> <modifier>
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004550 Specify a method to use for mapping hashes to servers
4551 May be used in sections : defaults | frontend | listen | backend
4552 yes | no | yes | yes
4553 Arguments :
Bhaskar98634f02013-10-29 23:30:51 -04004554 <method> is the method used to select a server from the hash computed by
4555 the <function> :
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004556
Bhaskar98634f02013-10-29 23:30:51 -04004557 map-based the hash table is a static array containing all alive servers.
4558 The hashes will be very smooth, will consider weights, but
4559 will be static in that weight changes while a server is up
4560 will be ignored. This means that there will be no slow start.
4561 Also, since a server is selected by its position in the array,
4562 most mappings are changed when the server count changes. This
4563 means that when a server goes up or down, or when a server is
4564 added to a farm, most connections will be redistributed to
4565 different servers. This can be inconvenient with caches for
4566 instance.
Willy Tarreau798a39c2010-11-24 15:04:29 +01004567
Bhaskar98634f02013-10-29 23:30:51 -04004568 consistent the hash table is a tree filled with many occurrences of each
4569 server. The hash key is looked up in the tree and the closest
4570 server is chosen. This hash is dynamic, it supports changing
4571 weights while the servers are up, so it is compatible with the
4572 slow start feature. It has the advantage that when a server
4573 goes up or down, only its associations are moved. When a
4574 server is added to the farm, only a few part of the mappings
4575 are redistributed, making it an ideal method for caches.
4576 However, due to its principle, the distribution will never be
4577 very smooth and it may sometimes be necessary to adjust a
4578 server's weight or its ID to get a more balanced distribution.
4579 In order to get the same distribution on multiple load
4580 balancers, it is important that all servers have the exact
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004581 same IDs. Note: consistent hash uses sdbm and avalanche if no
4582 hash function is specified.
Bhaskar98634f02013-10-29 23:30:51 -04004583
4584 <function> is the hash function to be used :
4585
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03004586 sdbm this function was created initially for sdbm (a public-domain
Bhaskar98634f02013-10-29 23:30:51 -04004587 reimplementation of ndbm) database library. It was found to do
4588 well in scrambling bits, causing better distribution of the keys
4589 and fewer splits. It also happens to be a good general hashing
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004590 function with good distribution, unless the total server weight
4591 is a multiple of 64, in which case applying the avalanche
4592 modifier may help.
Bhaskar98634f02013-10-29 23:30:51 -04004593
4594 djb2 this function was first proposed by Dan Bernstein many years ago
4595 on comp.lang.c. Studies have shown that for certain workload this
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004596 function provides a better distribution than sdbm. It generally
4597 works well with text-based inputs though it can perform extremely
4598 poorly with numeric-only input or when the total server weight is
4599 a multiple of 33, unless the avalanche modifier is also used.
4600
Willy Tarreaua0f42712013-11-14 14:30:35 +01004601 wt6 this function was designed for haproxy while testing other
4602 functions in the past. It is not as smooth as the other ones, but
4603 is much less sensible to the input data set or to the number of
4604 servers. It can make sense as an alternative to sdbm+avalanche or
4605 djb2+avalanche for consistent hashing or when hashing on numeric
4606 data such as a source IP address or a visitor identifier in a URL
4607 parameter.
4608
Willy Tarreau324f07f2015-01-20 19:44:50 +01004609 crc32 this is the most common CRC32 implementation as used in Ethernet,
4610 gzip, PNG, etc. It is slower than the other ones but may provide
4611 a better distribution or less predictable results especially when
4612 used on strings.
4613
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004614 <modifier> indicates an optional method applied after hashing the key :
4615
4616 avalanche This directive indicates that the result from the hash
4617 function above should not be used in its raw form but that
4618 a 4-byte full avalanche hash must be applied first. The
4619 purpose of this step is to mix the resulting bits from the
4620 previous hash in order to avoid any undesired effect when
4621 the input contains some limited values or when the number of
4622 servers is a multiple of one of the hash's components (64
4623 for SDBM, 33 for DJB2). Enabling avalanche tends to make the
4624 result less predictable, but it's also not as smooth as when
4625 using the original function. Some testing might be needed
4626 with some workloads. This hash is one of the many proposed
4627 by Bob Jenkins.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004628
Bhaskar98634f02013-10-29 23:30:51 -04004629 The default hash type is "map-based" and is recommended for most usages. The
4630 default function is "sdbm", the selection of a function should be based on
4631 the range of the values being hashed.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004632
Andrew Rodland17be45e2016-10-25 17:04:12 -04004633 See also : "balance", "hash-balance-factor", "server"
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004634
4635
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01004636http-after-response <action> <options...> [ { if | unless } <condition> ]
4637 Access control for all Layer 7 responses (server, applet/service and internal
4638 ones).
4639
4640 May be used in sections: defaults | frontend | listen | backend
4641 no | yes | yes | yes
4642
4643 The http-after-response statement defines a set of rules which apply to layer
4644 7 processing. The rules are evaluated in their declaration order when they
4645 are met in a frontend, listen or backend section. Any rule may optionally be
4646 followed by an ACL-based condition, in which case it will only be evaluated
4647 if the condition is true. Since these rules apply on responses, the backend
4648 rules are applied first, followed by the frontend's rules.
4649
4650 Unlike http-response rules, these ones are applied on all responses, the
4651 server ones but also to all responses generated by HAProxy. These rules are
4652 evaluated at the end of the responses analysis, before the data forwarding.
4653
4654 The first keyword is the rule's action. The supported actions are described
4655 below.
4656
4657 There is no limit to the number of http-after-response statements per
4658 instance.
4659
4660 Example:
4661 http-after-response set-header Strict-Transport-Security "max-age=31536000"
4662 http-after-response set-header Cache-Control "no-store,no-cache,private"
4663 http-after-response set-header Pragma "no-cache"
4664
4665http-after-response add-header <name> <fmt> [ { if | unless } <condition> ]
4666
4667 This appends an HTTP header field whose name is specified in <name> and whose
4668 value is defined by <fmt> which follows the log-format rules (see Custom Log
4669 Format in section 8.2.4). This may be used to send a cookie to a client for
4670 example, or to pass some internal information.
4671 This rule is not final, so it is possible to add other similar rules.
4672 Note that header addition is performed immediately, so one rule might reuse
4673 the resulting header from a previous rule.
4674
4675http-after-response allow [ { if | unless } <condition> ]
4676
4677 This stops the evaluation of the rules and lets the response pass the check.
4678 No further "http-after-response" rules are evaluated.
4679
4680http-after-response del-header <name> [ { if | unless } <condition> ]
4681
4682 This removes all HTTP header fields whose name is specified in <name>.
4683
4684http-after-response replace-header <name> <regex-match> <replace-fmt>
4685 [ { if | unless } <condition> ]
4686
4687 This works like "http-response replace-header".
4688
4689 Example:
4690 http-after-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2
4691
4692 # applied to:
4693 Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT
4694
4695 # outputs:
4696 Set-Cookie: C=1;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT
4697
4698 # assuming the backend IP is 192.168.1.20.
4699
4700http-after-response replace-value <name> <regex-match> <replace-fmt>
4701 [ { if | unless } <condition> ]
4702
4703 This works like "http-response replace-value".
4704
4705 Example:
4706 http-after-response replace-value Cache-control ^public$ private
4707
4708 # applied to:
4709 Cache-Control: max-age=3600, public
4710
4711 # outputs:
4712 Cache-Control: max-age=3600, private
4713
4714http-after-response set-header <name> <fmt> [ { if | unless } <condition> ]
4715
4716 This does the same as "add-header" except that the header name is first
4717 removed if it existed. This is useful when passing security information to
4718 the server, where the header must not be manipulated by external users.
4719
4720http-after-response set-status <status> [reason <str>]
4721 [ { if | unless } <condition> ]
4722
4723 This replaces the response status code with <status> which must be an integer
4724 between 100 and 999. Optionally, a custom reason text can be provided defined
4725 by <str>, or the default reason for the specified code will be used as a
4726 fallback.
4727
4728 Example:
4729 # return "431 Request Header Fields Too Large"
4730 http-response set-status 431
4731 # return "503 Slow Down", custom reason
4732 http-response set-status 503 reason "Slow Down"
4733
4734http-after-response set-var(<var-name>) <expr> [ { if | unless } <condition> ]
4735
4736 This is used to set the contents of a variable. The variable is declared
4737 inline.
4738
4739 Arguments:
4740 <var-name> The name of the variable starts with an indication about its
4741 scope. The scopes allowed are:
4742 "proc" : the variable is shared with the whole process
4743 "sess" : the variable is shared with the whole session
4744 "txn" : the variable is shared with the transaction
4745 (request and response)
4746 "req" : the variable is shared only during request
4747 processing
4748 "res" : the variable is shared only during response
4749 processing
4750 This prefix is followed by a name. The separator is a '.'.
4751 The name may only contain characters 'a-z', 'A-Z', '0-9', '.'
4752 and '_'.
4753
4754 <expr> Is a standard HAProxy expression formed by a sample-fetch
4755 followed by some converters.
4756
4757 Example:
4758 http-after-response set-var(sess.last_redir) res.hdr(location)
4759
4760http-after-response strict-mode { on | off }
4761
4762 This enables or disables the strict rewriting mode for following rules. It
4763 does not affect rules declared before it and it is only applicable on rules
4764 performing a rewrite on the responses. When the strict mode is enabled, any
4765 rewrite failure triggers an internal error. Otherwise, such errors are
4766 silently ignored. The purpose of the strict rewriting mode is to make some
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05004767 rewrites optional while others must be performed to continue the response
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01004768 processing.
4769
4770 By default, the strict rewriting mode is enabled. Its value is also reset
4771 when a ruleset evaluation ends. So, for instance, if you change the mode on
Daniel Corbett67a82712020-07-06 23:01:19 -04004772 the backend, the default mode is restored when HAProxy starts the frontend
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01004773 rules evaluation.
4774
4775http-after-response unset-var(<var-name>) [ { if | unless } <condition> ]
4776
4777 This is used to unset a variable. See "http-after-response set-var" for
4778 details about <var-name>.
4779
4780 Example:
4781 http-after-response unset-var(sess.last_redir)
4782
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02004783
4784http-check comment <string>
4785 Defines a comment for the following the http-check rule, reported in logs if
4786 it fails.
4787 May be used in sections : defaults | frontend | listen | backend
4788 yes | no | yes | yes
4789
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02004790 Arguments :
4791 <string> is the comment message to add in logs if the following http-check
4792 rule fails.
4793
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02004794 It only works for connect, send and expect rules. It is useful to make
4795 user-friendly error reporting.
4796
Daniel Corbett67a82712020-07-06 23:01:19 -04004797 See also : "option httpchk", "http-check connect", "http-check send" and
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02004798 "http-check expect".
4799
4800
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02004801http-check connect [default] [port <expr>] [addr <ip>] [send-proxy]
4802 [via-socks4] [ssl] [sni <sni>] [alpn <alpn>] [linger]
Christopher Fauletedc6ed92020-04-23 16:27:59 +02004803 [proto <name>] [comment <msg>]
Christopher Faulete5870d82020-04-15 11:32:03 +02004804 Opens a new connection to perform an HTTP health check
4805 May be used in sections : defaults | frontend | listen | backend
4806 yes | no | yes | yes
4807
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02004808 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02004809 comment <msg> defines a message to report if the rule evaluation fails.
4810
Christopher Faulete5870d82020-04-15 11:32:03 +02004811 default Use default options of the server line to do the health
Daniel Corbett67a82712020-07-06 23:01:19 -04004812 checks. The server options are used only if not redefined.
Christopher Faulete5870d82020-04-15 11:32:03 +02004813
4814 port <expr> if not set, check port or server port is used.
4815 It tells HAProxy where to open the connection to.
4816 <port> must be a valid TCP port source integer, from 1 to
4817 65535 or an sample-fetch expression.
4818
4819 addr <ip> defines the IP address to do the health check.
4820
4821 send-proxy send a PROXY protocol string
4822
4823 via-socks4 enables outgoing health checks using upstream socks4 proxy.
4824
4825 ssl opens a ciphered connection
4826
4827 sni <sni> specifies the SNI to use to do health checks over SSL.
4828
4829 alpn <alpn> defines which protocols to advertise with ALPN. The protocol
4830 list consists in a comma-delimited list of protocol names,
4831 for instance: "h2,http/1.1". If it is not set, the server ALPN
4832 is used.
4833
Christopher Fauletedc6ed92020-04-23 16:27:59 +02004834 proto <name> forces the multiplexer's protocol to use for this connection.
4835 It must be an HTTP mux protocol and it must be usable on the
4836 backend side. The list of available protocols is reported in
4837 haproxy -vv.
4838
Christopher Faulete5870d82020-04-15 11:32:03 +02004839 linger cleanly close the connection instead of using a single RST.
4840
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02004841 Just like tcp-check health checks, it is possible to configure the connection
4842 to use to perform HTTP health check. This directive should also be used to
4843 describe a scenario involving several request/response exchanges, possibly on
4844 different ports or with different servers.
4845
4846 When there are no TCP port configured on the server line neither server port
4847 directive, then the first step of the http-check sequence must be to specify
4848 the port with a "http-check connect".
4849
4850 In an http-check ruleset a 'connect' is required, it is also mandatory to start
4851 the ruleset with a 'connect' rule. Purpose is to ensure admin know what they
4852 do.
4853
4854 When a connect must start the ruleset, if may still be preceded by set-var,
4855 unset-var or comment rules.
4856
4857 Examples :
Christopher Faulete5870d82020-04-15 11:32:03 +02004858 # check HTTP and HTTPs services on a server.
4859 # first open port 80 thanks to server line port directive, then
4860 # tcp-check opens port 443, ciphered and run a request on it:
4861 option httpchk
4862
4863 http-check connect
Christopher Fauleta5c14ef2020-04-29 14:19:13 +02004864 http-check send meth GET uri / ver HTTP/1.1 hdr host haproxy.1wt.eu
Christopher Faulet8021a5f2020-04-24 13:53:12 +02004865 http-check expect status 200-399
Christopher Faulete5870d82020-04-15 11:32:03 +02004866 http-check connect port 443 ssl sni haproxy.1wt.eu
Christopher Fauleta5c14ef2020-04-29 14:19:13 +02004867 http-check send meth GET uri / ver HTTP/1.1 hdr host haproxy.1wt.eu
Christopher Faulet8021a5f2020-04-24 13:53:12 +02004868 http-check expect status 200-399
Christopher Faulete5870d82020-04-15 11:32:03 +02004869
4870 server www 10.0.0.1 check port 80
4871
4872 See also : "option httpchk", "http-check send", "http-check expect"
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01004873
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02004874
Willy Tarreau0ba27502007-12-24 16:55:16 +01004875http-check disable-on-404
4876 Enable a maintenance mode upon HTTP/404 response to health-checks
4877 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau2769aa02007-12-27 18:26:09 +01004878 yes | no | yes | yes
Willy Tarreau0ba27502007-12-24 16:55:16 +01004879 Arguments : none
4880
4881 When this option is set, a server which returns an HTTP code 404 will be
4882 excluded from further load-balancing, but will still receive persistent
4883 connections. This provides a very convenient method for Web administrators
4884 to perform a graceful shutdown of their servers. It is also important to note
4885 that a server which is detected as failed while it was in this mode will not
4886 generate an alert, just a notice. If the server responds 2xx or 3xx again, it
4887 will immediately be reinserted into the farm. The status on the stats page
4888 reports "NOLB" for a server in this mode. It is important to note that this
Willy Tarreaubd741542010-03-16 18:46:54 +01004889 option only works in conjunction with the "httpchk" option. If this option
4890 is used with "http-check expect", then it has precedence over it so that 404
4891 responses will still be considered as soft-stop.
4892
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02004893 See also : "option httpchk" and "http-check expect".
Willy Tarreaubd741542010-03-16 18:46:54 +01004894
4895
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02004896http-check expect [min-recv <int>] [comment <msg>]
Christopher Faulete5870d82020-04-15 11:32:03 +02004897 [ok-status <st>] [error-status <st>] [tout-status <st>]
4898 [on-success <fmt>] [on-error <fmt>] [status-code <expr>]
4899 [!] <match> <pattern>
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004900 Make HTTP health checks consider response contents or specific status codes
Willy Tarreaubd741542010-03-16 18:46:54 +01004901 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau1ee51a62011-08-19 20:04:17 +02004902 yes | no | yes | yes
Christopher Faulete5870d82020-04-15 11:32:03 +02004903
Willy Tarreaubd741542010-03-16 18:46:54 +01004904 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02004905 comment <msg> defines a message to report if the rule evaluation fails.
4906
Christopher Faulete5870d82020-04-15 11:32:03 +02004907 min-recv is optional and can define the minimum amount of data required to
4908 evaluate the current expect rule. If the number of received bytes
4909 is under this limit, the check will wait for more data. This
4910 option can be used to resolve some ambiguous matching rules or to
4911 avoid executing costly regex matches on content known to be still
4912 incomplete. If an exact string is used, the minimum between the
4913 string length and this parameter is used. This parameter is
4914 ignored if it is set to -1. If the expect rule does not match,
4915 the check will wait for more data. If set to 0, the evaluation
4916 result is always conclusive.
4917
4918 ok-status <st> is optional and can be used to set the check status if
4919 the expect rule is successfully evaluated and if it is
4920 the last rule in the tcp-check ruleset. "L7OK", "L7OKC",
Christopher Fauletd888f0f2020-05-07 07:40:17 +02004921 "L6OK" and "L4OK" are supported :
4922 - L7OK : check passed on layer 7
4923 - L7OKC : check conditionally passed on layer 7, for
4924 example 404 with disable-on-404
4925 - L6OK : check passed on layer 6
4926 - L4OK : check passed on layer 4
4927 By default "L7OK" is used.
Christopher Faulete5870d82020-04-15 11:32:03 +02004928
4929 error-status <st> is optional and can be used to set the check status if
4930 an error occurred during the expect rule evaluation.
Christopher Fauletd888f0f2020-05-07 07:40:17 +02004931 "L7RSP", "L7STS", "L6RSP" and "L4CON" are supported :
4932 - L7RSP : layer 7 invalid response - protocol error
4933 - L7STS : layer 7 response error, for example HTTP 5xx
4934 - L6RSP : layer 6 invalid response - protocol error
4935 - L4CON : layer 1-4 connection problem
4936 By default "L7RSP" is used.
Christopher Faulete5870d82020-04-15 11:32:03 +02004937
4938 tout-status <st> is optional and can be used to set the check status if
4939 a timeout occurred during the expect rule evaluation.
Christopher Fauletd888f0f2020-05-07 07:40:17 +02004940 "L7TOUT", "L6TOUT", and "L4TOUT" are supported :
4941 - L7TOUT : layer 7 (HTTP/SMTP) timeout
4942 - L6TOUT : layer 6 (SSL) timeout
4943 - L4TOUT : layer 1-4 timeout
Christopher Faulete5870d82020-04-15 11:32:03 +02004944 By default "L7TOUT" is used.
4945
4946 on-success <fmt> is optional and can be used to customize the
4947 informational message reported in logs if the expect
4948 rule is successfully evaluated and if it is the last rule
4949 in the tcp-check ruleset. <fmt> is a log-format string.
4950
4951 on-error <fmt> is optional and can be used to customize the
4952 informational message reported in logs if an error
4953 occurred during the expect rule evaluation. <fmt> is a
4954 log-format string.
4955
Willy Tarreaubd741542010-03-16 18:46:54 +01004956 <match> is a keyword indicating how to look for a specific pattern in the
Christopher Fauletb5594262020-05-05 20:23:13 +02004957 response. The keyword may be one of "status", "rstatus", "hdr",
4958 "fhdr", "string", or "rstring". The keyword may be preceded by an
Willy Tarreaubd741542010-03-16 18:46:54 +01004959 exclamation mark ("!") to negate the match. Spaces are allowed
4960 between the exclamation mark and the keyword. See below for more
4961 details on the supported keywords.
4962
Christopher Faulet39708192020-05-05 10:47:36 +02004963 <pattern> is the pattern to look for. It may be a string, a regular
4964 expression or a more complex pattern with several arguments. If
4965 the string pattern contains spaces, they must be escaped with the
4966 usual backslash ('\').
Willy Tarreaubd741542010-03-16 18:46:54 +01004967
4968 By default, "option httpchk" considers that response statuses 2xx and 3xx
4969 are valid, and that others are invalid. When "http-check expect" is used,
4970 it defines what is considered valid or invalid. Only one "http-check"
4971 statement is supported in a backend. If a server fails to respond or times
4972 out, the check obviously fails. The available matches are :
4973
Christopher Faulet8021a5f2020-04-24 13:53:12 +02004974 status <codes> : test the status codes found parsing <codes> string. it
4975 must be a comma-separated list of status codes or range
4976 codes. A health check response will be considered as
4977 valid if the response's status code matches any status
4978 code or is inside any range of the list. If the "status"
4979 keyword is prefixed with "!", then the response will be
4980 considered invalid if the status code matches.
Willy Tarreaubd741542010-03-16 18:46:54 +01004981
4982 rstatus <regex> : test a regular expression for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004983 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01004984 response's status code matches the expression. If the
4985 "rstatus" keyword is prefixed with "!", then the response
4986 will be considered invalid if the status code matches.
4987 This is mostly used to check for multiple codes.
4988
Christopher Fauletb5594262020-05-05 20:23:13 +02004989 hdr { name | name-lf } [ -m <meth> ] <name>
4990 [ { value | value-lf } [ -m <meth> ] <value> :
Christopher Faulet39708192020-05-05 10:47:36 +02004991 test the specified header pattern on the HTTP response
4992 headers. The name pattern is mandatory but the value
4993 pattern is optional. If not specified, only the header
4994 presence is verified. <meth> is the matching method,
4995 applied on the header name or the header value. Supported
4996 matching methods are "str" (exact match), "beg" (prefix
4997 match), "end" (suffix match), "sub" (substring match) or
4998 "reg" (regex match). If not specified, exact matching
Christopher Fauletb5594262020-05-05 20:23:13 +02004999 method is used. If the "name-lf" parameter is used,
5000 <name> is evaluated as a log-format string. If "value-lf"
5001 parameter is used, <value> is evaluated as a log-format
5002 string. These parameters cannot be used with the regex
5003 matching method. Finally, the header value is considered
5004 as comma-separated list. Note that matchings are case
5005 insensitive on the header names.
5006
5007 fhdr { name | name-lf } [ -m <meth> ] <name>
5008 [ { value | value-lf } [ -m <meth> ] <value> :
5009 test the specified full header pattern on the HTTP
5010 response headers. It does exactly the same than "hdr"
5011 keyword, except the full header value is tested, commas
5012 are not considered as delimiters.
Christopher Faulet39708192020-05-05 10:47:36 +02005013
Willy Tarreaubd741542010-03-16 18:46:54 +01005014 string <string> : test the exact string match in the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04005015 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01005016 response's body contains this exact string. If the
5017 "string" keyword is prefixed with "!", then the response
5018 will be considered invalid if the body contains this
5019 string. This can be used to look for a mandatory word at
5020 the end of a dynamic page, or to detect a failure when a
Davor Ocelice9ed2812017-12-25 17:49:28 +01005021 specific error appears on the check page (e.g. a stack
Willy Tarreaubd741542010-03-16 18:46:54 +01005022 trace).
5023
5024 rstring <regex> : test a regular expression on the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04005025 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01005026 response's body matches this expression. If the "rstring"
5027 keyword is prefixed with "!", then the response will be
5028 considered invalid if the body matches the expression.
5029 This can be used to look for a mandatory word at the end
5030 of a dynamic page, or to detect a failure when a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01005031 error appears on the check page (e.g. a stack trace).
Willy Tarreaubd741542010-03-16 18:46:54 +01005032
Christopher Fauletaaab0832020-05-05 15:54:22 +02005033 string-lf <fmt> : test a log-format string match in the HTTP response body.
5034 A health check response will be considered valid if the
5035 response's body contains the string resulting of the
5036 evaluation of <fmt>, which follows the log-format rules.
5037 If prefixed with "!", then the response will be
5038 considered invalid if the body contains the string.
5039
Willy Tarreaubd741542010-03-16 18:46:54 +01005040 It is important to note that the responses will be limited to a certain size
5041 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
5042 Thus, too large responses may not contain the mandatory pattern when using
5043 "string" or "rstring". If a large response is absolutely required, it is
5044 possible to change the default max size by setting the global variable.
5045 However, it is worth keeping in mind that parsing very large responses can
5046 waste some CPU cycles, especially when regular expressions are used, and that
5047 it is always better to focus the checks on smaller resources.
5048
Christopher Faulete5870d82020-04-15 11:32:03 +02005049 In an http-check ruleset, the last expect rule may be implicit. If no expect
5050 rule is specified after the last "http-check send", an implicit expect rule
5051 is defined to match on 2xx or 3xx status codes. It means this rule is also
5052 defined if there is no "http-check" rule at all, when only "option httpchk"
5053 is set.
Cyril Bonté32602d22015-01-30 00:07:07 +01005054
Willy Tarreaubd741542010-03-16 18:46:54 +01005055 Last, if "http-check expect" is combined with "http-check disable-on-404",
5056 then this last one has precedence when the server responds with 404.
5057
5058 Examples :
5059 # only accept status 200 as valid
Christopher Faulet8021a5f2020-04-24 13:53:12 +02005060 http-check expect status 200,201,300-310
Willy Tarreaubd741542010-03-16 18:46:54 +01005061
Christopher Faulet39708192020-05-05 10:47:36 +02005062 # be sure a sessid coookie is set
5063 http-check expect header name "set-cookie" value -m beg "sessid="
5064
Willy Tarreaubd741542010-03-16 18:46:54 +01005065 # consider SQL errors as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01005066 http-check expect ! string SQL\ Error
Willy Tarreaubd741542010-03-16 18:46:54 +01005067
5068 # consider status 5xx only as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01005069 http-check expect ! rstatus ^5
Willy Tarreaubd741542010-03-16 18:46:54 +01005070
5071 # check that we have a correct hexadecimal tag before /html
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03005072 http-check expect rstring <!--tag:[0-9a-f]*--></html>
Willy Tarreau0ba27502007-12-24 16:55:16 +01005073
Christopher Faulete5870d82020-04-15 11:32:03 +02005074 See also : "option httpchk", "http-check connect", "http-check disable-on-404"
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005075 and "http-check send".
Willy Tarreau2769aa02007-12-27 18:26:09 +01005076
5077
Christopher Faulet7c95f5f2020-05-06 15:06:34 +02005078http-check send [meth <method>] [{ uri <uri> | uri-lf <fmt> }>] [ver <version>]
Christopher Faulet574e7bd2020-05-06 15:38:58 +02005079 [hdr <name> <fmt>]* [{ body <string> | body-lf <fmt> }]
5080 [comment <msg>]
Christopher Faulet8acb1282020-04-09 08:44:06 +02005081 Add a possible list of headers and/or a body to the request sent during HTTP
5082 health checks.
5083 May be used in sections : defaults | frontend | listen | backend
5084 yes | no | yes | yes
5085 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005086 comment <msg> defines a message to report if the rule evaluation fails.
5087
Christopher Faulete5870d82020-04-15 11:32:03 +02005088 meth <method> is the optional HTTP method used with the requests. When not
5089 set, the "OPTIONS" method is used, as it generally requires
5090 low server processing and is easy to filter out from the
5091 logs. Any method may be used, though it is not recommended
5092 to invent non-standard ones.
5093
Christopher Faulet7c95f5f2020-05-06 15:06:34 +02005094 uri <uri> is optional and set the URI referenced in the HTTP requests
5095 to the string <uri>. It defaults to "/" which is accessible
5096 by default on almost any server, but may be changed to any
5097 other URI. Query strings are permitted.
5098
5099 uri-lf <fmt> is optional and set the URI referenced in the HTTP requests
5100 using the log-format string <fmt>. It defaults to "/" which
5101 is accessible by default on almost any server, but may be
5102 changed to any other URI. Query strings are permitted.
Christopher Faulet8acb1282020-04-09 08:44:06 +02005103
Christopher Faulet907701b2020-04-28 09:37:00 +02005104 ver <version> is the optional HTTP version string. It defaults to
Christopher Faulete5870d82020-04-15 11:32:03 +02005105 "HTTP/1.0" but some servers might behave incorrectly in HTTP
Daniel Corbett67a82712020-07-06 23:01:19 -04005106 1.0, so turning it to HTTP/1.1 may sometimes help. Note that
Christopher Faulete5870d82020-04-15 11:32:03 +02005107 the Host field is mandatory in HTTP/1.1, use "hdr" argument
5108 to add it.
5109
5110 hdr <name> <fmt> adds the HTTP header field whose name is specified in
5111 <name> and whose value is defined by <fmt>, which follows
5112 to the log-format rules.
5113
5114 body <string> add the body defined by <string> to the request sent during
5115 HTTP health checks. If defined, the "Content-Length" header
5116 is thus automatically added to the request.
Christopher Faulet8acb1282020-04-09 08:44:06 +02005117
Christopher Faulet574e7bd2020-05-06 15:38:58 +02005118 body-lf <fmt> add the body defined by the log-format string <fmt> to the
5119 request sent during HTTP health checks. If defined, the
5120 "Content-Length" header is thus automatically added to the
5121 request.
5122
Christopher Faulet8acb1282020-04-09 08:44:06 +02005123 In addition to the request line defined by the "option httpchk" directive,
5124 this one is the valid way to add some headers and optionally a body to the
5125 request sent during HTTP health checks. If a body is defined, the associate
Christopher Faulet9df910c2020-04-29 14:20:47 +02005126 "Content-Length" header is automatically added. Thus, this header or
5127 "Transfer-encoding" header should not be present in the request provided by
5128 "http-check send". If so, it will be ignored. The old trick consisting to add
5129 headers after the version string on the "option httpchk" line is now
Christopher Faulet8acb1282020-04-09 08:44:06 +02005130 deprecated. Note also the "Connection: close" header is still added if a
Daniel Corbett67a82712020-07-06 23:01:19 -04005131 "http-check expect" directive is defined independently of this directive, just
Christopher Faulet8acb1282020-04-09 08:44:06 +02005132 like the state header if the directive "http-check send-state" is defined.
5133
Christopher Faulete5870d82020-04-15 11:32:03 +02005134 Also "http-check send" doesn't support HTTP keep-alive. Keep in mind that it
5135 will automatically append a "Connection: close" header, meaning that this
Christopher Faulet9df910c2020-04-29 14:20:47 +02005136 header should not be present in the request provided by "http-check send". If
5137 so, it will be ignored.
5138
5139 Note that the Host header and the request authority, when both defined, are
5140 automatically synchronized. It means when the HTTP request is sent, when a
5141 Host is inserted in the request, the request authority is accordingly
5142 updated. Thus, don't be surprised if the Host header value overwrites the
5143 configured request authority.
5144
5145 Note also for now, no Host header is automatically added in HTTP/1.1 or above
5146 requests. You should add it explicitly.
Christopher Faulete5870d82020-04-15 11:32:03 +02005147
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005148 See also : "option httpchk", "http-check send-state" and "http-check expect".
Christopher Faulet8acb1282020-04-09 08:44:06 +02005149
5150
Willy Tarreauef781042010-01-27 11:53:01 +01005151http-check send-state
5152 Enable emission of a state header with HTTP health checks
5153 May be used in sections : defaults | frontend | listen | backend
5154 yes | no | yes | yes
5155 Arguments : none
5156
5157 When this option is set, haproxy will systematically send a special header
5158 "X-Haproxy-Server-State" with a list of parameters indicating to each server
5159 how they are seen by haproxy. This can be used for instance when a server is
5160 manipulated without access to haproxy and the operator needs to know whether
5161 haproxy still sees it up or not, or if the server is the last one in a farm.
5162
5163 The header is composed of fields delimited by semi-colons, the first of which
5164 is a word ("UP", "DOWN", "NOLB"), possibly followed by a number of valid
5165 checks on the total number before transition, just as appears in the stats
5166 interface. Next headers are in the form "<variable>=<value>", indicating in
5167 no specific order some values available in the stats interface :
Joseph Lynch514061c2015-01-15 17:52:59 -08005168 - a variable "address", containing the address of the backend server.
5169 This corresponds to the <address> field in the server declaration. For
5170 unix domain sockets, it will read "unix".
5171
5172 - a variable "port", containing the port of the backend server. This
5173 corresponds to the <port> field in the server declaration. For unix
5174 domain sockets, it will read "unix".
5175
Willy Tarreauef781042010-01-27 11:53:01 +01005176 - a variable "name", containing the name of the backend followed by a slash
5177 ("/") then the name of the server. This can be used when a server is
5178 checked in multiple backends.
5179
5180 - a variable "node" containing the name of the haproxy node, as set in the
5181 global "node" variable, otherwise the system's hostname if unspecified.
5182
5183 - a variable "weight" indicating the weight of the server, a slash ("/")
5184 and the total weight of the farm (just counting usable servers). This
5185 helps to know if other servers are available to handle the load when this
5186 one fails.
5187
5188 - a variable "scur" indicating the current number of concurrent connections
5189 on the server, followed by a slash ("/") then the total number of
5190 connections on all servers of the same backend.
5191
5192 - a variable "qcur" indicating the current number of requests in the
5193 server's queue.
5194
5195 Example of a header received by the application server :
5196 >>> X-Haproxy-Server-State: UP 2/3; name=bck/srv2; node=lb1; weight=1/2; \
5197 scur=13/22; qcur=0
5198
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005199 See also : "option httpchk", "http-check disable-on-404" and
5200 "http-check send".
Willy Tarreauef781042010-01-27 11:53:01 +01005201
Christopher Faulete5870d82020-04-15 11:32:03 +02005202
5203http-check set-var(<var-name>) <expr>
Christopher Faulete5870d82020-04-15 11:32:03 +02005204 This operation sets the content of a variable. The variable is declared inline.
Christopher Faulete5870d82020-04-15 11:32:03 +02005205 May be used in sections: defaults | frontend | listen | backend
5206 yes | no | yes | yes
5207
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005208 Arguments :
Christopher Faulete5870d82020-04-15 11:32:03 +02005209 <var-name> The name of the variable starts with an indication about its
5210 scope. The scopes allowed for http-check are:
5211 "proc" : the variable is shared with the whole process.
5212 "sess" : the variable is shared with the tcp-check session.
5213 "check": the variable is declared for the lifetime of the tcp-check.
5214 This prefix is followed by a name. The separator is a '.'.
5215 The name may only contain characters 'a-z', 'A-Z', '0-9', '.',
5216 and '-'.
5217
5218 <expr> Is a sample-fetch expression potentially followed by converters.
5219
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005220 Examples :
5221 http-check set-var(check.port) int(1234)
Christopher Faulete5870d82020-04-15 11:32:03 +02005222
5223
5224http-check unset-var(<var-name>)
Christopher Faulete5870d82020-04-15 11:32:03 +02005225 Free a reference to a variable within its scope.
Christopher Faulete5870d82020-04-15 11:32:03 +02005226 May be used in sections: defaults | frontend | listen | backend
5227 yes | no | yes | yes
5228
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005229 Arguments :
Christopher Faulete5870d82020-04-15 11:32:03 +02005230 <var-name> The name of the variable starts with an indication about its
5231 scope. The scopes allowed for http-check are:
5232 "proc" : the variable is shared with the whole process.
5233 "sess" : the variable is shared with the tcp-check session.
5234 "check": the variable is declared for the lifetime of the tcp-check.
5235 This prefix is followed by a name. The separator is a '.'.
5236 The name may only contain characters 'a-z', 'A-Z', '0-9', '.',
5237 and '-'.
5238
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005239 Examples :
5240 http-check unset-var(check.port)
Christopher Faulete5870d82020-04-15 11:32:03 +02005241
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005242
Christopher Faulet3b967c12020-05-15 15:47:44 +02005243http-error status <code> [content-type <type>]
5244 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
5245 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
5246 [ hdr <name> <fmt> ]*
5247 Defines a custom error message to use instead of errors generated by HAProxy.
5248 May be used in sections : defaults | frontend | listen | backend
5249 yes | yes | yes | yes
5250 Arguments :
Ilya Shipitsin11057a32020-06-21 21:18:27 +05005251 status <code> is the HTTP status code. It must be specified.
Christopher Faulet3b967c12020-05-15 15:47:44 +02005252 Currently, HAProxy is capable of generating codes
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02005253 200, 400, 401, 403, 404, 405, 407, 408, 410, 413, 425,
5254 429, 500, 502, 503, and 504.
Christopher Faulet3b967c12020-05-15 15:47:44 +02005255
5256 content-type <type> is the response content type, for instance
5257 "text/plain". This parameter is ignored and should be
5258 omitted when an errorfile is configured or when the
5259 payload is empty. Otherwise, it must be defined.
5260
5261 default-errorfiles Reset the previously defined error message for current
5262 proxy for the status <code>. If used on a backend, the
5263 frontend error message is used, if defined. If used on
5264 a frontend, the default error message is used.
5265
5266 errorfile <file> designates a file containing the full HTTP response.
5267 It is recommended to follow the common practice of
5268 appending ".http" to the filename so that people do
5269 not confuse the response with HTML error pages, and to
5270 use absolute paths, since files are read before any
5271 chroot is performed.
5272
5273 errorfiles <name> designates the http-errors section to use to import
5274 the error message with the status code <code>. If no
5275 such message is found, the proxy's error messages are
5276 considered.
5277
5278 file <file> specifies the file to use as response payload. If the
5279 file is not empty, its content-type must be set as
5280 argument to "content-type", otherwise, any
5281 "content-type" argument is ignored. <file> is
5282 considered as a raw string.
5283
5284 string <str> specifies the raw string to use as response payload.
5285 The content-type must always be set as argument to
5286 "content-type".
5287
5288 lf-file <file> specifies the file to use as response payload. If the
5289 file is not empty, its content-type must be set as
5290 argument to "content-type", otherwise, any
5291 "content-type" argument is ignored. <file> is
5292 evaluated as a log-format string.
5293
5294 lf-string <str> specifies the log-format string to use as response
5295 payload. The content-type must always be set as
5296 argument to "content-type".
5297
5298 hdr <name> <fmt> adds to the response the HTTP header field whose name
5299 is specified in <name> and whose value is defined by
5300 <fmt>, which follows to the log-format rules.
5301 This parameter is ignored if an errorfile is used.
5302
5303 This directive may be used instead of "errorfile", to define a custom error
5304 message. As "errorfile" directive, it is used for errors detected and
5305 returned by HAProxy. If an errorfile is defined, it is parsed when HAProxy
5306 starts and must be valid according to the HTTP standards. The generated
5307 response must not exceed the configured buffer size (BUFFSIZE), otherwise an
5308 internal error will be returned. Finally, if you consider to use some
5309 http-after-response rules to rewrite these errors, the reserved buffer space
5310 should be available (see "tune.maxrewrite").
5311
5312 The files are read at the same time as the configuration and kept in memory.
5313 For this reason, the errors continue to be returned even when the process is
5314 chrooted, and no file change is considered while the process is running.
5315
5316 See also : "errorfile", "errorfiles", "errorloc", "errorloc302",
5317 "errorloc303" and section 3.8 about http-errors.
5318
5319
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005320http-request <action> [options...] [ { if | unless } <condition> ]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01005321 Access control for Layer 7 requests
5322
5323 May be used in sections: defaults | frontend | listen | backend
5324 no | yes | yes | yes
5325
Willy Tarreau20b0de52012-12-24 15:45:22 +01005326 The http-request statement defines a set of rules which apply to layer 7
5327 processing. The rules are evaluated in their declaration order when they are
5328 met in a frontend, listen or backend section. Any rule may optionally be
5329 followed by an ACL-based condition, in which case it will only be evaluated
5330 if the condition is true.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01005331
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005332 The first keyword is the rule's action. The supported actions are described
5333 below.
Willy Tarreau20b0de52012-12-24 15:45:22 +01005334
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005335 There is no limit to the number of http-request statements per instance.
Willy Tarreau20b0de52012-12-24 15:45:22 +01005336
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005337 Example:
5338 acl nagios src 192.168.129.3
5339 acl local_net src 192.168.0.0/16
5340 acl auth_ok http_auth(L1)
Willy Tarreau20b0de52012-12-24 15:45:22 +01005341
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005342 http-request allow if nagios
5343 http-request allow if local_net auth_ok
5344 http-request auth realm Gimme if local_net auth_ok
5345 http-request deny
Willy Tarreau81499eb2012-12-27 12:19:02 +01005346
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005347 Example:
5348 acl key req.hdr(X-Add-Acl-Key) -m found
5349 acl add path /addacl
5350 acl del path /delacl
Willy Tarreau20b0de52012-12-24 15:45:22 +01005351
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005352 acl myhost hdr(Host) -f myhost.lst
Willy Tarreau20b0de52012-12-24 15:45:22 +01005353
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005354 http-request add-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key add
5355 http-request del-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key del
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02005356
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005357 Example:
5358 acl value req.hdr(X-Value) -m found
5359 acl setmap path /setmap
5360 acl delmap path /delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06005361
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005362 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06005363
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005364 http-request set-map(map.lst) %[src] %[req.hdr(X-Value)] if setmap value
5365 http-request del-map(map.lst) %[src] if delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06005366
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005367 See also : "stats http-request", section 3.4 about userlists and section 7
5368 about ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06005369
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005370http-request add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005371
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005372 This is used to add a new entry into an ACL. The ACL must be loaded from a
5373 file (even a dummy empty file). The file name of the ACL to be updated is
5374 passed between parentheses. It takes one argument: <key fmt>, which follows
5375 log-format rules, to collect content of the new entry. It performs a lookup
5376 in the ACL before insertion, to avoid duplicated (or more) values. This
5377 lookup is done by a linear search and can be expensive with large lists!
5378 It is the equivalent of the "add acl" command from the stats socket, but can
5379 be triggered by an HTTP request.
Sasha Pachev218f0642014-06-16 12:05:59 -06005380
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005381http-request add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005382
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005383 This appends an HTTP header field whose name is specified in <name> and
5384 whose value is defined by <fmt> which follows the log-format rules (see
5385 Custom Log Format in section 8.2.4). This is particularly useful to pass
5386 connection-specific information to the server (e.g. the client's SSL
5387 certificate), or to combine several headers into one. This rule is not
5388 final, so it is possible to add other similar rules. Note that header
5389 addition is performed immediately, so one rule might reuse the resulting
5390 header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06005391
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005392http-request allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005393
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005394 This stops the evaluation of the rules and lets the request pass the check.
5395 No further "http-request" rules are evaluated.
Sasha Pachev218f0642014-06-16 12:05:59 -06005396
Sasha Pachev218f0642014-06-16 12:05:59 -06005397
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005398http-request auth [realm <realm>] [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005399
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005400 This stops the evaluation of the rules and immediately responds with an
5401 HTTP 401 or 407 error code to invite the user to present a valid user name
5402 and password. No further "http-request" rules are evaluated. An optional
5403 "realm" parameter is supported, it sets the authentication realm that is
5404 returned with the response (typically the application's name).
Sasha Pachev218f0642014-06-16 12:05:59 -06005405
Christopher Faulet612f2ea2020-05-27 09:57:28 +02005406 The corresponding proxy's error message is used. It may be customized using
5407 an "errorfile" or an "http-error" directive. For 401 responses, all
5408 occurrences of the WWW-Authenticate header are removed and replaced by a new
5409 one with a basic authentication challenge for realm "<realm>". For 407
5410 responses, the same is done on the Proxy-Authenticate header. If the error
5411 message must not be altered, consider to use "http-request return" rule
5412 instead.
5413
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005414 Example:
5415 acl auth_ok http_auth_group(L1) G1
5416 http-request auth unless auth_ok
Sasha Pachev218f0642014-06-16 12:05:59 -06005417
Jarno Huuskonen251a6b72019-01-04 14:05:02 +02005418http-request cache-use <name> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005419
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02005420 See section 6.2 about cache setup.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005421
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005422http-request capture <sample> [ len <length> | id <id> ]
5423 [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005424
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005425 This captures sample expression <sample> from the request buffer, and
5426 converts it to a string of at most <len> characters. The resulting string is
5427 stored into the next request "capture" slot, so it will possibly appear next
5428 to some captured HTTP headers. It will then automatically appear in the logs,
5429 and it will be possible to extract it using sample fetch rules to feed it
5430 into headers or anything. The length should be limited given that this size
5431 will be allocated for each capture during the whole session life.
5432 Please check section 7.3 (Fetching samples) and "capture request header" for
5433 more information.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005434
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005435 If the keyword "id" is used instead of "len", the action tries to store the
5436 captured string in a previously declared capture slot. This is useful to run
5437 captures in backends. The slot id can be declared by a previous directive
Baptiste Assmann19a69b32020-01-16 14:34:22 +01005438 "http-request capture" or with the "declare capture" keyword.
5439
5440 When using this action in a backend, double check that the relevant
5441 frontend(s) have the required capture slots otherwise, this rule will be
5442 ignored at run time. This can't be detected at configuration parsing time
5443 due to HAProxy's ability to dynamically resolve backend name at runtime.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005444
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005445http-request del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005446
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005447 This is used to delete an entry from an ACL. The ACL must be loaded from a
5448 file (even a dummy empty file). The file name of the ACL to be updated is
5449 passed between parentheses. It takes one argument: <key fmt>, which follows
5450 log-format rules, to collect content of the entry to delete.
5451 It is the equivalent of the "del acl" command from the stats socket, but can
5452 be triggered by an HTTP request.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005453
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005454http-request del-header <name> [ { if | unless } <condition> ]
Willy Tarreauf4c43c12013-06-11 17:01:13 +02005455
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005456 This removes all HTTP header fields whose name is specified in <name>.
Willy Tarreau9a355ec2013-06-11 17:45:46 +02005457
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005458http-request del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau42cf39e2013-06-11 18:51:32 +02005459
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005460 This is used to delete an entry from a MAP. The MAP must be loaded from a
5461 file (even a dummy empty file). The file name of the MAP to be updated is
5462 passed between parentheses. It takes one argument: <key fmt>, which follows
5463 log-format rules, to collect content of the entry to delete.
5464 It takes one argument: "file name" It is the equivalent of the "del map"
5465 command from the stats socket, but can be triggered by an HTTP request.
Willy Tarreau51347ed2013-06-11 19:34:13 +02005466
Christopher Faulet5cb513a2020-05-13 17:56:56 +02005467http-request deny [deny_status <status>] [ { if | unless } <condition> ]
5468http-request deny [ { status | deny_status } <code>] [content-type <type>]
5469 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
5470 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
5471 [ hdr <name> <fmt> ]*
5472 [ { if | unless } <condition> ]
Patrick Hemmer268a7072018-05-11 12:52:31 -04005473
Christopher Faulet5cb513a2020-05-13 17:56:56 +02005474 This stops the evaluation of the rules and immediately rejects the request.
5475 By default an HTTP 403 error is returned. But the response may be customized
5476 using same syntax than "http-request return" rules. Thus, see "http-request
Ilya Shipitsin11057a32020-06-21 21:18:27 +05005477 return" for details. For compatibility purpose, when no argument is defined,
Christopher Faulet5cb513a2020-05-13 17:56:56 +02005478 or only "deny_status", the argument "default-errorfiles" is implied. It means
5479 "http-request deny [deny_status <status>]" is an alias of
5480 "http-request deny [status <status>] default-errorfiles".
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005481 No further "http-request" rules are evaluated.
Christopher Faulet5cb513a2020-05-13 17:56:56 +02005482 See also "http-request return".
Patrick Hemmer268a7072018-05-11 12:52:31 -04005483
Olivier Houchard602bf7d2019-05-10 13:59:15 +02005484http-request disable-l7-retry [ { if | unless } <condition> ]
5485 This disables any attempt to retry the request if it fails for any other
5486 reason than a connection failure. This can be useful for example to make
5487 sure POST requests aren't retried on failure.
5488
Baptiste Assmann333939c2019-01-21 08:34:50 +01005489http-request do-resolve(<var>,<resolvers>,[ipv4,ipv6]) <expr> :
5490
5491 This action performs a DNS resolution of the output of <expr> and stores
5492 the result in the variable <var>. It uses the DNS resolvers section
5493 pointed by <resolvers>.
5494 It is possible to choose a resolution preference using the optional
5495 arguments 'ipv4' or 'ipv6'.
5496 When performing the DNS resolution, the client side connection is on
5497 pause waiting till the end of the resolution.
5498 If an IP address can be found, it is stored into <var>. If any kind of
5499 error occurs, then <var> is not set.
5500 One can use this action to discover a server IP address at run time and
5501 based on information found in the request (IE a Host header).
5502 If this action is used to find the server's IP address (using the
5503 "set-dst" action), then the server IP address in the backend must be set
5504 to 0.0.0.0.
5505
5506 Example:
5507 resolvers mydns
5508 nameserver local 127.0.0.53:53
5509 nameserver google 8.8.8.8:53
5510 timeout retry 1s
5511 hold valid 10s
5512 hold nx 3s
5513 hold other 3s
5514 hold obsolete 0s
5515 accepted_payload_size 8192
5516
5517 frontend fe
5518 bind 10.42.0.1:80
5519 http-request do-resolve(txn.myip,mydns,ipv4) hdr(Host),lower
5520 http-request capture var(txn.myip) len 40
5521
5522 # return 503 when the variable is not set,
5523 # which mean DNS resolution error
5524 use_backend b_503 unless { var(txn.myip) -m found }
5525
5526 default_backend be
5527
5528 backend b_503
5529 # dummy backend used to return 503.
5530 # one can use the errorfile directive to send a nice
5531 # 503 error page to end users
5532
5533 backend be
5534 # rule to prevent HAProxy from reconnecting to services
5535 # on the local network (forged DNS name used to scan the network)
5536 http-request deny if { var(txn.myip) -m ip 127.0.0.0/8 10.0.0.0/8 }
5537 http-request set-dst var(txn.myip)
5538 server clear 0.0.0.0:0
5539
5540 NOTE: Don't forget to set the "protection" rules to ensure HAProxy won't
5541 be used to scan the network or worst won't loop over itself...
5542
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01005543http-request early-hint <name> <fmt> [ { if | unless } <condition> ]
5544
5545 This is used to build an HTTP 103 Early Hints response prior to any other one.
5546 This appends an HTTP header field to this response whose name is specified in
5547 <name> and whose value is defined by <fmt> which follows the log-format rules
5548 (see Custom Log Format in section 8.2.4). This is particularly useful to pass
Frédéric Lécaille3aac1062018-11-13 09:42:13 +01005549 to the client some Link headers to preload resources required to render the
5550 HTML documents.
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01005551
5552 See RFC 8297 for more information.
5553
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005554http-request redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005555
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005556 This performs an HTTP redirection based on a redirect rule. This is exactly
5557 the same as the "redirect" statement except that it inserts a redirect rule
5558 which can be processed in the middle of other "http-request" rules and that
5559 these rules use the "log-format" strings. See the "redirect" keyword for the
5560 rule's syntax.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005561
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005562http-request reject [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005563
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005564 This stops the evaluation of the rules and immediately closes the connection
5565 without sending any response. It acts similarly to the
5566 "tcp-request content reject" rules. It can be useful to force an immediate
5567 connection closure on HTTP/2 connections.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005568
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005569http-request replace-header <name> <match-regex> <replace-fmt>
5570 [ { if | unless } <condition> ]
Willy Tarreaua9083d02015-05-08 15:27:59 +02005571
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05005572 This matches the value of all occurrences of header field <name> against
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005573 <match-regex>. Matching is performed case-sensitively. Matching values are
5574 completely replaced by <replace-fmt>. Format characters are allowed in
5575 <replace-fmt> and work like <fmt> arguments in "http-request add-header".
5576 Standard back-references using the backslash ('\') followed by a number are
5577 supported.
Thierry FOURNIER82bf70d2015-05-26 17:58:29 +02005578
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005579 This action acts on whole header lines, regardless of the number of values
5580 they may contain. Thus it is well-suited to process headers naturally
5581 containing commas in their value, such as If-Modified-Since. Headers that
5582 contain a comma-separated list of values, such as Accept, should be processed
5583 using "http-request replace-value".
William Lallemand86d0df02017-11-24 21:36:45 +01005584
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005585 Example:
5586 http-request replace-header Cookie foo=([^;]*);(.*) foo=\1;ip=%bi;\2
5587
5588 # applied to:
5589 Cookie: foo=foobar; expires=Tue, 14-Jun-2016 01:40:45 GMT;
5590
5591 # outputs:
5592 Cookie: foo=foobar;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT;
5593
5594 # assuming the backend IP is 192.168.1.20
Willy Tarreau09448f72014-06-25 18:12:15 +02005595
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005596 http-request replace-header User-Agent curl foo
5597
5598 # applied to:
5599 User-Agent: curl/7.47.0
Willy Tarreau09448f72014-06-25 18:12:15 +02005600
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005601 # outputs:
5602 User-Agent: foo
Willy Tarreau09448f72014-06-25 18:12:15 +02005603
Willy Tarreau262c3f12019-12-17 06:52:51 +01005604http-request replace-path <match-regex> <replace-fmt>
5605 [ { if | unless } <condition> ]
5606
5607 This works like "replace-header" except that it works on the request's path
5608 component instead of a header. The path component starts at the first '/'
Christopher Faulet82c83322020-09-02 14:16:59 +02005609 after an optional scheme+authority and ends before the question mark. Thus,
5610 the replacement does not modify the scheme, the authority and the
5611 query-string.
Willy Tarreau262c3f12019-12-17 06:52:51 +01005612
5613 It is worth noting that regular expressions may be more expensive to evaluate
5614 than certain ACLs, so rare replacements may benefit from a condition to avoid
5615 performing the evaluation at all if it does not match.
5616
5617 Example:
5618 # prefix /foo : turn /bar?q=1 into /foo/bar?q=1 :
5619 http-request replace-path (.*) /foo\1
5620
Willy Tarreau262c3f12019-12-17 06:52:51 +01005621 # strip /foo : turn /foo/bar?q=1 into /bar?q=1
5622 http-request replace-path /foo/(.*) /\1
5623 # or more efficient if only some requests match :
5624 http-request replace-path /foo/(.*) /\1 if { url_beg /foo/ }
5625
Christopher Faulet312294f2020-09-02 17:17:44 +02005626http-request replace-pathq <match-regex> <replace-fmt>
5627 [ { if | unless } <condition> ]
5628
5629 This does the same as "http-request replace-path" except that the path
5630 contains the query-string if any is present. Thus, the path and the
5631 query-string are replaced.
5632
5633 Example:
5634 # suffix /foo : turn /bar?q=1 into /bar/foo?q=1 :
5635 http-request replace-pathq ([^?]*)(\?(.*))? \1/foo\2
5636
Willy Tarreau33810222019-06-12 17:44:02 +02005637http-request replace-uri <match-regex> <replace-fmt>
5638 [ { if | unless } <condition> ]
5639
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005640 This works like "replace-header" except that it works on the request's URI part
5641 instead of a header. The URI part may contain an optional scheme, authority or
5642 query string. These are considered to be part of the value that is matched
5643 against.
5644
5645 It is worth noting that regular expressions may be more expensive to evaluate
5646 than certain ACLs, so rare replacements may benefit from a condition to avoid
5647 performing the evaluation at all if it does not match.
Willy Tarreau33810222019-06-12 17:44:02 +02005648
Willy Tarreau62b59132019-12-17 06:51:20 +01005649 IMPORTANT NOTE: historically in HTTP/1.x, the vast majority of requests sent
5650 by browsers use the "origin form", which differs from the "absolute form" in
5651 that they do not contain a scheme nor authority in the URI portion. Mostly
5652 only requests sent to proxies, those forged by hand and some emitted by
5653 certain applications use the absolute form. As such, "replace-uri" usually
5654 works fine most of the time in HTTP/1.x with rules starting with a "/". But
5655 with HTTP/2, clients are encouraged to send absolute URIs only, which look
5656 like the ones HTTP/1 clients use to talk to proxies. Such partial replace-uri
5657 rules may then fail in HTTP/2 when they work in HTTP/1. Either the rules need
Willy Tarreau262c3f12019-12-17 06:52:51 +01005658 to be adapted to optionally match a scheme and authority, or replace-path
5659 should be used.
Willy Tarreau33810222019-06-12 17:44:02 +02005660
Willy Tarreau62b59132019-12-17 06:51:20 +01005661 Example:
5662 # rewrite all "http" absolute requests to "https":
5663 http-request replace-uri ^http://(.*) https://\1
Willy Tarreau33810222019-06-12 17:44:02 +02005664
Willy Tarreau62b59132019-12-17 06:51:20 +01005665 # prefix /foo : turn /bar?q=1 into /foo/bar?q=1 :
5666 http-request replace-uri ([^/:]*://[^/]*)?(.*) \1/foo\2
Willy Tarreau33810222019-06-12 17:44:02 +02005667
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005668http-request replace-value <name> <match-regex> <replace-fmt>
5669 [ { if | unless } <condition> ]
Willy Tarreau09448f72014-06-25 18:12:15 +02005670
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005671 This works like "replace-header" except that it matches the regex against
5672 every comma-delimited value of the header field <name> instead of the
5673 entire header. This is suited for all headers which are allowed to carry
5674 more than one value. An example could be the Accept header.
Willy Tarreau09448f72014-06-25 18:12:15 +02005675
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005676 Example:
5677 http-request replace-value X-Forwarded-For ^192\.168\.(.*)$ 172.16.\1
Thierry FOURNIER236657b2015-08-19 08:25:14 +02005678
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005679 # applied to:
5680 X-Forwarded-For: 192.168.10.1, 192.168.13.24, 10.0.0.37
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02005681
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005682 # outputs:
5683 X-Forwarded-For: 172.16.10.1, 172.16.13.24, 10.0.0.37
Frédéric Lécaille6778b272018-01-29 15:22:53 +01005684
Christopher Faulet24231ab2020-01-24 17:44:23 +01005685http-request return [status <code>] [content-type <type>]
5686 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
5687 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
Christopher Faulet4a2c1422020-01-31 17:36:01 +01005688 [ hdr <name> <fmt> ]*
Christopher Faulet24231ab2020-01-24 17:44:23 +01005689 [ { if | unless } <condition> ]
5690
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05005691 This stops the evaluation of the rules and immediately returns a response. The
Christopher Faulet24231ab2020-01-24 17:44:23 +01005692 default status code used for the response is 200. It can be optionally
5693 specified as an arguments to "status". The response content-type may also be
Daniel Corbett67a82712020-07-06 23:01:19 -04005694 specified as an argument to "content-type". Finally the response itself may
Sébastien Grossab877122020-10-08 10:06:03 +02005695 be defined. It can be a full HTTP response specifying the errorfile to use,
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05005696 or the response payload specifying the file or the string to use. These rules
Christopher Faulet24231ab2020-01-24 17:44:23 +01005697 are followed to create the response :
5698
5699 * If neither the errorfile nor the payload to use is defined, a dummy
5700 response is returned. Only the "status" argument is considered. It can be
5701 any code in the range [200, 599]. The "content-type" argument, if any, is
5702 ignored.
5703
5704 * If "default-errorfiles" argument is set, the proxy's errorfiles are
5705 considered. If the "status" argument is defined, it must be one of the
Daniel Corbett67a82712020-07-06 23:01:19 -04005706 status code handled by haproxy (200, 400, 403, 404, 405, 408, 410, 413,
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02005707 425, 429, 500, 502, 503, and 504). The "content-type" argument, if any,
5708 is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01005709
5710 * If a specific errorfile is defined, with an "errorfile" argument, the
5711 corresponding file, containing a full HTTP response, is returned. Only the
5712 "status" argument is considered. It must be one of the status code handled
Daniel Corbett67a82712020-07-06 23:01:19 -04005713 by haproxy (200, 400, 403, 404, 405, 408, 410, 413, 425, 429, 500, 502, 503,
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02005714 and 504). The "content-type" argument, if any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01005715
5716 * If an http-errors section is defined, with an "errorfiles" argument, the
5717 corresponding file in the specified http-errors section, containing a full
5718 HTTP response, is returned. Only the "status" argument is considered. It
Daniel Corbett67a82712020-07-06 23:01:19 -04005719 must be one of the status code handled by haproxy (200, 400, 403, 404, 405,
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02005720 408, 410, 413, 425, 429, 500, 502, 503, and 504). The "content-type"
5721 argument, if any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01005722
5723 * If a "file" or a "lf-file" argument is specified, the file's content is
5724 used as the response payload. If the file is not empty, its content-type
5725 must be set as argument to "content-type". Otherwise, any "content-type"
5726 argument is ignored. With a "lf-file" argument, the file's content is
5727 evaluated as a log-format string. With a "file" argument, it is considered
5728 as a raw content.
5729
5730 * If a "string" or "lf-string" argument is specified, the defined string is
5731 used as the response payload. The content-type must always be set as
5732 argument to "content-type". With a "lf-string" argument, the string is
5733 evaluated as a log-format string. With a "string" argument, it is
5734 considered as a raw string.
5735
Sébastien Grossab877122020-10-08 10:06:03 +02005736 When the response is not based on an errorfile, it is possible to append HTTP
Christopher Faulet4a2c1422020-01-31 17:36:01 +01005737 header fields to the response using "hdr" arguments. Otherwise, all "hdr"
5738 arguments are ignored. For each one, the header name is specified in <name>
5739 and its value is defined by <fmt> which follows the log-format rules.
5740
Christopher Faulet24231ab2020-01-24 17:44:23 +01005741 Note that the generated response must be smaller than a buffer. And to avoid
5742 any warning, when an errorfile or a raw file is loaded, the buffer space
Sébastien Grossab877122020-10-08 10:06:03 +02005743 reserved for the headers rewriting should also be free.
Christopher Faulet24231ab2020-01-24 17:44:23 +01005744
5745 No further "http-request" rules are evaluated.
5746
5747 Example:
Daniel Corbett67a82712020-07-06 23:01:19 -04005748 http-request return errorfile /etc/haproxy/errorfiles/200.http \
Christopher Faulet24231ab2020-01-24 17:44:23 +01005749 if { path /ping }
5750
5751 http-request return content-type image/x-icon file /var/www/favicon.ico \
5752 if { path /favicon.ico }
5753
5754 http-request return status 403 content-type text/plain \
5755 lf-string "Access denied. IP %[src] is blacklisted." \
5756 if { src -f /etc/haproxy/blacklist.lst }
5757
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005758http-request sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
5759http-request sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02005760
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005761 This actions increments the GPC0 or GPC1 counter according with the sticky
5762 counter designated by <sc-id>. If an error occurs, this action silently fails
5763 and the actions evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02005764
Cédric Dufour0d7712d2019-11-06 18:38:53 +01005765http-request sc-set-gpt0(<sc-id>) { <int> | <expr> }
5766 [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02005767
Cédric Dufour0d7712d2019-11-06 18:38:53 +01005768 This action sets the 32-bit unsigned GPT0 tag according to the sticky counter
5769 designated by <sc-id> and the value of <int>/<expr>. The expected result is a
5770 boolean. If an error occurs, this action silently fails and the actions
5771 evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02005772
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005773http-request set-dst <expr> [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02005774
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005775 This is used to set the destination IP address to the value of specified
5776 expression. Useful when a proxy in front of HAProxy rewrites destination IP,
5777 but provides the correct IP in a HTTP header; or you want to mask the IP for
5778 privacy. If you want to connect to the new address/port, use '0.0.0.0:0' as a
5779 server address in the backend.
Christopher Faulet85d79c92016-11-09 16:54:56 +01005780
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005781 Arguments:
5782 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
5783 by some converters.
Christopher Faulet85d79c92016-11-09 16:54:56 +01005784
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005785 Example:
5786 http-request set-dst hdr(x-dst)
5787 http-request set-dst dst,ipmask(24)
Christopher Faulet85d79c92016-11-09 16:54:56 +01005788
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005789 When possible, set-dst preserves the original destination port as long as the
5790 address family allows it, otherwise the destination port is set to 0.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02005791
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005792http-request set-dst-port <expr> [ { if | unless } <condition> ]
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02005793
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005794 This is used to set the destination port address to the value of specified
5795 expression. If you want to connect to the new address/port, use '0.0.0.0:0'
5796 as a server address in the backend.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02005797
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005798 Arguments:
5799 <expr> Is a standard HAProxy expression formed by a sample-fetch
5800 followed by some converters.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02005801
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005802 Example:
5803 http-request set-dst-port hdr(x-port)
5804 http-request set-dst-port int(4000)
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02005805
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005806 When possible, set-dst-port preserves the original destination address as
5807 long as the address family supports a port, otherwise it forces the
5808 destination address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +02005809
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005810http-request set-header <name> <fmt> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02005811
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005812 This does the same as "http-request add-header" except that the header name
5813 is first removed if it existed. This is useful when passing security
5814 information to the server, where the header must not be manipulated by
5815 external users. Note that the new value is computed before the removal so it
5816 is possible to concatenate a value to an existing header.
William Lallemand44be6402016-05-25 01:51:35 +02005817
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005818 Example:
5819 http-request set-header X-Haproxy-Current-Date %T
5820 http-request set-header X-SSL %[ssl_fc]
5821 http-request set-header X-SSL-Session_ID %[ssl_fc_session_id,hex]
5822 http-request set-header X-SSL-Client-Verify %[ssl_c_verify]
5823 http-request set-header X-SSL-Client-DN %{+Q}[ssl_c_s_dn]
5824 http-request set-header X-SSL-Client-CN %{+Q}[ssl_c_s_dn(cn)]
5825 http-request set-header X-SSL-Issuer %{+Q}[ssl_c_i_dn]
5826 http-request set-header X-SSL-Client-NotBefore %{+Q}[ssl_c_notbefore]
5827 http-request set-header X-SSL-Client-NotAfter %{+Q}[ssl_c_notafter]
William Lallemand44be6402016-05-25 01:51:35 +02005828
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005829http-request set-log-level <level> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02005830
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005831 This is used to change the log level of the current request when a certain
5832 condition is met. Valid levels are the 8 syslog levels (see the "log"
5833 keyword) plus the special level "silent" which disables logging for this
5834 request. This rule is not final so the last matching rule wins. This rule
5835 can be useful to disable health checks coming from another equipment.
William Lallemand13e9b0c2016-05-25 02:34:07 +02005836
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005837http-request set-map(<file-name>) <key fmt> <value fmt>
5838 [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02005839
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005840 This is used to add a new entry into a MAP. The MAP must be loaded from a
5841 file (even a dummy empty file). The file name of the MAP to be updated is
5842 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
5843 log-format rules, used to collect MAP key, and <value fmt>, which follows
5844 log-format rules, used to collect content for the new entry.
5845 It performs a lookup in the MAP before insertion, to avoid duplicated (or
5846 more) values. This lookup is done by a linear search and can be expensive
5847 with large lists! It is the equivalent of the "set map" command from the
5848 stats socket, but can be triggered by an HTTP request.
William Lallemand13e9b0c2016-05-25 02:34:07 +02005849
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005850http-request set-mark <mark> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02005851
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005852 This is used to set the Netfilter MARK on all packets sent to the client to
5853 the value passed in <mark> on platforms which support it. This value is an
5854 unsigned 32 bit value which can be matched by netfilter and by the routing
5855 table. It can be expressed both in decimal or hexadecimal format (prefixed by
5856 "0x"). This can be useful to force certain packets to take a different route
5857 (for example a cheaper network path for bulk downloads). This works on Linux
5858 kernels 2.6.32 and above and requires admin privileges.
Willy Tarreau00005ce2016-10-21 15:07:45 +02005859
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005860http-request set-method <fmt> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02005861
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005862 This rewrites the request method with the result of the evaluation of format
5863 string <fmt>. There should be very few valid reasons for having to do so as
5864 this is more likely to break something than to fix it.
William Lallemand13e9b0c2016-05-25 02:34:07 +02005865
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005866http-request set-nice <nice> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02005867
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005868 This sets the "nice" factor of the current request being processed. It only
5869 has effect against the other requests being processed at the same time.
5870 The default value is 0, unless altered by the "nice" setting on the "bind"
5871 line. The accepted range is -1024..1024. The higher the value, the nicest
5872 the request will be. Lower values will make the request more important than
5873 other ones. This can be useful to improve the speed of some requests, or
5874 lower the priority of non-important requests. Using this setting without
5875 prior experimentation can cause some major slowdown.
William Lallemand13e9b0c2016-05-25 02:34:07 +02005876
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005877http-request set-path <fmt> [ { if | unless } <condition> ]
Willy Tarreau00005ce2016-10-21 15:07:45 +02005878
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005879 This rewrites the request path with the result of the evaluation of format
5880 string <fmt>. The query string, if any, is left intact. If a scheme and
5881 authority is found before the path, they are left intact as well. If the
5882 request doesn't have a path ("*"), this one is replaced with the format.
5883 This can be used to prepend a directory component in front of a path for
5884 example. See also "http-request set-query" and "http-request set-uri".
Willy Tarreau2d392c22015-08-24 01:43:45 +02005885
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005886 Example :
5887 # prepend the host name before the path
5888 http-request set-path /%[hdr(host)]%[path]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02005889
Christopher Faulet312294f2020-09-02 17:17:44 +02005890http-request set-pathq <fmt> [ { if | unless } <condition> ]
5891
5892 This does the same as "http-request set-path" except that the query-string is
5893 also rewritten. It may be used to remove the query-string, including the
5894 question mark (it is not possible using "http-request set-query").
5895
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005896http-request set-priority-class <expr> [ { if | unless } <condition> ]
Olivier Houchardccaa7de2017-10-02 11:51:03 +02005897
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005898 This is used to set the queue priority class of the current request.
5899 The value must be a sample expression which converts to an integer in the
5900 range -2047..2047. Results outside this range will be truncated.
5901 The priority class determines the order in which queued requests are
5902 processed. Lower values have higher priority.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02005903
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005904http-request set-priority-offset <expr> [ { if | unless } <condition> ]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02005905
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005906 This is used to set the queue priority timestamp offset of the current
5907 request. The value must be a sample expression which converts to an integer
5908 in the range -524287..524287. Results outside this range will be truncated.
5909 When a request is queued, it is ordered first by the priority class, then by
5910 the current timestamp adjusted by the given offset in milliseconds. Lower
5911 values have higher priority.
5912 Note that the resulting timestamp is is only tracked with enough precision
5913 for 524,287ms (8m44s287ms). If the request is queued long enough to where the
5914 adjusted timestamp exceeds this value, it will be misidentified as highest
5915 priority. Thus it is important to set "timeout queue" to a value, where when
5916 combined with the offset, does not exceed this limit.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02005917
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005918http-request set-query <fmt> [ { if | unless } <condition> ]
Willy Tarreau20b0de52012-12-24 15:45:22 +01005919
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005920 This rewrites the request's query string which appears after the first
5921 question mark ("?") with the result of the evaluation of format string <fmt>.
5922 The part prior to the question mark is left intact. If the request doesn't
5923 contain a question mark and the new value is not empty, then one is added at
5924 the end of the URI, followed by the new value. If a question mark was
5925 present, it will never be removed even if the value is empty. This can be
5926 used to add or remove parameters from the query string.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08005927
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005928 See also "http-request set-query" and "http-request set-uri".
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01005929
5930 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005931 # replace "%3D" with "=" in the query string
5932 http-request set-query %[query,regsub(%3D,=,g)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01005933
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005934http-request set-src <expr> [ { if | unless } <condition> ]
5935 This is used to set the source IP address to the value of specified
5936 expression. Useful when a proxy in front of HAProxy rewrites source IP, but
5937 provides the correct IP in a HTTP header; or you want to mask source IP for
Olivier Doucet56e31202020-04-21 09:32:56 +02005938 privacy. All subsequent calls to "src" fetch will return this value
5939 (see example).
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005940
5941 Arguments :
5942 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
5943 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01005944
Olivier Doucet56e31202020-04-21 09:32:56 +02005945 See also "option forwardfor".
5946
Cyril Bonté78caf842010-03-10 22:41:43 +01005947 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005948 http-request set-src hdr(x-forwarded-for)
5949 http-request set-src src,ipmask(24)
5950
Olivier Doucet56e31202020-04-21 09:32:56 +02005951 # After the masking this will track connections
5952 # based on the IP address with the last byte zeroed out.
5953 http-request track-sc0 src
5954
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005955 When possible, set-src preserves the original source port as long as the
5956 address family allows it, otherwise the source port is set to 0.
5957
5958http-request set-src-port <expr> [ { if | unless } <condition> ]
5959
5960 This is used to set the source port address to the value of specified
5961 expression.
5962
5963 Arguments:
5964 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
5965 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01005966
Willy Tarreau20b0de52012-12-24 15:45:22 +01005967 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005968 http-request set-src-port hdr(x-port)
5969 http-request set-src-port int(4000)
5970
5971 When possible, set-src-port preserves the original source address as long as
5972 the address family supports a port, otherwise it forces the source address to
5973 IPv4 "0.0.0.0" before rewriting the port.
5974
5975http-request set-tos <tos> [ { if | unless } <condition> ]
5976
5977 This is used to set the TOS or DSCP field value of packets sent to the client
5978 to the value passed in <tos> on platforms which support this. This value
5979 represents the whole 8 bits of the IP TOS field, and can be expressed both in
5980 decimal or hexadecimal format (prefixed by "0x"). Note that only the 6 higher
5981 bits are used in DSCP or TOS, and the two lower bits are always 0. This can
5982 be used to adjust some routing behavior on border routers based on some
5983 information from the request.
5984
5985 See RFC 2474, 2597, 3260 and 4594 for more information.
5986
5987http-request set-uri <fmt> [ { if | unless } <condition> ]
5988
5989 This rewrites the request URI with the result of the evaluation of format
5990 string <fmt>. The scheme, authority, path and query string are all replaced
5991 at once. This can be used to rewrite hosts in front of proxies, or to
5992 perform complex modifications to the URI such as moving parts between the
5993 path and the query string.
5994 See also "http-request set-path" and "http-request set-query".
5995
5996http-request set-var(<var-name>) <expr> [ { if | unless } <condition> ]
5997
5998 This is used to set the contents of a variable. The variable is declared
5999 inline.
6000
6001 Arguments:
6002 <var-name> The name of the variable starts with an indication about its
6003 scope. The scopes allowed are:
6004 "proc" : the variable is shared with the whole process
6005 "sess" : the variable is shared with the whole session
6006 "txn" : the variable is shared with the transaction
6007 (request and response)
6008 "req" : the variable is shared only during request
6009 processing
6010 "res" : the variable is shared only during response
6011 processing
6012 This prefix is followed by a name. The separator is a '.'.
6013 The name may only contain characters 'a-z', 'A-Z', '0-9'
6014 and '_'.
6015
6016 <expr> Is a standard HAProxy expression formed by a sample-fetch
6017 followed by some converters.
Willy Tarreau20b0de52012-12-24 15:45:22 +01006018
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006019 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006020 http-request set-var(req.my_var) req.fhdr(user-agent),lower
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006021
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006022http-request send-spoe-group <engine-name> <group-name>
6023 [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006024
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006025 This action is used to trigger sending of a group of SPOE messages. To do so,
6026 the SPOE engine used to send messages must be defined, as well as the SPOE
6027 group to send. Of course, the SPOE engine must refer to an existing SPOE
6028 filter. If not engine name is provided on the SPOE filter line, the SPOE
6029 agent name must be used.
6030
6031 Arguments:
6032 <engine-name> The SPOE engine name.
6033
6034 <group-name> The SPOE group name as specified in the engine
6035 configuration.
6036
6037http-request silent-drop [ { if | unless } <condition> ]
6038
6039 This stops the evaluation of the rules and makes the client-facing connection
6040 suddenly disappear using a system-dependent way that tries to prevent the
6041 client from being notified. The effect it then that the client still sees an
6042 established connection while there's none on HAProxy. The purpose is to
6043 achieve a comparable effect to "tarpit" except that it doesn't use any local
6044 resource at all on the machine running HAProxy. It can resist much higher
6045 loads than "tarpit", and slow down stronger attackers. It is important to
6046 understand the impact of using this mechanism. All stateful equipment placed
6047 between the client and HAProxy (firewalls, proxies, load balancers) will also
6048 keep the established connection for a long time and may suffer from this
6049 action.
6050 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
6051 option is used to block the emission of a TCP reset. On other systems, the
6052 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
6053 router, though it's still delivered to local networks. Do not use it unless
6054 you fully understand how it works.
6055
Christopher Faulet46f95542019-12-20 10:07:22 +01006056http-request strict-mode { on | off }
6057
6058 This enables or disables the strict rewriting mode for following rules. It
6059 does not affect rules declared before it and it is only applicable on rules
6060 performing a rewrite on the requests. When the strict mode is enabled, any
6061 rewrite failure triggers an internal error. Otherwise, such errors are
6062 silently ignored. The purpose of the strict rewriting mode is to make some
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05006063 rewrites optional while others must be performed to continue the request
Christopher Faulet46f95542019-12-20 10:07:22 +01006064 processing.
6065
Christopher Faulet1aea50e2020-01-17 16:03:53 +01006066 By default, the strict rewriting mode is enabled. Its value is also reset
Christopher Faulet46f95542019-12-20 10:07:22 +01006067 when a ruleset evaluation ends. So, for instance, if you change the mode on
6068 the frontend, the default mode is restored when HAProxy starts the backend
6069 rules evaluation.
6070
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006071http-request tarpit [deny_status <status>] [ { if | unless } <condition> ]
6072http-request tarpit [ { status | deny_status } <code>] [content-type <type>]
6073 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
6074 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
6075 [ hdr <name> <fmt> ]*
6076 [ { if | unless } <condition> ]
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006077
6078 This stops the evaluation of the rules and immediately blocks the request
6079 without responding for a delay specified by "timeout tarpit" or
6080 "timeout connect" if the former is not set. After that delay, if the client
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006081 is still connected, a response is returned so that the client does not
6082 suspect it has been tarpitted. Logs will report the flags "PT". The goal of
6083 the tarpit rule is to slow down robots during an attack when they're limited
6084 on the number of concurrent requests. It can be very efficient against very
6085 dumb robots, and will significantly reduce the load on firewalls compared to
6086 a "deny" rule. But when facing "correctly" developed robots, it can make
6087 things worse by forcing haproxy and the front firewall to support insane
6088 number of concurrent connections. By default an HTTP error 500 is returned.
6089 But the response may be customized using same syntax than
6090 "http-request return" rules. Thus, see "http-request return" for details.
Ilya Shipitsin11057a32020-06-21 21:18:27 +05006091 For compatibility purpose, when no argument is defined, or only "deny_status",
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006092 the argument "default-errorfiles" is implied. It means
6093 "http-request tarpit [deny_status <status>]" is an alias of
6094 "http-request tarpit [status <status>] default-errorfiles".
6095 No further "http-request" rules are evaluated.
6096 See also "http-request return" and "http-request silent-drop".
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006097
6098http-request track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
6099http-request track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
6100http-request track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
6101
6102 This enables tracking of sticky counters from current request. These rules do
6103 not stop evaluation and do not change default action. The number of counters
6104 that may be simultaneously tracked by the same connection is set in
6105 MAX_SESS_STKCTR at build time (reported in haproxy -vv) which defaults to 3,
6106 so the track-sc number is between 0 and (MAX_SESS_STCKTR-1). The first
6107 "track-sc0" rule executed enables tracking of the counters of the specified
6108 table as the first set. The first "track-sc1" rule executed enables tracking
6109 of the counters of the specified table as the second set. The first
6110 "track-sc2" rule executed enables tracking of the counters of the specified
6111 table as the third set. It is a recommended practice to use the first set of
6112 counters for the per-frontend counters and the second set for the per-backend
6113 ones. But this is just a guideline, all may be used everywhere.
6114
6115 Arguments :
6116 <key> is mandatory, and is a sample expression rule as described in
6117 section 7.3. It describes what elements of the incoming request or
6118 connection will be analyzed, extracted, combined, and used to
6119 select which table entry to update the counters.
6120
6121 <table> is an optional table to be used instead of the default one, which
6122 is the stick-table declared in the current proxy. All the counters
6123 for the matches and updates for the key will then be performed in
6124 that table until the session ends.
6125
6126 Once a "track-sc*" rule is executed, the key is looked up in the table and if
6127 it is not found, an entry is allocated for it. Then a pointer to that entry
6128 is kept during all the session's life, and this entry's counters are updated
6129 as often as possible, every time the session's counters are updated, and also
6130 systematically when the session ends. Counters are only updated for events
6131 that happen after the tracking has been started. As an exception, connection
6132 counters and request counters are systematically updated so that they reflect
6133 useful information.
6134
6135 If the entry tracks concurrent connection counters, one connection is counted
6136 for as long as the entry is tracked, and the entry will not expire during
6137 that time. Tracking counters also provides a performance advantage over just
6138 checking the keys, because only one table lookup is performed for all ACL
6139 checks that make use of it.
6140
6141http-request unset-var(<var-name>) [ { if | unless } <condition> ]
6142
6143 This is used to unset a variable. See above for details about <var-name>.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006144
6145 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006146 http-request unset-var(req.my_var)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006147
Christopher Faulet579d83b2019-11-22 15:34:17 +01006148http-request use-service <service-name> [ { if | unless } <condition> ]
6149
6150 This directive executes the configured HTTP service to reply to the request
6151 and stops the evaluation of the rules. An HTTP service may choose to reply by
6152 sending any valid HTTP response or it may immediately close the connection
6153 without sending any response. Outside natives services, for instance the
6154 Prometheus exporter, it is possible to write your own services in Lua. No
6155 further "http-request" rules are evaluated.
6156
6157 Arguments :
6158 <service-name> is mandatory. It is the service to call
6159
6160 Example:
6161 http-request use-service prometheus-exporter if { path /metrics }
6162
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006163http-request wait-for-handshake [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006164
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006165 This will delay the processing of the request until the SSL handshake
6166 happened. This is mostly useful to delay processing early data until we're
6167 sure they are valid.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006168
Willy Tarreauef781042010-01-27 11:53:01 +01006169
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006170http-response <action> <options...> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02006171 Access control for Layer 7 responses
6172
6173 May be used in sections: defaults | frontend | listen | backend
6174 no | yes | yes | yes
6175
6176 The http-response statement defines a set of rules which apply to layer 7
6177 processing. The rules are evaluated in their declaration order when they are
6178 met in a frontend, listen or backend section. Any rule may optionally be
6179 followed by an ACL-based condition, in which case it will only be evaluated
6180 if the condition is true. Since these rules apply on responses, the backend
6181 rules are applied first, followed by the frontend's rules.
6182
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006183 The first keyword is the rule's action. The supported actions are described
6184 below.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02006185
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006186 There is no limit to the number of http-response statements per instance.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02006187
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006188 Example:
6189 acl key_acl res.hdr(X-Acl-Key) -m found
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02006190
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006191 acl myhost hdr(Host) -f myhost.lst
Sasha Pachev218f0642014-06-16 12:05:59 -06006192
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006193 http-response add-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
6194 http-response del-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
Sasha Pachev218f0642014-06-16 12:05:59 -06006195
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006196 Example:
6197 acl value res.hdr(X-Value) -m found
Sasha Pachev218f0642014-06-16 12:05:59 -06006198
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006199 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06006200
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006201 http-response set-map(map.lst) %[src] %[res.hdr(X-Value)] if value
6202 http-response del-map(map.lst) %[src] if ! value
Sasha Pachev218f0642014-06-16 12:05:59 -06006203
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006204 See also : "http-request", section 3.4 about userlists and section 7 about
6205 ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06006206
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006207http-response add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006208
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006209 This is used to add a new entry into an ACL. The ACL must be loaded from a
6210 file (even a dummy empty file). The file name of the ACL to be updated is
6211 passed between parentheses. It takes one argument: <key fmt>, which follows
6212 log-format rules, to collect content of the new entry. It performs a lookup
6213 in the ACL before insertion, to avoid duplicated (or more) values.
6214 This lookup is done by a linear search and can be expensive with large lists!
6215 It is the equivalent of the "add acl" command from the stats socket, but can
6216 be triggered by an HTTP response.
Sasha Pachev218f0642014-06-16 12:05:59 -06006217
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006218http-response add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006219
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006220 This appends an HTTP header field whose name is specified in <name> and whose
6221 value is defined by <fmt> which follows the log-format rules (see Custom Log
6222 Format in section 8.2.4). This may be used to send a cookie to a client for
6223 example, or to pass some internal information.
6224 This rule is not final, so it is possible to add other similar rules.
6225 Note that header addition is performed immediately, so one rule might reuse
6226 the resulting header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06006227
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006228http-response allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006229
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006230 This stops the evaluation of the rules and lets the response pass the check.
6231 No further "http-response" rules are evaluated for the current section.
Sasha Pachev218f0642014-06-16 12:05:59 -06006232
Jarno Huuskonen251a6b72019-01-04 14:05:02 +02006233http-response cache-store <name> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006234
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02006235 See section 6.2 about cache setup.
Sasha Pachev218f0642014-06-16 12:05:59 -06006236
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006237http-response capture <sample> id <id> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006238
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006239 This captures sample expression <sample> from the response buffer, and
6240 converts it to a string. The resulting string is stored into the next request
6241 "capture" slot, so it will possibly appear next to some captured HTTP
6242 headers. It will then automatically appear in the logs, and it will be
6243 possible to extract it using sample fetch rules to feed it into headers or
6244 anything. Please check section 7.3 (Fetching samples) and
6245 "capture response header" for more information.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02006246
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006247 The keyword "id" is the id of the capture slot which is used for storing the
6248 string. The capture slot must be defined in an associated frontend.
6249 This is useful to run captures in backends. The slot id can be declared by a
6250 previous directive "http-response capture" or with the "declare capture"
6251 keyword.
Baptiste Assmann19a69b32020-01-16 14:34:22 +01006252
6253 When using this action in a backend, double check that the relevant
6254 frontend(s) have the required capture slots otherwise, this rule will be
6255 ignored at run time. This can't be detected at configuration parsing time
6256 due to HAProxy's ability to dynamically resolve backend name at runtime.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02006257
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006258http-response del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02006259
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006260 This is used to delete an entry from an ACL. The ACL must be loaded from a
6261 file (even a dummy empty file). The file name of the ACL to be updated is
6262 passed between parentheses. It takes one argument: <key fmt>, which follows
6263 log-format rules, to collect content of the entry to delete.
6264 It is the equivalent of the "del acl" command from the stats socket, but can
6265 be triggered by an HTTP response.
Willy Tarreauf4c43c12013-06-11 17:01:13 +02006266
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006267http-response del-header <name> [ { if | unless } <condition> ]
Willy Tarreau9a355ec2013-06-11 17:45:46 +02006268
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006269 This removes all HTTP header fields whose name is specified in <name>.
Willy Tarreau42cf39e2013-06-11 18:51:32 +02006270
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006271http-response del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau51347ed2013-06-11 19:34:13 +02006272
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006273 This is used to delete an entry from a MAP. The MAP must be loaded from a
6274 file (even a dummy empty file). The file name of the MAP to be updated is
6275 passed between parentheses. It takes one argument: <key fmt>, which follows
6276 log-format rules, to collect content of the entry to delete.
6277 It takes one argument: "file name" It is the equivalent of the "del map"
6278 command from the stats socket, but can be triggered by an HTTP response.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006279
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006280http-response deny [deny_status <status>] [ { if | unless } <condition> ]
6281http-response deny [ { status | deny_status } <code>] [content-type <type>]
6282 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
6283 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
6284 [ hdr <name> <fmt> ]*
6285 [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006286
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006287 This stops the evaluation of the rules and immediately rejects the response.
6288 By default an HTTP 502 error is returned. But the response may be customized
6289 using same syntax than "http-response return" rules. Thus, see
Ilya Shipitsin11057a32020-06-21 21:18:27 +05006290 "http-response return" for details. For compatibility purpose, when no
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006291 argument is defined, or only "deny_status", the argument "default-errorfiles"
6292 is implied. It means "http-response deny [deny_status <status>]" is an alias
6293 of "http-response deny [status <status>] default-errorfiles".
Christopher Faulet040c8cd2020-01-13 16:43:45 +01006294 No further "http-response" rules are evaluated.
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006295 See also "http-response return".
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006296
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006297http-response redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006298
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006299 This performs an HTTP redirection based on a redirect rule.
6300 This supports a format string similarly to "http-request redirect" rules,
6301 with the exception that only the "location" type of redirect is possible on
6302 the response. See the "redirect" keyword for the rule's syntax. When a
6303 redirect rule is applied during a response, connections to the server are
6304 closed so that no data can be forwarded from the server to the client.
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02006305
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006306http-response replace-header <name> <regex-match> <replace-fmt>
6307 [ { if | unless } <condition> ]
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02006308
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006309 This works like "http-request replace-header" except that it works on the
6310 server's response instead of the client's request.
William Lallemand86d0df02017-11-24 21:36:45 +01006311
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006312 Example:
6313 http-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2
Willy Tarreau51d861a2015-05-22 17:30:48 +02006314
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006315 # applied to:
6316 Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006317
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006318 # outputs:
6319 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 +02006320
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006321 # assuming the backend IP is 192.168.1.20.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006322
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006323http-response replace-value <name> <regex-match> <replace-fmt>
6324 [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006325
Tim Duesterhus6bd909b2020-01-17 15:53:18 +01006326 This works like "http-request replace-value" except that it works on the
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006327 server's response instead of the client's request.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006328
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006329 Example:
6330 http-response replace-value Cache-control ^public$ private
Christopher Faulet85d79c92016-11-09 16:54:56 +01006331
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006332 # applied to:
6333 Cache-Control: max-age=3600, public
Christopher Faulet85d79c92016-11-09 16:54:56 +01006334
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006335 # outputs:
6336 Cache-Control: max-age=3600, private
Christopher Faulet85d79c92016-11-09 16:54:56 +01006337
Christopher Faulet24231ab2020-01-24 17:44:23 +01006338http-response return [status <code>] [content-type <type>]
6339 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
6340 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
Christopher Faulet4a2c1422020-01-31 17:36:01 +01006341 [ hdr <name> <value> ]*
Christopher Faulet24231ab2020-01-24 17:44:23 +01006342 [ { if | unless } <condition> ]
6343
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05006344 This stops the evaluation of the rules and immediately returns a response. The
Christopher Faulet24231ab2020-01-24 17:44:23 +01006345 default status code used for the response is 200. It can be optionally
6346 specified as an arguments to "status". The response content-type may also be
Daniel Corbett67a82712020-07-06 23:01:19 -04006347 specified as an argument to "content-type". Finally the response itself may
Christopher Faulet24231ab2020-01-24 17:44:23 +01006348 be defined. If can be a full HTTP response specifying the errorfile to use,
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05006349 or the response payload specifying the file or the string to use. These rules
Christopher Faulet24231ab2020-01-24 17:44:23 +01006350 are followed to create the response :
6351
6352 * If neither the errorfile nor the payload to use is defined, a dummy
6353 response is returned. Only the "status" argument is considered. It can be
6354 any code in the range [200, 599]. The "content-type" argument, if any, is
6355 ignored.
6356
6357 * If "default-errorfiles" argument is set, the proxy's errorfiles are
6358 considered. If the "status" argument is defined, it must be one of the
Daniel Corbett67a82712020-07-06 23:01:19 -04006359 status code handled by haproxy (200, 400, 403, 404, 405, 408, 410, 413,
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02006360 425, 429, 500, 502, 503, and 504). The "content-type" argument, if any,
6361 is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006362
6363 * If a specific errorfile is defined, with an "errorfile" argument, the
6364 corresponding file, containing a full HTTP response, is returned. Only the
6365 "status" argument is considered. It must be one of the status code handled
Daniel Corbett67a82712020-07-06 23:01:19 -04006366 by haproxy (200, 400, 403, 404, 405, 408, 410, 413, 425, 429, 500, 502, 503,
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02006367 and 504). The "content-type" argument, if any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006368
6369 * If an http-errors section is defined, with an "errorfiles" argument, the
6370 corresponding file in the specified http-errors section, containing a full
6371 HTTP response, is returned. Only the "status" argument is considered. It
Daniel Corbett67a82712020-07-06 23:01:19 -04006372 must be one of the status code handled by haproxy (200, 400, 403, 404, 405,
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02006373 408, 410, 413, 425, 429, 500, 502, 503, and 504). The "content-type"
6374 argument, if any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006375
6376 * If a "file" or a "lf-file" argument is specified, the file's content is
6377 used as the response payload. If the file is not empty, its content-type
6378 must be set as argument to "content-type". Otherwise, any "content-type"
6379 argument is ignored. With a "lf-file" argument, the file's content is
6380 evaluated as a log-format string. With a "file" argument, it is considered
6381 as a raw content.
6382
6383 * If a "string" or "lf-string" argument is specified, the defined string is
6384 used as the response payload. The content-type must always be set as
6385 argument to "content-type". With a "lf-string" argument, the string is
6386 evaluated as a log-format string. With a "string" argument, it is
6387 considered as a raw string.
6388
Christopher Faulet4a2c1422020-01-31 17:36:01 +01006389 When the response is not based an errorfile, it is possible to appends HTTP
6390 header fields to the response using "hdr" arguments. Otherwise, all "hdr"
6391 arguments are ignored. For each one, the header name is specified in <name>
6392 and its value is defined by <fmt> which follows the log-format rules.
6393
Christopher Faulet24231ab2020-01-24 17:44:23 +01006394 Note that the generated response must be smaller than a buffer. And to avoid
6395 any warning, when an errorfile or a raw file is loaded, the buffer space
Ilya Shipitsin11057a32020-06-21 21:18:27 +05006396 reserved to the headers rewriting should also be free.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006397
6398 No further "http-response" rules are evaluated.
6399
6400 Example:
Daniel Corbett67a82712020-07-06 23:01:19 -04006401 http-response return errorfile /etc/haproxy/errorfiles/200.http \
Christopher Faulet24231ab2020-01-24 17:44:23 +01006402 if { status eq 404 }
6403
6404 http-response return content-type text/plain \
6405 string "This is the end !" \
6406 if { status eq 500 }
6407
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006408http-response sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
6409http-response sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Ruoshan Huange4edc6b2016-07-14 15:07:45 +08006410
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006411 This action increments the GPC0 or GPC1 counter according with the sticky
6412 counter designated by <sc-id>. If an error occurs, this action silently fails
6413 and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +02006414
Cédric Dufour0d7712d2019-11-06 18:38:53 +01006415http-response sc-set-gpt0(<sc-id>) { <int> | <expr> }
6416 [ { if | unless } <condition> ]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02006417
Cédric Dufour0d7712d2019-11-06 18:38:53 +01006418 This action sets the 32-bit unsigned GPT0 tag according to the sticky counter
6419 designated by <sc-id> and the value of <int>/<expr>. The expected result is a
6420 boolean. If an error occurs, this action silently fails and the actions
6421 evaluation continues.
Frédéric Lécaille6778b272018-01-29 15:22:53 +01006422
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006423http-response send-spoe-group [ { if | unless } <condition> ]
Willy Tarreau2d392c22015-08-24 01:43:45 +02006424
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006425 This action is used to trigger sending of a group of SPOE messages. To do so,
6426 the SPOE engine used to send messages must be defined, as well as the SPOE
6427 group to send. Of course, the SPOE engine must refer to an existing SPOE
6428 filter. If not engine name is provided on the SPOE filter line, the SPOE
6429 agent name must be used.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006430
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006431 Arguments:
6432 <engine-name> The SPOE engine name.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006433
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006434 <group-name> The SPOE group name as specified in the engine
6435 configuration.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006436
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006437http-response set-header <name> <fmt> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02006438
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006439 This does the same as "add-header" except that the header name is first
6440 removed if it existed. This is useful when passing security information to
6441 the server, where the header must not be manipulated by external users.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02006442
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006443http-response set-log-level <level> [ { if | unless } <condition> ]
6444
6445 This is used to change the log level of the current request when a certain
6446 condition is met. Valid levels are the 8 syslog levels (see the "log"
6447 keyword) plus the special level "silent" which disables logging for this
6448 request. This rule is not final so the last matching rule wins. This rule can
6449 be useful to disable health checks coming from another equipment.
6450
6451http-response set-map(<file-name>) <key fmt> <value fmt>
6452
6453 This is used to add a new entry into a MAP. The MAP must be loaded from a
6454 file (even a dummy empty file). The file name of the MAP to be updated is
6455 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
6456 log-format rules, used to collect MAP key, and <value fmt>, which follows
6457 log-format rules, used to collect content for the new entry. It performs a
6458 lookup in the MAP before insertion, to avoid duplicated (or more) values.
6459 This lookup is done by a linear search and can be expensive with large lists!
6460 It is the equivalent of the "set map" command from the stats socket, but can
6461 be triggered by an HTTP response.
6462
6463http-response set-mark <mark> [ { if | unless } <condition> ]
6464
6465 This is used to set the Netfilter MARK on all packets sent to the client to
6466 the value passed in <mark> on platforms which support it. This value is an
6467 unsigned 32 bit value which can be matched by netfilter and by the routing
6468 table. It can be expressed both in decimal or hexadecimal format (prefixed
6469 by "0x"). This can be useful to force certain packets to take a different
6470 route (for example a cheaper network path for bulk downloads). This works on
6471 Linux kernels 2.6.32 and above and requires admin privileges.
6472
6473http-response set-nice <nice> [ { if | unless } <condition> ]
6474
6475 This sets the "nice" factor of the current request being processed.
6476 It only has effect against the other requests being processed at the same
6477 time. The default value is 0, unless altered by the "nice" setting on the
6478 "bind" line. The accepted range is -1024..1024. The higher the value, the
6479 nicest the request will be. Lower values will make the request more important
6480 than other ones. This can be useful to improve the speed of some requests, or
6481 lower the priority of non-important requests. Using this setting without
6482 prior experimentation can cause some major slowdown.
6483
6484http-response set-status <status> [reason <str>]
6485 [ { if | unless } <condition> ]
6486
6487 This replaces the response status code with <status> which must be an integer
6488 between 100 and 999. Optionally, a custom reason text can be provided defined
6489 by <str>, or the default reason for the specified code will be used as a
6490 fallback.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08006491
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006492 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006493 # return "431 Request Header Fields Too Large"
6494 http-response set-status 431
6495 # return "503 Slow Down", custom reason
6496 http-response set-status 503 reason "Slow Down".
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006497
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006498http-response set-tos <tos> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006499
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006500 This is used to set the TOS or DSCP field value of packets sent to the client
6501 to the value passed in <tos> on platforms which support this.
6502 This value represents the whole 8 bits of the IP TOS field, and can be
6503 expressed both in decimal or hexadecimal format (prefixed by "0x"). Note that
6504 only the 6 higher bits are used in DSCP or TOS, and the two lower bits are
6505 always 0. This can be used to adjust some routing behavior on border routers
6506 based on some information from the request.
6507
6508 See RFC 2474, 2597, 3260 and 4594 for more information.
6509
6510http-response set-var(<var-name>) <expr> [ { if | unless } <condition> ]
6511
6512 This is used to set the contents of a variable. The variable is declared
6513 inline.
6514
6515 Arguments:
6516 <var-name> The name of the variable starts with an indication about its
6517 scope. The scopes allowed are:
6518 "proc" : the variable is shared with the whole process
6519 "sess" : the variable is shared with the whole session
6520 "txn" : the variable is shared with the transaction
6521 (request and response)
6522 "req" : the variable is shared only during request
6523 processing
6524 "res" : the variable is shared only during response
6525 processing
6526 This prefix is followed by a name. The separator is a '.'.
6527 The name may only contain characters 'a-z', 'A-Z', '0-9', '.'
6528 and '_'.
6529
6530 <expr> Is a standard HAProxy expression formed by a sample-fetch
6531 followed by some converters.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006532
6533 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006534 http-response set-var(sess.last_redir) res.hdr(location)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006535
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006536http-response silent-drop [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006537
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006538 This stops the evaluation of the rules and makes the client-facing connection
6539 suddenly disappear using a system-dependent way that tries to prevent the
6540 client from being notified. The effect it then that the client still sees an
6541 established connection while there's none on HAProxy. The purpose is to
6542 achieve a comparable effect to "tarpit" except that it doesn't use any local
6543 resource at all on the machine running HAProxy. It can resist much higher
6544 loads than "tarpit", and slow down stronger attackers. It is important to
6545 understand the impact of using this mechanism. All stateful equipment placed
6546 between the client and HAProxy (firewalls, proxies, load balancers) will also
6547 keep the established connection for a long time and may suffer from this
6548 action.
6549 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
6550 option is used to block the emission of a TCP reset. On other systems, the
6551 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
6552 router, though it's still delivered to local networks. Do not use it unless
6553 you fully understand how it works.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006554
Christopher Faulet46f95542019-12-20 10:07:22 +01006555http-response strict-mode { on | off }
6556
6557 This enables or disables the strict rewriting mode for following rules. It
6558 does not affect rules declared before it and it is only applicable on rules
6559 performing a rewrite on the responses. When the strict mode is enabled, any
6560 rewrite failure triggers an internal error. Otherwise, such errors are
6561 silently ignored. The purpose of the strict rewriting mode is to make some
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05006562 rewrites optional while others must be performed to continue the response
Christopher Faulet46f95542019-12-20 10:07:22 +01006563 processing.
6564
Christopher Faulet1aea50e2020-01-17 16:03:53 +01006565 By default, the strict rewriting mode is enabled. Its value is also reset
Christopher Faulet46f95542019-12-20 10:07:22 +01006566 when a ruleset evaluation ends. So, for instance, if you change the mode on
Daniel Corbett67a82712020-07-06 23:01:19 -04006567 the backend, the default mode is restored when HAProxy starts the frontend
Christopher Faulet46f95542019-12-20 10:07:22 +01006568 rules evaluation.
6569
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006570http-response track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
6571http-response track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
6572http-response track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02006573
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006574 This enables tracking of sticky counters from current response. Please refer
6575 to "http-request track-sc" for a complete description. The only difference
6576 from "http-request track-sc" is the <key> sample expression can only make use
6577 of samples in response (e.g. res.*, status etc.) and samples below Layer 6
6578 (e.g. SSL-related samples, see section 7.3.4). If the sample is not
6579 supported, haproxy will fail and warn while parsing the config.
6580
6581http-response unset-var(<var-name>) [ { if | unless } <condition> ]
6582
6583 This is used to unset a variable. See "http-response set-var" for details
6584 about <var-name>.
6585
6586 Example:
6587 http-response unset-var(sess.last_redir)
6588
Baptiste Assmann5ecb77f2013-10-06 23:24:13 +02006589
Willy Tarreau30631952015-08-06 15:05:24 +02006590http-reuse { never | safe | aggressive | always }
6591 Declare how idle HTTP connections may be shared between requests
6592
6593 May be used in sections: defaults | frontend | listen | backend
6594 yes | no | yes | yes
6595
6596 By default, a connection established between haproxy and the backend server
Olivier Houchard86006a52018-12-14 19:37:49 +01006597 which is considered safe for reuse is moved back to the server's idle
6598 connections pool so that any other request can make use of it. This is the
6599 "safe" strategy below.
Willy Tarreau30631952015-08-06 15:05:24 +02006600
6601 The argument indicates the desired connection reuse strategy :
6602
Olivier Houchard86006a52018-12-14 19:37:49 +01006603 - "never" : idle connections are never shared between sessions. This mode
6604 may be enforced to cancel a different strategy inherited from
6605 a defaults section or for troubleshooting. For example, if an
6606 old bogus application considers that multiple requests over
6607 the same connection come from the same client and it is not
6608 possible to fix the application, it may be desirable to
6609 disable connection sharing in a single backend. An example of
6610 such an application could be an old haproxy using cookie
6611 insertion in tunnel mode and not checking any request past the
6612 first one.
Willy Tarreau30631952015-08-06 15:05:24 +02006613
Olivier Houchard86006a52018-12-14 19:37:49 +01006614 - "safe" : this is the default and the recommended strategy. The first
6615 request of a session is always sent over its own connection,
6616 and only subsequent requests may be dispatched over other
6617 existing connections. This ensures that in case the server
6618 closes the connection when the request is being sent, the
6619 browser can decide to silently retry it. Since it is exactly
6620 equivalent to regular keep-alive, there should be no side
6621 effects.
Willy Tarreau30631952015-08-06 15:05:24 +02006622
6623 - "aggressive" : this mode may be useful in webservices environments where
6624 all servers are not necessarily known and where it would be
6625 appreciable to deliver most first requests over existing
6626 connections. In this case, first requests are only delivered
6627 over existing connections that have been reused at least once,
6628 proving that the server correctly supports connection reuse.
6629 It should only be used when it's sure that the client can
6630 retry a failed request once in a while and where the benefit
Michael Prokop4438c602019-05-24 10:25:45 +02006631 of aggressive connection reuse significantly outweighs the
Willy Tarreau30631952015-08-06 15:05:24 +02006632 downsides of rare connection failures.
6633
6634 - "always" : this mode is only recommended when the path to the server is
6635 known for never breaking existing connections quickly after
6636 releasing them. It allows the first request of a session to be
6637 sent to an existing connection. This can provide a significant
6638 performance increase over the "safe" strategy when the backend
6639 is a cache farm, since such components tend to show a
Davor Ocelice9ed2812017-12-25 17:49:28 +01006640 consistent behavior and will benefit from the connection
Willy Tarreau30631952015-08-06 15:05:24 +02006641 sharing. It is recommended that the "http-keep-alive" timeout
6642 remains low in this mode so that no dead connections remain
6643 usable. In most cases, this will lead to the same performance
6644 gains as "aggressive" but with more risks. It should only be
6645 used when it improves the situation over "aggressive".
6646
6647 When http connection sharing is enabled, a great care is taken to respect the
Davor Ocelice9ed2812017-12-25 17:49:28 +01006648 connection properties and compatibility. Specifically :
6649 - connections made with "usesrc" followed by a client-dependent value
6650 ("client", "clientip", "hdr_ip") are marked private and never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02006651
6652 - connections sent to a server with a TLS SNI extension are marked private
Davor Ocelice9ed2812017-12-25 17:49:28 +01006653 and are never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02006654
Lukas Tribusfd9b68c2018-10-27 20:06:59 +02006655 - connections with certain bogus authentication schemes (relying on the
6656 connection) like NTLM are detected, marked private and are never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02006657
Lukas Tribuse8adfeb2019-11-06 11:50:25 +01006658 A connection pool is involved and configurable with "pool-max-conn".
Willy Tarreau30631952015-08-06 15:05:24 +02006659
6660 Note: connection reuse improves the accuracy of the "server maxconn" setting,
6661 because almost no new connection will be established while idle connections
6662 remain available. This is particularly true with the "always" strategy.
6663
6664 See also : "option http-keep-alive", "server maxconn"
6665
6666
Mark Lamourinec2247f02012-01-04 13:02:01 -05006667http-send-name-header [<header>]
6668 Add the server name to a request. Use the header string given by <header>
Mark Lamourinec2247f02012-01-04 13:02:01 -05006669 May be used in sections: defaults | frontend | listen | backend
6670 yes | no | yes | yes
Mark Lamourinec2247f02012-01-04 13:02:01 -05006671 Arguments :
Mark Lamourinec2247f02012-01-04 13:02:01 -05006672 <header> The header string to use to send the server name
6673
Willy Tarreau81bef7e2019-10-07 14:58:02 +02006674 The "http-send-name-header" statement causes the header field named <header>
6675 to be set to the name of the target server at the moment the request is about
6676 to be sent on the wire. Any existing occurrences of this header are removed.
6677 Upon retries and redispatches, the header field is updated to always reflect
6678 the server being attempted to connect to. Given that this header is modified
6679 very late in the connection setup, it may have unexpected effects on already
6680 modified headers. For example using it with transport-level header such as
6681 connection, content-length, transfer-encoding and so on will likely result in
6682 invalid requests being sent to the server. Additionally it has been reported
6683 that this directive is currently being used as a way to overwrite the Host
6684 header field in outgoing requests; while this trick has been known to work
6685 as a side effect of the feature for some time, it is not officially supported
6686 and might possibly not work anymore in a future version depending on the
6687 technical difficulties this feature induces. A long-term solution instead
6688 consists in fixing the application which required this trick so that it binds
6689 to the correct host name.
Mark Lamourinec2247f02012-01-04 13:02:01 -05006690
6691 See also : "server"
6692
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01006693id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +02006694 Set a persistent ID to a proxy.
6695 May be used in sections : defaults | frontend | listen | backend
6696 no | yes | yes | yes
6697 Arguments : none
6698
6699 Set a persistent ID for the proxy. This ID must be unique and positive.
6700 An unused ID will automatically be assigned if unset. The first assigned
6701 value will be 1. This ID is currently only returned in statistics.
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01006702
6703
Cyril Bonté0d4bf012010-04-25 23:21:46 +02006704ignore-persist { if | unless } <condition>
6705 Declare a condition to ignore persistence
6706 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01006707 no | no | yes | yes
Cyril Bonté0d4bf012010-04-25 23:21:46 +02006708
6709 By default, when cookie persistence is enabled, every requests containing
6710 the cookie are unconditionally persistent (assuming the target server is up
6711 and running).
6712
6713 The "ignore-persist" statement allows one to declare various ACL-based
6714 conditions which, when met, will cause a request to ignore persistence.
6715 This is sometimes useful to load balance requests for static files, which
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03006716 often don't require persistence. This can also be used to fully disable
Cyril Bonté0d4bf012010-04-25 23:21:46 +02006717 persistence for a specific User-Agent (for example, some web crawler bots).
6718
Cyril Bonté0d4bf012010-04-25 23:21:46 +02006719 The persistence is ignored when an "if" condition is met, or unless an
6720 "unless" condition is met.
6721
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03006722 Example:
6723 acl url_static path_beg /static /images /img /css
6724 acl url_static path_end .gif .png .jpg .css .js
6725 ignore-persist if url_static
6726
Cyril Bonté0d4bf012010-04-25 23:21:46 +02006727 See also : "force-persist", "cookie", and section 7 about ACL usage.
6728
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02006729load-server-state-from-file { global | local | none }
6730 Allow seamless reload of HAProxy
6731 May be used in sections: defaults | frontend | listen | backend
6732 yes | no | yes | yes
6733
6734 This directive points HAProxy to a file where server state from previous
6735 running process has been saved. That way, when starting up, before handling
6736 traffic, the new process can apply old states to servers exactly has if no
Davor Ocelice9ed2812017-12-25 17:49:28 +01006737 reload occurred. The purpose of the "load-server-state-from-file" directive is
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02006738 to tell haproxy which file to use. For now, only 2 arguments to either prevent
6739 loading state or load states from a file containing all backends and servers.
6740 The state file can be generated by running the command "show servers state"
6741 over the stats socket and redirect output.
6742
Davor Ocelice9ed2812017-12-25 17:49:28 +01006743 The format of the file is versioned and is very specific. To understand it,
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02006744 please read the documentation of the "show servers state" command (chapter
Willy Tarreau1af20c72017-06-23 16:01:14 +02006745 9.3 of Management Guide).
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02006746
6747 Arguments:
6748 global load the content of the file pointed by the global directive
6749 named "server-state-file".
6750
6751 local load the content of the file pointed by the directive
6752 "server-state-file-name" if set. If not set, then the backend
6753 name is used as a file name.
6754
6755 none don't load any stat for this backend
6756
6757 Notes:
Willy Tarreaue5a60682016-11-09 14:54:53 +01006758 - server's IP address is preserved across reloads by default, but the
6759 order can be changed thanks to the server's "init-addr" setting. This
6760 means that an IP address change performed on the CLI at run time will
Davor Ocelice9ed2812017-12-25 17:49:28 +01006761 be preserved, and that any change to the local resolver (e.g. /etc/hosts)
Willy Tarreaue5a60682016-11-09 14:54:53 +01006762 will possibly not have any effect if the state file is in use.
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02006763
6764 - server's weight is applied from previous running process unless it has
6765 has changed between previous and new configuration files.
6766
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02006767 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02006768
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02006769 global
6770 stats socket /tmp/socket
6771 server-state-file /tmp/server_state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02006772
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02006773 defaults
6774 load-server-state-from-file global
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02006775
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02006776 backend bk
6777 server s1 127.0.0.1:22 check weight 11
6778 server s2 127.0.0.1:22 check weight 12
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02006779
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02006780
6781 Then one can run :
6782
6783 socat /tmp/socket - <<< "show servers state" > /tmp/server_state
6784
6785 Content of the file /tmp/server_state would be like this:
6786
6787 1
6788 # <field names skipped for the doc example>
6789 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
6790 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
6791
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02006792 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02006793
6794 global
6795 stats socket /tmp/socket
6796 server-state-base /etc/haproxy/states
6797
6798 defaults
6799 load-server-state-from-file local
6800
6801 backend bk
6802 server s1 127.0.0.1:22 check weight 11
6803 server s2 127.0.0.1:22 check weight 12
6804
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02006805
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02006806 Then one can run :
6807
6808 socat /tmp/socket - <<< "show servers state bk" > /etc/haproxy/states/bk
6809
6810 Content of the file /etc/haproxy/states/bk would be like this:
6811
6812 1
6813 # <field names skipped for the doc example>
6814 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
6815 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
6816
6817 See also: "server-state-file", "server-state-file-name", and
6818 "show servers state"
6819
Cyril Bonté0d4bf012010-04-25 23:21:46 +02006820
Willy Tarreau2769aa02007-12-27 18:26:09 +01006821log global
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02006822log <address> [len <length>] [format <format>] [sample <ranges>:<smp_size>]
6823 <facility> [<level> [<minlevel>]]
William Lallemand0f99e342011-10-12 17:50:54 +02006824no log
Willy Tarreau2769aa02007-12-27 18:26:09 +01006825 Enable per-instance logging of events and traffic.
6826 May be used in sections : defaults | frontend | listen | backend
6827 yes | yes | yes | yes
William Lallemand0f99e342011-10-12 17:50:54 +02006828
6829 Prefix :
6830 no should be used when the logger list must be flushed. For example,
6831 if you don't want to inherit from the default logger list. This
6832 prefix does not allow arguments.
6833
Willy Tarreau2769aa02007-12-27 18:26:09 +01006834 Arguments :
6835 global should be used when the instance's logging parameters are the
6836 same as the global ones. This is the most common usage. "global"
6837 replaces <address>, <facility> and <level> with those of the log
6838 entries found in the "global" section. Only one "log global"
6839 statement may be used per instance, and this form takes no other
6840 parameter.
6841
6842 <address> indicates where to send the logs. It takes the same format as
6843 for the "global" section's logs, and can be one of :
6844
6845 - An IPv4 address optionally followed by a colon (':') and a UDP
6846 port. If no port is specified, 514 is used by default (the
6847 standard syslog port).
6848
David du Colombier24bb5f52011-03-17 10:40:23 +01006849 - An IPv6 address followed by a colon (':') and optionally a UDP
6850 port. If no port is specified, 514 is used by default (the
6851 standard syslog port).
6852
Willy Tarreau2769aa02007-12-27 18:26:09 +01006853 - A filesystem path to a UNIX domain socket, keeping in mind
6854 considerations for chroot (be sure the path is accessible
6855 inside the chroot) and uid/gid (be sure the path is
Davor Ocelice9ed2812017-12-25 17:49:28 +01006856 appropriately writable).
Willy Tarreau2769aa02007-12-27 18:26:09 +01006857
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01006858 - A file descriptor number in the form "fd@<number>", which may
6859 point to a pipe, terminal, or socket. In this case unbuffered
6860 logs are used and one writev() call per log is performed. This
6861 is a bit expensive but acceptable for most workloads. Messages
6862 sent this way will not be truncated but may be dropped, in
6863 which case the DroppedLogs counter will be incremented. The
6864 writev() call is atomic even on pipes for messages up to
6865 PIPE_BUF size, which POSIX recommends to be at least 512 and
6866 which is 4096 bytes on most modern operating systems. Any
6867 larger message may be interleaved with messages from other
6868 processes. Exceptionally for debugging purposes the file
6869 descriptor may also be directed to a file, but doing so will
6870 significantly slow haproxy down as non-blocking calls will be
6871 ignored. Also there will be no way to purge nor rotate this
6872 file without restarting the process. Note that the configured
6873 syslog format is preserved, so the output is suitable for use
Willy Tarreauc1b06452018-11-12 11:57:56 +01006874 with a TCP syslog server. See also the "short" and "raw"
6875 formats below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01006876
6877 - "stdout" / "stderr", which are respectively aliases for "fd@1"
6878 and "fd@2", see above.
6879
Willy Tarreauc046d162019-08-30 15:24:59 +02006880 - A ring buffer in the form "ring@<name>", which will correspond
6881 to an in-memory ring buffer accessible over the CLI using the
6882 "show events" command, which will also list existing rings and
6883 their sizes. Such buffers are lost on reload or restart but
6884 when used as a complement this can help troubleshooting by
6885 having the logs instantly available.
6886
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01006887 You may want to reference some environment variables in the
6888 address parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01006889
Willy Tarreau18324f52014-06-27 18:10:07 +02006890 <length> is an optional maximum line length. Log lines larger than this
6891 value will be truncated before being sent. The reason is that
6892 syslog servers act differently on log line length. All servers
6893 support the default value of 1024, but some servers simply drop
6894 larger lines while others do log them. If a server supports long
6895 lines, it may make sense to set this value here in order to avoid
6896 truncating long lines. Similarly, if a server drops long lines,
6897 it is preferable to truncate them before sending them. Accepted
6898 values are 80 to 65535 inclusive. The default value of 1024 is
6899 generally fine for all standard usages. Some specific cases of
Davor Ocelice9ed2812017-12-25 17:49:28 +01006900 long captures or JSON-formatted logs may require larger values.
Willy Tarreau18324f52014-06-27 18:10:07 +02006901
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02006902 <ranges> A list of comma-separated ranges to identify the logs to sample.
6903 This is used to balance the load of the logs to send to the log
6904 server. The limits of the ranges cannot be null. They are numbered
6905 from 1. The size or period (in number of logs) of the sample must
6906 be set with <sample_size> parameter.
6907
6908 <sample_size>
6909 The size of the sample in number of logs to consider when balancing
6910 their logging loads. It is used to balance the load of the logs to
6911 send to the syslog server. This size must be greater or equal to the
6912 maximum of the high limits of the ranges.
6913 (see also <ranges> parameter).
6914
Willy Tarreauadb345d2018-11-12 07:56:13 +01006915 <format> is the log format used when generating syslog messages. It may be
6916 one of the following :
6917
6918 rfc3164 The RFC3164 syslog message format. This is the default.
6919 (https://tools.ietf.org/html/rfc3164)
6920
6921 rfc5424 The RFC5424 syslog message format.
6922 (https://tools.ietf.org/html/rfc5424)
6923
Emeric Brun54648852020-07-06 15:54:06 +02006924 priority A message containing only a level plus syslog facility between
6925 angle brackets such as '<63>', followed by the text. The PID,
6926 date, time, process name and system name are omitted. This is
6927 designed to be used with a local log server.
6928
Willy Tarreaue8746a02018-11-12 08:45:00 +01006929 short A message containing only a level between angle brackets such as
6930 '<3>', followed by the text. The PID, date, time, process name
6931 and system name are omitted. This is designed to be used with a
6932 local log server. This format is compatible with what the
6933 systemd logger consumes.
6934
Emeric Brun54648852020-07-06 15:54:06 +02006935 timed A message containing only a level between angle brackets such as
6936 '<3>', followed by ISO date and by the text. The PID, process
6937 name and system name are omitted. This is designed to be
6938 used with a local log server.
6939
6940 iso A message containing only the ISO date, followed by the text.
6941 The PID, process name and system name are omitted. This is
6942 designed to be used with a local log server.
6943
Willy Tarreauc1b06452018-11-12 11:57:56 +01006944 raw A message containing only the text. The level, PID, date, time,
6945 process name and system name are omitted. This is designed to
6946 be used in containers or during development, where the severity
6947 only depends on the file descriptor used (stdout/stderr).
6948
Willy Tarreau2769aa02007-12-27 18:26:09 +01006949 <facility> must be one of the 24 standard syslog facilities :
6950
Willy Tarreaue8746a02018-11-12 08:45:00 +01006951 kern user mail daemon auth syslog lpr news
6952 uucp cron auth2 ftp ntp audit alert cron2
6953 local0 local1 local2 local3 local4 local5 local6 local7
6954
Willy Tarreauc1b06452018-11-12 11:57:56 +01006955 Note that the facility is ignored for the "short" and "raw"
6956 formats, but still required as a positional field. It is
6957 recommended to use "daemon" in this case to make it clear that
6958 it's only supposed to be used locally.
Willy Tarreau2769aa02007-12-27 18:26:09 +01006959
6960 <level> is optional and can be specified to filter outgoing messages. By
6961 default, all messages are sent. If a level is specified, only
6962 messages with a severity at least as important as this level
Willy Tarreauf7edefa2009-05-10 17:20:05 +02006963 will be sent. An optional minimum level can be specified. If it
6964 is set, logs emitted with a more severe level than this one will
6965 be capped to this level. This is used to avoid sending "emerg"
6966 messages on all terminals on some default syslog configurations.
6967 Eight levels are known :
Willy Tarreau2769aa02007-12-27 18:26:09 +01006968
6969 emerg alert crit err warning notice info debug
6970
William Lallemand0f99e342011-10-12 17:50:54 +02006971 It is important to keep in mind that it is the frontend which decides what to
6972 log from a connection, and that in case of content switching, the log entries
6973 from the backend will be ignored. Connections are logged at level "info".
Willy Tarreaucc6c8912009-02-22 10:53:55 +01006974
6975 However, backend log declaration define how and where servers status changes
6976 will be logged. Level "notice" will be used to indicate a server going up,
6977 "warning" will be used for termination signals and definitive service
6978 termination, and "alert" will be used for when a server goes down.
6979
6980 Note : According to RFC3164, messages are truncated to 1024 bytes before
6981 being emitted.
Willy Tarreau2769aa02007-12-27 18:26:09 +01006982
6983 Example :
6984 log global
Willy Tarreauc1b06452018-11-12 11:57:56 +01006985 log stdout format short daemon # send log to systemd
6986 log stdout format raw daemon # send everything to stdout
6987 log stderr format raw daemon notice # send important events to stderr
Willy Tarreauf7edefa2009-05-10 17:20:05 +02006988 log 127.0.0.1:514 local0 notice # only send important events
6989 log 127.0.0.1:514 local0 notice notice # same but limit output level
William Lallemandb2f07452015-05-12 14:27:13 +02006990 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
Willy Tarreaudad36a32013-03-11 01:20:04 +01006991
Willy Tarreau2769aa02007-12-27 18:26:09 +01006992
William Lallemand48940402012-01-30 16:47:22 +01006993log-format <string>
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01006994 Specifies the log format string to use for traffic logs
6995 May be used in sections: defaults | frontend | listen | backend
6996 yes | yes | yes | no
William Lallemand48940402012-01-30 16:47:22 +01006997
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01006998 This directive specifies the log format string that will be used for all logs
6999 resulting from traffic passing through the frontend using this line. If the
7000 directive is used in a defaults section, all subsequent frontends will use
7001 the same log format. Please see section 8.2.4 which covers the log format
7002 string in depth.
William Lallemand48940402012-01-30 16:47:22 +01007003
Guillaume de Lafond29f45602017-03-31 19:52:15 +02007004 "log-format" directive overrides previous "option tcplog", "log-format" and
7005 "option httplog" directives.
7006
Dragan Dosen7ad31542015-09-28 17:16:47 +02007007log-format-sd <string>
7008 Specifies the RFC5424 structured-data log format string
7009 May be used in sections: defaults | frontend | listen | backend
7010 yes | yes | yes | no
7011
7012 This directive specifies the RFC5424 structured-data log format string that
7013 will be used for all logs resulting from traffic passing through the frontend
7014 using this line. If the directive is used in a defaults section, all
7015 subsequent frontends will use the same log format. Please see section 8.2.4
7016 which covers the log format string in depth.
7017
7018 See https://tools.ietf.org/html/rfc5424#section-6.3 for more information
7019 about the RFC5424 structured-data part.
7020
7021 Note : This log format string will be used only for loggers that have set
7022 log format to "rfc5424".
7023
7024 Example :
7025 log-format-sd [exampleSDID@1234\ bytes=\"%B\"\ status=\"%ST\"]
7026
7027
Willy Tarreau094af4e2015-01-07 15:03:42 +01007028log-tag <string>
7029 Specifies the log tag to use for all outgoing logs
7030 May be used in sections: defaults | frontend | listen | backend
7031 yes | yes | yes | yes
7032
7033 Sets the tag field in the syslog header to this string. It defaults to the
7034 log-tag set in the global section, otherwise the program name as launched
7035 from the command line, which usually is "haproxy". Sometimes it can be useful
7036 to differentiate between multiple processes running on the same host, or to
7037 differentiate customer instances running in the same process. In the backend,
7038 logs about servers up/down will use this tag. As a hint, it can be convenient
7039 to set a log-tag related to a hosted customer in a defaults section then put
7040 all the frontends and backends for that customer, then start another customer
7041 in a new defaults section. See also the global "log-tag" directive.
Willy Tarreau2769aa02007-12-27 18:26:09 +01007042
Willy Tarreauc35362a2014-04-25 13:58:37 +02007043max-keep-alive-queue <value>
7044 Set the maximum server queue size for maintaining keep-alive connections
7045 May be used in sections: defaults | frontend | listen | backend
7046 yes | no | yes | yes
7047
7048 HTTP keep-alive tries to reuse the same server connection whenever possible,
7049 but sometimes it can be counter-productive, for example if a server has a lot
7050 of connections while other ones are idle. This is especially true for static
7051 servers.
7052
7053 The purpose of this setting is to set a threshold on the number of queued
7054 connections at which haproxy stops trying to reuse the same server and prefers
7055 to find another one. The default value, -1, means there is no limit. A value
7056 of zero means that keep-alive requests will never be queued. For very close
7057 servers which can be reached with a low latency and which are not sensible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01007058 breaking keep-alive, a low value is recommended (e.g. local static server can
Willy Tarreauc35362a2014-04-25 13:58:37 +02007059 use a value of 10 or less). For remote servers suffering from a high latency,
7060 higher values might be needed to cover for the latency and/or the cost of
7061 picking a different server.
7062
7063 Note that this has no impact on responses which are maintained to the same
7064 server consecutively to a 401 response. They will still go to the same server
7065 even if they have to be queued.
7066
7067 See also : "option http-server-close", "option prefer-last-server", server
7068 "maxconn" and cookie persistence.
7069
Olivier Houcharda4d4fdf2018-12-14 19:27:06 +01007070max-session-srv-conns <nb>
7071 Set the maximum number of outgoing connections we can keep idling for a given
7072 client session. The default is 5 (it precisely equals MAX_SRV_LIST which is
7073 defined at build time).
Willy Tarreauc35362a2014-04-25 13:58:37 +02007074
Willy Tarreau2769aa02007-12-27 18:26:09 +01007075maxconn <conns>
7076 Fix the maximum number of concurrent connections on a frontend
7077 May be used in sections : defaults | frontend | listen | backend
7078 yes | yes | yes | no
7079 Arguments :
7080 <conns> is the maximum number of concurrent connections the frontend will
7081 accept to serve. Excess connections will be queued by the system
7082 in the socket's listen queue and will be served once a connection
7083 closes.
7084
7085 If the system supports it, it can be useful on big sites to raise this limit
7086 very high so that haproxy manages connection queues, instead of leaving the
7087 clients with unanswered connection attempts. This value should not exceed the
7088 global maxconn. Also, keep in mind that a connection contains two buffers
Baptiste Assmann79fb45d2016-03-06 23:34:31 +01007089 of tune.bufsize (16kB by default) each, as well as some other data resulting
7090 in about 33 kB of RAM being consumed per established connection. That means
7091 that a medium system equipped with 1GB of RAM can withstand around
7092 20000-25000 concurrent connections if properly tuned.
Willy Tarreau2769aa02007-12-27 18:26:09 +01007093
7094 Also, when <conns> is set to large values, it is possible that the servers
7095 are not sized to accept such loads, and for this reason it is generally wise
7096 to assign them some reasonable connection limits.
7097
Willy Tarreauc8d5b952019-02-27 17:25:52 +01007098 When this value is set to zero, which is the default, the global "maxconn"
7099 value is used.
Vincent Bernat6341be52012-06-27 17:18:30 +02007100
Willy Tarreau2769aa02007-12-27 18:26:09 +01007101 See also : "server", global section's "maxconn", "fullconn"
7102
7103
7104mode { tcp|http|health }
7105 Set the running mode or protocol of the instance
7106 May be used in sections : defaults | frontend | listen | backend
7107 yes | yes | yes | yes
7108 Arguments :
7109 tcp The instance will work in pure TCP mode. A full-duplex connection
7110 will be established between clients and servers, and no layer 7
7111 examination will be performed. This is the default mode. It
7112 should be used for SSL, SSH, SMTP, ...
7113
7114 http The instance will work in HTTP mode. The client request will be
7115 analyzed in depth before connecting to any server. Any request
7116 which is not RFC-compliant will be rejected. Layer 7 filtering,
7117 processing and switching will be possible. This is the mode which
7118 brings HAProxy most of its value.
7119
7120 health The instance will work in "health" mode. It will just reply "OK"
Willy Tarreau82569f92012-09-27 23:48:56 +02007121 to incoming connections and close the connection. Alternatively,
7122 If the "httpchk" option is set, "HTTP/1.0 200 OK" will be sent
7123 instead. Nothing will be logged in either case. This mode is used
7124 to reply to external components health checks. This mode is
7125 deprecated and should not be used anymore as it is possible to do
7126 the same and even better by combining TCP or HTTP modes with the
7127 "monitor" keyword.
Willy Tarreau2769aa02007-12-27 18:26:09 +01007128
Cyril Bonté108cf6e2012-04-21 23:30:29 +02007129 When doing content switching, it is mandatory that the frontend and the
7130 backend are in the same mode (generally HTTP), otherwise the configuration
7131 will be refused.
Willy Tarreau2769aa02007-12-27 18:26:09 +01007132
Cyril Bonté108cf6e2012-04-21 23:30:29 +02007133 Example :
Willy Tarreau2769aa02007-12-27 18:26:09 +01007134 defaults http_instances
7135 mode http
7136
Cyril Bonté108cf6e2012-04-21 23:30:29 +02007137 See also : "monitor", "monitor-net"
Willy Tarreau2769aa02007-12-27 18:26:09 +01007138
Willy Tarreau0ba27502007-12-24 16:55:16 +01007139
Cyril Bontéf0c60612010-02-06 14:44:47 +01007140monitor fail { if | unless } <condition>
Willy Tarreau2769aa02007-12-27 18:26:09 +01007141 Add a condition to report a failure to a monitor HTTP request.
Willy Tarreau0ba27502007-12-24 16:55:16 +01007142 May be used in sections : defaults | frontend | listen | backend
7143 no | yes | yes | no
Willy Tarreau0ba27502007-12-24 16:55:16 +01007144 Arguments :
7145 if <cond> the monitor request will fail if the condition is satisfied,
7146 and will succeed otherwise. The condition should describe a
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007147 combined test which must induce a failure if all conditions
Willy Tarreau0ba27502007-12-24 16:55:16 +01007148 are met, for instance a low number of servers both in a
7149 backend and its backup.
7150
7151 unless <cond> the monitor request will succeed only if the condition is
7152 satisfied, and will fail otherwise. Such a condition may be
7153 based on a test on the presence of a minimum number of active
7154 servers in a list of backends.
7155
7156 This statement adds a condition which can force the response to a monitor
7157 request to report a failure. By default, when an external component queries
7158 the URI dedicated to monitoring, a 200 response is returned. When one of the
7159 conditions above is met, haproxy will return 503 instead of 200. This is
7160 very useful to report a site failure to an external component which may base
7161 routing advertisements between multiple sites on the availability reported by
7162 haproxy. In this case, one would rely on an ACL involving the "nbsrv"
Willy Tarreauae94d4d2011-05-11 16:28:49 +02007163 criterion. Note that "monitor fail" only works in HTTP mode. Both status
7164 messages may be tweaked using "errorfile" or "errorloc" if needed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01007165
7166 Example:
7167 frontend www
Willy Tarreau2769aa02007-12-27 18:26:09 +01007168 mode http
Willy Tarreau0ba27502007-12-24 16:55:16 +01007169 acl site_dead nbsrv(dynamic) lt 2
7170 acl site_dead nbsrv(static) lt 2
7171 monitor-uri /site_alive
7172 monitor fail if site_dead
7173
Willy Tarreauae94d4d2011-05-11 16:28:49 +02007174 See also : "monitor-net", "monitor-uri", "errorfile", "errorloc"
Willy Tarreau2769aa02007-12-27 18:26:09 +01007175
7176
7177monitor-net <source>
7178 Declare a source network which is limited to monitor requests
7179 May be used in sections : defaults | frontend | listen | backend
7180 yes | yes | yes | no
7181 Arguments :
7182 <source> is the source IPv4 address or network which will only be able to
7183 get monitor responses to any request. It can be either an IPv4
7184 address, a host name, or an address followed by a slash ('/')
7185 followed by a mask.
7186
7187 In TCP mode, any connection coming from a source matching <source> will cause
7188 the connection to be immediately closed without any log. This allows another
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007189 equipment to probe the port and verify that it is still listening, without
Willy Tarreau2769aa02007-12-27 18:26:09 +01007190 forwarding the connection to a remote server.
7191
7192 In HTTP mode, a connection coming from a source matching <source> will be
7193 accepted, the following response will be sent without waiting for a request,
7194 then the connection will be closed : "HTTP/1.0 200 OK". This is normally
7195 enough for any front-end HTTP probe to detect that the service is UP and
Willy Tarreau82569f92012-09-27 23:48:56 +02007196 running without forwarding the request to a backend server. Note that this
7197 response is sent in raw format, without any transformation. This is important
7198 as it means that it will not be SSL-encrypted on SSL listeners.
Willy Tarreau2769aa02007-12-27 18:26:09 +01007199
Willy Tarreau82569f92012-09-27 23:48:56 +02007200 Monitor requests are processed very early, just after tcp-request connection
7201 ACLs which are the only ones able to block them. These connections are short
7202 lived and never wait for any data from the client. They cannot be logged, and
7203 it is the intended purpose. They are only used to report HAProxy's health to
7204 an upper component, nothing more. Please note that "monitor fail" rules do
7205 not apply to connections intercepted by "monitor-net".
Willy Tarreau2769aa02007-12-27 18:26:09 +01007206
Willy Tarreau95cd2832010-03-04 23:36:33 +01007207 Last, please note that only one "monitor-net" statement can be specified in
7208 a frontend. If more than one is found, only the last one will be considered.
Cyril Bonté108cf6e2012-04-21 23:30:29 +02007209
Willy Tarreau2769aa02007-12-27 18:26:09 +01007210 Example :
7211 # addresses .252 and .253 are just probing us.
7212 frontend www
7213 monitor-net 192.168.0.252/31
7214
7215 See also : "monitor fail", "monitor-uri"
7216
7217
7218monitor-uri <uri>
7219 Intercept a URI used by external components' monitor requests
7220 May be used in sections : defaults | frontend | listen | backend
7221 yes | yes | yes | no
7222 Arguments :
7223 <uri> is the exact URI which we want to intercept to return HAProxy's
7224 health status instead of forwarding the request.
7225
7226 When an HTTP request referencing <uri> will be received on a frontend,
7227 HAProxy will not forward it nor log it, but instead will return either
7228 "HTTP/1.0 200 OK" or "HTTP/1.0 503 Service unavailable", depending on failure
7229 conditions defined with "monitor fail". This is normally enough for any
7230 front-end HTTP probe to detect that the service is UP and running without
7231 forwarding the request to a backend server. Note that the HTTP method, the
7232 version and all headers are ignored, but the request must at least be valid
7233 at the HTTP level. This keyword may only be used with an HTTP-mode frontend.
7234
Willy Tarreau721d8e02017-12-01 18:25:08 +01007235 Monitor requests are processed very early, just after the request is parsed
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02007236 and even before any "http-request". The only rulesets applied before are the
7237 tcp-request ones. They cannot be logged either, and it is the intended
7238 purpose. They are only used to report HAProxy's health to an upper component,
7239 nothing more. However, it is possible to add any number of conditions using
7240 "monitor fail" and ACLs so that the result can be adjusted to whatever check
7241 can be imagined (most often the number of available servers in a backend).
Willy Tarreau2769aa02007-12-27 18:26:09 +01007242
Christopher Faulet6072beb2020-02-18 15:34:58 +01007243 Note: if <uri> starts by a slash ('/'), the matching is performed against the
7244 request's path instead of the request's uri. It is a workaround to let
7245 the HTTP/2 requests match the monitor-uri. Indeed, in HTTP/2, clients
7246 are encouraged to send absolute URIs only.
7247
Willy Tarreau2769aa02007-12-27 18:26:09 +01007248 Example :
7249 # Use /haproxy_test to report haproxy's status
7250 frontend www
7251 mode http
7252 monitor-uri /haproxy_test
7253
7254 See also : "monitor fail", "monitor-net"
7255
Willy Tarreau0ba27502007-12-24 16:55:16 +01007256
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007257option abortonclose
7258no option abortonclose
7259 Enable or disable early dropping of aborted requests pending in queues.
7260 May be used in sections : defaults | frontend | listen | backend
7261 yes | no | yes | yes
7262 Arguments : none
7263
7264 In presence of very high loads, the servers will take some time to respond.
7265 The per-instance connection queue will inflate, and the response time will
7266 increase respective to the size of the queue times the average per-session
7267 response time. When clients will wait for more than a few seconds, they will
Willy Tarreau198a7442008-01-17 12:05:32 +01007268 often hit the "STOP" button on their browser, leaving a useless request in
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007269 the queue, and slowing down other users, and the servers as well, because the
7270 request will eventually be served, then aborted at the first error
7271 encountered while delivering the response.
7272
7273 As there is no way to distinguish between a full STOP and a simple output
7274 close on the client side, HTTP agents should be conservative and consider
7275 that the client might only have closed its output channel while waiting for
7276 the response. However, this introduces risks of congestion when lots of users
7277 do the same, and is completely useless nowadays because probably no client at
7278 all will close the session while waiting for the response. Some HTTP agents
Davor Ocelice9ed2812017-12-25 17:49:28 +01007279 support this behavior (Squid, Apache, HAProxy), and others do not (TUX, most
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007280 hardware-based load balancers). So the probability for a closed input channel
Willy Tarreau198a7442008-01-17 12:05:32 +01007281 to represent a user hitting the "STOP" button is close to 100%, and the risk
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007282 of being the single component to break rare but valid traffic is extremely
7283 low, which adds to the temptation to be able to abort a session early while
7284 still not served and not pollute the servers.
7285
Davor Ocelice9ed2812017-12-25 17:49:28 +01007286 In HAProxy, the user can choose the desired behavior using the option
7287 "abortonclose". By default (without the option) the behavior is HTTP
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007288 compliant and aborted requests will be served. But when the option is
7289 specified, a session with an incoming channel closed will be aborted while
7290 it is still possible, either pending in the queue for a connection slot, or
7291 during the connection establishment if the server has not yet acknowledged
7292 the connection request. This considerably reduces the queue size and the load
7293 on saturated servers when users are tempted to click on STOP, which in turn
Willy Tarreaud72758d2010-01-12 10:42:19 +01007294 reduces the response time for other users.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007295
7296 If this option has been enabled in a "defaults" section, it can be disabled
7297 in a specific instance by prepending the "no" keyword before it.
7298
7299 See also : "timeout queue" and server's "maxconn" and "maxqueue" parameters
7300
7301
Willy Tarreau4076a152009-04-02 15:18:36 +02007302option accept-invalid-http-request
7303no option accept-invalid-http-request
7304 Enable or disable relaxing of HTTP request parsing
7305 May be used in sections : defaults | frontend | listen | backend
7306 yes | yes | yes | no
7307 Arguments : none
7308
Willy Tarreau91852eb2015-05-01 13:26:00 +02007309 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02007310 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01007311 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02007312 forbidden characters are essentially used to build attacks exploiting server
7313 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
7314 server will emit invalid header names for whatever reason (configuration,
7315 implementation) and the issue will not be immediately fixed. In such a case,
7316 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau422246e2012-01-07 23:54:13 +01007317 even if that does not make sense, by specifying this option. Similarly, the
7318 list of characters allowed to appear in a URI is well defined by RFC3986, and
7319 chars 0-31, 32 (space), 34 ('"'), 60 ('<'), 62 ('>'), 92 ('\'), 94 ('^'), 96
7320 ('`'), 123 ('{'), 124 ('|'), 125 ('}'), 127 (delete) and anything above are
Davor Ocelice9ed2812017-12-25 17:49:28 +01007321 not allowed at all. HAProxy always blocks a number of them (0..32, 127). The
Willy Tarreau91852eb2015-05-01 13:26:00 +02007322 remaining ones are blocked by default unless this option is enabled. This
Willy Tarreau13317662015-05-01 13:47:08 +02007323 option also relaxes the test on the HTTP version, it allows HTTP/0.9 requests
7324 to pass through (no version specified) and multiple digits for both the major
7325 and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02007326
7327 This option should never be enabled by default as it hides application bugs
7328 and open security breaches. It should only be deployed after a problem has
7329 been confirmed.
7330
7331 When this option is enabled, erroneous header names will still be accepted in
7332 requests, but the complete request will be captured in order to permit later
Willy Tarreau422246e2012-01-07 23:54:13 +01007333 analysis using the "show errors" request on the UNIX stats socket. Similarly,
7334 requests containing invalid chars in the URI part will be logged. Doing this
Willy Tarreau4076a152009-04-02 15:18:36 +02007335 also helps confirming that the issue has been solved.
7336
7337 If this option has been enabled in a "defaults" section, it can be disabled
7338 in a specific instance by prepending the "no" keyword before it.
7339
7340 See also : "option accept-invalid-http-response" and "show errors" on the
7341 stats socket.
7342
7343
7344option accept-invalid-http-response
7345no option accept-invalid-http-response
7346 Enable or disable relaxing of HTTP response parsing
7347 May be used in sections : defaults | frontend | listen | backend
7348 yes | no | yes | yes
7349 Arguments : none
7350
Willy Tarreau91852eb2015-05-01 13:26:00 +02007351 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02007352 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01007353 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02007354 forbidden characters are essentially used to build attacks exploiting server
7355 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
7356 server will emit invalid header names for whatever reason (configuration,
7357 implementation) and the issue will not be immediately fixed. In such a case,
7358 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau91852eb2015-05-01 13:26:00 +02007359 even if that does not make sense, by specifying this option. This option also
7360 relaxes the test on the HTTP version format, it allows multiple digits for
7361 both the major and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02007362
7363 This option should never be enabled by default as it hides application bugs
7364 and open security breaches. It should only be deployed after a problem has
7365 been confirmed.
7366
7367 When this option is enabled, erroneous header names will still be accepted in
7368 responses, but the complete response will be captured in order to permit
7369 later analysis using the "show errors" request on the UNIX stats socket.
7370 Doing this also helps confirming that the issue has been solved.
7371
7372 If this option has been enabled in a "defaults" section, it can be disabled
7373 in a specific instance by prepending the "no" keyword before it.
7374
7375 See also : "option accept-invalid-http-request" and "show errors" on the
7376 stats socket.
7377
7378
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007379option allbackups
7380no option allbackups
7381 Use either all backup servers at a time or only the first one
7382 May be used in sections : defaults | frontend | listen | backend
7383 yes | no | yes | yes
7384 Arguments : none
7385
7386 By default, the first operational backup server gets all traffic when normal
7387 servers are all down. Sometimes, it may be preferred to use multiple backups
7388 at once, because one will not be enough. When "option allbackups" is enabled,
7389 the load balancing will be performed among all backup servers when all normal
7390 ones are unavailable. The same load balancing algorithm will be used and the
7391 servers' weights will be respected. Thus, there will not be any priority
7392 order between the backup servers anymore.
7393
7394 This option is mostly used with static server farms dedicated to return a
7395 "sorry" page when an application is completely offline.
7396
7397 If this option has been enabled in a "defaults" section, it can be disabled
7398 in a specific instance by prepending the "no" keyword before it.
7399
7400
7401option checkcache
7402no option checkcache
Godbach7056a352013-12-11 20:01:07 +08007403 Analyze all server responses and block responses with cacheable cookies
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007404 May be used in sections : defaults | frontend | listen | backend
7405 yes | no | yes | yes
7406 Arguments : none
7407
7408 Some high-level frameworks set application cookies everywhere and do not
7409 always let enough control to the developer to manage how the responses should
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007410 be cached. When a session cookie is returned on a cacheable object, there is a
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007411 high risk of session crossing or stealing between users traversing the same
7412 caches. In some situations, it is better to block the response than to let
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02007413 some sensitive session information go in the wild.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007414
7415 The option "checkcache" enables deep inspection of all server responses for
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007416 strict compliance with HTTP specification in terms of cacheability. It
Willy Tarreau198a7442008-01-17 12:05:32 +01007417 carefully checks "Cache-control", "Pragma" and "Set-cookie" headers in server
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007418 response to check if there's a risk of caching a cookie on a client-side
7419 proxy. When this option is enabled, the only responses which can be delivered
Willy Tarreau198a7442008-01-17 12:05:32 +01007420 to the client are :
Davor Ocelice9ed2812017-12-25 17:49:28 +01007421 - all those without "Set-Cookie" header;
Willy Tarreauc55ddce2017-12-21 11:41:38 +01007422 - all those with a return code other than 200, 203, 204, 206, 300, 301,
7423 404, 405, 410, 414, 501, provided that the server has not set a
Davor Ocelice9ed2812017-12-25 17:49:28 +01007424 "Cache-control: public" header field;
Willy Tarreau24ea0bc2017-12-21 11:32:55 +01007425 - all those that result from a request using a method other than GET, HEAD,
7426 OPTIONS, TRACE, provided that the server has not set a 'Cache-Control:
Davor Ocelice9ed2812017-12-25 17:49:28 +01007427 public' header field;
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007428 - those with a 'Pragma: no-cache' header
7429 - those with a 'Cache-control: private' header
7430 - those with a 'Cache-control: no-store' header
7431 - those with a 'Cache-control: max-age=0' header
7432 - those with a 'Cache-control: s-maxage=0' header
7433 - those with a 'Cache-control: no-cache' header
7434 - those with a 'Cache-control: no-cache="set-cookie"' header
7435 - those with a 'Cache-control: no-cache="set-cookie,' header
7436 (allowing other fields after set-cookie)
7437
7438 If a response doesn't respect these requirements, then it will be blocked
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02007439 just as if it was from an "http-response deny" rule, with an "HTTP 502 bad
7440 gateway". The session state shows "PH--" meaning that the proxy blocked the
7441 response during headers processing. Additionally, an alert will be sent in
7442 the logs so that admins are informed that there's something to be fixed.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007443
7444 Due to the high impact on the application, the application should be tested
7445 in depth with the option enabled before going to production. It is also a
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01007446 good practice to always activate it during tests, even if it is not used in
Davor Ocelice9ed2812017-12-25 17:49:28 +01007447 production, as it will report potentially dangerous application behaviors.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007448
7449 If this option has been enabled in a "defaults" section, it can be disabled
7450 in a specific instance by prepending the "no" keyword before it.
7451
7452
7453option clitcpka
7454no option clitcpka
7455 Enable or disable the sending of TCP keepalive packets on the client side
7456 May be used in sections : defaults | frontend | listen | backend
7457 yes | yes | yes | no
7458 Arguments : none
7459
7460 When there is a firewall or any session-aware component between a client and
7461 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01007462 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007463 components decides to expire a session which has remained idle for too long.
7464
7465 Enabling socket-level TCP keep-alives makes the system regularly send packets
7466 to the other end of the connection, leaving it active. The delay between
7467 keep-alive probes is controlled by the system only and depends both on the
7468 operating system and its tuning parameters.
7469
7470 It is important to understand that keep-alive packets are neither emitted nor
7471 received at the application level. It is only the network stacks which sees
7472 them. For this reason, even if one side of the proxy already uses keep-alives
7473 to maintain its connection alive, those keep-alive packets will not be
7474 forwarded to the other side of the proxy.
7475
7476 Please note that this has nothing to do with HTTP keep-alive.
7477
7478 Using option "clitcpka" enables the emission of TCP keep-alive probes on the
7479 client side of a connection, which should help when session expirations are
7480 noticed between HAProxy and a client.
7481
7482 If this option has been enabled in a "defaults" section, it can be disabled
7483 in a specific instance by prepending the "no" keyword before it.
7484
7485 See also : "option srvtcpka", "option tcpka"
7486
7487
Willy Tarreau0ba27502007-12-24 16:55:16 +01007488option contstats
7489 Enable continuous traffic statistics updates
7490 May be used in sections : defaults | frontend | listen | backend
7491 yes | yes | yes | no
7492 Arguments : none
7493
7494 By default, counters used for statistics calculation are incremented
7495 only when a session finishes. It works quite well when serving small
7496 objects, but with big ones (for example large images or archives) or
7497 with A/V streaming, a graph generated from haproxy counters looks like
Willy Tarreaudef0d222016-11-08 22:03:00 +01007498 a hedgehog. With this option enabled counters get incremented frequently
7499 along the session, typically every 5 seconds, which is often enough to
7500 produce clean graphs. Recounting touches a hotpath directly so it is not
7501 not enabled by default, as it can cause a lot of wakeups for very large
7502 session counts and cause a small performance drop.
Willy Tarreau0ba27502007-12-24 16:55:16 +01007503
Christopher Faulet89aed322020-06-02 17:33:56 +02007504option disable-h2-upgrade
7505no option disable-h2-upgrade
7506 Enable or disable the implicit HTTP/2 upgrade from an HTTP/1.x client
7507 connection.
7508 May be used in sections : defaults | frontend | listen | backend
7509 yes | yes | yes | no
7510 Arguments : none
7511
7512 By default, HAProxy is able to implicitly upgrade an HTTP/1.x client
7513 connection to an HTTP/2 connection if the first request it receives from a
7514 given HTTP connection matches the HTTP/2 connection preface (i.e. the string
7515 "PRI * HTTP/2.0\r\n\r\nSM\r\n\r\n"). This way, it is possible to support
7516 HTTP/1.x and HTTP/2 clients on a non-SSL connections. This option must be used to
7517 disable the implicit upgrade. Note this implicit upgrade is only supported
7518 for HTTP proxies, thus this option too. Note also it is possible to force the
7519 HTTP/2 on clear connections by specifying "proto h2" on the bind line.
7520
7521 If this option has been enabled in a "defaults" section, it can be disabled
7522 in a specific instance by prepending the "no" keyword before it.
Willy Tarreau0ba27502007-12-24 16:55:16 +01007523
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02007524option dontlog-normal
7525no option dontlog-normal
7526 Enable or disable logging of normal, successful connections
7527 May be used in sections : defaults | frontend | listen | backend
7528 yes | yes | yes | no
7529 Arguments : none
7530
7531 There are large sites dealing with several thousand connections per second
7532 and for which logging is a major pain. Some of them are even forced to turn
7533 logs off and cannot debug production issues. Setting this option ensures that
7534 normal connections, those which experience no error, no timeout, no retry nor
7535 redispatch, will not be logged. This leaves disk space for anomalies. In HTTP
7536 mode, the response status code is checked and return codes 5xx will still be
7537 logged.
7538
7539 It is strongly discouraged to use this option as most of the time, the key to
7540 complex issues is in the normal logs which will not be logged here. If you
7541 need to separate logs, see the "log-separate-errors" option instead.
7542
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007543 See also : "log", "dontlognull", "log-separate-errors" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02007544 logging.
7545
7546
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007547option dontlognull
7548no option dontlognull
7549 Enable or disable logging of null connections
7550 May be used in sections : defaults | frontend | listen | backend
7551 yes | yes | yes | no
7552 Arguments : none
7553
7554 In certain environments, there are components which will regularly connect to
7555 various systems to ensure that they are still alive. It can be the case from
7556 another load balancer as well as from monitoring systems. By default, even a
7557 simple port probe or scan will produce a log. If those connections pollute
7558 the logs too much, it is possible to enable option "dontlognull" to indicate
7559 that a connection on which no data has been transferred will not be logged,
Willy Tarreau0f228a02015-05-01 15:37:53 +02007560 which typically corresponds to those probes. Note that errors will still be
7561 returned to the client and accounted for in the stats. If this is not what is
7562 desired, option http-ignore-probes can be used instead.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007563
7564 It is generally recommended not to use this option in uncontrolled
Davor Ocelice9ed2812017-12-25 17:49:28 +01007565 environments (e.g. internet), otherwise scans and other malicious activities
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007566 would not be logged.
7567
7568 If this option has been enabled in a "defaults" section, it can be disabled
7569 in a specific instance by prepending the "no" keyword before it.
7570
Willy Tarreau0f228a02015-05-01 15:37:53 +02007571 See also : "log", "http-ignore-probes", "monitor-net", "monitor-uri", and
7572 section 8 about logging.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007573
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007574
Willy Tarreau87cf5142011-08-19 22:57:24 +02007575option forwardfor [ except <network> ] [ header <name> ] [ if-none ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01007576 Enable insertion of the X-Forwarded-For header to requests sent to servers
7577 May be used in sections : defaults | frontend | listen | backend
7578 yes | yes | yes | yes
7579 Arguments :
7580 <network> is an optional argument used to disable this option for sources
7581 matching <network>
Ross Westaf72a1d2008-08-03 10:51:45 +02007582 <name> an optional argument to specify a different "X-Forwarded-For"
Willy Tarreaud72758d2010-01-12 10:42:19 +01007583 header name.
Willy Tarreauc27debf2008-01-06 08:57:02 +01007584
7585 Since HAProxy works in reverse-proxy mode, the servers see its IP address as
7586 their client address. This is sometimes annoying when the client's IP address
7587 is expected in server logs. To solve this problem, the well-known HTTP header
7588 "X-Forwarded-For" may be added by HAProxy to all requests sent to the server.
7589 This header contains a value representing the client's IP address. Since this
7590 header is always appended at the end of the existing header list, the server
7591 must be configured to always use the last occurrence of this header only. See
Ross Westaf72a1d2008-08-03 10:51:45 +02007592 the server's manual to find how to enable use of this standard header. Note
7593 that only the last occurrence of the header must be used, since it is really
7594 possible that the client has already brought one.
7595
Willy Tarreaud72758d2010-01-12 10:42:19 +01007596 The keyword "header" may be used to supply a different header name to replace
Ross Westaf72a1d2008-08-03 10:51:45 +02007597 the default "X-Forwarded-For". This can be useful where you might already
Davor Ocelice9ed2812017-12-25 17:49:28 +01007598 have a "X-Forwarded-For" header from a different application (e.g. stunnel),
Willy Tarreaud72758d2010-01-12 10:42:19 +01007599 and you need preserve it. Also if your backend server doesn't use the
Davor Ocelice9ed2812017-12-25 17:49:28 +01007600 "X-Forwarded-For" header and requires different one (e.g. Zeus Web Servers
Ross Westaf72a1d2008-08-03 10:51:45 +02007601 require "X-Cluster-Client-IP").
Willy Tarreauc27debf2008-01-06 08:57:02 +01007602
7603 Sometimes, a same HAProxy instance may be shared between a direct client
7604 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
7605 used to decrypt HTTPS traffic). It is possible to disable the addition of the
7606 header for a known source address or network by adding the "except" keyword
7607 followed by the network address. In this case, any source IP matching the
7608 network will not cause an addition of this header. Most common uses are with
7609 private networks or 127.0.0.1.
7610
Willy Tarreau87cf5142011-08-19 22:57:24 +02007611 Alternatively, the keyword "if-none" states that the header will only be
7612 added if it is not present. This should only be used in perfectly trusted
7613 environment, as this might cause a security issue if headers reaching haproxy
7614 are under the control of the end-user.
7615
Willy Tarreauc27debf2008-01-06 08:57:02 +01007616 This option may be specified either in the frontend or in the backend. If at
Ross Westaf72a1d2008-08-03 10:51:45 +02007617 least one of them uses it, the header will be added. Note that the backend's
7618 setting of the header subargument takes precedence over the frontend's if
Willy Tarreau87cf5142011-08-19 22:57:24 +02007619 both are defined. In the case of the "if-none" argument, if at least one of
7620 the frontend or the backend does not specify it, it wants the addition to be
7621 mandatory, so it wins.
Willy Tarreauc27debf2008-01-06 08:57:02 +01007622
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007623 Example :
Willy Tarreauc27debf2008-01-06 08:57:02 +01007624 # Public HTTP address also used by stunnel on the same machine
7625 frontend www
7626 mode http
7627 option forwardfor except 127.0.0.1 # stunnel already adds the header
7628
Ross Westaf72a1d2008-08-03 10:51:45 +02007629 # Those servers want the IP Address in X-Client
7630 backend www
7631 mode http
7632 option forwardfor header X-Client
7633
Willy Tarreau87cf5142011-08-19 22:57:24 +02007634 See also : "option httpclose", "option http-server-close",
Christopher Faulet315b39c2018-09-21 16:26:19 +02007635 "option http-keep-alive"
Willy Tarreauc27debf2008-01-06 08:57:02 +01007636
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02007637
Christopher Faulet98fbe952019-07-22 16:18:24 +02007638option h1-case-adjust-bogus-client
7639no option h1-case-adjust-bogus-client
7640 Enable or disable the case adjustment of HTTP/1 headers sent to bogus clients
7641 May be used in sections : defaults | frontend | listen | backend
7642 yes | yes | yes | no
7643 Arguments : none
7644
7645 There is no standard case for header names because, as stated in RFC7230,
7646 they are case-insensitive. So applications must handle them in a case-
7647 insensitive manner. But some bogus applications violate the standards and
7648 erroneously rely on the cases most commonly used by browsers. This problem
7649 becomes critical with HTTP/2 because all header names must be exchanged in
7650 lower case, and HAProxy follows the same convention. All header names are
7651 sent in lower case to clients and servers, regardless of the HTTP version.
7652
7653 When HAProxy receives an HTTP/1 response, its header names are converted to
7654 lower case and manipulated and sent this way to the clients. If a client is
7655 known to violate the HTTP standards and to fail to process a response coming
7656 from HAProxy, it is possible to transform the lower case header names to a
7657 different format when the response is formatted and sent to the client, by
7658 enabling this option and specifying the list of headers to be reformatted
7659 using the global directives "h1-case-adjust" or "h1-case-adjust-file". This
7660 must only be a temporary workaround for the time it takes the client to be
7661 fixed, because clients which require such workarounds might be vulnerable to
7662 content smuggling attacks and must absolutely be fixed.
7663
7664 Please note that this option will not affect standards-compliant clients.
7665
7666 If this option has been enabled in a "defaults" section, it can be disabled
7667 in a specific instance by prepending the "no" keyword before it.
7668
7669 See also: "option h1-case-adjust-bogus-server", "h1-case-adjust",
7670 "h1-case-adjust-file".
7671
7672
7673option h1-case-adjust-bogus-server
7674no option h1-case-adjust-bogus-server
7675 Enable or disable the case adjustment of HTTP/1 headers sent to bogus servers
7676 May be used in sections : defaults | frontend | listen | backend
7677 yes | no | yes | yes
7678 Arguments : none
7679
7680 There is no standard case for header names because, as stated in RFC7230,
7681 they are case-insensitive. So applications must handle them in a case-
7682 insensitive manner. But some bogus applications violate the standards and
7683 erroneously rely on the cases most commonly used by browsers. This problem
7684 becomes critical with HTTP/2 because all header names must be exchanged in
7685 lower case, and HAProxy follows the same convention. All header names are
7686 sent in lower case to clients and servers, regardless of the HTTP version.
7687
7688 When HAProxy receives an HTTP/1 request, its header names are converted to
7689 lower case and manipulated and sent this way to the servers. If a server is
7690 known to violate the HTTP standards and to fail to process a request coming
7691 from HAProxy, it is possible to transform the lower case header names to a
7692 different format when the request is formatted and sent to the server, by
7693 enabling this option and specifying the list of headers to be reformatted
7694 using the global directives "h1-case-adjust" or "h1-case-adjust-file". This
7695 must only be a temporary workaround for the time it takes the server to be
7696 fixed, because servers which require such workarounds might be vulnerable to
7697 content smuggling attacks and must absolutely be fixed.
7698
7699 Please note that this option will not affect standards-compliant servers.
7700
7701 If this option has been enabled in a "defaults" section, it can be disabled
7702 in a specific instance by prepending the "no" keyword before it.
7703
7704 See also: "option h1-case-adjust-bogus-client", "h1-case-adjust",
7705 "h1-case-adjust-file".
7706
7707
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02007708option http-buffer-request
7709no option http-buffer-request
7710 Enable or disable waiting for whole HTTP request body before proceeding
7711 May be used in sections : defaults | frontend | listen | backend
7712 yes | yes | yes | yes
7713 Arguments : none
7714
7715 It is sometimes desirable to wait for the body of an HTTP request before
7716 taking a decision. This is what is being done by "balance url_param" for
7717 example. The first use case is to buffer requests from slow clients before
7718 connecting to the server. Another use case consists in taking the routing
7719 decision based on the request body's contents. This option placed in a
7720 frontend or backend forces the HTTP processing to wait until either the whole
Christopher Faulet6db8a2e2019-11-19 16:27:25 +01007721 body is received or the request buffer is full. It can have undesired side
7722 effects with some applications abusing HTTP by expecting unbuffered
7723 transmissions between the frontend and the backend, so this should definitely
7724 not be used by default.
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02007725
Baptiste Assmanneccdf432015-10-28 13:49:01 +01007726 See also : "option http-no-delay", "timeout http-request"
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02007727
7728
Willy Tarreau0f228a02015-05-01 15:37:53 +02007729option http-ignore-probes
7730no option http-ignore-probes
7731 Enable or disable logging of null connections and request timeouts
7732 May be used in sections : defaults | frontend | listen | backend
7733 yes | yes | yes | no
7734 Arguments : none
7735
7736 Recently some browsers started to implement a "pre-connect" feature
7737 consisting in speculatively connecting to some recently visited web sites
7738 just in case the user would like to visit them. This results in many
7739 connections being established to web sites, which end up in 408 Request
7740 Timeout if the timeout strikes first, or 400 Bad Request when the browser
7741 decides to close them first. These ones pollute the log and feed the error
7742 counters. There was already "option dontlognull" but it's insufficient in
7743 this case. Instead, this option does the following things :
7744 - prevent any 400/408 message from being sent to the client if nothing
Davor Ocelice9ed2812017-12-25 17:49:28 +01007745 was received over a connection before it was closed;
7746 - prevent any log from being emitted in this situation;
Willy Tarreau0f228a02015-05-01 15:37:53 +02007747 - prevent any error counter from being incremented
7748
7749 That way the empty connection is silently ignored. Note that it is better
7750 not to use this unless it is clear that it is needed, because it will hide
7751 real problems. The most common reason for not receiving a request and seeing
7752 a 408 is due to an MTU inconsistency between the client and an intermediary
7753 element such as a VPN, which blocks too large packets. These issues are
7754 generally seen with POST requests as well as GET with large cookies. The logs
7755 are often the only way to detect them.
7756
7757 If this option has been enabled in a "defaults" section, it can be disabled
7758 in a specific instance by prepending the "no" keyword before it.
7759
7760 See also : "log", "dontlognull", "errorfile", and section 8 about logging.
7761
7762
Willy Tarreau16bfb022010-01-16 19:48:41 +01007763option http-keep-alive
7764no option http-keep-alive
7765 Enable or disable HTTP keep-alive from client to server
7766 May be used in sections : defaults | frontend | listen | backend
7767 yes | yes | yes | yes
7768 Arguments : none
7769
Willy Tarreau70dffda2014-01-30 03:07:23 +01007770 By default HAProxy operates in keep-alive mode with regards to persistent
7771 connections: for each connection it processes each request and response, and
Christopher Faulet315b39c2018-09-21 16:26:19 +02007772 leaves the connection idle on both sides between the end of a response and
7773 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02007774 as "option http-server-close" or "option httpclose". This option allows to
7775 set back the keep-alive mode, which can be useful when another mode was used
7776 in a defaults section.
Willy Tarreau70dffda2014-01-30 03:07:23 +01007777
7778 Setting "option http-keep-alive" enables HTTP keep-alive mode on the client-
7779 and server- sides. This provides the lowest latency on the client side (slow
Willy Tarreau16bfb022010-01-16 19:48:41 +01007780 network) and the fastest session reuse on the server side at the expense
7781 of maintaining idle connections to the servers. In general, it is possible
7782 with this option to achieve approximately twice the request rate that the
7783 "http-server-close" option achieves on small objects. There are mainly two
7784 situations where this option may be useful :
7785
7786 - when the server is non-HTTP compliant and authenticates the connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01007787 instead of requests (e.g. NTLM authentication)
Willy Tarreau16bfb022010-01-16 19:48:41 +01007788
7789 - when the cost of establishing the connection to the server is significant
7790 compared to the cost of retrieving the associated object from the server.
7791
7792 This last case can happen when the server is a fast static server of cache.
7793 In this case, the server will need to be properly tuned to support high enough
7794 connection counts because connections will last until the client sends another
7795 request.
7796
7797 If the client request has to go to another backend or another server due to
7798 content switching or the load balancing algorithm, the idle connection will
Willy Tarreau9420b122013-12-15 18:58:25 +01007799 immediately be closed and a new one re-opened. Option "prefer-last-server" is
7800 available to try optimize server selection so that if the server currently
7801 attached to an idle connection is usable, it will be used.
Willy Tarreau16bfb022010-01-16 19:48:41 +01007802
Willy Tarreau16bfb022010-01-16 19:48:41 +01007803 At the moment, logs will not indicate whether requests came from the same
7804 session or not. The accept date reported in the logs corresponds to the end
7805 of the previous request, and the request time corresponds to the time spent
7806 waiting for a new request. The keep-alive request time is still bound to the
7807 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
7808 not set.
7809
Christopher Faulet159e6672019-07-16 15:09:52 +02007810 This option disables and replaces any previous "option httpclose" or "option
7811 http-server-close". When backend and frontend options differ, all of these 4
7812 options have precedence over "option http-keep-alive".
Willy Tarreau16bfb022010-01-16 19:48:41 +01007813
Christopher Faulet315b39c2018-09-21 16:26:19 +02007814 See also : "option httpclose",, "option http-server-close",
Willy Tarreau9420b122013-12-15 18:58:25 +01007815 "option prefer-last-server", "option http-pretend-keepalive",
Frédéric Lécaille93d33162019-03-06 09:35:59 +01007816 and "1.1. The HTTP transaction model".
Willy Tarreau16bfb022010-01-16 19:48:41 +01007817
7818
Willy Tarreau96e31212011-05-30 18:10:30 +02007819option http-no-delay
7820no option http-no-delay
7821 Instruct the system to favor low interactive delays over performance in HTTP
7822 May be used in sections : defaults | frontend | listen | backend
7823 yes | yes | yes | yes
7824 Arguments : none
7825
7826 In HTTP, each payload is unidirectional and has no notion of interactivity.
7827 Any agent is expected to queue data somewhat for a reasonably low delay.
7828 There are some very rare server-to-server applications that abuse the HTTP
7829 protocol and expect the payload phase to be highly interactive, with many
7830 interleaved data chunks in both directions within a single request. This is
7831 absolutely not supported by the HTTP specification and will not work across
7832 most proxies or servers. When such applications attempt to do this through
7833 haproxy, it works but they will experience high delays due to the network
7834 optimizations which favor performance by instructing the system to wait for
7835 enough data to be available in order to only send full packets. Typical
7836 delays are around 200 ms per round trip. Note that this only happens with
7837 abnormal uses. Normal uses such as CONNECT requests nor WebSockets are not
7838 affected.
7839
7840 When "option http-no-delay" is present in either the frontend or the backend
7841 used by a connection, all such optimizations will be disabled in order to
7842 make the exchanges as fast as possible. Of course this offers no guarantee on
7843 the functionality, as it may break at any other place. But if it works via
7844 HAProxy, it will work as fast as possible. This option should never be used
7845 by default, and should never be used at all unless such a buggy application
7846 is discovered. The impact of using this option is an increase of bandwidth
7847 usage and CPU usage, which may significantly lower performance in high
7848 latency environments.
7849
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02007850 See also : "option http-buffer-request"
7851
Willy Tarreau96e31212011-05-30 18:10:30 +02007852
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02007853option http-pretend-keepalive
7854no option http-pretend-keepalive
7855 Define whether haproxy will announce keepalive to the server or not
7856 May be used in sections : defaults | frontend | listen | backend
Christopher Faulet98db9762018-09-21 10:25:19 +02007857 yes | no | yes | yes
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02007858 Arguments : none
7859
Christopher Faulet315b39c2018-09-21 16:26:19 +02007860 When running with "option http-server-close" or "option httpclose", haproxy
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02007861 adds a "Connection: close" header to the request forwarded to the server.
7862 Unfortunately, when some servers see this header, they automatically refrain
7863 from using the chunked encoding for responses of unknown length, while this
7864 is totally unrelated. The immediate effect is that this prevents haproxy from
7865 maintaining the client connection alive. A second effect is that a client or
7866 a cache could receive an incomplete response without being aware of it, and
7867 consider the response complete.
7868
7869 By setting "option http-pretend-keepalive", haproxy will make the server
7870 believe it will keep the connection alive. The server will then not fall back
7871 to the abnormal undesired above. When haproxy gets the whole response, it
7872 will close the connection with the server just as it would do with the
Christopher Faulet315b39c2018-09-21 16:26:19 +02007873 "option httpclose". That way the client gets a normal response and the
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02007874 connection is correctly closed on the server side.
7875
7876 It is recommended not to enable this option by default, because most servers
7877 will more efficiently close the connection themselves after the last packet,
7878 and release its buffers slightly earlier. Also, the added packet on the
7879 network could slightly reduce the overall peak performance. However it is
7880 worth noting that when this option is enabled, haproxy will have slightly
7881 less work to do. So if haproxy is the bottleneck on the whole architecture,
7882 enabling this option might save a few CPU cycles.
7883
Christopher Faulet98db9762018-09-21 10:25:19 +02007884 This option may be set in backend and listen sections. Using it in a frontend
7885 section will be ignored and a warning will be reported during startup. It is
7886 a backend related option, so there is no real reason to set it on a
7887 frontend. This option may be combined with "option httpclose", which will
7888 cause keepalive to be announced to the server and close to be announced to
7889 the client. This practice is discouraged though.
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02007890
7891 If this option has been enabled in a "defaults" section, it can be disabled
7892 in a specific instance by prepending the "no" keyword before it.
7893
Christopher Faulet315b39c2018-09-21 16:26:19 +02007894 See also : "option httpclose", "option http-server-close", and
Willy Tarreau16bfb022010-01-16 19:48:41 +01007895 "option http-keep-alive"
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02007896
Willy Tarreauc27debf2008-01-06 08:57:02 +01007897
Willy Tarreaub608feb2010-01-02 22:47:18 +01007898option http-server-close
7899no option http-server-close
7900 Enable or disable HTTP connection closing on the server side
7901 May be used in sections : defaults | frontend | listen | backend
7902 yes | yes | yes | yes
7903 Arguments : none
7904
Willy Tarreau70dffda2014-01-30 03:07:23 +01007905 By default HAProxy operates in keep-alive mode with regards to persistent
7906 connections: for each connection it processes each request and response, and
7907 leaves the connection idle on both sides between the end of a response and
7908 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02007909 as "option http-server-close" or "option httpclose". Setting "option
7910 http-server-close" enables HTTP connection-close mode on the server side
7911 while keeping the ability to support HTTP keep-alive and pipelining on the
7912 client side. This provides the lowest latency on the client side (slow
7913 network) and the fastest session reuse on the server side to save server
7914 resources, similarly to "option httpclose". It also permits non-keepalive
7915 capable servers to be served in keep-alive mode to the clients if they
7916 conform to the requirements of RFC7230. Please note that some servers do not
7917 always conform to those requirements when they see "Connection: close" in the
7918 request. The effect will be that keep-alive will never be used. A workaround
7919 consists in enabling "option http-pretend-keepalive".
Willy Tarreaub608feb2010-01-02 22:47:18 +01007920
7921 At the moment, logs will not indicate whether requests came from the same
7922 session or not. The accept date reported in the logs corresponds to the end
7923 of the previous request, and the request time corresponds to the time spent
7924 waiting for a new request. The keep-alive request time is still bound to the
Willy Tarreaub16a5742010-01-10 14:46:16 +01007925 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
7926 not set.
Willy Tarreaub608feb2010-01-02 22:47:18 +01007927
7928 This option may be set both in a frontend and in a backend. It is enabled if
7929 at least one of the frontend or backend holding a connection has it enabled.
Christopher Faulet159e6672019-07-16 15:09:52 +02007930 It disables and replaces any previous "option httpclose" or "option
7931 http-keep-alive". Please check section 4 ("Proxies") to see how this option
7932 combines with others when frontend and backend options differ.
Willy Tarreaub608feb2010-01-02 22:47:18 +01007933
7934 If this option has been enabled in a "defaults" section, it can be disabled
7935 in a specific instance by prepending the "no" keyword before it.
7936
Christopher Faulet315b39c2018-09-21 16:26:19 +02007937 See also : "option httpclose", "option http-pretend-keepalive",
7938 "option http-keep-alive", and "1.1. The HTTP transaction model".
Willy Tarreaub608feb2010-01-02 22:47:18 +01007939
Willy Tarreau88d349d2010-01-25 12:15:43 +01007940option http-use-proxy-header
Cyril Bontéf0c60612010-02-06 14:44:47 +01007941no option http-use-proxy-header
Willy Tarreau88d349d2010-01-25 12:15:43 +01007942 Make use of non-standard Proxy-Connection header instead of Connection
7943 May be used in sections : defaults | frontend | listen | backend
7944 yes | yes | yes | no
7945 Arguments : none
7946
Lukas Tribus23953682017-04-28 13:24:30 +00007947 While RFC7230 explicitly states that HTTP/1.1 agents must use the
Willy Tarreau88d349d2010-01-25 12:15:43 +01007948 Connection header to indicate their wish of persistent or non-persistent
7949 connections, both browsers and proxies ignore this header for proxied
7950 connections and make use of the undocumented, non-standard Proxy-Connection
7951 header instead. The issue begins when trying to put a load balancer between
7952 browsers and such proxies, because there will be a difference between what
7953 haproxy understands and what the client and the proxy agree on.
7954
7955 By setting this option in a frontend, haproxy can automatically switch to use
7956 that non-standard header if it sees proxied requests. A proxied request is
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01007957 defined here as one where the URI begins with neither a '/' nor a '*'. This
7958 is incompatible with the HTTP tunnel mode. Note that this option can only be
7959 specified in a frontend and will affect the request along its whole life.
Willy Tarreau88d349d2010-01-25 12:15:43 +01007960
Willy Tarreau844a7e72010-01-31 21:46:18 +01007961 Also, when this option is set, a request which requires authentication will
7962 automatically switch to use proxy authentication headers if it is itself a
7963 proxied request. That makes it possible to check or enforce authentication in
7964 front of an existing proxy.
7965
Willy Tarreau88d349d2010-01-25 12:15:43 +01007966 This option should normally never be used, except in front of a proxy.
7967
Christopher Faulet315b39c2018-09-21 16:26:19 +02007968 See also : "option httpclose", and "option http-server-close".
Willy Tarreau88d349d2010-01-25 12:15:43 +01007969
Willy Tarreaud63335a2010-02-26 12:56:52 +01007970option httpchk
7971option httpchk <uri>
7972option httpchk <method> <uri>
7973option httpchk <method> <uri> <version>
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02007974 Enables HTTP protocol to check on the servers health
Willy Tarreaud63335a2010-02-26 12:56:52 +01007975 May be used in sections : defaults | frontend | listen | backend
7976 yes | no | yes | yes
7977 Arguments :
7978 <method> is the optional HTTP method used with the requests. When not set,
7979 the "OPTIONS" method is used, as it generally requires low server
7980 processing and is easy to filter out from the logs. Any method
7981 may be used, though it is not recommended to invent non-standard
7982 ones.
7983
7984 <uri> is the URI referenced in the HTTP requests. It defaults to " / "
7985 which is accessible by default on almost any server, but may be
7986 changed to any other URI. Query strings are permitted.
7987
7988 <version> is the optional HTTP version string. It defaults to "HTTP/1.0"
7989 but some servers might behave incorrectly in HTTP 1.0, so turning
7990 it to HTTP/1.1 may sometimes help. Note that the Host field is
Christopher Faulet8acb1282020-04-09 08:44:06 +02007991 mandatory in HTTP/1.1, use "http-check send" directive to add it.
Willy Tarreaud63335a2010-02-26 12:56:52 +01007992
7993 By default, server health checks only consist in trying to establish a TCP
7994 connection. When "option httpchk" is specified, a complete HTTP request is
7995 sent once the TCP connection is established, and responses 2xx and 3xx are
7996 considered valid, while all other ones indicate a server failure, including
7997 the lack of any response.
7998
Christopher Faulete5870d82020-04-15 11:32:03 +02007999 Combined with "http-check" directives, it is possible to customize the
8000 request sent during the HTTP health checks or the matching rules on the
8001 response. It is also possible to configure a send/expect sequence, just like
8002 with the directive "tcp-check" for TCP health checks.
8003
8004 The server configuration is used by default to open connections to perform
8005 HTTP health checks. By it is also possible to overwrite server parameters
8006 using "http-check connect" rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008007
Christopher Faulete5870d82020-04-15 11:32:03 +02008008 "httpchk" option does not necessarily require an HTTP backend, it also works
8009 with plain TCP backends. This is particularly useful to check simple scripts
Christopher Faulet14cd3162020-04-16 14:50:06 +02008010 bound to some dedicated ports using the inetd daemon. However, it will always
Daniel Corbett67a82712020-07-06 23:01:19 -04008011 internally relies on an HTX multiplexer. Thus, it means the request
Christopher Faulet14cd3162020-04-16 14:50:06 +02008012 formatting and the response parsing will be strict.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008013
Christopher Faulet8acb1282020-04-09 08:44:06 +02008014 Note : For a while, there was no way to add headers or body in the request
8015 used for HTTP health checks. So a workaround was to hide it at the end
8016 of the version string with a "\r\n" after the version. It is now
8017 deprecated. The directive "http-check send" must be used instead.
8018
Willy Tarreaud63335a2010-02-26 12:56:52 +01008019 Examples :
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02008020 # Relay HTTPS traffic to Apache instance and check service availability
8021 # using HTTP request "OPTIONS * HTTP/1.1" on port 80.
8022 backend https_relay
8023 mode tcp
8024 option httpchk OPTIONS * HTTP/1.1
8025 http-check send hdr Host www
8026 server apache1 192.168.1.1:443 check port 80
Willy Tarreaud63335a2010-02-26 12:56:52 +01008027
Simon Hormanafc47ee2013-11-25 10:46:35 +09008028 See also : "option ssl-hello-chk", "option smtpchk", "option mysql-check",
8029 "option pgsql-check", "http-check" and the "check", "port" and
8030 "inter" server options.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008031
8032
Willy Tarreauc27debf2008-01-06 08:57:02 +01008033option httpclose
8034no option httpclose
Christopher Faulet315b39c2018-09-21 16:26:19 +02008035 Enable or disable HTTP connection closing
Willy Tarreauc27debf2008-01-06 08:57:02 +01008036 May be used in sections : defaults | frontend | listen | backend
8037 yes | yes | yes | yes
8038 Arguments : none
8039
Willy Tarreau70dffda2014-01-30 03:07:23 +01008040 By default HAProxy operates in keep-alive mode with regards to persistent
8041 connections: for each connection it processes each request and response, and
8042 leaves the connection idle on both sides between the end of a response and
8043 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02008044 as "option http-server-close" or "option httpclose".
Willy Tarreau70dffda2014-01-30 03:07:23 +01008045
Christopher Faulet315b39c2018-09-21 16:26:19 +02008046 If "option httpclose" is set, HAProxy will close connections with the server
8047 and the client as soon as the request and the response are received. It will
John Roeslerfb2fce12019-07-10 15:45:51 -05008048 also check if a "Connection: close" header is already set in each direction,
Christopher Faulet315b39c2018-09-21 16:26:19 +02008049 and will add one if missing. Any "Connection" header different from "close"
8050 will also be removed.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008051
Christopher Faulet315b39c2018-09-21 16:26:19 +02008052 This option may also be combined with "option http-pretend-keepalive", which
8053 will disable sending of the "Connection: close" header, but will still cause
8054 the connection to be closed once the whole response is received.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008055
8056 This option may be set both in a frontend and in a backend. It is enabled if
8057 at least one of the frontend or backend holding a connection has it enabled.
Christopher Faulet159e6672019-07-16 15:09:52 +02008058 It disables and replaces any previous "option http-server-close" or "option
8059 http-keep-alive". Please check section 4 ("Proxies") to see how this option
8060 combines with others when frontend and backend options differ.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008061
8062 If this option has been enabled in a "defaults" section, it can be disabled
8063 in a specific instance by prepending the "no" keyword before it.
8064
Christopher Faulet315b39c2018-09-21 16:26:19 +02008065 See also : "option http-server-close" and "1.1. The HTTP transaction model".
Willy Tarreauc27debf2008-01-06 08:57:02 +01008066
8067
Emeric Brun3a058f32009-06-30 18:26:00 +02008068option httplog [ clf ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01008069 Enable logging of HTTP request, session state and timers
8070 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01008071 yes | yes | yes | no
Emeric Brun3a058f32009-06-30 18:26:00 +02008072 Arguments :
8073 clf if the "clf" argument is added, then the output format will be
8074 the CLF format instead of HAProxy's default HTTP format. You can
8075 use this when you need to feed HAProxy's logs through a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01008076 log analyzer which only support the CLF format and which is not
Emeric Brun3a058f32009-06-30 18:26:00 +02008077 extensible.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008078
8079 By default, the log output format is very poor, as it only contains the
8080 source and destination addresses, and the instance name. By specifying
8081 "option httplog", each log line turns into a much richer format including,
8082 but not limited to, the HTTP request, the connection timers, the session
8083 status, the connections numbers, the captured headers and cookies, the
8084 frontend, backend and server name, and of course the source address and
8085 ports.
8086
PiBa-NLbd556bf2014-12-11 21:31:54 +01008087 Specifying only "option httplog" will automatically clear the 'clf' mode
8088 if it was set by default.
Emeric Brun3a058f32009-06-30 18:26:00 +02008089
Guillaume de Lafond29f45602017-03-31 19:52:15 +02008090 "option httplog" overrides any previous "log-format" directive.
8091
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008092 See also : section 8 about logging.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008093
Willy Tarreau55165fe2009-05-10 12:02:55 +02008094
8095option http_proxy
8096no option http_proxy
8097 Enable or disable plain HTTP proxy mode
8098 May be used in sections : defaults | frontend | listen | backend
8099 yes | yes | yes | yes
8100 Arguments : none
8101
8102 It sometimes happens that people need a pure HTTP proxy which understands
8103 basic proxy requests without caching nor any fancy feature. In this case,
8104 it may be worth setting up an HAProxy instance with the "option http_proxy"
8105 set. In this mode, no server is declared, and the connection is forwarded to
8106 the IP address and port found in the URL after the "http://" scheme.
8107
8108 No host address resolution is performed, so this only works when pure IP
8109 addresses are passed. Since this option's usage perimeter is rather limited,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01008110 it will probably be used only by experts who know they need exactly it. This
8111 is incompatible with the HTTP tunnel mode.
Willy Tarreau55165fe2009-05-10 12:02:55 +02008112
8113 If this option has been enabled in a "defaults" section, it can be disabled
8114 in a specific instance by prepending the "no" keyword before it.
8115
8116 Example :
8117 # this backend understands HTTP proxy requests and forwards them directly.
8118 backend direct_forward
8119 option httpclose
8120 option http_proxy
8121
8122 See also : "option httpclose"
8123
Willy Tarreau211ad242009-10-03 21:45:07 +02008124
Jamie Gloudon801a0a32012-08-25 00:18:33 -04008125option independent-streams
8126no option independent-streams
8127 Enable or disable independent timeout processing for both directions
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02008128 May be used in sections : defaults | frontend | listen | backend
8129 yes | yes | yes | yes
8130 Arguments : none
8131
8132 By default, when data is sent over a socket, both the write timeout and the
8133 read timeout for that socket are refreshed, because we consider that there is
8134 activity on that socket, and we have no other means of guessing if we should
8135 receive data or not.
8136
Davor Ocelice9ed2812017-12-25 17:49:28 +01008137 While this default behavior is desirable for almost all applications, there
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02008138 exists a situation where it is desirable to disable it, and only refresh the
8139 read timeout if there are incoming data. This happens on sessions with large
8140 timeouts and low amounts of exchanged data such as telnet session. If the
8141 server suddenly disappears, the output data accumulates in the system's
8142 socket buffers, both timeouts are correctly refreshed, and there is no way
8143 to know the server does not receive them, so we don't timeout. However, when
8144 the underlying protocol always echoes sent data, it would be enough by itself
8145 to detect the issue using the read timeout. Note that this problem does not
8146 happen with more verbose protocols because data won't accumulate long in the
8147 socket buffers.
8148
8149 When this option is set on the frontend, it will disable read timeout updates
8150 on data sent to the client. There probably is little use of this case. When
8151 the option is set on the backend, it will disable read timeout updates on
8152 data sent to the server. Doing so will typically break large HTTP posts from
8153 slow lines, so use it with caution.
8154
Willy Tarreauce887fd2012-05-12 12:50:00 +02008155 See also : "timeout client", "timeout server" and "timeout tunnel"
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02008156
8157
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02008158option ldap-check
8159 Use LDAPv3 health checks for server testing
8160 May be used in sections : defaults | frontend | listen | backend
8161 yes | no | yes | yes
8162 Arguments : none
8163
8164 It is possible to test that the server correctly talks LDAPv3 instead of just
8165 testing that it accepts the TCP connection. When this option is set, an
8166 LDAPv3 anonymous simple bind message is sent to the server, and the response
8167 is analyzed to find an LDAPv3 bind response message.
8168
8169 The server is considered valid only when the LDAP response contains success
8170 resultCode (http://tools.ietf.org/html/rfc4511#section-4.1.9).
8171
8172 Logging of bind requests is server dependent see your documentation how to
8173 configure it.
8174
8175 Example :
8176 option ldap-check
8177
8178 See also : "option httpchk"
8179
8180
Simon Horman98637e52014-06-20 12:30:16 +09008181option external-check
8182 Use external processes for server health checks
8183 May be used in sections : defaults | frontend | listen | backend
8184 yes | no | yes | yes
8185
8186 It is possible to test the health of a server using an external command.
8187 This is achieved by running the executable set using "external-check
8188 command".
8189
8190 Requires the "external-check" global to be set.
8191
8192 See also : "external-check", "external-check command", "external-check path"
8193
8194
Willy Tarreau211ad242009-10-03 21:45:07 +02008195option log-health-checks
8196no option log-health-checks
Willy Tarreaubef1b322014-05-13 21:01:39 +02008197 Enable or disable logging of health checks status updates
Willy Tarreau211ad242009-10-03 21:45:07 +02008198 May be used in sections : defaults | frontend | listen | backend
8199 yes | no | yes | yes
8200 Arguments : none
8201
Willy Tarreaubef1b322014-05-13 21:01:39 +02008202 By default, failed health check are logged if server is UP and successful
8203 health checks are logged if server is DOWN, so the amount of additional
8204 information is limited.
Willy Tarreau211ad242009-10-03 21:45:07 +02008205
Willy Tarreaubef1b322014-05-13 21:01:39 +02008206 When this option is enabled, any change of the health check status or to
8207 the server's health will be logged, so that it becomes possible to know
8208 that a server was failing occasional checks before crashing, or exactly when
8209 it failed to respond a valid HTTP status, then when the port started to
8210 reject connections, then when the server stopped responding at all.
8211
Davor Ocelice9ed2812017-12-25 17:49:28 +01008212 Note that status changes not caused by health checks (e.g. enable/disable on
Willy Tarreaubef1b322014-05-13 21:01:39 +02008213 the CLI) are intentionally not logged by this option.
Willy Tarreau211ad242009-10-03 21:45:07 +02008214
Willy Tarreaubef1b322014-05-13 21:01:39 +02008215 See also: "option httpchk", "option ldap-check", "option mysql-check",
8216 "option pgsql-check", "option redis-check", "option smtpchk",
8217 "option tcp-check", "log" and section 8 about logging.
Willy Tarreau211ad242009-10-03 21:45:07 +02008218
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02008219
8220option log-separate-errors
8221no option log-separate-errors
8222 Change log level for non-completely successful connections
8223 May be used in sections : defaults | frontend | listen | backend
8224 yes | yes | yes | no
8225 Arguments : none
8226
8227 Sometimes looking for errors in logs is not easy. This option makes haproxy
8228 raise the level of logs containing potentially interesting information such
8229 as errors, timeouts, retries, redispatches, or HTTP status codes 5xx. The
8230 level changes from "info" to "err". This makes it possible to log them
8231 separately to a different file with most syslog daemons. Be careful not to
8232 remove them from the original file, otherwise you would lose ordering which
8233 provides very important information.
8234
8235 Using this option, large sites dealing with several thousand connections per
8236 second may log normal traffic to a rotating buffer and only archive smaller
8237 error logs.
8238
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008239 See also : "log", "dontlognull", "dontlog-normal" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02008240 logging.
8241
Willy Tarreauc27debf2008-01-06 08:57:02 +01008242
8243option logasap
8244no option logasap
Jerome Magnin95fb57b2020-04-23 19:01:17 +02008245 Enable or disable early logging.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008246 May be used in sections : defaults | frontend | listen | backend
8247 yes | yes | yes | no
8248 Arguments : none
8249
Jerome Magnin95fb57b2020-04-23 19:01:17 +02008250 By default, logs are emitted when all the log format variables and sample
8251 fetches used in the definition of the log-format string return a value, or
8252 when the session is terminated. This allows the built in log-format strings
8253 to account for the transfer time, or the number of bytes in log messages.
8254
8255 When handling long lived connections such as large file transfers or RDP,
8256 it may take a while for the request or connection to appear in the logs.
8257 Using "option logasap", the log message is created as soon as the server
8258 connection is established in mode tcp, or as soon as the server sends the
8259 complete headers in mode http. Missing information in the logs will be the
Ilya Shipitsin4329a9a2020-05-05 21:17:10 +05008260 total number of bytes which will only indicate the amount of data transferred
Jerome Magnin95fb57b2020-04-23 19:01:17 +02008261 before the message was created and the total time which will not take the
8262 remainder of the connection life or transfer time into account. For the case
8263 of HTTP, it is good practice to capture the Content-Length response header
8264 so that the logs at least indicate how many bytes are expected to be
Ilya Shipitsin4329a9a2020-05-05 21:17:10 +05008265 transferred.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008266
Willy Tarreaucc6c8912009-02-22 10:53:55 +01008267 Examples :
8268 listen http_proxy 0.0.0.0:80
8269 mode http
8270 option httplog
8271 option logasap
8272 log 192.168.2.200 local3
8273
8274 >>> Feb 6 12:14:14 localhost \
8275 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
8276 static/srv1 9/10/7/14/+30 200 +243 - - ---- 3/1/1/1/0 1/0 \
8277 "GET /image.iso HTTP/1.0"
8278
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008279 See also : "option httplog", "capture response header", and section 8 about
Willy Tarreauc27debf2008-01-06 08:57:02 +01008280 logging.
8281
8282
Christopher Faulet62f79fe2020-05-18 18:13:03 +02008283option mysql-check [ user <username> [ { post-41 | pre-41 } ] ]
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02008284 Use MySQL health checks for server testing
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01008285 May be used in sections : defaults | frontend | listen | backend
8286 yes | no | yes | yes
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02008287 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008288 <username> This is the username which will be used when connecting to MySQL
8289 server.
Christopher Faulet62f79fe2020-05-18 18:13:03 +02008290 post-41 Send post v4.1 client compatible checks (the default)
8291 pre-41 Send pre v4.1 client compatible checks
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02008292
8293 If you specify a username, the check consists of sending two MySQL packet,
8294 one Client Authentication packet, and one QUIT packet, to correctly close
Davor Ocelice9ed2812017-12-25 17:49:28 +01008295 MySQL session. We then parse the MySQL Handshake Initialization packet and/or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02008296 Error packet. It is a basic but useful test which does not produce error nor
8297 aborted connect on the server. However, it requires adding an authorization
8298 in the MySQL table, like this :
8299
8300 USE mysql;
8301 INSERT INTO user (Host,User) values ('<ip_of_haproxy>','<username>');
8302 FLUSH PRIVILEGES;
8303
8304 If you don't specify a username (it is deprecated and not recommended), the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008305 check only consists in parsing the Mysql Handshake Initialization packet or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02008306 Error packet, we don't send anything in this mode. It was reported that it
8307 can generate lockout if check is too frequent and/or if there is not enough
8308 traffic. In fact, you need in this case to check MySQL "max_connect_errors"
8309 value as if a connection is established successfully within fewer than MySQL
8310 "max_connect_errors" attempts after a previous connection was interrupted,
8311 the error count for the host is cleared to zero. If HAProxy's server get
8312 blocked, the "FLUSH HOSTS" statement is the only way to unblock it.
8313
8314 Remember that this does not check database presence nor database consistency.
8315 To do this, you can use an external check with xinetd for example.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01008316
Hervé COMMOWICK212f7782011-06-10 14:05:59 +02008317 The check requires MySQL >=3.22, for older version, please use TCP check.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01008318
8319 Most often, an incoming MySQL server needs to see the client's IP address for
8320 various purposes, including IP privilege matching and connection logging.
8321 When possible, it is often wise to masquerade the client's IP address when
8322 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02008323 which requires the transparent proxy feature to be compiled in, and the MySQL
8324 server to route the client via the machine hosting haproxy.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01008325
8326 See also: "option httpchk"
8327
8328
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008329option nolinger
8330no option nolinger
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01008331 Enable or disable immediate session resource cleaning after close
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008332 May be used in sections: defaults | frontend | listen | backend
8333 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008334 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008335
Davor Ocelice9ed2812017-12-25 17:49:28 +01008336 When clients or servers abort connections in a dirty way (e.g. they are
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008337 physically disconnected), the session timeouts triggers and the session is
8338 closed. But it will remain in FIN_WAIT1 state for some time in the system,
8339 using some resources and possibly limiting the ability to establish newer
8340 connections.
8341
8342 When this happens, it is possible to activate "option nolinger" which forces
8343 the system to immediately remove any socket's pending data on close. Thus,
8344 the session is instantly purged from the system's tables. This usually has
8345 side effects such as increased number of TCP resets due to old retransmits
8346 getting immediately rejected. Some firewalls may sometimes complain about
8347 this too.
8348
8349 For this reason, it is not recommended to use this option when not absolutely
8350 needed. You know that you need it when you have thousands of FIN_WAIT1
8351 sessions on your system (TIME_WAIT ones do not count).
8352
8353 This option may be used both on frontends and backends, depending on the side
8354 where it is required. Use it on the frontend for clients, and on the backend
8355 for servers.
8356
8357 If this option has been enabled in a "defaults" section, it can be disabled
8358 in a specific instance by prepending the "no" keyword before it.
8359
8360
Willy Tarreau55165fe2009-05-10 12:02:55 +02008361option originalto [ except <network> ] [ header <name> ]
8362 Enable insertion of the X-Original-To header to requests sent to servers
8363 May be used in sections : defaults | frontend | listen | backend
8364 yes | yes | yes | yes
8365 Arguments :
8366 <network> is an optional argument used to disable this option for sources
8367 matching <network>
8368 <name> an optional argument to specify a different "X-Original-To"
8369 header name.
8370
8371 Since HAProxy can work in transparent mode, every request from a client can
8372 be redirected to the proxy and HAProxy itself can proxy every request to a
8373 complex SQUID environment and the destination host from SO_ORIGINAL_DST will
8374 be lost. This is annoying when you want access rules based on destination ip
8375 addresses. To solve this problem, a new HTTP header "X-Original-To" may be
8376 added by HAProxy to all requests sent to the server. This header contains a
8377 value representing the original destination IP address. Since this must be
8378 configured to always use the last occurrence of this header only. Note that
8379 only the last occurrence of the header must be used, since it is really
8380 possible that the client has already brought one.
8381
8382 The keyword "header" may be used to supply a different header name to replace
8383 the default "X-Original-To". This can be useful where you might already
8384 have a "X-Original-To" header from a different application, and you need
8385 preserve it. Also if your backend server doesn't use the "X-Original-To"
8386 header and requires different one.
8387
8388 Sometimes, a same HAProxy instance may be shared between a direct client
8389 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
8390 used to decrypt HTTPS traffic). It is possible to disable the addition of the
8391 header for a known source address or network by adding the "except" keyword
8392 followed by the network address. In this case, any source IP matching the
8393 network will not cause an addition of this header. Most common uses are with
8394 private networks or 127.0.0.1.
8395
8396 This option may be specified either in the frontend or in the backend. If at
8397 least one of them uses it, the header will be added. Note that the backend's
8398 setting of the header subargument takes precedence over the frontend's if
8399 both are defined.
8400
Willy Tarreau55165fe2009-05-10 12:02:55 +02008401 Examples :
8402 # Original Destination address
8403 frontend www
8404 mode http
8405 option originalto except 127.0.0.1
8406
8407 # Those servers want the IP Address in X-Client-Dst
8408 backend www
8409 mode http
8410 option originalto header X-Client-Dst
8411
Christopher Faulet315b39c2018-09-21 16:26:19 +02008412 See also : "option httpclose", "option http-server-close".
Willy Tarreau55165fe2009-05-10 12:02:55 +02008413
8414
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008415option persist
8416no option persist
8417 Enable or disable forced persistence on down servers
8418 May be used in sections: defaults | frontend | listen | backend
8419 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008420 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008421
8422 When an HTTP request reaches a backend with a cookie which references a dead
8423 server, by default it is redispatched to another server. It is possible to
8424 force the request to be sent to the dead server first using "option persist"
8425 if absolutely needed. A common use case is when servers are under extreme
8426 load and spend their time flapping. In this case, the users would still be
8427 directed to the server they opened the session on, in the hope they would be
8428 correctly served. It is recommended to use "option redispatch" in conjunction
8429 with this option so that in the event it would not be possible to connect to
8430 the server at all (server definitely dead), the client would finally be
8431 redirected to another valid server.
8432
8433 If this option has been enabled in a "defaults" section, it can be disabled
8434 in a specific instance by prepending the "no" keyword before it.
8435
Willy Tarreau4de91492010-01-22 19:10:05 +01008436 See also : "option redispatch", "retries", "force-persist"
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008437
8438
Willy Tarreau0c122822013-12-15 18:49:01 +01008439option pgsql-check [ user <username> ]
8440 Use PostgreSQL health checks for server testing
8441 May be used in sections : defaults | frontend | listen | backend
8442 yes | no | yes | yes
8443 Arguments :
8444 <username> This is the username which will be used when connecting to
8445 PostgreSQL server.
8446
8447 The check sends a PostgreSQL StartupMessage and waits for either
8448 Authentication request or ErrorResponse message. It is a basic but useful
8449 test which does not produce error nor aborted connect on the server.
8450 This check is identical with the "mysql-check".
8451
8452 See also: "option httpchk"
8453
8454
Willy Tarreau9420b122013-12-15 18:58:25 +01008455option prefer-last-server
8456no option prefer-last-server
8457 Allow multiple load balanced requests to remain on the same server
8458 May be used in sections: defaults | frontend | listen | backend
8459 yes | no | yes | yes
8460 Arguments : none
8461
8462 When the load balancing algorithm in use is not deterministic, and a previous
8463 request was sent to a server to which haproxy still holds a connection, it is
8464 sometimes desirable that subsequent requests on a same session go to the same
8465 server as much as possible. Note that this is different from persistence, as
8466 we only indicate a preference which haproxy tries to apply without any form
8467 of warranty. The real use is for keep-alive connections sent to servers. When
8468 this option is used, haproxy will try to reuse the same connection that is
8469 attached to the server instead of rebalancing to another server, causing a
8470 close of the connection. This can make sense for static file servers. It does
Willy Tarreau068621e2013-12-23 15:11:25 +01008471 not make much sense to use this in combination with hashing algorithms. Note,
8472 haproxy already automatically tries to stick to a server which sends a 401 or
Lukas Tribus80512b12018-10-27 20:07:40 +02008473 to a proxy which sends a 407 (authentication required), when the load
8474 balancing algorithm is not deterministic. This is mandatory for use with the
8475 broken NTLM authentication challenge, and significantly helps in
Willy Tarreau068621e2013-12-23 15:11:25 +01008476 troubleshooting some faulty applications. Option prefer-last-server might be
8477 desirable in these environments as well, to avoid redistributing the traffic
8478 after every other response.
Willy Tarreau9420b122013-12-15 18:58:25 +01008479
8480 If this option has been enabled in a "defaults" section, it can be disabled
8481 in a specific instance by prepending the "no" keyword before it.
8482
8483 See also: "option http-keep-alive"
8484
8485
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008486option redispatch
Joseph Lynch726ab712015-05-11 23:25:34 -07008487option redispatch <interval>
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008488no option redispatch
8489 Enable or disable session redistribution in case of connection failure
8490 May be used in sections: defaults | frontend | listen | backend
8491 yes | no | yes | yes
Joseph Lynch726ab712015-05-11 23:25:34 -07008492 Arguments :
8493 <interval> The optional integer value that controls how often redispatches
8494 occur when retrying connections. Positive value P indicates a
8495 redispatch is desired on every Pth retry, and negative value
Davor Ocelice9ed2812017-12-25 17:49:28 +01008496 N indicate a redispatch is desired on the Nth retry prior to the
Joseph Lynch726ab712015-05-11 23:25:34 -07008497 last retry. For example, the default of -1 preserves the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008498 historical behavior of redispatching on the last retry, a
Joseph Lynch726ab712015-05-11 23:25:34 -07008499 positive value of 1 would indicate a redispatch on every retry,
8500 and a positive value of 3 would indicate a redispatch on every
8501 third retry. You can disable redispatches with a value of 0.
8502
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008503
8504 In HTTP mode, if a server designated by a cookie is down, clients may
8505 definitely stick to it because they cannot flush the cookie, so they will not
8506 be able to access the service anymore.
8507
Willy Tarreau59884a62019-01-02 14:48:31 +01008508 Specifying "option redispatch" will allow the proxy to break cookie or
8509 consistent hash based persistence and redistribute them to a working server.
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008510
Olivier Carrère6e6f59b2020-04-15 11:30:18 +02008511 Active servers are selected from a subset of the list of available
8512 servers. Active servers that are not down or in maintenance (i.e., whose
8513 health is not checked or that have been checked as "up"), are selected in the
8514 following order:
8515
8516 1. Any active, non-backup server, if any, or,
8517
8518 2. If the "allbackups" option is not set, the first backup server in the
8519 list, or
8520
8521 3. If the "allbackups" option is set, any backup server.
8522
8523 When a retry occurs, HAProxy tries to select another server than the last
8524 one. The new server is selected from the current list of servers.
8525
8526 Sometimes, if the list is updated between retries (e.g., if numerous retries
8527 occur and last longer than the time needed to check that a server is down,
8528 remove it from the list and fall back on the list of backup servers),
8529 connections may be redirected to a backup server, though.
8530
Joseph Lynch726ab712015-05-11 23:25:34 -07008531 It also allows to retry connections to another server in case of multiple
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008532 connection failures. Of course, it requires having "retries" set to a nonzero
8533 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01008534
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008535 If this option has been enabled in a "defaults" section, it can be disabled
8536 in a specific instance by prepending the "no" keyword before it.
8537
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02008538 See also : "retries", "force-persist"
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008539
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008540
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02008541option redis-check
8542 Use redis health checks for server testing
8543 May be used in sections : defaults | frontend | listen | backend
8544 yes | no | yes | yes
8545 Arguments : none
8546
8547 It is possible to test that the server correctly talks REDIS protocol instead
8548 of just testing that it accepts the TCP connection. When this option is set,
8549 a PING redis command is sent to the server, and the response is analyzed to
8550 find the "+PONG" response message.
8551
8552 Example :
8553 option redis-check
8554
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03008555 See also : "option httpchk", "option tcp-check", "tcp-check expect"
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02008556
8557
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008558option smtpchk
8559option smtpchk <hello> <domain>
8560 Use SMTP health checks for server testing
8561 May be used in sections : defaults | frontend | listen | backend
8562 yes | no | yes | yes
Willy Tarreaud72758d2010-01-12 10:42:19 +01008563 Arguments :
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008564 <hello> is an optional argument. It is the "hello" command to use. It can
Lukas Tribus27935782018-10-01 02:00:16 +02008565 be either "HELO" (for SMTP) or "EHLO" (for ESMTP). All other
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008566 values will be turned into the default command ("HELO").
8567
8568 <domain> is the domain name to present to the server. It may only be
8569 specified (and is mandatory) if the hello command has been
8570 specified. By default, "localhost" is used.
8571
8572 When "option smtpchk" is set, the health checks will consist in TCP
8573 connections followed by an SMTP command. By default, this command is
8574 "HELO localhost". The server's return code is analyzed and only return codes
8575 starting with a "2" will be considered as valid. All other responses,
8576 including a lack of response will constitute an error and will indicate a
8577 dead server.
8578
8579 This test is meant to be used with SMTP servers or relays. Depending on the
8580 request, it is possible that some servers do not log each connection attempt,
Davor Ocelice9ed2812017-12-25 17:49:28 +01008581 so you may want to experiment to improve the behavior. Using telnet on port
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008582 25 is often easier than adjusting the configuration.
8583
8584 Most often, an incoming SMTP server needs to see the client's IP address for
8585 various purposes, including spam filtering, anti-spoofing and logging. When
8586 possible, it is often wise to masquerade the client's IP address when
8587 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02008588 which requires the transparent proxy feature to be compiled in.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008589
8590 Example :
8591 option smtpchk HELO mydomain.org
8592
8593 See also : "option httpchk", "source"
8594
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008595
Krzysztof Piotr Oledzkiaeebf9b2009-10-04 15:43:17 +02008596option socket-stats
8597no option socket-stats
8598
8599 Enable or disable collecting & providing separate statistics for each socket.
8600 May be used in sections : defaults | frontend | listen | backend
8601 yes | yes | yes | no
8602
8603 Arguments : none
8604
8605
Willy Tarreauff4f82d2009-02-06 11:28:13 +01008606option splice-auto
8607no option splice-auto
8608 Enable or disable automatic kernel acceleration on sockets in both directions
8609 May be used in sections : defaults | frontend | listen | backend
8610 yes | yes | yes | yes
8611 Arguments : none
8612
8613 When this option is enabled either on a frontend or on a backend, haproxy
8614 will automatically evaluate the opportunity to use kernel tcp splicing to
Davor Ocelice9ed2812017-12-25 17:49:28 +01008615 forward data between the client and the server, in either direction. HAProxy
Willy Tarreauff4f82d2009-02-06 11:28:13 +01008616 uses heuristics to estimate if kernel splicing might improve performance or
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01008617 not. Both directions are handled independently. Note that the heuristics used
Willy Tarreauff4f82d2009-02-06 11:28:13 +01008618 are not much aggressive in order to limit excessive use of splicing. This
8619 option requires splicing to be enabled at compile time, and may be globally
8620 disabled with the global option "nosplice". Since splice uses pipes, using it
8621 requires that there are enough spare pipes.
8622
8623 Important note: kernel-based TCP splicing is a Linux-specific feature which
8624 first appeared in kernel 2.6.25. It offers kernel-based acceleration to
8625 transfer data between sockets without copying these data to user-space, thus
8626 providing noticeable performance gains and CPU cycles savings. Since many
8627 early implementations are buggy, corrupt data and/or are inefficient, this
8628 feature is not enabled by default, and it should be used with extreme care.
8629 While it is not possible to detect the correctness of an implementation,
8630 2.6.29 is the first version offering a properly working implementation. In
8631 case of doubt, splicing may be globally disabled using the global "nosplice"
8632 keyword.
8633
8634 Example :
8635 option splice-auto
8636
8637 If this option has been enabled in a "defaults" section, it can be disabled
8638 in a specific instance by prepending the "no" keyword before it.
8639
8640 See also : "option splice-request", "option splice-response", and global
8641 options "nosplice" and "maxpipes"
8642
8643
8644option splice-request
8645no option splice-request
8646 Enable or disable automatic kernel acceleration on sockets for requests
8647 May be used in sections : defaults | frontend | listen | backend
8648 yes | yes | yes | yes
8649 Arguments : none
8650
8651 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04008652 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01008653 the client to the server. It might still use the recv/send scheme if there
8654 are no spare pipes left. This option requires splicing to be enabled at
8655 compile time, and may be globally disabled with the global option "nosplice".
8656 Since splice uses pipes, using it requires that there are enough spare pipes.
8657
8658 Important note: see "option splice-auto" for usage limitations.
8659
8660 Example :
8661 option splice-request
8662
8663 If this option has been enabled in a "defaults" section, it can be disabled
8664 in a specific instance by prepending the "no" keyword before it.
8665
8666 See also : "option splice-auto", "option splice-response", and global options
8667 "nosplice" and "maxpipes"
8668
8669
8670option splice-response
8671no option splice-response
8672 Enable or disable automatic kernel acceleration on sockets for responses
8673 May be used in sections : defaults | frontend | listen | backend
8674 yes | yes | yes | yes
8675 Arguments : none
8676
8677 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04008678 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01008679 the server to the client. It might still use the recv/send scheme if there
8680 are no spare pipes left. This option requires splicing to be enabled at
8681 compile time, and may be globally disabled with the global option "nosplice".
8682 Since splice uses pipes, using it requires that there are enough spare pipes.
8683
8684 Important note: see "option splice-auto" for usage limitations.
8685
8686 Example :
8687 option splice-response
8688
8689 If this option has been enabled in a "defaults" section, it can be disabled
8690 in a specific instance by prepending the "no" keyword before it.
8691
8692 See also : "option splice-auto", "option splice-request", and global options
8693 "nosplice" and "maxpipes"
8694
8695
Christopher Fauletba7bc162016-11-07 21:07:38 +01008696option spop-check
8697 Use SPOP health checks for server testing
8698 May be used in sections : defaults | frontend | listen | backend
8699 no | no | no | yes
8700 Arguments : none
8701
8702 It is possible to test that the server correctly talks SPOP protocol instead
8703 of just testing that it accepts the TCP connection. When this option is set,
8704 a HELLO handshake is performed between HAProxy and the server, and the
8705 response is analyzed to check no error is reported.
8706
8707 Example :
8708 option spop-check
8709
8710 See also : "option httpchk"
8711
8712
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008713option srvtcpka
8714no option srvtcpka
8715 Enable or disable the sending of TCP keepalive packets on the server side
8716 May be used in sections : defaults | frontend | listen | backend
8717 yes | no | yes | yes
8718 Arguments : none
8719
8720 When there is a firewall or any session-aware component between a client and
8721 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01008722 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008723 components decides to expire a session which has remained idle for too long.
8724
8725 Enabling socket-level TCP keep-alives makes the system regularly send packets
8726 to the other end of the connection, leaving it active. The delay between
8727 keep-alive probes is controlled by the system only and depends both on the
8728 operating system and its tuning parameters.
8729
8730 It is important to understand that keep-alive packets are neither emitted nor
8731 received at the application level. It is only the network stacks which sees
8732 them. For this reason, even if one side of the proxy already uses keep-alives
8733 to maintain its connection alive, those keep-alive packets will not be
8734 forwarded to the other side of the proxy.
8735
8736 Please note that this has nothing to do with HTTP keep-alive.
8737
8738 Using option "srvtcpka" enables the emission of TCP keep-alive probes on the
8739 server side of a connection, which should help when session expirations are
8740 noticed between HAProxy and a server.
8741
8742 If this option has been enabled in a "defaults" section, it can be disabled
8743 in a specific instance by prepending the "no" keyword before it.
8744
8745 See also : "option clitcpka", "option tcpka"
8746
8747
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008748option ssl-hello-chk
8749 Use SSLv3 client hello health checks for server testing
8750 May be used in sections : defaults | frontend | listen | backend
8751 yes | no | yes | yes
8752 Arguments : none
8753
8754 When some SSL-based protocols are relayed in TCP mode through HAProxy, it is
8755 possible to test that the server correctly talks SSL instead of just testing
8756 that it accepts the TCP connection. When "option ssl-hello-chk" is set, pure
8757 SSLv3 client hello messages are sent once the connection is established to
8758 the server, and the response is analyzed to find an SSL server hello message.
8759 The server is considered valid only when the response contains this server
8760 hello message.
8761
8762 All servers tested till there correctly reply to SSLv3 client hello messages,
8763 and most servers tested do not even log the requests containing only hello
8764 messages, which is appreciable.
8765
Willy Tarreau763a95b2012-10-04 23:15:39 +02008766 Note that this check works even when SSL support was not built into haproxy
8767 because it forges the SSL message. When SSL support is available, it is best
8768 to use native SSL health checks instead of this one.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008769
Willy Tarreau763a95b2012-10-04 23:15:39 +02008770 See also: "option httpchk", "check-ssl"
8771
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008772
Willy Tarreaued179852013-12-16 01:07:00 +01008773option tcp-check
8774 Perform health checks using tcp-check send/expect sequences
8775 May be used in sections: defaults | frontend | listen | backend
8776 yes | no | yes | yes
8777
8778 This health check method is intended to be combined with "tcp-check" command
8779 lists in order to support send/expect types of health check sequences.
8780
8781 TCP checks currently support 4 modes of operations :
8782 - no "tcp-check" directive : the health check only consists in a connection
8783 attempt, which remains the default mode.
8784
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008785 - "tcp-check send" or "tcp-check send-binary" only is mentioned : this is
Willy Tarreaued179852013-12-16 01:07:00 +01008786 used to send a string along with a connection opening. With some
8787 protocols, it helps sending a "QUIT" message for example that prevents
8788 the server from logging a connection error for each health check. The
8789 check result will still be based on the ability to open the connection
8790 only.
8791
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008792 - "tcp-check expect" only is mentioned : this is used to test a banner.
Willy Tarreaued179852013-12-16 01:07:00 +01008793 The connection is opened and haproxy waits for the server to present some
8794 contents which must validate some rules. The check result will be based
8795 on the matching between the contents and the rules. This is suited for
8796 POP, IMAP, SMTP, FTP, SSH, TELNET.
8797
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008798 - both "tcp-check send" and "tcp-check expect" are mentioned : this is
Davor Ocelice9ed2812017-12-25 17:49:28 +01008799 used to test a hello-type protocol. HAProxy sends a message, the server
8800 responds and its response is analyzed. the check result will be based on
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008801 the matching between the response contents and the rules. This is often
Willy Tarreaued179852013-12-16 01:07:00 +01008802 suited for protocols which require a binding or a request/response model.
8803 LDAP, MySQL, Redis and SSL are example of such protocols, though they
8804 already all have their dedicated checks with a deeper understanding of
8805 the respective protocols.
8806 In this mode, many questions may be sent and many answers may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01008807 analyzed.
Willy Tarreaued179852013-12-16 01:07:00 +01008808
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02008809 A fifth mode can be used to insert comments in different steps of the script.
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02008810
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02008811 For each tcp-check rule you create, you can add a "comment" directive,
8812 followed by a string. This string will be reported in the log and stderr in
8813 debug mode. It is useful to make user-friendly error reporting. The
8814 "comment" is of course optional.
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02008815
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02008816 During the execution of a health check, a variable scope is made available to
8817 store data samples, using the "tcp-check set-var" operation. Freeing those
8818 variable is possible using "tcp-check unset-var".
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +01008819
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02008820
Willy Tarreaued179852013-12-16 01:07:00 +01008821 Examples :
Davor Ocelice9ed2812017-12-25 17:49:28 +01008822 # perform a POP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01008823 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02008824 tcp-check expect string +OK\ POP3\ ready comment POP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01008825
Davor Ocelice9ed2812017-12-25 17:49:28 +01008826 # perform an IMAP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01008827 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02008828 tcp-check expect string *\ OK\ IMAP4\ ready comment IMAP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01008829
8830 # look for the redis master server after ensuring it speaks well
8831 # redis protocol, then it exits properly.
Davor Ocelice9ed2812017-12-25 17:49:28 +01008832 # (send a command then analyze the response 3 times)
Willy Tarreaued179852013-12-16 01:07:00 +01008833 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02008834 tcp-check comment PING\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01008835 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02008836 tcp-check expect string +PONG
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02008837 tcp-check comment role\ check
Willy Tarreaued179852013-12-16 01:07:00 +01008838 tcp-check send info\ replication\r\n
8839 tcp-check expect string role:master
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02008840 tcp-check comment QUIT\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01008841 tcp-check send QUIT\r\n
8842 tcp-check expect string +OK
8843
Davor Ocelice9ed2812017-12-25 17:49:28 +01008844 forge a HTTP request, then analyze the response
Willy Tarreaued179852013-12-16 01:07:00 +01008845 (send many headers before analyzing)
8846 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02008847 tcp-check comment forge\ and\ send\ HTTP\ request
Willy Tarreaued179852013-12-16 01:07:00 +01008848 tcp-check send HEAD\ /\ HTTP/1.1\r\n
8849 tcp-check send Host:\ www.mydomain.com\r\n
8850 tcp-check send User-Agent:\ HAProxy\ tcpcheck\r\n
8851 tcp-check send \r\n
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02008852 tcp-check expect rstring HTTP/1\..\ (2..|3..) comment check\ HTTP\ response
Willy Tarreaued179852013-12-16 01:07:00 +01008853
8854
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02008855 See also : "tcp-check connect", "tcp-check expect" and "tcp-check send".
Willy Tarreaued179852013-12-16 01:07:00 +01008856
8857
Willy Tarreau9ea05a72009-06-14 12:07:01 +02008858option tcp-smart-accept
8859no option tcp-smart-accept
8860 Enable or disable the saving of one ACK packet during the accept sequence
8861 May be used in sections : defaults | frontend | listen | backend
8862 yes | yes | yes | no
8863 Arguments : none
8864
8865 When an HTTP connection request comes in, the system acknowledges it on
8866 behalf of HAProxy, then the client immediately sends its request, and the
8867 system acknowledges it too while it is notifying HAProxy about the new
8868 connection. HAProxy then reads the request and responds. This means that we
8869 have one TCP ACK sent by the system for nothing, because the request could
8870 very well be acknowledged by HAProxy when it sends its response.
8871
8872 For this reason, in HTTP mode, HAProxy automatically asks the system to avoid
8873 sending this useless ACK on platforms which support it (currently at least
8874 Linux). It must not cause any problem, because the system will send it anyway
8875 after 40 ms if the response takes more time than expected to come.
8876
8877 During complex network debugging sessions, it may be desirable to disable
8878 this optimization because delayed ACKs can make troubleshooting more complex
8879 when trying to identify where packets are delayed. It is then possible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01008880 fall back to normal behavior by specifying "no option tcp-smart-accept".
Willy Tarreau9ea05a72009-06-14 12:07:01 +02008881
8882 It is also possible to force it for non-HTTP proxies by simply specifying
8883 "option tcp-smart-accept". For instance, it can make sense with some services
8884 such as SMTP where the server speaks first.
8885
8886 It is recommended to avoid forcing this option in a defaults section. In case
8887 of doubt, consider setting it back to automatic values by prepending the
8888 "default" keyword before it, or disabling it using the "no" keyword.
8889
Willy Tarreaud88edf22009-06-14 15:48:17 +02008890 See also : "option tcp-smart-connect"
8891
8892
8893option tcp-smart-connect
8894no option tcp-smart-connect
8895 Enable or disable the saving of one ACK packet during the connect sequence
8896 May be used in sections : defaults | frontend | listen | backend
8897 yes | no | yes | yes
8898 Arguments : none
8899
8900 On certain systems (at least Linux), HAProxy can ask the kernel not to
8901 immediately send an empty ACK upon a connection request, but to directly
8902 send the buffer request instead. This saves one packet on the network and
8903 thus boosts performance. It can also be useful for some servers, because they
8904 immediately get the request along with the incoming connection.
8905
8906 This feature is enabled when "option tcp-smart-connect" is set in a backend.
8907 It is not enabled by default because it makes network troubleshooting more
8908 complex.
8909
8910 It only makes sense to enable it with protocols where the client speaks first
8911 such as HTTP. In other situations, if there is no data to send in place of
8912 the ACK, a normal ACK is sent.
8913
8914 If this option has been enabled in a "defaults" section, it can be disabled
8915 in a specific instance by prepending the "no" keyword before it.
8916
8917 See also : "option tcp-smart-accept"
8918
Willy Tarreau9ea05a72009-06-14 12:07:01 +02008919
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008920option tcpka
8921 Enable or disable the sending of TCP keepalive packets on both sides
8922 May be used in sections : defaults | frontend | listen | backend
8923 yes | yes | yes | yes
8924 Arguments : none
8925
8926 When there is a firewall or any session-aware component between a client and
8927 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01008928 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008929 components decides to expire a session which has remained idle for too long.
8930
8931 Enabling socket-level TCP keep-alives makes the system regularly send packets
8932 to the other end of the connection, leaving it active. The delay between
8933 keep-alive probes is controlled by the system only and depends both on the
8934 operating system and its tuning parameters.
8935
8936 It is important to understand that keep-alive packets are neither emitted nor
8937 received at the application level. It is only the network stacks which sees
8938 them. For this reason, even if one side of the proxy already uses keep-alives
8939 to maintain its connection alive, those keep-alive packets will not be
8940 forwarded to the other side of the proxy.
8941
8942 Please note that this has nothing to do with HTTP keep-alive.
8943
8944 Using option "tcpka" enables the emission of TCP keep-alive probes on both
8945 the client and server sides of a connection. Note that this is meaningful
8946 only in "defaults" or "listen" sections. If this option is used in a
8947 frontend, only the client side will get keep-alives, and if this option is
8948 used in a backend, only the server side will get keep-alives. For this
8949 reason, it is strongly recommended to explicitly use "option clitcpka" and
8950 "option srvtcpka" when the configuration is split between frontends and
8951 backends.
8952
8953 See also : "option clitcpka", "option srvtcpka"
8954
Willy Tarreau844e3c52008-01-11 16:28:18 +01008955
8956option tcplog
8957 Enable advanced logging of TCP connections with session state and timers
8958 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01008959 yes | yes | yes | no
Willy Tarreau844e3c52008-01-11 16:28:18 +01008960 Arguments : none
8961
8962 By default, the log output format is very poor, as it only contains the
8963 source and destination addresses, and the instance name. By specifying
8964 "option tcplog", each log line turns into a much richer format including, but
8965 not limited to, the connection timers, the session status, the connections
8966 numbers, the frontend, backend and server name, and of course the source
8967 address and ports. This option is useful for pure TCP proxies in order to
8968 find which of the client or server disconnects or times out. For normal HTTP
8969 proxies, it's better to use "option httplog" which is even more complete.
8970
Guillaume de Lafond29f45602017-03-31 19:52:15 +02008971 "option tcplog" overrides any previous "log-format" directive.
8972
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008973 See also : "option httplog", and section 8 about logging.
Willy Tarreau844e3c52008-01-11 16:28:18 +01008974
8975
Willy Tarreau844e3c52008-01-11 16:28:18 +01008976option transparent
8977no option transparent
8978 Enable client-side transparent proxying
8979 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +01008980 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +01008981 Arguments : none
8982
8983 This option was introduced in order to provide layer 7 persistence to layer 3
8984 load balancers. The idea is to use the OS's ability to redirect an incoming
8985 connection for a remote address to a local process (here HAProxy), and let
8986 this process know what address was initially requested. When this option is
8987 used, sessions without cookies will be forwarded to the original destination
8988 IP address of the incoming request (which should match that of another
8989 equipment), while requests with cookies will still be forwarded to the
8990 appropriate server.
8991
8992 Note that contrary to a common belief, this option does NOT make HAProxy
8993 present the client's IP to the server when establishing the connection.
8994
Willy Tarreaua1146052011-03-01 09:51:54 +01008995 See also: the "usesrc" argument of the "source" keyword, and the
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008996 "transparent" option of the "bind" keyword.
Willy Tarreau844e3c52008-01-11 16:28:18 +01008997
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008998
Simon Horman98637e52014-06-20 12:30:16 +09008999external-check command <command>
9000 Executable to run when performing an external-check
9001 May be used in sections : defaults | frontend | listen | backend
9002 yes | no | yes | yes
9003
9004 Arguments :
9005 <command> is the external command to run
9006
Simon Horman98637e52014-06-20 12:30:16 +09009007 The arguments passed to the to the command are:
9008
Cyril Bonté777be862014-12-02 21:21:35 +01009009 <proxy_address> <proxy_port> <server_address> <server_port>
Simon Horman98637e52014-06-20 12:30:16 +09009010
Cyril Bonté777be862014-12-02 21:21:35 +01009011 The <proxy_address> and <proxy_port> are derived from the first listener
9012 that is either IPv4, IPv6 or a UNIX socket. In the case of a UNIX socket
9013 listener the proxy_address will be the path of the socket and the
9014 <proxy_port> will be the string "NOT_USED". In a backend section, it's not
9015 possible to determine a listener, and both <proxy_address> and <proxy_port>
9016 will have the string value "NOT_USED".
Simon Horman98637e52014-06-20 12:30:16 +09009017
Cyril Bonté72cda2a2014-12-27 22:28:39 +01009018 Some values are also provided through environment variables.
9019
9020 Environment variables :
9021 HAPROXY_PROXY_ADDR The first bind address if available (or empty if not
9022 applicable, for example in a "backend" section).
9023
9024 HAPROXY_PROXY_ID The backend id.
9025
9026 HAPROXY_PROXY_NAME The backend name.
9027
9028 HAPROXY_PROXY_PORT The first bind port if available (or empty if not
9029 applicable, for example in a "backend" section or
9030 for a UNIX socket).
9031
9032 HAPROXY_SERVER_ADDR The server address.
9033
9034 HAPROXY_SERVER_CURCONN The current number of connections on the server.
9035
9036 HAPROXY_SERVER_ID The server id.
9037
9038 HAPROXY_SERVER_MAXCONN The server max connections.
9039
9040 HAPROXY_SERVER_NAME The server name.
9041
9042 HAPROXY_SERVER_PORT The server port if available (or empty for a UNIX
9043 socket).
9044
9045 PATH The PATH environment variable used when executing
9046 the command may be set using "external-check path".
9047
William Lallemand4d03e432019-06-14 15:35:37 +02009048 See also "2.3. Environment variables" for other variables.
9049
Simon Horman98637e52014-06-20 12:30:16 +09009050 If the command executed and exits with a zero status then the check is
9051 considered to have passed, otherwise the check is considered to have
9052 failed.
9053
9054 Example :
9055 external-check command /bin/true
9056
9057 See also : "external-check", "option external-check", "external-check path"
9058
9059
9060external-check path <path>
9061 The value of the PATH environment variable used when running an external-check
9062 May be used in sections : defaults | frontend | listen | backend
9063 yes | no | yes | yes
9064
9065 Arguments :
9066 <path> is the path used when executing external command to run
9067
9068 The default path is "".
9069
9070 Example :
9071 external-check path "/usr/bin:/bin"
9072
9073 See also : "external-check", "option external-check",
9074 "external-check command"
9075
9076
Emeric Brun647caf12009-06-30 17:57:00 +02009077persist rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02009078persist rdp-cookie(<name>)
Emeric Brun647caf12009-06-30 17:57:00 +02009079 Enable RDP cookie-based persistence
9080 May be used in sections : defaults | frontend | listen | backend
9081 yes | no | yes | yes
9082 Arguments :
9083 <name> is the optional name of the RDP cookie to check. If omitted, the
Willy Tarreau61e28f22010-05-16 22:31:05 +02009084 default cookie name "msts" will be used. There currently is no
9085 valid reason to change this name.
Emeric Brun647caf12009-06-30 17:57:00 +02009086
9087 This statement enables persistence based on an RDP cookie. The RDP cookie
9088 contains all information required to find the server in the list of known
Davor Ocelice9ed2812017-12-25 17:49:28 +01009089 servers. So when this option is set in the backend, the request is analyzed
Emeric Brun647caf12009-06-30 17:57:00 +02009090 and if an RDP cookie is found, it is decoded. If it matches a known server
9091 which is still UP (or if "option persist" is set), then the connection is
9092 forwarded to this server.
9093
9094 Note that this only makes sense in a TCP backend, but for this to work, the
9095 frontend must have waited long enough to ensure that an RDP cookie is present
9096 in the request buffer. This is the same requirement as with the "rdp-cookie"
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009097 load-balancing method. Thus it is highly recommended to put all statements in
Emeric Brun647caf12009-06-30 17:57:00 +02009098 a single "listen" section.
9099
Willy Tarreau61e28f22010-05-16 22:31:05 +02009100 Also, it is important to understand that the terminal server will emit this
9101 RDP cookie only if it is configured for "token redirection mode", which means
9102 that the "IP address redirection" option is disabled.
9103
Emeric Brun647caf12009-06-30 17:57:00 +02009104 Example :
9105 listen tse-farm
9106 bind :3389
9107 # wait up to 5s for an RDP cookie in the request
9108 tcp-request inspect-delay 5s
9109 tcp-request content accept if RDP_COOKIE
9110 # apply RDP cookie persistence
9111 persist rdp-cookie
9112 # if server is unknown, let's balance on the same cookie.
Cyril Bontédc4d9032012-04-08 21:57:39 +02009113 # alternatively, "balance leastconn" may be useful too.
Emeric Brun647caf12009-06-30 17:57:00 +02009114 balance rdp-cookie
9115 server srv1 1.1.1.1:3389
9116 server srv2 1.1.1.2:3389
9117
Simon Hormanab814e02011-06-24 14:50:20 +09009118 See also : "balance rdp-cookie", "tcp-request", the "req_rdp_cookie" ACL and
9119 the rdp_cookie pattern fetch function.
Emeric Brun647caf12009-06-30 17:57:00 +02009120
9121
Willy Tarreau3a7d2072009-03-05 23:48:25 +01009122rate-limit sessions <rate>
9123 Set a limit on the number of new sessions accepted per second on a frontend
9124 May be used in sections : defaults | frontend | listen | backend
9125 yes | yes | yes | no
9126 Arguments :
9127 <rate> The <rate> parameter is an integer designating the maximum number
9128 of new sessions per second to accept on the frontend.
9129
9130 When the frontend reaches the specified number of new sessions per second, it
9131 stops accepting new connections until the rate drops below the limit again.
9132 During this time, the pending sessions will be kept in the socket's backlog
9133 (in system buffers) and haproxy will not even be aware that sessions are
9134 pending. When applying very low limit on a highly loaded service, it may make
9135 sense to increase the socket's backlog using the "backlog" keyword.
9136
9137 This feature is particularly efficient at blocking connection-based attacks
9138 or service abuse on fragile servers. Since the session rate is measured every
9139 millisecond, it is extremely accurate. Also, the limit applies immediately,
9140 no delay is needed at all to detect the threshold.
9141
9142 Example : limit the connection rate on SMTP to 10 per second max
9143 listen smtp
9144 mode tcp
9145 bind :25
9146 rate-limit sessions 10
Panagiotis Panagiotopoulos7282d8e2016-02-11 16:37:15 +02009147 server smtp1 127.0.0.1:1025
Willy Tarreau3a7d2072009-03-05 23:48:25 +01009148
Willy Tarreaua17c2d92011-07-25 08:16:20 +02009149 Note : when the maximum rate is reached, the frontend's status is not changed
9150 but its sockets appear as "WAITING" in the statistics if the
9151 "socket-stats" option is enabled.
Willy Tarreau3a7d2072009-03-05 23:48:25 +01009152
9153 See also : the "backlog" keyword and the "fe_sess_rate" ACL criterion.
9154
9155
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009156redirect location <loc> [code <code>] <option> [{if | unless} <condition>]
9157redirect prefix <pfx> [code <code>] <option> [{if | unless} <condition>]
9158redirect scheme <sch> [code <code>] <option> [{if | unless} <condition>]
Willy Tarreaub463dfb2008-06-07 23:08:56 +02009159 Return an HTTP redirection if/unless a condition is matched
9160 May be used in sections : defaults | frontend | listen | backend
9161 no | yes | yes | yes
9162
9163 If/unless the condition is matched, the HTTP request will lead to a redirect
Willy Tarreauf285f542010-01-03 20:03:03 +01009164 response. If no condition is specified, the redirect applies unconditionally.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02009165
Willy Tarreau0140f252008-11-19 21:07:09 +01009166 Arguments :
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009167 <loc> With "redirect location", the exact value in <loc> is placed into
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01009168 the HTTP "Location" header. When used in an "http-request" rule,
9169 <loc> value follows the log-format rules and can include some
9170 dynamic values (see Custom Log Format in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009171
9172 <pfx> With "redirect prefix", the "Location" header is built from the
9173 concatenation of <pfx> and the complete URI path, including the
9174 query string, unless the "drop-query" option is specified (see
9175 below). As a special case, if <pfx> equals exactly "/", then
9176 nothing is inserted before the original URI. It allows one to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01009177 redirect to the same URL (for instance, to insert a cookie). When
9178 used in an "http-request" rule, <pfx> value follows the log-format
9179 rules and can include some dynamic values (see Custom Log Format
9180 in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009181
9182 <sch> With "redirect scheme", then the "Location" header is built by
9183 concatenating <sch> with "://" then the first occurrence of the
9184 "Host" header, and then the URI path, including the query string
9185 unless the "drop-query" option is specified (see below). If no
9186 path is found or if the path is "*", then "/" is used instead. If
9187 no "Host" header is found, then an empty host component will be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009188 returned, which most recent browsers interpret as redirecting to
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009189 the same host. This directive is mostly used to redirect HTTP to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01009190 HTTPS. When used in an "http-request" rule, <sch> value follows
9191 the log-format rules and can include some dynamic values (see
9192 Custom Log Format in section 8.2.4).
Willy Tarreau0140f252008-11-19 21:07:09 +01009193
9194 <code> The code is optional. It indicates which type of HTTP redirection
Willy Tarreaub67fdc42013-03-29 19:28:11 +01009195 is desired. Only codes 301, 302, 303, 307 and 308 are supported,
9196 with 302 used by default if no code is specified. 301 means
9197 "Moved permanently", and a browser may cache the Location. 302
Baptiste Assmannea849c02015-08-03 11:42:50 +02009198 means "Moved temporarily" and means that the browser should not
Willy Tarreaub67fdc42013-03-29 19:28:11 +01009199 cache the redirection. 303 is equivalent to 302 except that the
9200 browser will fetch the location with a GET method. 307 is just
9201 like 302 but makes it clear that the same method must be reused.
9202 Likewise, 308 replaces 301 if the same method must be used.
Willy Tarreau0140f252008-11-19 21:07:09 +01009203
9204 <option> There are several options which can be specified to adjust the
Davor Ocelice9ed2812017-12-25 17:49:28 +01009205 expected behavior of a redirection :
Willy Tarreau0140f252008-11-19 21:07:09 +01009206
9207 - "drop-query"
9208 When this keyword is used in a prefix-based redirection, then the
9209 location will be set without any possible query-string, which is useful
9210 for directing users to a non-secure page for instance. It has no effect
9211 with a location-type redirect.
9212
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01009213 - "append-slash"
9214 This keyword may be used in conjunction with "drop-query" to redirect
9215 users who use a URL not ending with a '/' to the same one with the '/'.
9216 It can be useful to ensure that search engines will only see one URL.
9217 For this, a return code 301 is preferred.
9218
Willy Tarreau0140f252008-11-19 21:07:09 +01009219 - "set-cookie NAME[=value]"
9220 A "Set-Cookie" header will be added with NAME (and optionally "=value")
9221 to the response. This is sometimes used to indicate that a user has
9222 been seen, for instance to protect against some types of DoS. No other
9223 cookie option is added, so the cookie will be a session cookie. Note
9224 that for a browser, a sole cookie name without an equal sign is
9225 different from a cookie with an equal sign.
9226
9227 - "clear-cookie NAME[=]"
9228 A "Set-Cookie" header will be added with NAME (and optionally "="), but
9229 with the "Max-Age" attribute set to zero. This will tell the browser to
9230 delete this cookie. It is useful for instance on logout pages. It is
9231 important to note that clearing the cookie "NAME" will not remove a
9232 cookie set with "NAME=value". You have to clear the cookie "NAME=" for
9233 that, because the browser makes the difference.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02009234
9235 Example: move the login URL only to HTTPS.
9236 acl clear dst_port 80
9237 acl secure dst_port 8080
9238 acl login_page url_beg /login
Willy Tarreau0140f252008-11-19 21:07:09 +01009239 acl logout url_beg /logout
Willy Tarreau79da4692008-11-19 20:03:04 +01009240 acl uid_given url_reg /login?userid=[^&]+
Willy Tarreau0140f252008-11-19 21:07:09 +01009241 acl cookie_set hdr_sub(cookie) SEEN=1
9242
9243 redirect prefix https://mysite.com set-cookie SEEN=1 if !cookie_set
Willy Tarreau79da4692008-11-19 20:03:04 +01009244 redirect prefix https://mysite.com if login_page !secure
9245 redirect prefix http://mysite.com drop-query if login_page !uid_given
9246 redirect location http://mysite.com/ if !login_page secure
Willy Tarreau0140f252008-11-19 21:07:09 +01009247 redirect location / clear-cookie USERID= if logout
Willy Tarreaub463dfb2008-06-07 23:08:56 +02009248
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01009249 Example: send redirects for request for articles without a '/'.
9250 acl missing_slash path_reg ^/article/[^/]*$
9251 redirect code 301 prefix / drop-query append-slash if missing_slash
9252
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009253 Example: redirect all HTTP traffic to HTTPS when SSL is handled by haproxy.
David BERARDe7153042012-11-03 00:11:31 +01009254 redirect scheme https if !{ ssl_fc }
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009255
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01009256 Example: append 'www.' prefix in front of all hosts not having it
Coen Rosdorff596659b2016-04-11 11:33:49 +02009257 http-request redirect code 301 location \
9258 http://www.%[hdr(host)]%[capture.req.uri] \
9259 unless { hdr_beg(host) -i www }
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01009260
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009261 See section 7 about ACL usage.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02009262
Willy Tarreau303c0352008-01-17 19:01:39 +01009263
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02009264retries <value>
9265 Set the number of retries to perform on a server after a connection failure
9266 May be used in sections: defaults | frontend | listen | backend
9267 yes | no | yes | yes
9268 Arguments :
9269 <value> is the number of times a connection attempt should be retried on
9270 a server when a connection either is refused or times out. The
9271 default value is 3.
9272
9273 It is important to understand that this value applies to the number of
9274 connection attempts, not full requests. When a connection has effectively
9275 been established to a server, there will be no more retry.
9276
9277 In order to avoid immediate reconnections to a server which is restarting,
Joseph Lynch726ab712015-05-11 23:25:34 -07009278 a turn-around timer of min("timeout connect", one second) is applied before
9279 a retry occurs.
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02009280
9281 When "option redispatch" is set, the last retry may be performed on another
9282 server even if a cookie references a different server.
9283
9284 See also : "option redispatch"
9285
9286
Olivier Houcharda254a372019-04-05 15:30:12 +02009287retry-on [list of keywords]
Jerome Magnin5ce3c142020-05-13 20:09:57 +02009288 Specify when to attempt to automatically retry a failed request.
9289 This setting is only valid when "mode" is set to http and is silently ignored
9290 otherwise.
Olivier Houcharda254a372019-04-05 15:30:12 +02009291 May be used in sections: defaults | frontend | listen | backend
9292 yes | no | yes | yes
9293 Arguments :
9294 <keywords> is a list of keywords or HTTP status codes, each representing a
9295 type of failure event on which an attempt to retry the request
9296 is desired. Please read the notes at the bottom before changing
9297 this setting. The following keywords are supported :
9298
9299 none never retry
9300
9301 conn-failure retry when the connection or the SSL handshake failed
9302 and the request could not be sent. This is the default.
9303
9304 empty-response retry when the server connection was closed after part
9305 of the request was sent, and nothing was received from
9306 the server. This type of failure may be caused by the
9307 request timeout on the server side, poor network
9308 condition, or a server crash or restart while
9309 processing the request.
9310
Olivier Houcharde3249a92019-05-03 23:01:47 +02009311 junk-response retry when the server returned something not looking
9312 like a complete HTTP response. This includes partial
9313 responses headers as well as non-HTTP contents. It
9314 usually is a bad idea to retry on such events, which
9315 may be caused a configuration issue (wrong server port)
9316 or by the request being harmful to the server (buffer
9317 overflow attack for example).
9318
Olivier Houcharda254a372019-04-05 15:30:12 +02009319 response-timeout the server timeout stroke while waiting for the server
9320 to respond to the request. This may be caused by poor
9321 network condition, the reuse of an idle connection
9322 which has expired on the path, or by the request being
9323 extremely expensive to process. It generally is a bad
9324 idea to retry on such events on servers dealing with
9325 heavy database processing (full scans, etc) as it may
9326 amplify denial of service attacks.
9327
Olivier Houchard865d8392019-05-03 22:46:27 +02009328 0rtt-rejected retry requests which were sent over early data and were
9329 rejected by the server. These requests are generally
9330 considered to be safe to retry.
9331
Olivier Houcharda254a372019-04-05 15:30:12 +02009332 <status> any HTTP status code among "404" (Not Found), "408"
9333 (Request Timeout), "425" (Too Early), "500" (Server
9334 Error), "501" (Not Implemented), "502" (Bad Gateway),
9335 "503" (Service Unavailable), "504" (Gateway Timeout).
9336
Olivier Houchardddf0e032019-05-10 18:05:40 +02009337 all-retryable-errors
9338 retry request for any error that are considered
9339 retryable. This currently activates "conn-failure",
9340 "empty-response", "junk-response", "response-timeout",
9341 "0rtt-rejected", "500", "502", "503", and "504".
9342
Olivier Houcharda254a372019-04-05 15:30:12 +02009343 Using this directive replaces any previous settings with the new ones; it is
9344 not cumulative.
9345
9346 Please note that using anything other than "none" and "conn-failure" requires
9347 to allocate a buffer and copy the whole request into it, so it has memory and
9348 performance impacts. Requests not fitting in a single buffer will never be
9349 retried (see the global tune.bufsize setting).
9350
9351 You have to make sure the application has a replay protection mechanism built
9352 in such as a unique transaction IDs passed in requests, or that replaying the
9353 same request has no consequence, or it is very dangerous to use any retry-on
9354 value beside "conn-failure" and "none". Static file servers and caches are
9355 generally considered safe against any type of retry. Using a status code can
9356 be useful to quickly leave a server showing an abnormal behavior (out of
9357 memory, file system issues, etc), but in this case it may be a good idea to
9358 immediately redispatch the connection to another server (please see "option
9359 redispatch" for this). Last, it is important to understand that most causes
9360 of failures are the requests themselves and that retrying a request causing a
9361 server to misbehave will often make the situation even worse for this server,
9362 or for the whole service in case of redispatch.
9363
9364 Unless you know exactly how the application deals with replayed requests, you
9365 should not use this directive.
9366
9367 The default is "conn-failure".
9368
9369 See also: "retries", "option redispatch", "tune.bufsize"
9370
David du Colombier486df472011-03-17 10:40:26 +01009371server <name> <address>[:[port]] [param*]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009372 Declare a server in a backend
9373 May be used in sections : defaults | frontend | listen | backend
9374 no | no | yes | yes
9375 Arguments :
9376 <name> is the internal name assigned to this server. This name will
Davor Ocelice9ed2812017-12-25 17:49:28 +01009377 appear in logs and alerts. If "http-send-name-header" is
Mark Lamourinec2247f02012-01-04 13:02:01 -05009378 set, it will be added to the request header sent to the server.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009379
David du Colombier486df472011-03-17 10:40:26 +01009380 <address> is the IPv4 or IPv6 address of the server. Alternatively, a
9381 resolvable hostname is supported, but this name will be resolved
9382 during start-up. Address "0.0.0.0" or "*" has a special meaning.
9383 It indicates that the connection will be forwarded to the same IP
Willy Tarreaud669a4f2010-07-13 14:49:50 +02009384 address as the one from the client connection. This is useful in
9385 transparent proxy architectures where the client's connection is
9386 intercepted and haproxy must forward to the original destination
9387 address. This is more or less what the "transparent" keyword does
9388 except that with a server it's possible to limit concurrency and
Willy Tarreau24709282013-03-10 21:32:12 +01009389 to report statistics. Optionally, an address family prefix may be
9390 used before the address to force the family regardless of the
9391 address format, which can be useful to specify a path to a unix
9392 socket with no slash ('/'). Currently supported prefixes are :
9393 - 'ipv4@' -> address is always IPv4
9394 - 'ipv6@' -> address is always IPv6
9395 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02009396 - 'abns@' -> address is in abstract namespace (Linux only)
William Lallemand2fe7dd02018-09-11 16:51:29 +02009397 - 'sockpair@' -> address is the FD of a connected unix
9398 socket or of a socketpair. During a connection, the
9399 backend creates a pair of connected sockets, and passes
9400 one of them over the FD. The bind part will use the
9401 received socket as the client FD. Should be used
9402 carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02009403 You may want to reference some environment variables in the
9404 address parameter, see section 2.3 about environment
Willy Tarreau6a031d12016-11-07 19:42:35 +01009405 variables. The "init-addr" setting can be used to modify the way
9406 IP addresses should be resolved upon startup.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009407
Willy Tarreaub6205fd2012-09-24 12:27:33 +02009408 <port> is an optional port specification. If set, all connections will
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009409 be sent to this port. If unset, the same port the client
9410 connected to will be used. The port may also be prefixed by a "+"
9411 or a "-". In this case, the server's port will be determined by
9412 adding this value to the client's port.
9413
9414 <param*> is a list of parameters for this server. The "server" keywords
9415 accepts an important number of options and has a complete section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009416 dedicated to it. Please refer to section 5 for more details.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009417
9418 Examples :
9419 server first 10.1.1.1:1080 cookie first check inter 1000
9420 server second 10.1.1.2:1080 cookie second check inter 1000
Willy Tarreau24709282013-03-10 21:32:12 +01009421 server transp ipv4@
William Lallemandb2f07452015-05-12 14:27:13 +02009422 server backup "${SRV_BACKUP}:1080" backup
9423 server www1_dc1 "${LAN_DC1}.101:80"
9424 server www1_dc2 "${LAN_DC2}.101:80"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009425
Willy Tarreau55dcaf62015-09-27 15:03:15 +02009426 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
9427 sun_path length is used for the address length. Some other programs
9428 such as socat use the string length only by default. Pass the option
9429 ",unix-tightsocklen=0" to any abstract socket definition in socat to
9430 make it compatible with HAProxy's.
9431
Mark Lamourinec2247f02012-01-04 13:02:01 -05009432 See also: "default-server", "http-send-name-header" and section 5 about
9433 server options
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009434
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02009435server-state-file-name [<file>]
9436 Set the server state file to read, load and apply to servers available in
9437 this backend. It only applies when the directive "load-server-state-from-file"
9438 is set to "local". When <file> is not provided or if this directive is not
9439 set, then backend name is used. If <file> starts with a slash '/', then it is
9440 considered as an absolute path. Otherwise, <file> is concatenated to the
9441 global directive "server-state-file-base".
9442
9443 Example: the minimal configuration below would make HAProxy look for the
9444 state server file '/etc/haproxy/states/bk':
9445
9446 global
9447 server-state-file-base /etc/haproxy/states
9448
Willy Tarreauc9c6cdb2020-03-05 16:03:58 +01009449 backend bk
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02009450 load-server-state-from-file
9451
9452 See also: "server-state-file-base", "load-server-state-from-file", and
9453 "show servers state"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009454
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02009455server-template <prefix> <num | range> <fqdn>[:<port>] [params*]
9456 Set a template to initialize servers with shared parameters.
9457 The names of these servers are built from <prefix> and <num | range> parameters.
9458 May be used in sections : defaults | frontend | listen | backend
9459 no | no | yes | yes
9460
9461 Arguments:
9462 <prefix> A prefix for the server names to be built.
9463
9464 <num | range>
9465 If <num> is provided, this template initializes <num> servers
9466 with 1 up to <num> as server name suffixes. A range of numbers
9467 <num_low>-<num_high> may also be used to use <num_low> up to
9468 <num_high> as server name suffixes.
9469
9470 <fqdn> A FQDN for all the servers this template initializes.
9471
9472 <port> Same meaning as "server" <port> argument (see "server" keyword).
9473
9474 <params*>
9475 Remaining server parameters among all those supported by "server"
9476 keyword.
9477
9478 Examples:
9479 # Initializes 3 servers with srv1, srv2 and srv3 as names,
9480 # google.com as FQDN, and health-check enabled.
9481 server-template srv 1-3 google.com:80 check
9482
9483 # or
9484 server-template srv 3 google.com:80 check
9485
9486 # would be equivalent to:
9487 server srv1 google.com:80 check
9488 server srv2 google.com:80 check
9489 server srv3 google.com:80 check
9490
9491
9492
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009493source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +02009494source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreaud53f96b2009-02-04 18:46:54 +01009495source <addr>[:<port>] [interface <name>]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009496 Set the source address for outgoing connections
9497 May be used in sections : defaults | frontend | listen | backend
9498 yes | no | yes | yes
9499 Arguments :
9500 <addr> is the IPv4 address HAProxy will bind to before connecting to a
9501 server. This address is also used as a source for health checks.
Willy Tarreau24709282013-03-10 21:32:12 +01009502
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009503 The default value of 0.0.0.0 means that the system will select
Willy Tarreau24709282013-03-10 21:32:12 +01009504 the most appropriate address to reach its destination. Optionally
9505 an address family prefix may be used before the address to force
9506 the family regardless of the address format, which can be useful
9507 to specify a path to a unix socket with no slash ('/'). Currently
9508 supported prefixes are :
9509 - 'ipv4@' -> address is always IPv4
9510 - 'ipv6@' -> address is always IPv6
9511 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02009512 - 'abns@' -> address is in abstract namespace (Linux only)
Cyril Bonté307ee1e2015-09-28 23:16:06 +02009513 You may want to reference some environment variables in the
9514 address parameter, see section 2.3 about environment variables.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009515
9516 <port> is an optional port. It is normally not needed but may be useful
9517 in some very specific contexts. The default value of zero means
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02009518 the system will select a free port. Note that port ranges are not
9519 supported in the backend. If you want to force port ranges, you
9520 have to specify them on each "server" line.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009521
9522 <addr2> is the IP address to present to the server when connections are
9523 forwarded in full transparent proxy mode. This is currently only
9524 supported on some patched Linux kernels. When this address is
9525 specified, clients connecting to the server will be presented
9526 with this address, while health checks will still use the address
9527 <addr>.
9528
9529 <port2> is the optional port to present to the server when connections
9530 are forwarded in full transparent proxy mode (see <addr2> above).
9531 The default value of zero means the system will select a free
9532 port.
9533
Willy Tarreaubce70882009-09-07 11:51:47 +02009534 <hdr> is the name of a HTTP header in which to fetch the IP to bind to.
9535 This is the name of a comma-separated header list which can
9536 contain multiple IP addresses. By default, the last occurrence is
9537 used. This is designed to work with the X-Forwarded-For header
Baptiste Assmannea3e73b2013-02-02 23:47:49 +01009538 and to automatically bind to the client's IP address as seen
Willy Tarreaubce70882009-09-07 11:51:47 +02009539 by previous proxy, typically Stunnel. In order to use another
9540 occurrence from the last one, please see the <occ> parameter
9541 below. When the header (or occurrence) is not found, no binding
9542 is performed so that the proxy's default IP address is used. Also
9543 keep in mind that the header name is case insensitive, as for any
9544 HTTP header.
9545
9546 <occ> is the occurrence number of a value to be used in a multi-value
9547 header. This is to be used in conjunction with "hdr_ip(<hdr>)",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04009548 in order to specify which occurrence to use for the source IP
Willy Tarreaubce70882009-09-07 11:51:47 +02009549 address. Positive values indicate a position from the first
9550 occurrence, 1 being the first one. Negative values indicate
9551 positions relative to the last one, -1 being the last one. This
9552 is helpful for situations where an X-Forwarded-For header is set
9553 at the entry point of an infrastructure and must be used several
9554 proxy layers away. When this value is not specified, -1 is
9555 assumed. Passing a zero here disables the feature.
9556
Willy Tarreaud53f96b2009-02-04 18:46:54 +01009557 <name> is an optional interface name to which to bind to for outgoing
9558 traffic. On systems supporting this features (currently, only
9559 Linux), this allows one to bind all traffic to the server to
9560 this interface even if it is not the one the system would select
9561 based on routing tables. This should be used with extreme care.
9562 Note that using this option requires root privileges.
9563
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009564 The "source" keyword is useful in complex environments where a specific
9565 address only is allowed to connect to the servers. It may be needed when a
9566 private address must be used through a public gateway for instance, and it is
9567 known that the system cannot determine the adequate source address by itself.
9568
9569 An extension which is available on certain patched Linux kernels may be used
9570 through the "usesrc" optional keyword. It makes it possible to connect to the
9571 servers with an IP address which does not belong to the system itself. This
9572 is called "full transparent proxy mode". For this to work, the destination
9573 servers have to route their traffic back to this address through the machine
9574 running HAProxy, and IP forwarding must generally be enabled on this machine.
9575
9576 In this "full transparent proxy" mode, it is possible to force a specific IP
9577 address to be presented to the servers. This is not much used in fact. A more
9578 common use is to tell HAProxy to present the client's IP address. For this,
9579 there are two methods :
9580
9581 - present the client's IP and port addresses. This is the most transparent
9582 mode, but it can cause problems when IP connection tracking is enabled on
9583 the machine, because a same connection may be seen twice with different
9584 states. However, this solution presents the huge advantage of not
9585 limiting the system to the 64k outgoing address+port couples, because all
9586 of the client ranges may be used.
9587
9588 - present only the client's IP address and select a spare port. This
9589 solution is still quite elegant but slightly less transparent (downstream
9590 firewalls logs will not match upstream's). It also presents the downside
9591 of limiting the number of concurrent connections to the usual 64k ports.
9592 However, since the upstream and downstream ports are different, local IP
9593 connection tracking on the machine will not be upset by the reuse of the
9594 same session.
9595
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009596 This option sets the default source for all servers in the backend. It may
9597 also be specified in a "defaults" section. Finer source address specification
9598 is possible at the server level using the "source" server option. Refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009599 section 5 for more information.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009600
Baptiste Assmann91bd3372015-07-17 21:59:42 +02009601 In order to work, "usesrc" requires root privileges.
9602
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009603 Examples :
9604 backend private
9605 # Connect to the servers using our 192.168.1.200 source address
9606 source 192.168.1.200
9607
9608 backend transparent_ssl1
9609 # Connect to the SSL farm from the client's source address
9610 source 192.168.1.200 usesrc clientip
9611
9612 backend transparent_ssl2
9613 # Connect to the SSL farm from the client's source address and port
9614 # not recommended if IP conntrack is present on the local machine.
9615 source 192.168.1.200 usesrc client
9616
9617 backend transparent_ssl3
9618 # Connect to the SSL farm from the client's source address. It
9619 # is more conntrack-friendly.
9620 source 192.168.1.200 usesrc clientip
9621
9622 backend transparent_smtp
9623 # Connect to the SMTP farm from the client's source address/port
9624 # with Tproxy version 4.
9625 source 0.0.0.0 usesrc clientip
9626
Willy Tarreaubce70882009-09-07 11:51:47 +02009627 backend transparent_http
9628 # Connect to the servers using the client's IP as seen by previous
9629 # proxy.
9630 source 0.0.0.0 usesrc hdr_ip(x-forwarded-for,-1)
9631
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009632 See also : the "source" server option in section 5, the Tproxy patches for
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009633 the Linux kernel on www.balabit.com, the "bind" keyword.
9634
Willy Tarreau844e3c52008-01-11 16:28:18 +01009635
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09009636srvtcpka-cnt <count>
9637 Sets the maximum number of keepalive probes TCP should send before dropping
9638 the connection on the server side.
9639 May be used in sections : defaults | frontend | listen | backend
9640 yes | no | yes | yes
9641 Arguments :
9642 <count> is the maximum number of keepalive probes.
9643
9644 This keyword corresponds to the socket option TCP_KEEPCNT. If this keyword
9645 is not specified, system-wide TCP parameter (tcp_keepalive_probes) is used.
Willy Tarreau52543212020-07-09 05:58:51 +02009646 The availability of this setting depends on the operating system. It is
9647 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09009648
9649 See also : "option srvtcpka", "srvtcpka-idle", "srvtcpka-intvl".
9650
9651
9652srvtcpka-idle <timeout>
9653 Sets the time the connection needs to remain idle before TCP starts sending
9654 keepalive probes, if enabled the sending of TCP keepalive packets on the
9655 server side.
9656 May be used in sections : defaults | frontend | listen | backend
9657 yes | no | yes | yes
9658 Arguments :
9659 <timeout> is the time the connection needs to remain idle before TCP starts
9660 sending keepalive probes. It is specified in seconds by default,
9661 but can be in any other unit if the number is suffixed by the
9662 unit, as explained at the top of this document.
9663
9664 This keyword corresponds to the socket option TCP_KEEPIDLE. If this keyword
9665 is not specified, system-wide TCP parameter (tcp_keepalive_time) is used.
Willy Tarreau52543212020-07-09 05:58:51 +02009666 The availability of this setting depends on the operating system. It is
9667 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09009668
9669 See also : "option srvtcpka", "srvtcpka-cnt", "srvtcpka-intvl".
9670
9671
9672srvtcpka-intvl <timeout>
9673 Sets the time between individual keepalive probes on the server side.
9674 May be used in sections : defaults | frontend | listen | backend
9675 yes | no | yes | yes
9676 Arguments :
9677 <timeout> is the time between individual keepalive probes. It is specified
9678 in seconds by default, but can be in any other unit if the number
9679 is suffixed by the unit, as explained at the top of this
9680 document.
9681
9682 This keyword corresponds to the socket option TCP_KEEPINTVL. If this keyword
9683 is not specified, system-wide TCP parameter (tcp_keepalive_intvl) is used.
Willy Tarreau52543212020-07-09 05:58:51 +02009684 The availability of this setting depends on the operating system. It is
9685 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09009686
9687 See also : "option srvtcpka", "srvtcpka-cnt", "srvtcpka-idle".
9688
9689
Cyril Bonté66c327d2010-10-12 00:14:37 +02009690stats admin { if | unless } <cond>
9691 Enable statistics admin level if/unless a condition is matched
9692 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02009693 no | yes | yes | yes
Cyril Bonté66c327d2010-10-12 00:14:37 +02009694
9695 This statement enables the statistics admin level if/unless a condition is
9696 matched.
9697
9698 The admin level allows to enable/disable servers from the web interface. By
9699 default, statistics page is read-only for security reasons.
9700
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009701 Note : Consider not using this feature in multi-process mode (nbproc > 1)
9702 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01009703 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009704
Cyril Bonté23b39d92011-02-10 22:54:44 +01009705 Currently, the POST request is limited to the buffer size minus the reserved
9706 buffer space, which means that if the list of servers is too long, the
9707 request won't be processed. It is recommended to alter few servers at a
9708 time.
Cyril Bonté66c327d2010-10-12 00:14:37 +02009709
9710 Example :
9711 # statistics admin level only for localhost
9712 backend stats_localhost
9713 stats enable
9714 stats admin if LOCALHOST
9715
9716 Example :
9717 # statistics admin level always enabled because of the authentication
9718 backend stats_auth
9719 stats enable
9720 stats auth admin:AdMiN123
9721 stats admin if TRUE
9722
9723 Example :
9724 # statistics admin level depends on the authenticated user
9725 userlist stats-auth
9726 group admin users admin
9727 user admin insecure-password AdMiN123
9728 group readonly users haproxy
9729 user haproxy insecure-password haproxy
9730
9731 backend stats_auth
9732 stats enable
9733 acl AUTH http_auth(stats-auth)
9734 acl AUTH_ADMIN http_auth_group(stats-auth) admin
9735 stats http-request auth unless AUTH
9736 stats admin if AUTH_ADMIN
9737
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009738 See also : "stats enable", "stats auth", "stats http-request", "nbproc",
9739 "bind-process", section 3.4 about userlists and section 7 about
9740 ACL usage.
Cyril Bonté66c327d2010-10-12 00:14:37 +02009741
9742
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009743stats auth <user>:<passwd>
9744 Enable statistics with authentication and grant access to an account
9745 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02009746 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009747 Arguments :
9748 <user> is a user name to grant access to
9749
9750 <passwd> is the cleartext password associated to this user
9751
9752 This statement enables statistics with default settings, and restricts access
9753 to declared users only. It may be repeated as many times as necessary to
9754 allow as many users as desired. When a user tries to access the statistics
9755 without a valid account, a "401 Forbidden" response will be returned so that
9756 the browser asks the user to provide a valid user and password. The real
9757 which will be returned to the browser is configurable using "stats realm".
9758
9759 Since the authentication method is HTTP Basic Authentication, the passwords
9760 circulate in cleartext on the network. Thus, it was decided that the
9761 configuration file would also use cleartext passwords to remind the users
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02009762 that those ones should not be sensitive and not shared with any other account.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009763
9764 It is also possible to reduce the scope of the proxies which appear in the
9765 report using "stats scope".
9766
9767 Though this statement alone is enough to enable statistics reporting, it is
9768 recommended to set all other settings in order to avoid relying on default
9769 unobvious parameters.
9770
9771 Example :
9772 # public access (limited to this backend only)
9773 backend public_www
9774 server srv1 192.168.0.1:80
9775 stats enable
9776 stats hide-version
9777 stats scope .
9778 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01009779 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009780 stats auth admin1:AdMiN123
9781 stats auth admin2:AdMiN321
9782
9783 # internal monitoring access (unlimited)
9784 backend private_monitoring
9785 stats enable
9786 stats uri /admin?stats
9787 stats refresh 5s
9788
9789 See also : "stats enable", "stats realm", "stats scope", "stats uri"
9790
9791
9792stats enable
9793 Enable statistics reporting with default settings
9794 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02009795 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009796 Arguments : none
9797
9798 This statement enables statistics reporting with default settings defined
9799 at build time. Unless stated otherwise, these settings are used :
9800 - stats uri : /haproxy?stats
9801 - stats realm : "HAProxy Statistics"
9802 - stats auth : no authentication
9803 - stats scope : no restriction
9804
9805 Though this statement alone is enough to enable statistics reporting, it is
9806 recommended to set all other settings in order to avoid relying on default
9807 unobvious parameters.
9808
9809 Example :
9810 # public access (limited to this backend only)
9811 backend public_www
9812 server srv1 192.168.0.1:80
9813 stats enable
9814 stats hide-version
9815 stats scope .
9816 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01009817 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009818 stats auth admin1:AdMiN123
9819 stats auth admin2:AdMiN321
9820
9821 # internal monitoring access (unlimited)
9822 backend private_monitoring
9823 stats enable
9824 stats uri /admin?stats
9825 stats refresh 5s
9826
9827 See also : "stats auth", "stats realm", "stats uri"
9828
9829
Willy Tarreaud63335a2010-02-26 12:56:52 +01009830stats hide-version
9831 Enable statistics and hide HAProxy version reporting
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02009832 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02009833 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01009834 Arguments : none
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02009835
Willy Tarreaud63335a2010-02-26 12:56:52 +01009836 By default, the stats page reports some useful status information along with
9837 the statistics. Among them is HAProxy's version. However, it is generally
9838 considered dangerous to report precise version to anyone, as it can help them
9839 target known weaknesses with specific attacks. The "stats hide-version"
9840 statement removes the version from the statistics report. This is recommended
9841 for public sites or any site with a weak login/password.
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02009842
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02009843 Though this statement alone is enough to enable statistics reporting, it is
9844 recommended to set all other settings in order to avoid relying on default
9845 unobvious parameters.
9846
Willy Tarreaud63335a2010-02-26 12:56:52 +01009847 Example :
9848 # public access (limited to this backend only)
9849 backend public_www
9850 server srv1 192.168.0.1:80
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02009851 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01009852 stats hide-version
9853 stats scope .
9854 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01009855 stats realm HAProxy\ Statistics
Willy Tarreaud63335a2010-02-26 12:56:52 +01009856 stats auth admin1:AdMiN123
9857 stats auth admin2:AdMiN321
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02009858
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02009859 # internal monitoring access (unlimited)
9860 backend private_monitoring
9861 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01009862 stats uri /admin?stats
9863 stats refresh 5s
Krzysztof Piotr Oledzki15514c22010-01-04 16:03:09 +01009864
Willy Tarreaud63335a2010-02-26 12:56:52 +01009865 See also : "stats auth", "stats enable", "stats realm", "stats uri"
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02009866
Willy Tarreau983e01e2010-01-11 18:42:06 +01009867
Cyril Bonté2be1b3f2010-09-30 23:46:30 +02009868stats http-request { allow | deny | auth [realm <realm>] }
9869 [ { if | unless } <condition> ]
9870 Access control for statistics
9871
9872 May be used in sections: defaults | frontend | listen | backend
9873 no | no | yes | yes
9874
9875 As "http-request", these set of options allow to fine control access to
9876 statistics. Each option may be followed by if/unless and acl.
9877 First option with matched condition (or option without condition) is final.
9878 For "deny" a 403 error will be returned, for "allow" normal processing is
9879 performed, for "auth" a 401/407 error code is returned so the client
9880 should be asked to enter a username and password.
9881
9882 There is no fixed limit to the number of http-request statements per
9883 instance.
9884
9885 See also : "http-request", section 3.4 about userlists and section 7
9886 about ACL usage.
9887
9888
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009889stats realm <realm>
9890 Enable statistics and set authentication realm
9891 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02009892 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009893 Arguments :
9894 <realm> is the name of the HTTP Basic Authentication realm reported to
9895 the browser. The browser uses it to display it in the pop-up
9896 inviting the user to enter a valid username and password.
9897
9898 The realm is read as a single word, so any spaces in it should be escaped
9899 using a backslash ('\').
9900
9901 This statement is useful only in conjunction with "stats auth" since it is
9902 only related to authentication.
9903
9904 Though this statement alone is enough to enable statistics reporting, it is
9905 recommended to set all other settings in order to avoid relying on default
9906 unobvious parameters.
9907
9908 Example :
9909 # public access (limited to this backend only)
9910 backend public_www
9911 server srv1 192.168.0.1:80
9912 stats enable
9913 stats hide-version
9914 stats scope .
9915 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01009916 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009917 stats auth admin1:AdMiN123
9918 stats auth admin2:AdMiN321
9919
9920 # internal monitoring access (unlimited)
9921 backend private_monitoring
9922 stats enable
9923 stats uri /admin?stats
9924 stats refresh 5s
9925
9926 See also : "stats auth", "stats enable", "stats uri"
9927
9928
9929stats refresh <delay>
9930 Enable statistics with automatic refresh
9931 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02009932 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009933 Arguments :
9934 <delay> is the suggested refresh delay, specified in seconds, which will
9935 be returned to the browser consulting the report page. While the
9936 browser is free to apply any delay, it will generally respect it
9937 and refresh the page this every seconds. The refresh interval may
9938 be specified in any other non-default time unit, by suffixing the
9939 unit after the value, as explained at the top of this document.
9940
9941 This statement is useful on monitoring displays with a permanent page
9942 reporting the load balancer's activity. When set, the HTML report page will
9943 include a link "refresh"/"stop refresh" so that the user can select whether
Jackie Tapia749f74c2020-07-22 18:59:40 -05009944 they want automatic refresh of the page or not.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009945
9946 Though this statement alone is enough to enable statistics reporting, it is
9947 recommended to set all other settings in order to avoid relying on default
9948 unobvious parameters.
9949
9950 Example :
9951 # public access (limited to this backend only)
9952 backend public_www
9953 server srv1 192.168.0.1:80
9954 stats enable
9955 stats hide-version
9956 stats scope .
9957 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01009958 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009959 stats auth admin1:AdMiN123
9960 stats auth admin2:AdMiN321
9961
9962 # internal monitoring access (unlimited)
9963 backend private_monitoring
9964 stats enable
9965 stats uri /admin?stats
9966 stats refresh 5s
9967
9968 See also : "stats auth", "stats enable", "stats realm", "stats uri"
9969
9970
9971stats scope { <name> | "." }
9972 Enable statistics and limit access scope
9973 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02009974 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009975 Arguments :
9976 <name> is the name of a listen, frontend or backend section to be
9977 reported. The special name "." (a single dot) designates the
9978 section in which the statement appears.
9979
9980 When this statement is specified, only the sections enumerated with this
9981 statement will appear in the report. All other ones will be hidden. This
9982 statement may appear as many times as needed if multiple sections need to be
9983 reported. Please note that the name checking is performed as simple string
9984 comparisons, and that it is never checked that a give section name really
9985 exists.
9986
9987 Though this statement alone is enough to enable statistics reporting, it is
9988 recommended to set all other settings in order to avoid relying on default
9989 unobvious parameters.
9990
9991 Example :
9992 # public access (limited to this backend only)
9993 backend public_www
9994 server srv1 192.168.0.1:80
9995 stats enable
9996 stats hide-version
9997 stats scope .
9998 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01009999 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010000 stats auth admin1:AdMiN123
10001 stats auth admin2:AdMiN321
10002
10003 # internal monitoring access (unlimited)
10004 backend private_monitoring
10005 stats enable
10006 stats uri /admin?stats
10007 stats refresh 5s
10008
10009 See also : "stats auth", "stats enable", "stats realm", "stats uri"
10010
Willy Tarreaud63335a2010-02-26 12:56:52 +010010011
Willy Tarreauc9705a12010-07-27 20:05:50 +020010012stats show-desc [ <desc> ]
Willy Tarreaud63335a2010-02-26 12:56:52 +010010013 Enable reporting of a description on the statistics page.
10014 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010015 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +010010016
Willy Tarreauc9705a12010-07-27 20:05:50 +020010017 <desc> is an optional description to be reported. If unspecified, the
Willy Tarreaud63335a2010-02-26 12:56:52 +010010018 description from global section is automatically used instead.
10019
10020 This statement is useful for users that offer shared services to their
10021 customers, where node or description should be different for each customer.
10022
10023 Though this statement alone is enough to enable statistics reporting, it is
10024 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +010010025 unobvious parameters. By default description is not shown.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010026
10027 Example :
10028 # internal monitoring access (unlimited)
10029 backend private_monitoring
10030 stats enable
10031 stats show-desc Master node for Europe, Asia, Africa
10032 stats uri /admin?stats
10033 stats refresh 5s
10034
10035 See also: "show-node", "stats enable", "stats uri" and "description" in
10036 global section.
10037
10038
10039stats show-legends
Willy Tarreaued2119c2014-04-24 22:10:39 +020010040 Enable reporting additional information on the statistics page
10041 May be used in sections : defaults | frontend | listen | backend
10042 yes | yes | yes | yes
10043 Arguments : none
10044
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010045 Enable reporting additional information on the statistics page :
Willy Tarreaud63335a2010-02-26 12:56:52 +010010046 - cap: capabilities (proxy)
10047 - mode: one of tcp, http or health (proxy)
10048 - id: SNMP ID (proxy, socket, server)
10049 - IP (socket, server)
10050 - cookie (backend, server)
10051
10052 Though this statement alone is enough to enable statistics reporting, it is
10053 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +010010054 unobvious parameters. Default behavior is not to show this information.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010055
10056 See also: "stats enable", "stats uri".
10057
10058
Amaury Denoyelle0b70a8a2020-10-05 11:49:45 +020010059stats show-modules
10060 Enable display of extra statistics module on the statistics page
10061 May be used in sections : defaults | frontend | listen | backend
10062 yes | yes | yes | yes
10063 Arguments : none
10064
10065 New columns are added at the end of the line containing the extra statistics
10066 values as a tooltip.
10067
10068 Though this statement alone is enough to enable statistics reporting, it is
10069 recommended to set all other settings in order to avoid relying on default
10070 unobvious parameters. Default behavior is not to show this information.
10071
10072 See also: "stats enable", "stats uri".
10073
10074
Willy Tarreaud63335a2010-02-26 12:56:52 +010010075stats show-node [ <name> ]
10076 Enable reporting of a host name on the statistics page.
10077 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010078 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +010010079 Arguments:
10080 <name> is an optional name to be reported. If unspecified, the
10081 node name from global section is automatically used instead.
10082
10083 This statement is useful for users that offer shared services to their
10084 customers, where node or description might be different on a stats page
Davor Ocelice9ed2812017-12-25 17:49:28 +010010085 provided for each customer. Default behavior is not to show host name.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010086
10087 Though this statement alone is enough to enable statistics reporting, it is
10088 recommended to set all other settings in order to avoid relying on default
10089 unobvious parameters.
10090
10091 Example:
10092 # internal monitoring access (unlimited)
10093 backend private_monitoring
10094 stats enable
10095 stats show-node Europe-1
10096 stats uri /admin?stats
10097 stats refresh 5s
10098
10099 See also: "show-desc", "stats enable", "stats uri", and "node" in global
10100 section.
10101
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010102
10103stats uri <prefix>
10104 Enable statistics and define the URI prefix to access them
10105 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010106 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010107 Arguments :
10108 <prefix> is the prefix of any URI which will be redirected to stats. This
10109 prefix may contain a question mark ('?') to indicate part of a
10110 query string.
10111
10112 The statistics URI is intercepted on the relayed traffic, so it appears as a
10113 page within the normal application. It is strongly advised to ensure that the
10114 selected URI will never appear in the application, otherwise it will never be
10115 possible to reach it in the application.
10116
10117 The default URI compiled in haproxy is "/haproxy?stats", but this may be
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010010118 changed at build time, so it's better to always explicitly specify it here.
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010119 It is generally a good idea to include a question mark in the URI so that
10120 intermediate proxies refrain from caching the results. Also, since any string
10121 beginning with the prefix will be accepted as a stats request, the question
10122 mark helps ensuring that no valid URI will begin with the same words.
10123
10124 It is sometimes very convenient to use "/" as the URI prefix, and put that
10125 statement in a "listen" instance of its own. That makes it easy to dedicate
10126 an address or a port to statistics only.
10127
10128 Though this statement alone is enough to enable statistics reporting, it is
10129 recommended to set all other settings in order to avoid relying on default
10130 unobvious parameters.
10131
10132 Example :
10133 # public access (limited to this backend only)
10134 backend public_www
10135 server srv1 192.168.0.1:80
10136 stats enable
10137 stats hide-version
10138 stats scope .
10139 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010140 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010141 stats auth admin1:AdMiN123
10142 stats auth admin2:AdMiN321
10143
10144 # internal monitoring access (unlimited)
10145 backend private_monitoring
10146 stats enable
10147 stats uri /admin?stats
10148 stats refresh 5s
10149
10150 See also : "stats auth", "stats enable", "stats realm"
10151
10152
Willy Tarreaud63335a2010-02-26 12:56:52 +010010153stick match <pattern> [table <table>] [{if | unless} <cond>]
10154 Define a request pattern matching condition to stick a user to a server
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010155 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaud63335a2010-02-26 12:56:52 +010010156 no | no | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010157
10158 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +020010159 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010160 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +010010161 will be analyzed in the hope to find a matching entry in a
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010162 stickiness table. This rule is mandatory.
10163
10164 <table> is an optional stickiness table name. If unspecified, the same
10165 backend's table is used. A stickiness table is declared using
10166 the "stick-table" statement.
10167
10168 <cond> is an optional matching condition. It makes it possible to match
10169 on a certain criterion only when other conditions are met (or
10170 not met). For instance, it could be used to match on a source IP
10171 address except when a request passes through a known proxy, in
10172 which case we'd match on a header containing that IP address.
10173
10174 Some protocols or applications require complex stickiness rules and cannot
10175 always simply rely on cookies nor hashing. The "stick match" statement
10176 describes a rule to extract the stickiness criterion from an incoming request
10177 or connection. See section 7 for a complete list of possible patterns and
10178 transformation rules.
10179
10180 The table has to be declared using the "stick-table" statement. It must be of
10181 a type compatible with the pattern. By default it is the one which is present
10182 in the same backend. It is possible to share a table with other backends by
10183 referencing it using the "table" keyword. If another table is referenced,
10184 the server's ID inside the backends are used. By default, all server IDs
10185 start at 1 in each backend, so the server ordering is enough. But in case of
10186 doubt, it is highly recommended to force server IDs using their "id" setting.
10187
10188 It is possible to restrict the conditions where a "stick match" statement
10189 will apply, using "if" or "unless" followed by a condition. See section 7 for
10190 ACL based conditions.
10191
10192 There is no limit on the number of "stick match" statements. The first that
10193 applies and matches will cause the request to be directed to the same server
10194 as was used for the request which created the entry. That way, multiple
10195 matches can be used as fallbacks.
10196
10197 The stick rules are checked after the persistence cookies, so they will not
10198 affect stickiness if a cookie has already been used to select a server. That
10199 way, it becomes very easy to insert cookies and match on IP addresses in
10200 order to maintain stickiness between HTTP and HTTPS.
10201
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010202 Note : Consider not using this feature in multi-process mode (nbproc > 1)
10203 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +010010204 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010205
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010206 Example :
10207 # forward SMTP users to the same server they just used for POP in the
10208 # last 30 minutes
10209 backend pop
10210 mode tcp
10211 balance roundrobin
10212 stick store-request src
10213 stick-table type ip size 200k expire 30m
10214 server s1 192.168.1.1:110
10215 server s2 192.168.1.1:110
10216
10217 backend smtp
10218 mode tcp
10219 balance roundrobin
10220 stick match src table pop
10221 server s1 192.168.1.1:25
10222 server s2 192.168.1.1:25
10223
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010224 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +020010225 about ACLs and samples fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010226
10227
10228stick on <pattern> [table <table>] [{if | unless} <condition>]
10229 Define a request pattern to associate a user to a server
10230 May be used in sections : defaults | frontend | listen | backend
10231 no | no | yes | yes
10232
10233 Note : This form is exactly equivalent to "stick match" followed by
10234 "stick store-request", all with the same arguments. Please refer
10235 to both keywords for details. It is only provided as a convenience
10236 for writing more maintainable configurations.
10237
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010238 Note : Consider not using this feature in multi-process mode (nbproc > 1)
10239 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +010010240 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010241
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010242 Examples :
10243 # The following form ...
Willy Tarreauec579d82010-02-26 19:15:04 +010010244 stick on src table pop if !localhost
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010245
10246 # ...is strictly equivalent to this one :
10247 stick match src table pop if !localhost
10248 stick store-request src table pop if !localhost
10249
10250
10251 # Use cookie persistence for HTTP, and stick on source address for HTTPS as
10252 # well as HTTP without cookie. Share the same table between both accesses.
10253 backend http
10254 mode http
10255 balance roundrobin
10256 stick on src table https
10257 cookie SRV insert indirect nocache
10258 server s1 192.168.1.1:80 cookie s1
10259 server s2 192.168.1.1:80 cookie s2
10260
10261 backend https
10262 mode tcp
10263 balance roundrobin
10264 stick-table type ip size 200k expire 30m
10265 stick on src
10266 server s1 192.168.1.1:443
10267 server s2 192.168.1.1:443
10268
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010269 See also : "stick match", "stick store-request", "nbproc" and "bind-process".
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010270
10271
10272stick store-request <pattern> [table <table>] [{if | unless} <condition>]
10273 Define a request pattern used to create an entry in a stickiness table
10274 May be used in sections : defaults | frontend | listen | backend
10275 no | no | yes | yes
10276
10277 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +020010278 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010279 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +010010280 will be analyzed, extracted and stored in the table once a
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010281 server is selected.
10282
10283 <table> is an optional stickiness table name. If unspecified, the same
10284 backend's table is used. A stickiness table is declared using
10285 the "stick-table" statement.
10286
10287 <cond> is an optional storage condition. It makes it possible to store
10288 certain criteria only when some conditions are met (or not met).
10289 For instance, it could be used to store the source IP address
10290 except when the request passes through a known proxy, in which
10291 case we'd store a converted form of a header containing that IP
10292 address.
10293
10294 Some protocols or applications require complex stickiness rules and cannot
10295 always simply rely on cookies nor hashing. The "stick store-request" statement
10296 describes a rule to decide what to extract from the request and when to do
10297 it, in order to store it into a stickiness table for further requests to
10298 match it using the "stick match" statement. Obviously the extracted part must
10299 make sense and have a chance to be matched in a further request. Storing a
10300 client's IP address for instance often makes sense. Storing an ID found in a
10301 URL parameter also makes sense. Storing a source port will almost never make
10302 any sense because it will be randomly matched. See section 7 for a complete
10303 list of possible patterns and transformation rules.
10304
10305 The table has to be declared using the "stick-table" statement. It must be of
10306 a type compatible with the pattern. By default it is the one which is present
10307 in the same backend. It is possible to share a table with other backends by
10308 referencing it using the "table" keyword. If another table is referenced,
10309 the server's ID inside the backends are used. By default, all server IDs
10310 start at 1 in each backend, so the server ordering is enough. But in case of
10311 doubt, it is highly recommended to force server IDs using their "id" setting.
10312
10313 It is possible to restrict the conditions where a "stick store-request"
10314 statement will apply, using "if" or "unless" followed by a condition. This
10315 condition will be evaluated while parsing the request, so any criteria can be
10316 used. See section 7 for ACL based conditions.
10317
10318 There is no limit on the number of "stick store-request" statements, but
10319 there is a limit of 8 simultaneous stores per request or response. This
10320 makes it possible to store up to 8 criteria, all extracted from either the
10321 request or the response, regardless of the number of rules. Only the 8 first
10322 ones which match will be kept. Using this, it is possible to feed multiple
10323 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +010010324 another protocol or access method. Using multiple store-request rules with
10325 the same table is possible and may be used to find the best criterion to rely
10326 on, by arranging the rules by decreasing preference order. Only the first
10327 extracted criterion for a given table will be stored. All subsequent store-
10328 request rules referencing the same table will be skipped and their ACLs will
10329 not be evaluated.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010330
10331 The "store-request" rules are evaluated once the server connection has been
10332 established, so that the table will contain the real server that processed
10333 the request.
10334
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010335 Note : Consider not using this feature in multi-process mode (nbproc > 1)
10336 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +010010337 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010338
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010339 Example :
10340 # forward SMTP users to the same server they just used for POP in the
10341 # last 30 minutes
10342 backend pop
10343 mode tcp
10344 balance roundrobin
10345 stick store-request src
10346 stick-table type ip size 200k expire 30m
10347 server s1 192.168.1.1:110
10348 server s2 192.168.1.1:110
10349
10350 backend smtp
10351 mode tcp
10352 balance roundrobin
10353 stick match src table pop
10354 server s1 192.168.1.1:25
10355 server s2 192.168.1.1:25
10356
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010357 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +020010358 about ACLs and sample fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010359
10360
Emeric Brun7c6b82e2010-09-24 16:34:28 +020010361stick-table type {ip | integer | string [len <length>] | binary [len <length>]}
Emeric Brunf099e792010-09-27 12:05:28 +020010362 size <size> [expire <expire>] [nopurge] [peers <peersect>]
10363 [store <data_type>]*
Godbach64cef792013-12-04 16:08:22 +080010364 Configure the stickiness table for the current section
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010365 May be used in sections : defaults | frontend | listen | backend
Willy Tarreauc00cdc22010-06-06 16:48:26 +020010366 no | yes | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010367
10368 Arguments :
10369 ip a table declared with "type ip" will only store IPv4 addresses.
10370 This form is very compact (about 50 bytes per entry) and allows
10371 very fast entry lookup and stores with almost no overhead. This
10372 is mainly used to store client source IP addresses.
10373
David du Colombier9a6d3c92011-03-17 10:40:24 +010010374 ipv6 a table declared with "type ipv6" will only store IPv6 addresses.
10375 This form is very compact (about 60 bytes per entry) and allows
10376 very fast entry lookup and stores with almost no overhead. This
10377 is mainly used to store client source IP addresses.
10378
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010379 integer a table declared with "type integer" will store 32bit integers
10380 which can represent a client identifier found in a request for
10381 instance.
10382
10383 string a table declared with "type string" will store substrings of up
10384 to <len> characters. If the string provided by the pattern
10385 extractor is larger than <len>, it will be truncated before
10386 being stored. During matching, at most <len> characters will be
10387 compared between the string in the table and the extracted
10388 pattern. When not specified, the string is automatically limited
Emeric Brun7c6b82e2010-09-24 16:34:28 +020010389 to 32 characters.
10390
10391 binary a table declared with "type binary" will store binary blocks
10392 of <len> bytes. If the block provided by the pattern
10393 extractor is larger than <len>, it will be truncated before
Willy Tarreaube722a22014-06-13 16:31:59 +020010394 being stored. If the block provided by the sample expression
Emeric Brun7c6b82e2010-09-24 16:34:28 +020010395 is shorter than <len>, it will be padded by 0. When not
10396 specified, the block is automatically limited to 32 bytes.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010397
10398 <length> is the maximum number of characters that will be stored in a
Emeric Brun7c6b82e2010-09-24 16:34:28 +020010399 "string" type table (See type "string" above). Or the number
10400 of bytes of the block in "binary" type table. Be careful when
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010401 changing this parameter as memory usage will proportionally
10402 increase.
10403
10404 <size> is the maximum number of entries that can fit in the table. This
Cyril Bonté78caf842010-03-10 22:41:43 +010010405 value directly impacts memory usage. Count approximately
10406 50 bytes per entry, plus the size of a string if any. The size
10407 supports suffixes "k", "m", "g" for 2^10, 2^20 and 2^30 factors.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010408
10409 [nopurge] indicates that we refuse to purge older entries when the table
10410 is full. When not specified and the table is full when haproxy
10411 wants to store an entry in it, it will flush a few of the oldest
10412 entries in order to release some space for the new ones. This is
Davor Ocelice9ed2812017-12-25 17:49:28 +010010413 most often the desired behavior. In some specific cases, it
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010414 be desirable to refuse new entries instead of purging the older
10415 ones. That may be the case when the amount of data to store is
10416 far above the hardware limits and we prefer not to offer access
10417 to new clients than to reject the ones already connected. When
10418 using this parameter, be sure to properly set the "expire"
10419 parameter (see below).
10420
Emeric Brunf099e792010-09-27 12:05:28 +020010421 <peersect> is the name of the peers section to use for replication. Entries
10422 which associate keys to server IDs are kept synchronized with
10423 the remote peers declared in this section. All entries are also
10424 automatically learned from the local peer (old process) during a
10425 soft restart.
10426
Willy Tarreau1abc6732015-05-01 19:21:02 +020010427 NOTE : each peers section may be referenced only by tables
10428 belonging to the same unique process.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010429
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010430 <expire> defines the maximum duration of an entry in the table since it
10431 was last created, refreshed or matched. The expiration delay is
10432 defined using the standard time format, similarly as the various
10433 timeouts. The maximum duration is slightly above 24 days. See
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +030010434 section 2.4 for more information. If this delay is not specified,
Cyril Bontédc4d9032012-04-08 21:57:39 +020010435 the session won't automatically expire, but older entries will
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010436 be removed once full. Be sure not to use the "nopurge" parameter
10437 if not expiration delay is specified.
10438
Willy Tarreau08d5f982010-06-06 13:34:54 +020010439 <data_type> is used to store additional information in the stick-table. This
10440 may be used by ACLs in order to control various criteria related
10441 to the activity of the client matching the stick-table. For each
10442 item specified here, the size of each entry will be inflated so
Willy Tarreauc9705a12010-07-27 20:05:50 +020010443 that the additional data can fit. Several data types may be
10444 stored with an entry. Multiple data types may be specified after
10445 the "store" keyword, as a comma-separated list. Alternatively,
10446 it is possible to repeat the "store" keyword followed by one or
10447 several data types. Except for the "server_id" type which is
10448 automatically detected and enabled, all data types must be
10449 explicitly declared to be stored. If an ACL references a data
10450 type which is not stored, the ACL will simply not match. Some
10451 data types require an argument which must be passed just after
10452 the type between parenthesis. See below for the supported data
10453 types and their arguments.
10454
10455 The data types that can be stored with an entry are the following :
10456 - server_id : this is an integer which holds the numeric ID of the server a
10457 request was assigned to. It is used by the "stick match", "stick store",
10458 and "stick on" rules. It is automatically enabled when referenced.
10459
10460 - gpc0 : first General Purpose Counter. It is a positive 32-bit integer
10461 integer which may be used for anything. Most of the time it will be used
10462 to put a special tag on some entries, for instance to note that a
Davor Ocelice9ed2812017-12-25 17:49:28 +010010463 specific behavior was detected and must be known for future matches.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010464
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010465 - gpc0_rate(<period>) : increment rate of the first General Purpose Counter
10466 over a period. It is a positive 32-bit integer integer which may be used
10467 for anything. Just like <gpc0>, it counts events, but instead of keeping
Davor Ocelice9ed2812017-12-25 17:49:28 +010010468 a cumulative number, it maintains the rate at which the counter is
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010469 incremented. Most of the time it will be used to measure the frequency of
Davor Ocelice9ed2812017-12-25 17:49:28 +010010470 occurrence of certain events (e.g. requests to a specific URL).
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010471
Frédéric Lécaille6778b272018-01-29 15:22:53 +010010472 - gpc1 : second General Purpose Counter. It is a positive 32-bit integer
10473 integer which may be used for anything. Most of the time it will be used
10474 to put a special tag on some entries, for instance to note that a
10475 specific behavior was detected and must be known for future matches.
10476
10477 - gpc1_rate(<period>) : increment rate of the second General Purpose Counter
10478 over a period. It is a positive 32-bit integer integer which may be used
10479 for anything. Just like <gpc1>, it counts events, but instead of keeping
10480 a cumulative number, it maintains the rate at which the counter is
10481 incremented. Most of the time it will be used to measure the frequency of
10482 occurrence of certain events (e.g. requests to a specific URL).
10483
Willy Tarreauc9705a12010-07-27 20:05:50 +020010484 - conn_cnt : Connection Count. It is a positive 32-bit integer which counts
10485 the absolute number of connections received from clients which matched
10486 this entry. It does not mean the connections were accepted, just that
10487 they were received.
10488
10489 - conn_cur : Current Connections. It is a positive 32-bit integer which
10490 stores the concurrent connection counts for the entry. It is incremented
10491 once an incoming connection matches the entry, and decremented once the
10492 connection leaves. That way it is possible to know at any time the exact
10493 number of concurrent connections for an entry.
10494
10495 - conn_rate(<period>) : frequency counter (takes 12 bytes). It takes an
10496 integer parameter <period> which indicates in milliseconds the length
10497 of the period over which the average is measured. It reports the average
10498 incoming connection rate over that period, in connections per period. The
10499 result is an integer which can be matched using ACLs.
10500
10501 - sess_cnt : Session Count. It is a positive 32-bit integer which counts
10502 the absolute number of sessions received from clients which matched this
10503 entry. A session is a connection that was accepted by the layer 4 rules.
10504
10505 - sess_rate(<period>) : frequency counter (takes 12 bytes). It takes an
10506 integer parameter <period> which indicates in milliseconds the length
10507 of the period over which the average is measured. It reports the average
10508 incoming session rate over that period, in sessions per period. The
10509 result is an integer which can be matched using ACLs.
10510
10511 - http_req_cnt : HTTP request Count. It is a positive 32-bit integer which
10512 counts the absolute number of HTTP requests received from clients which
10513 matched this entry. It does not matter whether they are valid requests or
10514 not. Note that this is different from sessions when keep-alive is used on
10515 the client side.
10516
10517 - http_req_rate(<period>) : frequency counter (takes 12 bytes). It takes an
10518 integer parameter <period> which indicates in milliseconds the length
10519 of the period over which the average is measured. It reports the average
10520 HTTP request rate over that period, in requests per period. The result is
10521 an integer which can be matched using ACLs. It does not matter whether
10522 they are valid requests or not. Note that this is different from sessions
10523 when keep-alive is used on the client side.
10524
10525 - http_err_cnt : HTTP Error Count. It is a positive 32-bit integer which
10526 counts the absolute number of HTTP requests errors induced by clients
10527 which matched this entry. Errors are counted on invalid and truncated
10528 requests, as well as on denied or tarpitted requests, and on failed
10529 authentications. If the server responds with 4xx, then the request is
10530 also counted as an error since it's an error triggered by the client
Davor Ocelice9ed2812017-12-25 17:49:28 +010010531 (e.g. vulnerability scan).
Willy Tarreauc9705a12010-07-27 20:05:50 +020010532
10533 - http_err_rate(<period>) : frequency counter (takes 12 bytes). It takes an
10534 integer parameter <period> which indicates in milliseconds the length
10535 of the period over which the average is measured. It reports the average
10536 HTTP request error rate over that period, in requests per period (see
10537 http_err_cnt above for what is accounted as an error). The result is an
10538 integer which can be matched using ACLs.
10539
10540 - bytes_in_cnt : client to server byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +010010541 integer which counts the cumulative number of bytes received from clients
Willy Tarreauc9705a12010-07-27 20:05:50 +020010542 which matched this entry. Headers are included in the count. This may be
10543 used to limit abuse of upload features on photo or video servers.
10544
10545 - bytes_in_rate(<period>) : frequency counter (takes 12 bytes). It takes an
10546 integer parameter <period> which indicates in milliseconds the length
10547 of the period over which the average is measured. It reports the average
10548 incoming bytes rate over that period, in bytes per period. It may be used
10549 to detect users which upload too much and too fast. Warning: with large
10550 uploads, it is possible that the amount of uploaded data will be counted
10551 once upon termination, thus causing spikes in the average transfer speed
10552 instead of having a smooth one. This may partially be smoothed with
10553 "option contstats" though this is not perfect yet. Use of byte_in_cnt is
10554 recommended for better fairness.
10555
10556 - bytes_out_cnt : server to client byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +010010557 integer which counts the cumulative number of bytes sent to clients which
Willy Tarreauc9705a12010-07-27 20:05:50 +020010558 matched this entry. Headers are included in the count. This may be used
10559 to limit abuse of bots sucking the whole site.
10560
10561 - bytes_out_rate(<period>) : frequency counter (takes 12 bytes). It takes
10562 an integer parameter <period> which indicates in milliseconds the length
10563 of the period over which the average is measured. It reports the average
10564 outgoing bytes rate over that period, in bytes per period. It may be used
10565 to detect users which download too much and too fast. Warning: with large
10566 transfers, it is possible that the amount of transferred data will be
10567 counted once upon termination, thus causing spikes in the average
10568 transfer speed instead of having a smooth one. This may partially be
10569 smoothed with "option contstats" though this is not perfect yet. Use of
10570 byte_out_cnt is recommended for better fairness.
Willy Tarreau08d5f982010-06-06 13:34:54 +020010571
Willy Tarreauc00cdc22010-06-06 16:48:26 +020010572 There is only one stick-table per proxy. At the moment of writing this doc,
10573 it does not seem useful to have multiple tables per proxy. If this happens
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010574 to be required, simply create a dummy backend with a stick-table in it and
10575 reference it.
10576
10577 It is important to understand that stickiness based on learning information
10578 has some limitations, including the fact that all learned associations are
Baptiste Assmann123ff042016-03-06 23:29:28 +010010579 lost upon restart unless peers are properly configured to transfer such
10580 information upon restart (recommended). In general it can be good as a
10581 complement but not always as an exclusive stickiness.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010582
Willy Tarreauc9705a12010-07-27 20:05:50 +020010583 Last, memory requirements may be important when storing many data types.
10584 Indeed, storing all indicators above at once in each entry requires 116 bytes
10585 per entry, or 116 MB for a 1-million entries table. This is definitely not
10586 something that can be ignored.
10587
10588 Example:
10589 # Keep track of counters of up to 1 million IP addresses over 5 minutes
10590 # and store a general purpose counter and the average connection rate
10591 # computed over a sliding window of 30 seconds.
10592 stick-table type ip size 1m expire 5m store gpc0,conn_rate(30s)
10593
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +030010594 See also : "stick match", "stick on", "stick store-request", section 2.4
David du Colombiera13d1b92011-03-17 10:40:22 +010010595 about time format and section 7 about ACLs.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010596
10597
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010598stick store-response <pattern> [table <table>] [{if | unless} <condition>]
Baptiste Assmann2f2d2ec2016-03-06 23:27:24 +010010599 Define a response pattern used to create an entry in a stickiness table
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010600 May be used in sections : defaults | frontend | listen | backend
10601 no | no | yes | yes
10602
10603 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +020010604 <pattern> is a sample expression rule as described in section 7.3. It
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010605 describes what elements of the response or connection will
Davor Ocelice9ed2812017-12-25 17:49:28 +010010606 be analyzed, extracted and stored in the table once a
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010607 server is selected.
10608
10609 <table> is an optional stickiness table name. If unspecified, the same
10610 backend's table is used. A stickiness table is declared using
10611 the "stick-table" statement.
10612
10613 <cond> is an optional storage condition. It makes it possible to store
10614 certain criteria only when some conditions are met (or not met).
10615 For instance, it could be used to store the SSL session ID only
10616 when the response is a SSL server hello.
10617
10618 Some protocols or applications require complex stickiness rules and cannot
10619 always simply rely on cookies nor hashing. The "stick store-response"
10620 statement describes a rule to decide what to extract from the response and
10621 when to do it, in order to store it into a stickiness table for further
10622 requests to match it using the "stick match" statement. Obviously the
10623 extracted part must make sense and have a chance to be matched in a further
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010624 request. Storing an ID found in a header of a response makes sense.
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010625 See section 7 for a complete list of possible patterns and transformation
10626 rules.
10627
10628 The table has to be declared using the "stick-table" statement. It must be of
10629 a type compatible with the pattern. By default it is the one which is present
10630 in the same backend. It is possible to share a table with other backends by
10631 referencing it using the "table" keyword. If another table is referenced,
10632 the server's ID inside the backends are used. By default, all server IDs
10633 start at 1 in each backend, so the server ordering is enough. But in case of
10634 doubt, it is highly recommended to force server IDs using their "id" setting.
10635
10636 It is possible to restrict the conditions where a "stick store-response"
10637 statement will apply, using "if" or "unless" followed by a condition. This
10638 condition will be evaluated while parsing the response, so any criteria can
10639 be used. See section 7 for ACL based conditions.
10640
10641 There is no limit on the number of "stick store-response" statements, but
10642 there is a limit of 8 simultaneous stores per request or response. This
10643 makes it possible to store up to 8 criteria, all extracted from either the
10644 request or the response, regardless of the number of rules. Only the 8 first
10645 ones which match will be kept. Using this, it is possible to feed multiple
10646 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +010010647 another protocol or access method. Using multiple store-response rules with
10648 the same table is possible and may be used to find the best criterion to rely
10649 on, by arranging the rules by decreasing preference order. Only the first
10650 extracted criterion for a given table will be stored. All subsequent store-
10651 response rules referencing the same table will be skipped and their ACLs will
10652 not be evaluated. However, even if a store-request rule references a table, a
10653 store-response rule may also use the same table. This means that each table
10654 may learn exactly one element from the request and one element from the
10655 response at once.
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010656
10657 The table will contain the real server that processed the request.
10658
10659 Example :
10660 # Learn SSL session ID from both request and response and create affinity.
10661 backend https
10662 mode tcp
10663 balance roundrobin
Cyril Bontédc4d9032012-04-08 21:57:39 +020010664 # maximum SSL session ID length is 32 bytes.
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010665 stick-table type binary len 32 size 30k expire 30m
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010666
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010667 acl clienthello req_ssl_hello_type 1
10668 acl serverhello rep_ssl_hello_type 2
10669
10670 # use tcp content accepts to detects ssl client and server hello.
10671 tcp-request inspect-delay 5s
10672 tcp-request content accept if clienthello
10673
10674 # no timeout on response inspect delay by default.
10675 tcp-response content accept if serverhello
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010676
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010677 # SSL session ID (SSLID) may be present on a client or server hello.
10678 # Its length is coded on 1 byte at offset 43 and its value starts
10679 # at offset 44.
10680
10681 # Match and learn on request if client hello.
10682 stick on payload_lv(43,1) if clienthello
10683
10684 # Learn on response if server hello.
10685 stick store-response payload_lv(43,1) if serverhello
Cyril Bontédc4d9032012-04-08 21:57:39 +020010686
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010687 server s1 192.168.1.1:443
10688 server s2 192.168.1.1:443
10689
10690 See also : "stick-table", "stick on", and section 7 about ACLs and pattern
10691 extraction.
10692
10693
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020010694tcp-check comment <string>
10695 Defines a comment for the following the tcp-check rule, reported in logs if
10696 it fails.
10697 May be used in sections : defaults | frontend | listen | backend
10698 yes | no | yes | yes
10699
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020010700 Arguments :
10701 <string> is the comment message to add in logs if the following tcp-check
10702 rule fails.
10703
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020010704 It only works for connect, send and expect rules. It is useful to make
10705 user-friendly error reporting.
10706
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020010707 See also : "option tcp-check", "tcp-check connect", "tcp-check send" and
10708 "tcp-check expect".
10709
10710
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020010711tcp-check connect [default] [port <expr>] [addr <ip>] [send-proxy] [via-socks4]
10712 [ssl] [sni <sni>] [alpn <alpn>] [linger]
Christopher Fauletedc6ed92020-04-23 16:27:59 +020010713 [proto <name>] [comment <msg>]
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010714 Opens a new connection
10715 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020010716 yes | no | yes | yes
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010717
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020010718 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020010719 comment <msg> defines a message to report if the rule evaluation fails.
10720
Christopher Faulet4dce5922020-03-30 13:54:42 +020010721 default Use default options of the server line to do the health
Daniel Corbett67a82712020-07-06 23:01:19 -040010722 checks. The server options are used only if not redefined.
Christopher Faulet4dce5922020-03-30 13:54:42 +020010723
Christopher Fauletb7d30092020-03-30 15:19:03 +020010724 port <expr> if not set, check port or server port is used.
Christopher Faulet5c288742020-03-31 08:15:58 +020010725 It tells HAProxy where to open the connection to.
10726 <port> must be a valid TCP port source integer, from 1 to
Christopher Fauletb7d30092020-03-30 15:19:03 +020010727 65535 or an sample-fetch expression.
Christopher Faulet5c288742020-03-31 08:15:58 +020010728
10729 addr <ip> defines the IP address to do the health check.
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010730
10731 send-proxy send a PROXY protocol string
10732
Christopher Faulet085426a2020-03-30 13:07:02 +020010733 via-socks4 enables outgoing health checks using upstream socks4 proxy.
10734
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010735 ssl opens a ciphered connection
10736
Christopher Faulet79b31d42020-03-30 13:00:05 +020010737 sni <sni> specifies the SNI to use to do health checks over SSL.
10738
Christopher Faulet98572322020-03-30 13:16:44 +020010739 alpn <alpn> defines which protocols to advertise with ALPN. The protocol
10740 list consists in a comma-delimited list of protocol names,
10741 for instance: "http/1.1,http/1.0" (without quotes).
10742 If it is not set, the server ALPN is used.
10743
Christopher Fauletedc6ed92020-04-23 16:27:59 +020010744 proto <name> forces the multiplexer's protocol to use for this connection.
10745 It must be a TCP mux protocol and it must be usable on the
10746 backend side. The list of available protocols is reported in
10747 haproxy -vv.
10748
Christopher Faulet5c288742020-03-31 08:15:58 +020010749 linger cleanly close the connection instead of using a single RST.
Gaetan Rivetf8ba6772020-02-07 15:37:17 +010010750
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020010751 When an application lies on more than a single TCP port or when HAProxy
10752 load-balance many services in a single backend, it makes sense to probe all
10753 the services individually before considering a server as operational.
10754
10755 When there are no TCP port configured on the server line neither server port
10756 directive, then the 'tcp-check connect port <port>' must be the first step
10757 of the sequence.
10758
10759 In a tcp-check ruleset a 'connect' is required, it is also mandatory to start
10760 the ruleset with a 'connect' rule. Purpose is to ensure admin know what they
10761 do.
10762
10763 When a connect must start the ruleset, if may still be preceded by set-var,
10764 unset-var or comment rules.
10765
10766 Examples :
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010767 # check HTTP and HTTPs services on a server.
10768 # first open port 80 thanks to server line port directive, then
10769 # tcp-check opens port 443, ciphered and run a request on it:
10770 option tcp-check
10771 tcp-check connect
10772 tcp-check send GET\ /\ HTTP/1.0\r\n
10773 tcp-check send Host:\ haproxy.1wt.eu\r\n
10774 tcp-check send \r\n
10775 tcp-check expect rstring (2..|3..)
10776 tcp-check connect port 443 ssl
10777 tcp-check send GET\ /\ HTTP/1.0\r\n
10778 tcp-check send Host:\ haproxy.1wt.eu\r\n
10779 tcp-check send \r\n
10780 tcp-check expect rstring (2..|3..)
10781 server www 10.0.0.1 check port 80
10782
10783 # check both POP and IMAP from a single server:
10784 option tcp-check
Gaetan Rivetf8ba6772020-02-07 15:37:17 +010010785 tcp-check connect port 110 linger
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010786 tcp-check expect string +OK\ POP3\ ready
10787 tcp-check connect port 143
10788 tcp-check expect string *\ OK\ IMAP4\ ready
10789 server mail 10.0.0.1 check
10790
10791 See also : "option tcp-check", "tcp-check send", "tcp-check expect"
10792
10793
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020010794tcp-check expect [min-recv <int>] [comment <msg>]
Christopher Fauletec07e382020-04-07 14:56:26 +020010795 [ok-status <st>] [error-status <st>] [tout-status <st>]
Christopher Faulet98cc57c2020-04-01 20:52:31 +020010796 [on-success <fmt>] [on-error <fmt>] [status-code <expr>]
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020010797 [!] <match> <pattern>
Davor Ocelice9ed2812017-12-25 17:49:28 +010010798 Specify data to be collected and analyzed during a generic health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010799 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020010800 yes | no | yes | yes
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010801
10802 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020010803 comment <msg> defines a message to report if the rule evaluation fails.
10804
Gaetan Rivet1afd8262020-02-07 15:37:17 +010010805 min-recv is optional and can define the minimum amount of data required to
10806 evaluate the current expect rule. If the number of received bytes
10807 is under this limit, the check will wait for more data. This
10808 option can be used to resolve some ambiguous matching rules or to
10809 avoid executing costly regex matches on content known to be still
10810 incomplete. If an exact string (string or binary) is used, the
10811 minimum between the string length and this parameter is used.
10812 This parameter is ignored if it is set to -1. If the expect rule
10813 does not match, the check will wait for more data. If set to 0,
10814 the evaluation result is always conclusive.
10815
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010816 <match> is a keyword indicating how to look for a specific pattern in the
Gaetan Rivetefab6c62020-02-07 15:37:17 +010010817 response. The keyword may be one of "string", "rstring", "binary" or
10818 "rbinary".
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010819 The keyword may be preceded by an exclamation mark ("!") to negate
10820 the match. Spaces are allowed between the exclamation mark and the
10821 keyword. See below for more details on the supported keywords.
10822
Christopher Fauletec07e382020-04-07 14:56:26 +020010823 ok-status <st> is optional and can be used to set the check status if
10824 the expect rule is successfully evaluated and if it is
10825 the last rule in the tcp-check ruleset. "L7OK", "L7OKC",
Christopher Fauletd888f0f2020-05-07 07:40:17 +020010826 "L6OK" and "L4OK" are supported :
10827 - L7OK : check passed on layer 7
10828 - L7OKC : check conditionally passed on layer 7, for
10829 example 404 with disable-on-404
10830 - L6OK : check passed on layer 6
10831 - L4OK : check passed on layer 4
Christopher Fauletec07e382020-04-07 14:56:26 +020010832 By default "L7OK" is used.
10833
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020010834 error-status <st> is optional and can be used to set the check status if
10835 an error occurred during the expect rule evaluation.
Christopher Fauletd888f0f2020-05-07 07:40:17 +020010836 "L7RSP", "L7STS", "L6RSP" and "L4CON" are supported :
10837 - L7RSP : layer 7 invalid response - protocol error
10838 - L7STS : layer 7 response error, for example HTTP 5xx
10839 - L6RSP : layer 6 invalid response - protocol error
10840 - L4CON : layer 1-4 connection problem
10841 By default "L7RSP" is used.
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020010842
Christopher Fauletec07e382020-04-07 14:56:26 +020010843 tout-status <st> is optional and can be used to set the check status if
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020010844 a timeout occurred during the expect rule evaluation.
Christopher Fauletd888f0f2020-05-07 07:40:17 +020010845 "L7TOUT", "L6TOUT", and "L4TOUT" are supported :
10846 - L7TOUT : layer 7 (HTTP/SMTP) timeout
10847 - L6TOUT : layer 6 (SSL) timeout
10848 - L4TOUT : layer 1-4 timeout
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020010849 By default "L7TOUT" is used.
10850
Christopher Fauletbe52b4d2020-04-01 16:30:22 +020010851 on-success <fmt> is optional and can be used to customize the
10852 informational message reported in logs if the expect
10853 rule is successfully evaluated and if it is the last rule
10854 in the tcp-check ruleset. <fmt> is a log-format string.
10855
10856 on-error <fmt> is optional and can be used to customize the
10857 informational message reported in logs if an error
10858 occurred during the expect rule evaluation. <fmt> is a
10859 log-format string.
10860
Christopher Faulet98cc57c2020-04-01 20:52:31 +020010861 status-code <expr> is optional and can be used to set the check status code
10862 reported in logs, on success or on error. <expr> is a
10863 standard HAProxy expression formed by a sample-fetch
10864 followed by some converters.
10865
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010866 <pattern> is the pattern to look for. It may be a string or a regular
10867 expression. If the pattern contains spaces, they must be escaped
10868 with the usual backslash ('\').
10869 If the match is set to binary, then the pattern must be passed as
Davor Ocelice9ed2812017-12-25 17:49:28 +010010870 a series of hexadecimal digits in an even number. Each sequence of
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010871 two digits will represent a byte. The hexadecimal digits may be
10872 used upper or lower case.
10873
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010874 The available matches are intentionally similar to their http-check cousins :
10875
10876 string <string> : test the exact string matches in the response buffer.
10877 A health check response will be considered valid if the
10878 response's buffer contains this exact string. If the
10879 "string" keyword is prefixed with "!", then the response
10880 will be considered invalid if the body contains this
10881 string. This can be used to look for a mandatory pattern
10882 in a protocol response, or to detect a failure when a
10883 specific error appears in a protocol banner.
10884
10885 rstring <regex> : test a regular expression on the response buffer.
10886 A health check response will be considered valid if the
10887 response's buffer matches this expression. If the
10888 "rstring" keyword is prefixed with "!", then the response
10889 will be considered invalid if the body matches the
10890 expression.
10891
Christopher Fauletaaab0832020-05-05 15:54:22 +020010892 string-lf <fmt> : test a log-format string match in the response's buffer.
10893 A health check response will be considered valid if the
10894 response's buffer contains the string resulting of the
10895 evaluation of <fmt>, which follows the log-format rules.
10896 If prefixed with "!", then the response will be
10897 considered invalid if the buffer contains the string.
10898
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010899 binary <hexstring> : test the exact string in its hexadecimal form matches
10900 in the response buffer. A health check response will
10901 be considered valid if the response's buffer contains
10902 this exact hexadecimal string.
10903 Purpose is to match data on binary protocols.
10904
Gaetan Rivetefab6c62020-02-07 15:37:17 +010010905 rbinary <regex> : test a regular expression on the response buffer, like
10906 "rstring". However, the response buffer is transformed
10907 into its hexadecimal form, including NUL-bytes. This
10908 allows using all regex engines to match any binary
10909 content. The hexadecimal transformation takes twice the
10910 size of the original response. As such, the expected
10911 pattern should work on at-most half the response buffer
10912 size.
10913
Christopher Fauletaaab0832020-05-05 15:54:22 +020010914 binary-lf <hexfmt> : test a log-format string in its hexadecimal form
10915 match in the response's buffer. A health check response
10916 will be considered valid if the response's buffer
10917 contains the hexadecimal string resulting of the
10918 evaluation of <fmt>, which follows the log-format
10919 rules. If prefixed with "!", then the response will be
10920 considered invalid if the buffer contains the
10921 hexadecimal string. The hexadecimal string is converted
10922 in a binary string before matching the response's
10923 buffer.
10924
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010925 It is important to note that the responses will be limited to a certain size
10926 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
10927 Thus, too large responses may not contain the mandatory pattern when using
10928 "string", "rstring" or binary. If a large response is absolutely required, it
10929 is possible to change the default max size by setting the global variable.
10930 However, it is worth keeping in mind that parsing very large responses can
10931 waste some CPU cycles, especially when regular expressions are used, and that
10932 it is always better to focus the checks on smaller resources. Also, in its
10933 current state, the check will not find any string nor regex past a null
10934 character in the response. Similarly it is not possible to request matching
10935 the null character.
10936
10937 Examples :
10938 # perform a POP check
10939 option tcp-check
10940 tcp-check expect string +OK\ POP3\ ready
10941
10942 # perform an IMAP check
10943 option tcp-check
10944 tcp-check expect string *\ OK\ IMAP4\ ready
10945
10946 # look for the redis master server
10947 option tcp-check
10948 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +020010949 tcp-check expect string +PONG
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010950 tcp-check send info\ replication\r\n
10951 tcp-check expect string role:master
10952 tcp-check send QUIT\r\n
10953 tcp-check expect string +OK
10954
10955
10956 See also : "option tcp-check", "tcp-check connect", "tcp-check send",
10957 "tcp-check send-binary", "http-check expect", tune.chksize
10958
10959
Christopher Fauletb50b3e62020-05-05 18:43:43 +020010960tcp-check send <data> [comment <msg>]
10961tcp-check send-lf <fmt> [comment <msg>]
10962 Specify a string or a log-format string to be sent as a question during a
10963 generic health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010964 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020010965 yes | no | yes | yes
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010966
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020010967 Arguments :
10968 comment <msg> defines a message to report if the rule evaluation fails.
10969
Christopher Fauletb50b3e62020-05-05 18:43:43 +020010970 <data> is the string that will be sent during a generic health
10971 check session.
Christopher Faulet16fff672020-04-30 07:50:54 +020010972
Christopher Fauletb50b3e62020-05-05 18:43:43 +020010973 <fmt> is the log-format string that will be sent, once evaluated,
10974 during a generic health check session.
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010975
10976 Examples :
10977 # look for the redis master server
10978 option tcp-check
10979 tcp-check send info\ replication\r\n
10980 tcp-check expect string role:master
10981
10982 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
10983 "tcp-check send-binary", tune.chksize
10984
10985
Christopher Fauletb50b3e62020-05-05 18:43:43 +020010986tcp-check send-binary <hexstring> [comment <msg>]
10987tcp-check send-binary-lf <hexfmt> [comment <msg>]
10988 Specify an hex digits string or an hex digits log-format string to be sent as
10989 a binary question during a raw tcp health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010990 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020010991 yes | no | yes | yes
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010992
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020010993 Arguments :
10994 comment <msg> defines a message to report if the rule evaluation fails.
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020010995
Christopher Fauletb50b3e62020-05-05 18:43:43 +020010996 <hexstring> is the hexadecimal string that will be send, once converted
10997 to binary, during a generic health check session.
Christopher Faulet16fff672020-04-30 07:50:54 +020010998
Christopher Fauletb50b3e62020-05-05 18:43:43 +020010999 <hexfmt> is the hexadecimal log-format string that will be send, once
11000 evaluated and converted to binary, during a generic health
11001 check session.
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011002
11003 Examples :
11004 # redis check in binary
11005 option tcp-check
11006 tcp-check send-binary 50494e470d0a # PING\r\n
11007 tcp-check expect binary 2b504F4e47 # +PONG
11008
11009
11010 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
11011 "tcp-check send", tune.chksize
11012
11013
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011014tcp-check set-var(<var-name>) <expr>
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011015 This operation sets the content of a variable. The variable is declared inline.
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011016 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020011017 yes | no | yes | yes
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011018
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011019 Arguments :
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011020 <var-name> The name of the variable starts with an indication about its
11021 scope. The scopes allowed for tcp-check are:
11022 "proc" : the variable is shared with the whole process.
11023 "sess" : the variable is shared with the tcp-check session.
11024 "check": the variable is declared for the lifetime of the tcp-check.
11025 This prefix is followed by a name. The separator is a '.'.
11026 The name may only contain characters 'a-z', 'A-Z', '0-9', '.',
11027 and '-'.
11028
11029 <expr> Is a sample-fetch expression potentially followed by converters.
11030
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011031 Examples :
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011032 tcp-check set-var(check.port) int(1234)
11033
11034
11035tcp-check unset-var(<var-name>)
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011036 Free a reference to a variable within its scope.
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011037 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020011038 yes | no | yes | yes
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011039
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011040 Arguments :
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011041 <var-name> The name of the variable starts with an indication about its
11042 scope. The scopes allowed for tcp-check are:
11043 "proc" : the variable is shared with the whole process.
11044 "sess" : the variable is shared with the tcp-check session.
11045 "check": the variable is declared for the lifetime of the tcp-check.
11046 This prefix is followed by a name. The separator is a '.'.
11047 The name may only contain characters 'a-z', 'A-Z', '0-9', '.',
11048 and '-'.
11049
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011050 Examples :
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011051 tcp-check unset-var(check.port)
11052
11053
Willy Tarreaue9656522010-08-17 15:40:09 +020011054tcp-request connection <action> [{if | unless} <condition>]
11055 Perform an action on an incoming connection depending on a layer 4 condition
Willy Tarreau1a687942010-05-23 22:40:30 +020011056 May be used in sections : defaults | frontend | listen | backend
11057 no | yes | yes | no
Willy Tarreaue9656522010-08-17 15:40:09 +020011058 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +020011059 <action> defines the action to perform if the condition applies. See
11060 below.
Willy Tarreau1a687942010-05-23 22:40:30 +020011061
Willy Tarreaue9656522010-08-17 15:40:09 +020011062 <condition> is a standard layer4-only ACL-based condition (see section 7).
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011063
11064 Immediately after acceptance of a new incoming connection, it is possible to
11065 evaluate some conditions to decide whether this connection must be accepted
Willy Tarreaue9656522010-08-17 15:40:09 +020011066 or dropped or have its counters tracked. Those conditions cannot make use of
11067 any data contents because the connection has not been read from yet, and the
11068 buffers are not yet allocated. This is used to selectively and very quickly
11069 accept or drop connections from various sources with a very low overhead. If
11070 some contents need to be inspected in order to take the decision, the
11071 "tcp-request content" statements must be used instead.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011072
Willy Tarreaue9656522010-08-17 15:40:09 +020011073 The "tcp-request connection" rules are evaluated in their exact declaration
11074 order. If no rule matches or if there is no rule, the default action is to
11075 accept the incoming connection. There is no specific limit to the number of
11076 rules which may be inserted.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011077
Willy Tarreaua9083d02015-05-08 15:27:59 +020011078 Four types of actions are supported :
Willy Tarreaue9656522010-08-17 15:40:09 +020011079 - accept :
11080 accepts the connection if the condition is true (when used with "if")
11081 or false (when used with "unless"). The first such rule executed ends
11082 the rules evaluation.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011083
Willy Tarreaue9656522010-08-17 15:40:09 +020011084 - reject :
11085 rejects the connection if the condition is true (when used with "if")
11086 or false (when used with "unless"). The first such rule executed ends
11087 the rules evaluation. Rejected connections do not even become a
11088 session, which is why they are accounted separately for in the stats,
11089 as "denied connections". They are not considered for the session
11090 rate-limit and are not logged either. The reason is that these rules
11091 should only be used to filter extremely high connection rates such as
11092 the ones encountered during a massive DDoS attack. Under these extreme
11093 conditions, the simple action of logging each event would make the
11094 system collapse and would considerably lower the filtering capacity. If
11095 logging is absolutely desired, then "tcp-request content" rules should
Willy Tarreau4f614292016-10-21 17:49:36 +020011096 be used instead, as "tcp-request session" rules will not log either.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011097
Willy Tarreau4f0d9192013-06-11 20:40:55 +020011098 - expect-proxy layer4 :
11099 configures the client-facing connection to receive a PROXY protocol
11100 header before any byte is read from the socket. This is equivalent to
11101 having the "accept-proxy" keyword on the "bind" line, except that using
11102 the TCP rule allows the PROXY protocol to be accepted only for certain
11103 IP address ranges using an ACL. This is convenient when multiple layers
11104 of load balancers are passed through by traffic coming from public
11105 hosts.
11106
Bertrand Jacquin90759682016-06-06 15:35:39 +010011107 - expect-netscaler-cip layer4 :
11108 configures the client-facing connection to receive a NetScaler Client
11109 IP insertion protocol header before any byte is read from the socket.
11110 This is equivalent to having the "accept-netscaler-cip" keyword on the
11111 "bind" line, except that using the TCP rule allows the PROXY protocol
11112 to be accepted only for certain IP address ranges using an ACL. This
11113 is convenient when multiple layers of load balancers are passed
11114 through by traffic coming from public hosts.
11115
Willy Tarreau18bf01e2014-06-13 16:18:52 +020011116 - capture <sample> len <length> :
11117 This only applies to "tcp-request content" rules. It captures sample
11118 expression <sample> from the request buffer, and converts it to a
11119 string of at most <len> characters. The resulting string is stored into
11120 the next request "capture" slot, so it will possibly appear next to
11121 some captured HTTP headers. It will then automatically appear in the
11122 logs, and it will be possible to extract it using sample fetch rules to
11123 feed it into headers or anything. The length should be limited given
11124 that this size will be allocated for each capture during the whole
Willy Tarreaua9083d02015-05-08 15:27:59 +020011125 session life. Please check section 7.3 (Fetching samples) and "capture
11126 request header" for more information.
Willy Tarreau18bf01e2014-06-13 16:18:52 +020011127
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011128 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] :
Willy Tarreaue9656522010-08-17 15:40:09 +020011129 enables tracking of sticky counters from current connection. These
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020011130 rules do not stop evaluation and do not change default action. The
11131 number of counters that may be simultaneously tracked by the same
11132 connection is set in MAX_SESS_STKCTR at build time (reported in
John Roeslerfb2fce12019-07-10 15:45:51 -050011133 haproxy -vv) which defaults to 3, so the track-sc number is between 0
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020011134 and (MAX_SESS_STCKTR-1). The first "track-sc0" rule executed enables
11135 tracking of the counters of the specified table as the first set. The
11136 first "track-sc1" rule executed enables tracking of the counters of the
11137 specified table as the second set. The first "track-sc2" rule executed
11138 enables tracking of the counters of the specified table as the third
11139 set. It is a recommended practice to use the first set of counters for
11140 the per-frontend counters and the second set for the per-backend ones.
11141 But this is just a guideline, all may be used everywhere.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011142
Willy Tarreaue9656522010-08-17 15:40:09 +020011143 These actions take one or two arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +020011144 <key> is mandatory, and is a sample expression rule as described
Willy Tarreau74ca5042013-06-11 23:12:07 +020011145 in section 7.3. It describes what elements of the incoming
Davor Ocelice9ed2812017-12-25 17:49:28 +010011146 request or connection will be analyzed, extracted, combined,
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011147 and used to select which table entry to update the counters.
11148 Note that "tcp-request connection" cannot use content-based
11149 fetches.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011150
Willy Tarreaue9656522010-08-17 15:40:09 +020011151 <table> is an optional table to be used instead of the default one,
11152 which is the stick-table declared in the current proxy. All
11153 the counters for the matches and updates for the key will
11154 then be performed in that table until the session ends.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011155
Willy Tarreaue9656522010-08-17 15:40:09 +020011156 Once a "track-sc*" rule is executed, the key is looked up in the table
11157 and if it is not found, an entry is allocated for it. Then a pointer to
11158 that entry is kept during all the session's life, and this entry's
11159 counters are updated as often as possible, every time the session's
11160 counters are updated, and also systematically when the session ends.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011161 Counters are only updated for events that happen after the tracking has
11162 been started. For example, connection counters will not be updated when
11163 tracking layer 7 information, since the connection event happens before
11164 layer7 information is extracted.
11165
Willy Tarreaue9656522010-08-17 15:40:09 +020011166 If the entry tracks concurrent connection counters, one connection is
11167 counted for as long as the entry is tracked, and the entry will not
11168 expire during that time. Tracking counters also provides a performance
11169 advantage over just checking the keys, because only one table lookup is
11170 performed for all ACL checks that make use of it.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011171
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +020011172 - sc-inc-gpc0(<sc-id>):
11173 The "sc-inc-gpc0" increments the GPC0 counter according to the sticky
11174 counter designated by <sc-id>. If an error occurs, this action silently
11175 fails and the actions evaluation continues.
11176
Frédéric Lécaille6778b272018-01-29 15:22:53 +010011177 - sc-inc-gpc1(<sc-id>):
11178 The "sc-inc-gpc1" increments the GPC1 counter according to the sticky
11179 counter designated by <sc-id>. If an error occurs, this action silently
11180 fails and the actions evaluation continues.
11181
Cédric Dufour0d7712d2019-11-06 18:38:53 +010011182 - sc-set-gpt0(<sc-id>) { <int> | <expr> }:
11183 This action sets the 32-bit unsigned GPT0 tag according to the sticky
11184 counter designated by <sc-id> and the value of <int>/<expr>. The
11185 expected result is a boolean. If an error occurs, this action silently
11186 fails and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +020011187
William Lallemand2e785f22016-05-25 01:48:42 +020011188 - set-src <expr> :
11189 Is used to set the source IP address to the value of specified
11190 expression. Useful if you want to mask source IP for privacy.
11191 If you want to provide an IP from a HTTP header use "http-request
Cyril Bonté6c81d5f2018-10-17 00:14:51 +020011192 set-src".
William Lallemand2e785f22016-05-25 01:48:42 +020011193
Cyril Bonté6c81d5f2018-10-17 00:14:51 +020011194 Arguments:
11195 <expr> Is a standard HAProxy expression formed by a sample-fetch
11196 followed by some converters.
William Lallemand2e785f22016-05-25 01:48:42 +020011197
11198 Example:
William Lallemand2e785f22016-05-25 01:48:42 +020011199 tcp-request connection set-src src,ipmask(24)
11200
Willy Tarreau0c630532016-10-21 17:52:58 +020011201 When possible, set-src preserves the original source port as long as the
11202 address family allows it, otherwise the source port is set to 0.
William Lallemand2e785f22016-05-25 01:48:42 +020011203
William Lallemand44be6402016-05-25 01:51:35 +020011204 - set-src-port <expr> :
11205 Is used to set the source port address to the value of specified
11206 expression.
11207
Cyril Bonté6c81d5f2018-10-17 00:14:51 +020011208 Arguments:
11209 <expr> Is a standard HAProxy expression formed by a sample-fetch
11210 followed by some converters.
William Lallemand44be6402016-05-25 01:51:35 +020011211
11212 Example:
William Lallemand44be6402016-05-25 01:51:35 +020011213 tcp-request connection set-src-port int(4000)
11214
Willy Tarreau0c630532016-10-21 17:52:58 +020011215 When possible, set-src-port preserves the original source address as long
11216 as the address family supports a port, otherwise it forces the source
11217 address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +020011218
William Lallemand13e9b0c2016-05-25 02:34:07 +020011219 - set-dst <expr> :
11220 Is used to set the destination IP address to the value of specified
11221 expression. Useful if you want to mask IP for privacy in log.
11222 If you want to provide an IP from a HTTP header use "http-request
11223 set-dst". If you want to connect to the new address/port, use
11224 '0.0.0.0:0' as a server address in the backend.
11225
11226 <expr> Is a standard HAProxy expression formed by a sample-fetch
11227 followed by some converters.
11228
11229 Example:
11230
11231 tcp-request connection set-dst dst,ipmask(24)
11232 tcp-request connection set-dst ipv4(10.0.0.1)
11233
Willy Tarreau0c630532016-10-21 17:52:58 +020011234 When possible, set-dst preserves the original destination port as long as
11235 the address family allows it, otherwise the destination port is set to 0.
11236
William Lallemand13e9b0c2016-05-25 02:34:07 +020011237 - set-dst-port <expr> :
11238 Is used to set the destination port address to the value of specified
11239 expression. If you want to connect to the new address/port, use
11240 '0.0.0.0:0' as a server address in the backend.
11241
11242
11243 <expr> Is a standard HAProxy expression formed by a sample-fetch
11244 followed by some converters.
11245
11246 Example:
11247
11248 tcp-request connection set-dst-port int(4000)
11249
Willy Tarreau0c630532016-10-21 17:52:58 +020011250 When possible, set-dst-port preserves the original destination address as
11251 long as the address family supports a port, otherwise it forces the
11252 destination address to IPv4 "0.0.0.0" before rewriting the port.
11253
Willy Tarreau2d392c22015-08-24 01:43:45 +020011254 - "silent-drop" :
11255 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +010011256 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +020011257 to prevent the client from being notified. The effect it then that the
11258 client still sees an established connection while there's none on
11259 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
11260 except that it doesn't use any local resource at all on the machine
11261 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +010011262 slow down stronger attackers. It is important to understand the impact
11263 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +020011264 client and HAProxy (firewalls, proxies, load balancers) will also keep
11265 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +010011266 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +020011267 TCP_REPAIR socket option is used to block the emission of a TCP
11268 reset. On other systems, the socket's TTL is reduced to 1 so that the
11269 TCP reset doesn't pass the first router, though it's still delivered to
11270 local networks. Do not use it unless you fully understand how it works.
11271
Willy Tarreaue9656522010-08-17 15:40:09 +020011272 Note that the "if/unless" condition is optional. If no condition is set on
11273 the action, it is simply performed unconditionally. That can be useful for
11274 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011275
Willy Tarreaue9656522010-08-17 15:40:09 +020011276 Example: accept all connections from white-listed hosts, reject too fast
11277 connection without counting them, and track accepted connections.
11278 This results in connection rate being capped from abusive sources.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011279
Willy Tarreaue9656522010-08-17 15:40:09 +020011280 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011281 tcp-request connection reject if { src_conn_rate gt 10 }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011282 tcp-request connection track-sc0 src
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011283
Willy Tarreaue9656522010-08-17 15:40:09 +020011284 Example: accept all connections from white-listed hosts, count all other
11285 connections and reject too fast ones. This results in abusive ones
11286 being blocked as long as they don't slow down.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011287
Willy Tarreaue9656522010-08-17 15:40:09 +020011288 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011289 tcp-request connection track-sc0 src
11290 tcp-request connection reject if { sc0_conn_rate gt 10 }
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011291
Willy Tarreau4f0d9192013-06-11 20:40:55 +020011292 Example: enable the PROXY protocol for traffic coming from all known proxies.
11293
11294 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
11295
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011296 See section 7 about ACL usage.
11297
Willy Tarreau4f614292016-10-21 17:49:36 +020011298 See also : "tcp-request session", "tcp-request content", "stick-table"
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011299
11300
Willy Tarreaue9656522010-08-17 15:40:09 +020011301tcp-request content <action> [{if | unless} <condition>]
11302 Perform an action on a new session depending on a layer 4-7 condition
Willy Tarreau62644772008-07-16 18:36:06 +020011303 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +020011304 no | yes | yes | yes
Willy Tarreaue9656522010-08-17 15:40:09 +020011305 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +020011306 <action> defines the action to perform if the condition applies. See
11307 below.
Willy Tarreau62644772008-07-16 18:36:06 +020011308
Willy Tarreaue9656522010-08-17 15:40:09 +020011309 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
Willy Tarreau62644772008-07-16 18:36:06 +020011310
Davor Ocelice9ed2812017-12-25 17:49:28 +010011311 A request's contents can be analyzed at an early stage of request processing
Willy Tarreaue9656522010-08-17 15:40:09 +020011312 called "TCP content inspection". During this stage, ACL-based rules are
11313 evaluated every time the request contents are updated, until either an
11314 "accept" or a "reject" rule matches, or the TCP request inspection delay
11315 expires with no matching rule.
Willy Tarreau62644772008-07-16 18:36:06 +020011316
Willy Tarreaue9656522010-08-17 15:40:09 +020011317 The first difference between these rules and "tcp-request connection" rules
11318 is that "tcp-request content" rules can make use of contents to take a
11319 decision. Most often, these decisions will consider a protocol recognition or
11320 validity. The second difference is that content-based rules can be used in
Willy Tarreauf3338342014-01-28 21:40:28 +010011321 both frontends and backends. In case of HTTP keep-alive with the client, all
11322 tcp-request content rules are evaluated again, so haproxy keeps a record of
11323 what sticky counters were assigned by a "tcp-request connection" versus a
11324 "tcp-request content" rule, and flushes all the content-related ones after
11325 processing an HTTP request, so that they may be evaluated again by the rules
11326 being evaluated again for the next request. This is of particular importance
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011327 when the rule tracks some L7 information or when it is conditioned by an
Willy Tarreauf3338342014-01-28 21:40:28 +010011328 L7-based ACL, since tracking may change between requests.
Willy Tarreau62644772008-07-16 18:36:06 +020011329
Willy Tarreaue9656522010-08-17 15:40:09 +020011330 Content-based rules are evaluated in their exact declaration order. If no
11331 rule matches or if there is no rule, the default action is to accept the
11332 contents. There is no specific limit to the number of rules which may be
11333 inserted.
Willy Tarreau62644772008-07-16 18:36:06 +020011334
Thierry FOURNIER236657b2015-08-19 08:25:14 +020011335 Several types of actions are supported :
Willy Tarreau18bf01e2014-06-13 16:18:52 +020011336 - accept : the request is accepted
Baptiste Assmann333939c2019-01-21 08:34:50 +010011337 - do-resolve: perform a DNS resolution
Willy Tarreau18bf01e2014-06-13 16:18:52 +020011338 - reject : the request is rejected and the connection is closed
11339 - capture : the specified sample expression is captured
Patrick Hemmer268a7072018-05-11 12:52:31 -040011340 - set-priority-class <expr> | set-priority-offset <expr>
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011341 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +020011342 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +010011343 - sc-inc-gpc1(<sc-id>)
Cédric Dufour0d7712d2019-11-06 18:38:53 +010011344 - sc-set-gpt0(<sc-id>) { <int> | <expr> }
Baptiste Assmanne1afd4f2019-04-18 16:21:13 +020011345 - set-dst <expr>
11346 - set-dst-port <expr>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011347 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +010011348 - unset-var(<var-name>)
Willy Tarreau2d392c22015-08-24 01:43:45 +020011349 - silent-drop
Davor Ocelice9ed2812017-12-25 17:49:28 +010011350 - send-spoe-group <engine-name> <group-name>
Christopher Faulet579d83b2019-11-22 15:34:17 +010011351 - use-service <service-name>
Willy Tarreau62644772008-07-16 18:36:06 +020011352
Willy Tarreaue9656522010-08-17 15:40:09 +020011353 They have the same meaning as their counter-parts in "tcp-request connection"
11354 so please refer to that section for a complete description.
Baptiste Assmann333939c2019-01-21 08:34:50 +010011355 For "do-resolve" action, please check the "http-request do-resolve"
11356 configuration section.
Willy Tarreau62644772008-07-16 18:36:06 +020011357
Willy Tarreauf3338342014-01-28 21:40:28 +010011358 While there is nothing mandatory about it, it is recommended to use the
11359 track-sc0 in "tcp-request connection" rules, track-sc1 for "tcp-request
11360 content" rules in the frontend, and track-sc2 for "tcp-request content"
11361 rules in the backend, because that makes the configuration more readable
11362 and easier to troubleshoot, but this is just a guideline and all counters
11363 may be used everywhere.
Willy Tarreau62644772008-07-16 18:36:06 +020011364
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011365 Note that the "if/unless" condition is optional. If no condition is set on
Willy Tarreaue9656522010-08-17 15:40:09 +020011366 the action, it is simply performed unconditionally. That can be useful for
11367 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau62644772008-07-16 18:36:06 +020011368
Christopher Faulet2079a4a2020-10-02 11:48:57 +020011369 Note also that it is recommended to use a "tcp-request session" rule to track
11370 information that does *not* depend on Layer 7 contents, especially for HTTP
11371 frontends. Some HTTP processing are performed at the session level and may
11372 lead to an early rejection of the requests. Thus, the tracking at the content
11373 level may be disturbed in such case. A warning is emitted during startup to
11374 prevent, as far as possible, such unreliable usage.
11375
Willy Tarreaue9656522010-08-17 15:40:09 +020011376 It is perfectly possible to match layer 7 contents with "tcp-request content"
Christopher Faulet7ea509e2020-10-02 11:38:46 +020011377 rules from a TCP proxy, since HTTP-specific ACL matches are able to
11378 preliminarily parse the contents of a buffer before extracting the required
11379 data. If the buffered contents do not parse as a valid HTTP message, then the
11380 ACL does not match. The parser which is involved there is exactly the same
11381 as for all other HTTP processing, so there is no risk of parsing something
11382 differently. In an HTTP frontend or an HTTP backend, it is guaranteed that
11383 HTTP contents will always be immediately present when the rule is evaluated
11384 first because the HTTP parsing is performed in the early stages of the
11385 connection processing, at the session level. But for such proxies, using
11386 "http-request" rules is much more natural and recommended.
Willy Tarreau62644772008-07-16 18:36:06 +020011387
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011388 Tracking layer7 information is also possible provided that the information
Willy Tarreau4d54c7c2014-09-16 15:48:15 +020011389 are present when the rule is processed. The rule processing engine is able to
11390 wait until the inspect delay expires when the data to be tracked is not yet
11391 available.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011392
Baptiste Assmanne1afd4f2019-04-18 16:21:13 +020011393 The "set-dst" and "set-dst-port" are used to set respectively the destination
11394 IP and port. More information on how to use it at "http-request set-dst".
11395
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011396 The "set-var" is used to set the content of a variable. The variable is
Willy Tarreau4f614292016-10-21 17:49:36 +020011397 declared inline. For "tcp-request session" rules, only session-level
11398 variables can be used, without any layer7 contents.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011399
Daniel Schneller0b547052016-03-21 20:46:57 +010011400 <var-name> The name of the variable starts with an indication about
11401 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010011402 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010011403 "sess" : the variable is shared with the whole session
11404 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011405 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +010011406 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011407 processing
Daniel Schneller0b547052016-03-21 20:46:57 +010011408 "res" : the variable is shared only during response
11409 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011410 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +010011411 The name may only contain characters 'a-z', 'A-Z', '0-9',
11412 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011413
11414 <expr> Is a standard HAProxy expression formed by a sample-fetch
11415 followed by some converters.
11416
Christopher Faulet85d79c92016-11-09 16:54:56 +010011417 The "unset-var" is used to unset a variable. See above for details about
11418 <var-name>.
11419
Patrick Hemmer268a7072018-05-11 12:52:31 -040011420 The "set-priority-class" is used to set the queue priority class of the
11421 current request. The value must be a sample expression which converts to an
11422 integer in the range -2047..2047. Results outside this range will be
11423 truncated. The priority class determines the order in which queued requests
11424 are processed. Lower values have higher priority.
11425
11426 The "set-priority-offset" is used to set the queue priority timestamp offset
11427 of the current request. The value must be a sample expression which converts
11428 to an integer in the range -524287..524287. Results outside this range will be
11429 truncated. When a request is queued, it is ordered first by the priority
11430 class, then by the current timestamp adjusted by the given offset in
11431 milliseconds. Lower values have higher priority.
11432 Note that the resulting timestamp is is only tracked with enough precision for
11433 524,287ms (8m44s287ms). If the request is queued long enough to where the
11434 adjusted timestamp exceeds this value, it will be misidentified as highest
11435 priority. Thus it is important to set "timeout queue" to a value, where when
11436 combined with the offset, does not exceed this limit.
11437
Christopher Faulet76c09ef2017-09-21 11:03:52 +020011438 The "send-spoe-group" is used to trigger sending of a group of SPOE
11439 messages. To do so, the SPOE engine used to send messages must be defined, as
11440 well as the SPOE group to send. Of course, the SPOE engine must refer to an
11441 existing SPOE filter. If not engine name is provided on the SPOE filter line,
11442 the SPOE agent name must be used.
11443
11444 <engine-name> The SPOE engine name.
11445
11446 <group-name> The SPOE group name as specified in the engine configuration.
11447
Christopher Faulet579d83b2019-11-22 15:34:17 +010011448 The "use-service" is used to executes a TCP service which will reply to the
11449 request and stop the evaluation of the rules. This service may choose to
11450 reply by sending any valid response or it may immediately close the
11451 connection without sending anything. Outside natives services, it is possible
11452 to write your own services in Lua. No further "tcp-request" rules are
11453 evaluated.
11454
11455 Example:
11456 tcp-request content use-service lua.deny { src -f /etc/haproxy/blacklist.lst }
11457
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011458 Example:
11459
11460 tcp-request content set-var(sess.my_var) src
Christopher Faulet85d79c92016-11-09 16:54:56 +010011461 tcp-request content unset-var(sess.my_var2)
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011462
Willy Tarreau62644772008-07-16 18:36:06 +020011463 Example:
Willy Tarreaue9656522010-08-17 15:40:09 +020011464 # Accept HTTP requests containing a Host header saying "example.com"
11465 # and reject everything else.
11466 acl is_host_com hdr(Host) -i example.com
11467 tcp-request inspect-delay 30s
Willy Tarreauc0239e02012-04-16 14:42:55 +020011468 tcp-request content accept if is_host_com
Willy Tarreaue9656522010-08-17 15:40:09 +020011469 tcp-request content reject
11470
11471 Example:
Willy Tarreau62644772008-07-16 18:36:06 +020011472 # reject SMTP connection if client speaks first
11473 tcp-request inspect-delay 30s
11474 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011475 tcp-request content reject if content_present
Willy Tarreau62644772008-07-16 18:36:06 +020011476
11477 # Forward HTTPS connection only if client speaks
11478 tcp-request inspect-delay 30s
11479 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011480 tcp-request content accept if content_present
Willy Tarreaue9656522010-08-17 15:40:09 +020011481 tcp-request content reject
11482
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011483 Example:
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030011484 # Track the last IP(stick-table type string) from X-Forwarded-For
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011485 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +020011486 tcp-request content track-sc0 hdr(x-forwarded-for,-1)
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030011487 # Or track the last IP(stick-table type ip|ipv6) from X-Forwarded-For
11488 tcp-request content track-sc0 req.hdr_ip(x-forwarded-for,-1)
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011489
11490 Example:
11491 # track request counts per "base" (concatenation of Host+URL)
11492 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +020011493 tcp-request content track-sc0 base table req-rate
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011494
Willy Tarreaue9656522010-08-17 15:40:09 +020011495 Example: track per-frontend and per-backend counters, block abusers at the
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030011496 frontend when the backend detects abuse(and marks gpc0).
Willy Tarreaue9656522010-08-17 15:40:09 +020011497
11498 frontend http
Davor Ocelice9ed2812017-12-25 17:49:28 +010011499 # Use General Purpose Counter 0 in SC0 as a global abuse counter
Willy Tarreaue9656522010-08-17 15:40:09 +020011500 # protecting all our sites
11501 stick-table type ip size 1m expire 5m store gpc0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011502 tcp-request connection track-sc0 src
11503 tcp-request connection reject if { sc0_get_gpc0 gt 0 }
Willy Tarreaue9656522010-08-17 15:40:09 +020011504 ...
11505 use_backend http_dynamic if { path_end .php }
11506
11507 backend http_dynamic
11508 # if a source makes too fast requests to this dynamic site (tracked
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011509 # by SC1), block it globally in the frontend.
Willy Tarreaue9656522010-08-17 15:40:09 +020011510 stick-table type ip size 1m expire 5m store http_req_rate(10s)
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011511 acl click_too_fast sc1_http_req_rate gt 10
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030011512 acl mark_as_abuser sc0_inc_gpc0(http) gt 0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011513 tcp-request content track-sc1 src
Willy Tarreaue9656522010-08-17 15:40:09 +020011514 tcp-request content reject if click_too_fast mark_as_abuser
Willy Tarreau62644772008-07-16 18:36:06 +020011515
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011516 See section 7 about ACL usage.
Willy Tarreau62644772008-07-16 18:36:06 +020011517
Jarno Huuskonen95b012b2017-04-06 13:59:14 +030011518 See also : "tcp-request connection", "tcp-request session",
11519 "tcp-request inspect-delay", and "http-request".
Willy Tarreau62644772008-07-16 18:36:06 +020011520
11521
11522tcp-request inspect-delay <timeout>
11523 Set the maximum allowed time to wait for data during content inspection
11524 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +020011525 no | yes | yes | yes
Willy Tarreau62644772008-07-16 18:36:06 +020011526 Arguments :
11527 <timeout> is the timeout value specified in milliseconds by default, but
11528 can be in any other unit if the number is suffixed by the unit,
11529 as explained at the top of this document.
11530
11531 People using haproxy primarily as a TCP relay are often worried about the
11532 risk of passing any type of protocol to a server without any analysis. In
11533 order to be able to analyze the request contents, we must first withhold
11534 the data then analyze them. This statement simply enables withholding of
11535 data for at most the specified amount of time.
11536
Willy Tarreaufb356202010-08-03 14:02:05 +020011537 TCP content inspection applies very early when a connection reaches a
11538 frontend, then very early when the connection is forwarded to a backend. This
11539 means that a connection may experience a first delay in the frontend and a
11540 second delay in the backend if both have tcp-request rules.
11541
Willy Tarreau62644772008-07-16 18:36:06 +020011542 Note that when performing content inspection, haproxy will evaluate the whole
11543 rules for every new chunk which gets in, taking into account the fact that
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011544 those data are partial. If no rule matches before the aforementioned delay,
Willy Tarreau62644772008-07-16 18:36:06 +020011545 a last check is performed upon expiration, this time considering that the
Willy Tarreaud869b242009-03-15 14:43:58 +010011546 contents are definitive. If no delay is set, haproxy will not wait at all
11547 and will immediately apply a verdict based on the available information.
11548 Obviously this is unlikely to be very useful and might even be racy, so such
11549 setups are not recommended.
Willy Tarreau62644772008-07-16 18:36:06 +020011550
11551 As soon as a rule matches, the request is released and continues as usual. If
11552 the timeout is reached and no rule matches, the default policy will be to let
11553 it pass through unaffected.
11554
11555 For most protocols, it is enough to set it to a few seconds, as most clients
11556 send the full request immediately upon connection. Add 3 or more seconds to
11557 cover TCP retransmits but that's all. For some protocols, it may make sense
Willy Tarreaud72758d2010-01-12 10:42:19 +010011558 to use large values, for instance to ensure that the client never talks
Davor Ocelice9ed2812017-12-25 17:49:28 +010011559 before the server (e.g. SMTP), or to wait for a client to talk before passing
11560 data to the server (e.g. SSL). Note that the client timeout must cover at
Willy Tarreaub824b002010-09-29 16:36:16 +020011561 least the inspection delay, otherwise it will expire first. If the client
11562 closes the connection or if the buffer is full, the delay immediately expires
11563 since the contents will not be able to change anymore.
Willy Tarreau62644772008-07-16 18:36:06 +020011564
Willy Tarreau55165fe2009-05-10 12:02:55 +020011565 See also : "tcp-request content accept", "tcp-request content reject",
Willy Tarreau62644772008-07-16 18:36:06 +020011566 "timeout client".
11567
11568
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011569tcp-response content <action> [{if | unless} <condition>]
11570 Perform an action on a session response depending on a layer 4-7 condition
11571 May be used in sections : defaults | frontend | listen | backend
11572 no | no | yes | yes
11573 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +020011574 <action> defines the action to perform if the condition applies. See
11575 below.
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011576
11577 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
11578
Davor Ocelice9ed2812017-12-25 17:49:28 +010011579 Response contents can be analyzed at an early stage of response processing
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011580 called "TCP content inspection". During this stage, ACL-based rules are
11581 evaluated every time the response contents are updated, until either an
Willy Tarreaucc1e04b2013-09-11 23:20:29 +020011582 "accept", "close" or a "reject" rule matches, or a TCP response inspection
11583 delay is set and expires with no matching rule.
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011584
11585 Most often, these decisions will consider a protocol recognition or validity.
11586
11587 Content-based rules are evaluated in their exact declaration order. If no
11588 rule matches or if there is no rule, the default action is to accept the
11589 contents. There is no specific limit to the number of rules which may be
11590 inserted.
11591
Thierry FOURNIER236657b2015-08-19 08:25:14 +020011592 Several types of actions are supported :
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011593 - accept :
11594 accepts the response if the condition is true (when used with "if")
11595 or false (when used with "unless"). The first such rule executed ends
11596 the rules evaluation.
11597
Willy Tarreaucc1e04b2013-09-11 23:20:29 +020011598 - close :
11599 immediately closes the connection with the server if the condition is
11600 true (when used with "if"), or false (when used with "unless"). The
11601 first such rule executed ends the rules evaluation. The main purpose of
11602 this action is to force a connection to be finished between a client
11603 and a server after an exchange when the application protocol expects
11604 some long time outs to elapse first. The goal is to eliminate idle
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011605 connections which take significant resources on servers with certain
Willy Tarreaucc1e04b2013-09-11 23:20:29 +020011606 protocols.
11607
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011608 - reject :
11609 rejects the response if the condition is true (when used with "if")
11610 or false (when used with "unless"). The first such rule executed ends
Jamie Gloudonaaa21002012-08-25 00:18:33 -040011611 the rules evaluation. Rejected session are immediately closed.
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011612
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011613 - set-var(<var-name>) <expr>
11614 Sets a variable.
11615
Christopher Faulet85d79c92016-11-09 16:54:56 +010011616 - unset-var(<var-name>)
11617 Unsets a variable.
11618
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +020011619 - sc-inc-gpc0(<sc-id>):
11620 This action increments the GPC0 counter according to the sticky
11621 counter designated by <sc-id>. If an error occurs, this action fails
11622 silently and the actions evaluation continues.
11623
Frédéric Lécaille6778b272018-01-29 15:22:53 +010011624 - sc-inc-gpc1(<sc-id>):
11625 This action increments the GPC1 counter according to the sticky
11626 counter designated by <sc-id>. If an error occurs, this action fails
11627 silently and the actions evaluation continues.
11628
Cédric Dufour0d7712d2019-11-06 18:38:53 +010011629 - sc-set-gpt0(<sc-id>) { <int> | <expr> }
11630 This action sets the 32-bit unsigned GPT0 tag according to the sticky
11631 counter designated by <sc-id> and the value of <int>/<expr>. The
11632 expected result is a boolean. If an error occurs, this action silently
11633 fails and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +020011634
Willy Tarreau2d392c22015-08-24 01:43:45 +020011635 - "silent-drop" :
11636 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +010011637 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +020011638 to prevent the client from being notified. The effect it then that the
11639 client still sees an established connection while there's none on
11640 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
11641 except that it doesn't use any local resource at all on the machine
11642 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +010011643 slow down stronger attackers. It is important to understand the impact
11644 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +020011645 client and HAProxy (firewalls, proxies, load balancers) will also keep
11646 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +010011647 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +020011648 TCP_REPAIR socket option is used to block the emission of a TCP
11649 reset. On other systems, the socket's TTL is reduced to 1 so that the
11650 TCP reset doesn't pass the first router, though it's still delivered to
11651 local networks. Do not use it unless you fully understand how it works.
11652
Christopher Faulet76c09ef2017-09-21 11:03:52 +020011653 - send-spoe-group <engine-name> <group-name>
11654 Send a group of SPOE messages.
11655
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011656 Note that the "if/unless" condition is optional. If no condition is set on
11657 the action, it is simply performed unconditionally. That can be useful for
11658 for changing the default action to a reject.
11659
Jamie Gloudonaaa21002012-08-25 00:18:33 -040011660 It is perfectly possible to match layer 7 contents with "tcp-response
11661 content" rules, but then it is important to ensure that a full response has
11662 been buffered, otherwise no contents will match. In order to achieve this,
11663 the best solution involves detecting the HTTP protocol during the inspection
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011664 period.
11665
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011666 The "set-var" is used to set the content of a variable. The variable is
11667 declared inline.
11668
Daniel Schneller0b547052016-03-21 20:46:57 +010011669 <var-name> The name of the variable starts with an indication about
11670 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010011671 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010011672 "sess" : the variable is shared with the whole session
11673 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011674 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +010011675 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011676 processing
Daniel Schneller0b547052016-03-21 20:46:57 +010011677 "res" : the variable is shared only during response
11678 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011679 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +010011680 The name may only contain characters 'a-z', 'A-Z', '0-9',
11681 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011682
11683 <expr> Is a standard HAProxy expression formed by a sample-fetch
11684 followed by some converters.
11685
11686 Example:
11687
11688 tcp-request content set-var(sess.my_var) src
11689
Christopher Faulet85d79c92016-11-09 16:54:56 +010011690 The "unset-var" is used to unset a variable. See above for details about
11691 <var-name>.
11692
11693 Example:
11694
11695 tcp-request content unset-var(sess.my_var)
11696
Christopher Faulet76c09ef2017-09-21 11:03:52 +020011697 The "send-spoe-group" is used to trigger sending of a group of SPOE
11698 messages. To do so, the SPOE engine used to send messages must be defined, as
11699 well as the SPOE group to send. Of course, the SPOE engine must refer to an
11700 existing SPOE filter. If not engine name is provided on the SPOE filter line,
11701 the SPOE agent name must be used.
11702
11703 <engine-name> The SPOE engine name.
11704
11705 <group-name> The SPOE group name as specified in the engine configuration.
11706
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011707 See section 7 about ACL usage.
11708
11709 See also : "tcp-request content", "tcp-response inspect-delay"
11710
11711
Willy Tarreau4f614292016-10-21 17:49:36 +020011712tcp-request session <action> [{if | unless} <condition>]
11713 Perform an action on a validated session depending on a layer 5 condition
11714 May be used in sections : defaults | frontend | listen | backend
11715 no | yes | yes | no
11716 Arguments :
11717 <action> defines the action to perform if the condition applies. See
11718 below.
11719
11720 <condition> is a standard layer5-only ACL-based condition (see section 7).
11721
Davor Ocelice9ed2812017-12-25 17:49:28 +010011722 Once a session is validated, (i.e. after all handshakes have been completed),
Willy Tarreau4f614292016-10-21 17:49:36 +020011723 it is possible to evaluate some conditions to decide whether this session
11724 must be accepted or dropped or have its counters tracked. Those conditions
11725 cannot make use of any data contents because no buffers are allocated yet and
11726 the processing cannot wait at this stage. The main use case it to copy some
11727 early information into variables (since variables are accessible in the
11728 session), or to keep track of some information collected after the handshake,
11729 such as SSL-level elements (SNI, ciphers, client cert's CN) or information
Davor Ocelice9ed2812017-12-25 17:49:28 +010011730 from the PROXY protocol header (e.g. track a source forwarded this way). The
Willy Tarreau4f614292016-10-21 17:49:36 +020011731 extracted information can thus be copied to a variable or tracked using
11732 "track-sc" rules. Of course it is also possible to decide to accept/reject as
11733 with other rulesets. Most operations performed here could also be performed
11734 in "tcp-request content" rules, except that in HTTP these rules are evaluated
11735 for each new request, and that might not always be acceptable. For example a
11736 rule might increment a counter on each evaluation. It would also be possible
11737 that a country is resolved by geolocation from the source IP address,
11738 assigned to a session-wide variable, then the source address rewritten from
11739 an HTTP header for all requests. If some contents need to be inspected in
11740 order to take the decision, the "tcp-request content" statements must be used
11741 instead.
11742
11743 The "tcp-request session" rules are evaluated in their exact declaration
11744 order. If no rule matches or if there is no rule, the default action is to
11745 accept the incoming session. There is no specific limit to the number of
11746 rules which may be inserted.
11747
11748 Several types of actions are supported :
11749 - accept : the request is accepted
11750 - reject : the request is rejected and the connection is closed
11751 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
11752 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +010011753 - sc-inc-gpc1(<sc-id>)
Cédric Dufour0d7712d2019-11-06 18:38:53 +010011754 - sc-set-gpt0(<sc-id>) { <int> | <expr> }
Willy Tarreau4f614292016-10-21 17:49:36 +020011755 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +010011756 - unset-var(<var-name>)
Willy Tarreau4f614292016-10-21 17:49:36 +020011757 - silent-drop
11758
11759 These actions have the same meaning as their respective counter-parts in
11760 "tcp-request connection" and "tcp-request content", so please refer to these
11761 sections for a complete description.
11762
11763 Note that the "if/unless" condition is optional. If no condition is set on
11764 the action, it is simply performed unconditionally. That can be useful for
11765 "track-sc*" actions as well as for changing the default action to a reject.
11766
11767 Example: track the original source address by default, or the one advertised
11768 in the PROXY protocol header for connection coming from the local
11769 proxies. The first connection-level rule enables receipt of the
11770 PROXY protocol for these ones, the second rule tracks whatever
11771 address we decide to keep after optional decoding.
11772
11773 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
11774 tcp-request session track-sc0 src
11775
11776 Example: accept all sessions from white-listed hosts, reject too fast
11777 sessions without counting them, and track accepted sessions.
11778 This results in session rate being capped from abusive sources.
11779
11780 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
11781 tcp-request session reject if { src_sess_rate gt 10 }
11782 tcp-request session track-sc0 src
11783
11784 Example: accept all sessions from white-listed hosts, count all other
11785 sessions and reject too fast ones. This results in abusive ones
11786 being blocked as long as they don't slow down.
11787
11788 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
11789 tcp-request session track-sc0 src
11790 tcp-request session reject if { sc0_sess_rate gt 10 }
11791
11792 See section 7 about ACL usage.
11793
11794 See also : "tcp-request connection", "tcp-request content", "stick-table"
11795
11796
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011797tcp-response inspect-delay <timeout>
11798 Set the maximum allowed time to wait for a response during content inspection
11799 May be used in sections : defaults | frontend | listen | backend
11800 no | no | yes | yes
11801 Arguments :
11802 <timeout> is the timeout value specified in milliseconds by default, but
11803 can be in any other unit if the number is suffixed by the unit,
11804 as explained at the top of this document.
11805
11806 See also : "tcp-response content", "tcp-request inspect-delay".
11807
11808
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010011809timeout check <timeout>
11810 Set additional check timeout, but only after a connection has been already
11811 established.
11812
11813 May be used in sections: defaults | frontend | listen | backend
11814 yes | no | yes | yes
11815 Arguments:
11816 <timeout> is the timeout value specified in milliseconds by default, but
11817 can be in any other unit if the number is suffixed by the unit,
11818 as explained at the top of this document.
11819
11820 If set, haproxy uses min("timeout connect", "inter") as a connect timeout
11821 for check and "timeout check" as an additional read timeout. The "min" is
Davor Ocelice9ed2812017-12-25 17:49:28 +010011822 used so that people running with *very* long "timeout connect" (e.g. those
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010011823 who needed this due to the queue or tarpit) do not slow down their checks.
Willy Tarreaud7550a22010-02-10 05:10:19 +010011824 (Please also note that there is no valid reason to have such long connect
11825 timeouts, because "timeout queue" and "timeout tarpit" can always be used to
11826 avoid that).
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010011827
11828 If "timeout check" is not set haproxy uses "inter" for complete check
11829 timeout (connect + read) exactly like all <1.3.15 version.
11830
11831 In most cases check request is much simpler and faster to handle than normal
11832 requests and people may want to kick out laggy servers so this timeout should
Willy Tarreau41a340d2008-01-22 12:25:31 +010011833 be smaller than "timeout server".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010011834
11835 This parameter is specific to backends, but can be specified once for all in
11836 "defaults" sections. This is in fact one of the easiest solutions not to
11837 forget about it.
11838
Willy Tarreau41a340d2008-01-22 12:25:31 +010011839 See also: "timeout connect", "timeout queue", "timeout server",
11840 "timeout tarpit".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010011841
11842
Willy Tarreau0ba27502007-12-24 16:55:16 +010011843timeout client <timeout>
Willy Tarreau0ba27502007-12-24 16:55:16 +010011844 Set the maximum inactivity time on the client side.
11845 May be used in sections : defaults | frontend | listen | backend
11846 yes | yes | yes | no
11847 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010011848 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010011849 can be in any other unit if the number is suffixed by the unit,
11850 as explained at the top of this document.
11851
11852 The inactivity timeout applies when the client is expected to acknowledge or
11853 send data. In HTTP mode, this timeout is particularly important to consider
11854 during the first phase, when the client sends the request, and during the
Baptiste Assmann2e1941e2016-03-06 23:24:12 +010011855 response while it is reading data sent by the server. That said, for the
11856 first phase, it is preferable to set the "timeout http-request" to better
11857 protect HAProxy from Slowloris like attacks. The value is specified in
11858 milliseconds by default, but can be in any other unit if the number is
Willy Tarreau0ba27502007-12-24 16:55:16 +010011859 suffixed by the unit, as specified at the top of this document. In TCP mode
11860 (and to a lesser extent, in HTTP mode), it is highly recommended that the
11861 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010011862 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +010011863 losses by specifying timeouts that are slightly above multiples of 3 seconds
Davor Ocelice9ed2812017-12-25 17:49:28 +010011864 (e.g. 4 or 5 seconds). If some long-lived sessions are mixed with short-lived
11865 sessions (e.g. WebSocket and HTTP), it's worth considering "timeout tunnel",
Willy Tarreau05cdd962014-05-10 14:30:07 +020011866 which overrides "timeout client" and "timeout server" for tunnels, as well as
11867 "timeout client-fin" for half-closed connections.
Willy Tarreau0ba27502007-12-24 16:55:16 +010011868
11869 This parameter is specific to frontends, but can be specified once for all in
11870 "defaults" sections. This is in fact one of the easiest solutions not to
11871 forget about it. An unspecified timeout results in an infinite timeout, which
11872 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050011873 during startup because it may result in accumulation of expired sessions in
Willy Tarreau0ba27502007-12-24 16:55:16 +010011874 the system if the system's timeouts are not configured either.
11875
Willy Tarreau95c4e142017-11-26 12:18:55 +010011876 This also applies to HTTP/2 connections, which will be closed with GOAWAY.
Lukas Tribus75df9d72017-11-24 19:05:12 +010011877
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020011878 See also : "timeout server", "timeout tunnel", "timeout http-request".
Willy Tarreau0ba27502007-12-24 16:55:16 +010011879
Willy Tarreau0ba27502007-12-24 16:55:16 +010011880
Willy Tarreau05cdd962014-05-10 14:30:07 +020011881timeout client-fin <timeout>
11882 Set the inactivity timeout on the client side for half-closed connections.
11883 May be used in sections : defaults | frontend | listen | backend
11884 yes | yes | yes | no
11885 Arguments :
11886 <timeout> is the timeout value specified in milliseconds by default, but
11887 can be in any other unit if the number is suffixed by the unit,
11888 as explained at the top of this document.
11889
11890 The inactivity timeout applies when the client is expected to acknowledge or
11891 send data while one direction is already shut down. This timeout is different
11892 from "timeout client" in that it only applies to connections which are closed
11893 in one direction. This is particularly useful to avoid keeping connections in
11894 FIN_WAIT state for too long when clients do not disconnect cleanly. This
11895 problem is particularly common long connections such as RDP or WebSocket.
11896 Note that this timeout can override "timeout tunnel" when a connection shuts
Willy Tarreau599391a2017-11-24 10:16:00 +010011897 down in one direction. It is applied to idle HTTP/2 connections once a GOAWAY
11898 frame was sent, often indicating an expectation that the connection quickly
11899 ends.
Willy Tarreau05cdd962014-05-10 14:30:07 +020011900
11901 This parameter is specific to frontends, but can be specified once for all in
11902 "defaults" sections. By default it is not set, so half-closed connections
11903 will use the other timeouts (timeout.client or timeout.tunnel).
11904
11905 See also : "timeout client", "timeout server-fin", and "timeout tunnel".
11906
11907
Willy Tarreau0ba27502007-12-24 16:55:16 +010011908timeout connect <timeout>
Willy Tarreau0ba27502007-12-24 16:55:16 +010011909 Set the maximum time to wait for a connection attempt to a server to succeed.
11910 May be used in sections : defaults | frontend | listen | backend
11911 yes | no | yes | yes
11912 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010011913 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010011914 can be in any other unit if the number is suffixed by the unit,
11915 as explained at the top of this document.
11916
11917 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010011918 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +010011919 cover one or several TCP packet losses by specifying timeouts that are
Davor Ocelice9ed2812017-12-25 17:49:28 +010011920 slightly above multiples of 3 seconds (e.g. 4 or 5 seconds). By default, the
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010011921 connect timeout also presets both queue and tarpit timeouts to the same value
11922 if these have not been specified.
Willy Tarreau0ba27502007-12-24 16:55:16 +010011923
11924 This parameter is specific to backends, but can be specified once for all in
11925 "defaults" sections. This is in fact one of the easiest solutions not to
11926 forget about it. An unspecified timeout results in an infinite timeout, which
11927 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050011928 during startup because it may result in accumulation of failed sessions in
Willy Tarreau0ba27502007-12-24 16:55:16 +010011929 the system if the system's timeouts are not configured either.
11930
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020011931 See also: "timeout check", "timeout queue", "timeout server", "timeout tarpit".
Willy Tarreau0ba27502007-12-24 16:55:16 +010011932
Willy Tarreau0ba27502007-12-24 16:55:16 +010011933
Willy Tarreaub16a5742010-01-10 14:46:16 +010011934timeout http-keep-alive <timeout>
11935 Set the maximum allowed time to wait for a new HTTP request to appear
11936 May be used in sections : defaults | frontend | listen | backend
11937 yes | yes | yes | yes
11938 Arguments :
11939 <timeout> is the timeout value specified in milliseconds by default, but
11940 can be in any other unit if the number is suffixed by the unit,
11941 as explained at the top of this document.
11942
11943 By default, the time to wait for a new request in case of keep-alive is set
11944 by "timeout http-request". However this is not always convenient because some
11945 people want very short keep-alive timeouts in order to release connections
11946 faster, and others prefer to have larger ones but still have short timeouts
11947 once the request has started to present itself.
11948
11949 The "http-keep-alive" timeout covers these needs. It will define how long to
11950 wait for a new HTTP request to start coming after a response was sent. Once
11951 the first byte of request has been seen, the "http-request" timeout is used
11952 to wait for the complete request to come. Note that empty lines prior to a
11953 new request do not refresh the timeout and are not counted as a new request.
11954
11955 There is also another difference between the two timeouts : when a connection
11956 expires during timeout http-keep-alive, no error is returned, the connection
11957 just closes. If the connection expires in "http-request" while waiting for a
11958 connection to complete, a HTTP 408 error is returned.
11959
11960 In general it is optimal to set this value to a few tens to hundreds of
11961 milliseconds, to allow users to fetch all objects of a page at once but
Davor Ocelice9ed2812017-12-25 17:49:28 +010011962 without waiting for further clicks. Also, if set to a very small value (e.g.
Willy Tarreaub16a5742010-01-10 14:46:16 +010011963 1 millisecond) it will probably only accept pipelined requests but not the
11964 non-pipelined ones. It may be a nice trade-off for very large sites running
Patrick Mézard2382ad62010-05-09 10:43:32 +020011965 with tens to hundreds of thousands of clients.
Willy Tarreaub16a5742010-01-10 14:46:16 +010011966
11967 If this parameter is not set, the "http-request" timeout applies, and if both
11968 are not set, "timeout client" still applies at the lower level. It should be
11969 set in the frontend to take effect, unless the frontend is in TCP mode, in
11970 which case the HTTP backend's timeout will be used.
11971
Willy Tarreau95c4e142017-11-26 12:18:55 +010011972 When using HTTP/2 "timeout client" is applied instead. This is so we can keep
11973 using short keep-alive timeouts in HTTP/1.1 while using longer ones in HTTP/2
Lukas Tribus75df9d72017-11-24 19:05:12 +010011974 (where we only have one connection per client and a connection setup).
11975
Willy Tarreaub16a5742010-01-10 14:46:16 +010011976 See also : "timeout http-request", "timeout client".
11977
11978
Willy Tarreau036fae02008-01-06 13:24:40 +010011979timeout http-request <timeout>
11980 Set the maximum allowed time to wait for a complete HTTP request
11981 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaucd7afc02009-07-12 10:03:17 +020011982 yes | yes | yes | yes
Willy Tarreau036fae02008-01-06 13:24:40 +010011983 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010011984 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau036fae02008-01-06 13:24:40 +010011985 can be in any other unit if the number is suffixed by the unit,
11986 as explained at the top of this document.
11987
11988 In order to offer DoS protection, it may be required to lower the maximum
11989 accepted time to receive a complete HTTP request without affecting the client
11990 timeout. This helps protecting against established connections on which
11991 nothing is sent. The client timeout cannot offer a good protection against
11992 this abuse because it is an inactivity timeout, which means that if the
11993 attacker sends one character every now and then, the timeout will not
11994 trigger. With the HTTP request timeout, no matter what speed the client
Willy Tarreau2705a612014-05-23 17:38:34 +020011995 types, the request will be aborted if it does not complete in time. When the
11996 timeout expires, an HTTP 408 response is sent to the client to inform it
11997 about the problem, and the connection is closed. The logs will report
11998 termination codes "cR". Some recent browsers are having problems with this
Davor Ocelice9ed2812017-12-25 17:49:28 +010011999 standard, well-documented behavior, so it might be needed to hide the 408
Willy Tarreau0f228a02015-05-01 15:37:53 +020012000 code using "option http-ignore-probes" or "errorfile 408 /dev/null". See
12001 more details in the explanations of the "cR" termination code in section 8.5.
Willy Tarreau036fae02008-01-06 13:24:40 +010012002
Baptiste Assmanneccdf432015-10-28 13:49:01 +010012003 By default, this timeout only applies to the header part of the request,
12004 and not to any data. As soon as the empty line is received, this timeout is
12005 not used anymore. When combined with "option http-buffer-request", this
12006 timeout also applies to the body of the request..
12007 It is used again on keep-alive connections to wait for a second
Willy Tarreaub16a5742010-01-10 14:46:16 +010012008 request if "timeout http-keep-alive" is not set.
Willy Tarreau036fae02008-01-06 13:24:40 +010012009
12010 Generally it is enough to set it to a few seconds, as most clients send the
12011 full request immediately upon connection. Add 3 or more seconds to cover TCP
Davor Ocelice9ed2812017-12-25 17:49:28 +010012012 retransmits but that's all. Setting it to very low values (e.g. 50 ms) will
Willy Tarreau036fae02008-01-06 13:24:40 +010012013 generally work on local networks as long as there are no packet losses. This
12014 will prevent people from sending bare HTTP requests using telnet.
12015
12016 If this parameter is not set, the client timeout still applies between each
Willy Tarreaucd7afc02009-07-12 10:03:17 +020012017 chunk of the incoming request. It should be set in the frontend to take
12018 effect, unless the frontend is in TCP mode, in which case the HTTP backend's
12019 timeout will be used.
Willy Tarreau036fae02008-01-06 13:24:40 +010012020
Willy Tarreau0f228a02015-05-01 15:37:53 +020012021 See also : "errorfile", "http-ignore-probes", "timeout http-keep-alive", and
Baptiste Assmanneccdf432015-10-28 13:49:01 +010012022 "timeout client", "option http-buffer-request".
Willy Tarreau036fae02008-01-06 13:24:40 +010012023
Willy Tarreau844e3c52008-01-11 16:28:18 +010012024
12025timeout queue <timeout>
12026 Set the maximum time to wait in the queue for a connection slot to be free
12027 May be used in sections : defaults | frontend | listen | backend
12028 yes | no | yes | yes
12029 Arguments :
12030 <timeout> is the timeout value specified in milliseconds by default, but
12031 can be in any other unit if the number is suffixed by the unit,
12032 as explained at the top of this document.
12033
12034 When a server's maxconn is reached, connections are left pending in a queue
12035 which may be server-specific or global to the backend. In order not to wait
12036 indefinitely, a timeout is applied to requests pending in the queue. If the
12037 timeout is reached, it is considered that the request will almost never be
12038 served, so it is dropped and a 503 error is returned to the client.
12039
12040 The "timeout queue" statement allows to fix the maximum time for a request to
12041 be left pending in a queue. If unspecified, the same value as the backend's
12042 connection timeout ("timeout connect") is used, for backwards compatibility
12043 with older versions with no "timeout queue" parameter.
12044
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020012045 See also : "timeout connect".
Willy Tarreau844e3c52008-01-11 16:28:18 +010012046
12047
12048timeout server <timeout>
Willy Tarreau844e3c52008-01-11 16:28:18 +010012049 Set the maximum inactivity time on the server side.
12050 May be used in sections : defaults | frontend | listen | backend
12051 yes | no | yes | yes
12052 Arguments :
12053 <timeout> is the timeout value specified in milliseconds by default, but
12054 can be in any other unit if the number is suffixed by the unit,
12055 as explained at the top of this document.
12056
12057 The inactivity timeout applies when the server is expected to acknowledge or
12058 send data. In HTTP mode, this timeout is particularly important to consider
12059 during the first phase of the server's response, when it has to send the
12060 headers, as it directly represents the server's processing time for the
12061 request. To find out what value to put there, it's often good to start with
12062 what would be considered as unacceptable response times, then check the logs
12063 to observe the response time distribution, and adjust the value accordingly.
12064
12065 The value is specified in milliseconds by default, but can be in any other
12066 unit if the number is suffixed by the unit, as specified at the top of this
12067 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
12068 recommended that the client timeout remains equal to the server timeout in
12069 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010012070 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +010012071 packet losses by specifying timeouts that are slightly above multiples of 3
Davor Ocelice9ed2812017-12-25 17:49:28 +010012072 seconds (e.g. 4 or 5 seconds minimum). If some long-lived sessions are mixed
12073 with short-lived sessions (e.g. WebSocket and HTTP), it's worth considering
Willy Tarreauce887fd2012-05-12 12:50:00 +020012074 "timeout tunnel", which overrides "timeout client" and "timeout server" for
12075 tunnels.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012076
12077 This parameter is specific to backends, but can be specified once for all in
12078 "defaults" sections. This is in fact one of the easiest solutions not to
12079 forget about it. An unspecified timeout results in an infinite timeout, which
12080 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050012081 during startup because it may result in accumulation of expired sessions in
Willy Tarreau844e3c52008-01-11 16:28:18 +010012082 the system if the system's timeouts are not configured either.
12083
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020012084 See also : "timeout client" and "timeout tunnel".
Willy Tarreau844e3c52008-01-11 16:28:18 +010012085
Willy Tarreau05cdd962014-05-10 14:30:07 +020012086
12087timeout server-fin <timeout>
12088 Set the inactivity timeout on the server side for half-closed connections.
12089 May be used in sections : defaults | frontend | listen | backend
12090 yes | no | yes | yes
12091 Arguments :
12092 <timeout> is the timeout value specified in milliseconds by default, but
12093 can be in any other unit if the number is suffixed by the unit,
12094 as explained at the top of this document.
12095
12096 The inactivity timeout applies when the server is expected to acknowledge or
12097 send data while one direction is already shut down. This timeout is different
12098 from "timeout server" in that it only applies to connections which are closed
12099 in one direction. This is particularly useful to avoid keeping connections in
12100 FIN_WAIT state for too long when a remote server does not disconnect cleanly.
12101 This problem is particularly common long connections such as RDP or WebSocket.
12102 Note that this timeout can override "timeout tunnel" when a connection shuts
12103 down in one direction. This setting was provided for completeness, but in most
12104 situations, it should not be needed.
12105
12106 This parameter is specific to backends, but can be specified once for all in
12107 "defaults" sections. By default it is not set, so half-closed connections
12108 will use the other timeouts (timeout.server or timeout.tunnel).
12109
12110 See also : "timeout client-fin", "timeout server", and "timeout tunnel".
12111
Willy Tarreau844e3c52008-01-11 16:28:18 +010012112
12113timeout tarpit <timeout>
Cyril Bonté78caf842010-03-10 22:41:43 +010012114 Set the duration for which tarpitted connections will be maintained
Willy Tarreau844e3c52008-01-11 16:28:18 +010012115 May be used in sections : defaults | frontend | listen | backend
12116 yes | yes | yes | yes
12117 Arguments :
12118 <timeout> is the tarpit duration specified in milliseconds by default, but
12119 can be in any other unit if the number is suffixed by the unit,
12120 as explained at the top of this document.
12121
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020012122 When a connection is tarpitted using "http-request tarpit", it is maintained
12123 open with no activity for a certain amount of time, then closed. "timeout
12124 tarpit" defines how long it will be maintained open.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012125
12126 The value is specified in milliseconds by default, but can be in any other
12127 unit if the number is suffixed by the unit, as specified at the top of this
12128 document. If unspecified, the same value as the backend's connection timeout
12129 ("timeout connect") is used, for backwards compatibility with older versions
Cyril Bonté78caf842010-03-10 22:41:43 +010012130 with no "timeout tarpit" parameter.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012131
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020012132 See also : "timeout connect".
Willy Tarreau844e3c52008-01-11 16:28:18 +010012133
12134
Willy Tarreauce887fd2012-05-12 12:50:00 +020012135timeout tunnel <timeout>
12136 Set the maximum inactivity time on the client and server side for tunnels.
12137 May be used in sections : defaults | frontend | listen | backend
12138 yes | no | yes | yes
12139 Arguments :
12140 <timeout> is the timeout value specified in milliseconds by default, but
12141 can be in any other unit if the number is suffixed by the unit,
12142 as explained at the top of this document.
12143
Jamie Gloudonaaa21002012-08-25 00:18:33 -040012144 The tunnel timeout applies when a bidirectional connection is established
Willy Tarreauce887fd2012-05-12 12:50:00 +020012145 between a client and a server, and the connection remains inactive in both
12146 directions. This timeout supersedes both the client and server timeouts once
12147 the connection becomes a tunnel. In TCP, this timeout is used as soon as no
Davor Ocelice9ed2812017-12-25 17:49:28 +010012148 analyzer remains attached to either connection (e.g. tcp content rules are
12149 accepted). In HTTP, this timeout is used when a connection is upgraded (e.g.
Willy Tarreauce887fd2012-05-12 12:50:00 +020012150 when switching to the WebSocket protocol, or forwarding a CONNECT request
12151 to a proxy), or after the first response when no keepalive/close option is
12152 specified.
12153
Willy Tarreau05cdd962014-05-10 14:30:07 +020012154 Since this timeout is usually used in conjunction with long-lived connections,
12155 it usually is a good idea to also set "timeout client-fin" to handle the
12156 situation where a client suddenly disappears from the net and does not
12157 acknowledge a close, or sends a shutdown and does not acknowledge pending
12158 data anymore. This can happen in lossy networks where firewalls are present,
12159 and is detected by the presence of large amounts of sessions in a FIN_WAIT
12160 state.
12161
Willy Tarreauce887fd2012-05-12 12:50:00 +020012162 The value is specified in milliseconds by default, but can be in any other
12163 unit if the number is suffixed by the unit, as specified at the top of this
12164 document. Whatever the expected normal idle time, it is a good practice to
12165 cover at least one or several TCP packet losses by specifying timeouts that
Davor Ocelice9ed2812017-12-25 17:49:28 +010012166 are slightly above multiples of 3 seconds (e.g. 4 or 5 seconds minimum).
Willy Tarreauce887fd2012-05-12 12:50:00 +020012167
12168 This parameter is specific to backends, but can be specified once for all in
12169 "defaults" sections. This is in fact one of the easiest solutions not to
12170 forget about it.
12171
12172 Example :
12173 defaults http
12174 option http-server-close
12175 timeout connect 5s
12176 timeout client 30s
Willy Tarreau05cdd962014-05-10 14:30:07 +020012177 timeout client-fin 30s
Willy Tarreauce887fd2012-05-12 12:50:00 +020012178 timeout server 30s
12179 timeout tunnel 1h # timeout to use with WebSocket and CONNECT
12180
Willy Tarreau05cdd962014-05-10 14:30:07 +020012181 See also : "timeout client", "timeout client-fin", "timeout server".
Willy Tarreauce887fd2012-05-12 12:50:00 +020012182
12183
Willy Tarreau844e3c52008-01-11 16:28:18 +010012184transparent (deprecated)
12185 Enable client-side transparent proxying
12186 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +010012187 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +010012188 Arguments : none
12189
12190 This keyword was introduced in order to provide layer 7 persistence to layer
12191 3 load balancers. The idea is to use the OS's ability to redirect an incoming
12192 connection for a remote address to a local process (here HAProxy), and let
12193 this process know what address was initially requested. When this option is
12194 used, sessions without cookies will be forwarded to the original destination
12195 IP address of the incoming request (which should match that of another
12196 equipment), while requests with cookies will still be forwarded to the
12197 appropriate server.
12198
12199 The "transparent" keyword is deprecated, use "option transparent" instead.
12200
12201 Note that contrary to a common belief, this option does NOT make HAProxy
12202 present the client's IP to the server when establishing the connection.
12203
Willy Tarreau844e3c52008-01-11 16:28:18 +010012204 See also: "option transparent"
12205
William Lallemanda73203e2012-03-12 12:48:57 +010012206unique-id-format <string>
12207 Generate a unique ID for each request.
12208 May be used in sections : defaults | frontend | listen | backend
12209 yes | yes | yes | no
12210 Arguments :
12211 <string> is a log-format string.
12212
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012213 This keyword creates a ID for each request using the custom log format. A
12214 unique ID is useful to trace a request passing through many components of
12215 a complex infrastructure. The newly created ID may also be logged using the
12216 %ID tag the log-format string.
William Lallemanda73203e2012-03-12 12:48:57 +010012217
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012218 The format should be composed from elements that are guaranteed to be
12219 unique when combined together. For instance, if multiple haproxy instances
12220 are involved, it might be important to include the node name. It is often
12221 needed to log the incoming connection's source and destination addresses
12222 and ports. Note that since multiple requests may be performed over the same
12223 connection, including a request counter may help differentiate them.
12224 Similarly, a timestamp may protect against a rollover of the counter.
12225 Logging the process ID will avoid collisions after a service restart.
William Lallemanda73203e2012-03-12 12:48:57 +010012226
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012227 It is recommended to use hexadecimal notation for many fields since it
12228 makes them more compact and saves space in logs.
William Lallemanda73203e2012-03-12 12:48:57 +010012229
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012230 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010012231
Julien Vehentf21be322014-03-07 08:27:34 -050012232 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010012233
12234 will generate:
12235
12236 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
12237
12238 See also: "unique-id-header"
12239
12240unique-id-header <name>
12241 Add a unique ID header in the HTTP request.
12242 May be used in sections : defaults | frontend | listen | backend
12243 yes | yes | yes | no
12244 Arguments :
12245 <name> is the name of the header.
12246
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012247 Add a unique-id header in the HTTP request sent to the server, using the
12248 unique-id-format. It can't work if the unique-id-format doesn't exist.
William Lallemanda73203e2012-03-12 12:48:57 +010012249
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012250 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010012251
Julien Vehentf21be322014-03-07 08:27:34 -050012252 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010012253 unique-id-header X-Unique-ID
12254
12255 will generate:
12256
12257 X-Unique-ID: 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
12258
12259 See also: "unique-id-format"
Willy Tarreau844e3c52008-01-11 16:28:18 +010012260
Willy Tarreauf51658d2014-04-23 01:21:56 +020012261use_backend <backend> [{if | unless} <condition>]
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020012262 Switch to a specific backend if/unless an ACL-based condition is matched.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012263 May be used in sections : defaults | frontend | listen | backend
12264 no | yes | yes | no
12265 Arguments :
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010012266 <backend> is the name of a valid backend or "listen" section, or a
12267 "log-format" string resolving to a backend name.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012268
Willy Tarreauf51658d2014-04-23 01:21:56 +020012269 <condition> is a condition composed of ACLs, as described in section 7. If
12270 it is omitted, the rule is unconditionally applied.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012271
12272 When doing content-switching, connections arrive on a frontend and are then
12273 dispatched to various backends depending on a number of conditions. The
12274 relation between the conditions and the backends is described with the
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020012275 "use_backend" keyword. While it is normally used with HTTP processing, it can
Davor Ocelice9ed2812017-12-25 17:49:28 +010012276 also be used in pure TCP, either without content using stateless ACLs (e.g.
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020012277 source address validation) or combined with a "tcp-request" rule to wait for
12278 some payload.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012279
12280 There may be as many "use_backend" rules as desired. All of these rules are
12281 evaluated in their declaration order, and the first one which matches will
12282 assign the backend.
12283
12284 In the first form, the backend will be used if the condition is met. In the
12285 second form, the backend will be used if the condition is not met. If no
12286 condition is valid, the backend defined with "default_backend" will be used.
12287 If no default backend is defined, either the servers in the same section are
12288 used (in case of a "listen" section) or, in case of a frontend, no server is
12289 used and a 503 service unavailable response is returned.
12290
Willy Tarreau51aecc72009-07-12 09:47:04 +020012291 Note that it is possible to switch from a TCP frontend to an HTTP backend. In
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012292 this case, either the frontend has already checked that the protocol is HTTP,
Willy Tarreau51aecc72009-07-12 09:47:04 +020012293 and backend processing will immediately follow, or the backend will wait for
12294 a complete HTTP request to get in. This feature is useful when a frontend
12295 must decode several protocols on a unique port, one of them being HTTP.
12296
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010012297 When <backend> is a simple name, it is resolved at configuration time, and an
12298 error is reported if the specified backend does not exist. If <backend> is
12299 a log-format string instead, no check may be done at configuration time, so
12300 the backend name is resolved dynamically at run time. If the resulting
12301 backend name does not correspond to any valid backend, no other rule is
12302 evaluated, and the default_backend directive is applied instead. Note that
12303 when using dynamic backend names, it is highly recommended to use a prefix
12304 that no other backend uses in order to ensure that an unauthorized backend
12305 cannot be forced from the request.
12306
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012307 It is worth mentioning that "use_backend" rules with an explicit name are
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010012308 used to detect the association between frontends and backends to compute the
12309 backend's "fullconn" setting. This cannot be done for dynamic names.
12310
12311 See also: "default_backend", "tcp-request", "fullconn", "log-format", and
12312 section 7 about ACLs.
Willy Tarreaud72758d2010-01-12 10:42:19 +010012313
Christopher Fauletb30b3102019-09-12 23:03:09 +020012314use-fcgi-app <name>
12315 Defines the FastCGI application to use for the backend.
12316 May be used in sections : defaults | frontend | listen | backend
12317 no | no | yes | yes
12318 Arguments :
12319 <name> is the name of the FastCGI application to use.
12320
12321 See section 10.1 about FastCGI application setup for details.
Willy Tarreau036fae02008-01-06 13:24:40 +010012322
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012323use-server <server> if <condition>
12324use-server <server> unless <condition>
12325 Only use a specific server if/unless an ACL-based condition is matched.
12326 May be used in sections : defaults | frontend | listen | backend
12327 no | no | yes | yes
12328 Arguments :
Jerome Magnin824186b2020-03-29 09:37:12 +020012329 <server> is the name of a valid server in the same backend section
12330 or a "log-format" string resolving to a server name.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012331
12332 <condition> is a condition composed of ACLs, as described in section 7.
12333
12334 By default, connections which arrive to a backend are load-balanced across
12335 the available servers according to the configured algorithm, unless a
12336 persistence mechanism such as a cookie is used and found in the request.
12337
12338 Sometimes it is desirable to forward a particular request to a specific
12339 server without having to declare a dedicated backend for this server. This
12340 can be achieved using the "use-server" rules. These rules are evaluated after
12341 the "redirect" rules and before evaluating cookies, and they have precedence
12342 on them. There may be as many "use-server" rules as desired. All of these
12343 rules are evaluated in their declaration order, and the first one which
12344 matches will assign the server.
12345
12346 If a rule designates a server which is down, and "option persist" is not used
12347 and no force-persist rule was validated, it is ignored and evaluation goes on
12348 with the next rules until one matches.
12349
12350 In the first form, the server will be used if the condition is met. In the
12351 second form, the server will be used if the condition is not met. If no
12352 condition is valid, the processing continues and the server will be assigned
12353 according to other persistence mechanisms.
12354
12355 Note that even if a rule is matched, cookie processing is still performed but
12356 does not assign the server. This allows prefixed cookies to have their prefix
12357 stripped.
12358
12359 The "use-server" statement works both in HTTP and TCP mode. This makes it
12360 suitable for use with content-based inspection. For instance, a server could
Lukas Tribusa267b5d2020-07-19 00:25:06 +020012361 be selected in a farm according to the TLS SNI field when using protocols with
12362 implicit TLS (also see "req_ssl_sni"). And if these servers have their weight
12363 set to zero, they will not be used for other traffic.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012364
12365 Example :
12366 # intercept incoming TLS requests based on the SNI field
12367 use-server www if { req_ssl_sni -i www.example.com }
12368 server www 192.168.0.1:443 weight 0
12369 use-server mail if { req_ssl_sni -i mail.example.com }
Lukas Tribusa267b5d2020-07-19 00:25:06 +020012370 server mail 192.168.0.1:465 weight 0
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012371 use-server imap if { req_ssl_sni -i imap.example.com }
Lukas Tribus98a3e3f2017-03-26 12:55:35 +000012372 server imap 192.168.0.1:993 weight 0
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012373 # all the rest is forwarded to this server
12374 server default 192.168.0.2:443 check
12375
Jerome Magnin824186b2020-03-29 09:37:12 +020012376 When <server> is a simple name, it is checked against existing servers in the
12377 configuration and an error is reported if the specified server does not exist.
12378 If it is a log-format, no check is performed when parsing the configuration,
12379 and if we can't resolve a valid server name at runtime but the use-server rule
Ilya Shipitsin11057a32020-06-21 21:18:27 +050012380 was conditioned by an ACL returning true, no other use-server rule is applied
Jerome Magnin824186b2020-03-29 09:37:12 +020012381 and we fall back to load balancing.
12382
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012383 See also: "use_backend", section 5 about server and section 7 about ACLs.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012384
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012385
Davor Ocelice9ed2812017-12-25 17:49:28 +0100123865. Bind and server options
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012387--------------------------
12388
12389The "bind", "server" and "default-server" keywords support a number of settings
12390depending on some build options and on the system HAProxy was built on. These
12391settings generally each consist in one word sometimes followed by a value,
12392written on the same line as the "bind" or "server" line. All these options are
12393described in this section.
12394
12395
123965.1. Bind options
12397-----------------
12398
12399The "bind" keyword supports a certain number of settings which are all passed
12400as arguments on the same line. The order in which those arguments appear makes
12401no importance, provided that they appear after the bind address. All of these
12402parameters are optional. Some of them consist in a single words (booleans),
12403while other ones expect a value after them. In this case, the value must be
12404provided immediately after the setting name.
12405
12406The currently supported settings are the following ones.
12407
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010012408accept-netscaler-cip <magic number>
12409 Enforces the use of the NetScaler Client IP insertion protocol over any
12410 connection accepted by any of the TCP sockets declared on the same line. The
12411 NetScaler Client IP insertion protocol dictates the layer 3/4 addresses of
12412 the incoming connection to be used everywhere an address is used, with the
12413 only exception of "tcp-request connection" rules which will only see the
12414 real connection address. Logs will reflect the addresses indicated in the
12415 protocol, unless it is violated, in which case the real address will still
12416 be used. This keyword combined with support from external components can be
12417 used as an efficient and reliable alternative to the X-Forwarded-For
Bertrand Jacquin90759682016-06-06 15:35:39 +010012418 mechanism which is not always reliable and not even always usable. See also
12419 "tcp-request connection expect-netscaler-cip" for a finer-grained setting of
12420 which client is allowed to use the protocol.
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010012421
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012422accept-proxy
12423 Enforces the use of the PROXY protocol over any connection accepted by any of
Willy Tarreau77992672014-06-14 11:06:17 +020012424 the sockets declared on the same line. Versions 1 and 2 of the PROXY protocol
12425 are supported and correctly detected. The PROXY protocol dictates the layer
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012426 3/4 addresses of the incoming connection to be used everywhere an address is
12427 used, with the only exception of "tcp-request connection" rules which will
12428 only see the real connection address. Logs will reflect the addresses
12429 indicated in the protocol, unless it is violated, in which case the real
Davor Ocelice9ed2812017-12-25 17:49:28 +010012430 address will still be used. This keyword combined with support from external
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012431 components can be used as an efficient and reliable alternative to the
12432 X-Forwarded-For mechanism which is not always reliable and not even always
Willy Tarreau4f0d9192013-06-11 20:40:55 +020012433 usable. See also "tcp-request connection expect-proxy" for a finer-grained
12434 setting of which client is allowed to use the protocol.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012435
Olivier Houchardc2aae742017-09-22 18:26:28 +020012436allow-0rtt
Bertrand Jacquina25282b2018-08-14 00:56:13 +010012437 Allow receiving early data when using TLSv1.3. This is disabled by default,
Olivier Houchard69752962019-01-08 15:35:32 +010012438 due to security considerations. Because it is vulnerable to replay attacks,
John Roeslerfb2fce12019-07-10 15:45:51 -050012439 you should only allow if for requests that are safe to replay, i.e. requests
Olivier Houchard69752962019-01-08 15:35:32 +010012440 that are idempotent. You can use the "wait-for-handshake" action for any
12441 request that wouldn't be safe with early data.
Olivier Houchardc2aae742017-09-22 18:26:28 +020012442
Willy Tarreauab861d32013-04-02 02:30:41 +020012443alpn <protocols>
12444 This enables the TLS ALPN extension and advertises the specified protocol
12445 list as supported on top of ALPN. The protocol list consists in a comma-
12446 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
John Roeslerfb2fce12019-07-10 15:45:51 -050012447 quotes). This requires that the SSL library is built with support for TLS
Willy Tarreauab861d32013-04-02 02:30:41 +020012448 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
Willy Tarreau95c4e142017-11-26 12:18:55 +010012449 initial NPN extension. ALPN is required to enable HTTP/2 on an HTTP frontend.
12450 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
12451 now obsolete NPN extension. At the time of writing this, most browsers still
12452 support both ALPN and NPN for HTTP/2 so a fallback to NPN may still work for
12453 a while. But ALPN must be used whenever possible. If both HTTP/2 and HTTP/1.1
12454 are expected to be supported, both versions can be advertised, in order of
12455 preference, like below :
12456
12457 bind :443 ssl crt pub.pem alpn h2,http/1.1
Willy Tarreauab861d32013-04-02 02:30:41 +020012458
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012459backlog <backlog>
Willy Tarreaue2711c72019-02-27 15:39:41 +010012460 Sets the socket's backlog to this value. If unspecified or 0, the frontend's
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012461 backlog is used instead, which generally defaults to the maxconn value.
12462
Emmanuel Hocdete7f2b732017-01-09 16:15:54 +010012463curves <curves>
12464 This setting is only available when support for OpenSSL was built in. It sets
12465 the string describing the list of elliptic curves algorithms ("curve suite")
12466 that are negotiated during the SSL/TLS handshake with ECDHE. The format of the
12467 string is a colon-delimited list of curve name.
12468 Example: "X25519:P-256" (without quote)
12469 When "curves" is set, "ecdhe" parameter is ignored.
12470
Emeric Brun7fb34422012-09-28 15:26:15 +020012471ecdhe <named curve>
12472 This setting is only available when support for OpenSSL was built in. It sets
Emeric Brun6924ef82013-03-06 14:08:53 +010012473 the named curve (RFC 4492) used to generate ECDH ephemeral keys. By default,
12474 used named curve is prime256v1.
Emeric Brun7fb34422012-09-28 15:26:15 +020012475
Emeric Brunfd33a262012-10-11 16:28:27 +020012476ca-file <cafile>
Emeric Brun1a073b42012-09-28 17:07:34 +020012477 This setting is only available when support for OpenSSL was built in. It
12478 designates a PEM file from which to load CA certificates used to verify
12479 client's certificate.
12480
Emeric Brunb6dc9342012-09-28 17:55:37 +020012481ca-ignore-err [all|<errorID>,...]
12482 This setting is only available when support for OpenSSL was built in.
12483 Sets a comma separated list of errorIDs to ignore during verify at depth > 0.
12484 If set to 'all', all errors are ignored. SSL handshake is not aborted if an
12485 error is ignored.
12486
Christopher Faulet31af49d2015-06-09 17:29:50 +020012487ca-sign-file <cafile>
12488 This setting is only available when support for OpenSSL was built in. It
12489 designates a PEM file containing both the CA certificate and the CA private
12490 key used to create and sign server's certificates. This is a mandatory
12491 setting when the dynamic generation of certificates is enabled. See
12492 'generate-certificates' for details.
12493
Bertrand Jacquind4d0a232016-11-13 16:37:12 +000012494ca-sign-pass <passphrase>
Christopher Faulet31af49d2015-06-09 17:29:50 +020012495 This setting is only available when support for OpenSSL was built in. It is
12496 the CA private key passphrase. This setting is optional and used only when
12497 the dynamic generation of certificates is enabled. See
12498 'generate-certificates' for details.
12499
Emmanuel Hocdet842e94e2019-12-16 16:39:17 +010012500ca-verify-file <cafile>
12501 This setting designates a PEM file from which to load CA certificates used to
12502 verify client's certificate. It designates CA certificates which must not be
12503 included in CA names sent in server hello message. Typically, "ca-file" must
12504 be defined with intermediate certificates, and "ca-verify-file" with
12505 certificates to ending the chain, like root CA.
12506
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012507ciphers <ciphers>
12508 This setting is only available when support for OpenSSL was built in. It sets
12509 the string describing the list of cipher algorithms ("cipher suite") that are
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +000012510 negotiated during the SSL/TLS handshake up to TLSv1.2. The format of the
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000012511 string is defined in "man 1 ciphers" from OpenSSL man pages. For background
Dirkjan Bussink415150f2018-09-14 11:14:21 +020012512 information and recommendations see e.g.
12513 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
12514 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
12515 cipher configuration, please check the "ciphersuites" keyword.
12516
12517ciphersuites <ciphersuites>
12518 This setting is only available when support for OpenSSL was built in and
12519 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the string describing
12520 the list of cipher algorithms ("cipher suite") that are negotiated during the
12521 TLSv1.3 handshake. The format of the string is defined in "man 1 ciphers" from
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000012522 OpenSSL man pages under the "ciphersuites" section. For cipher configuration
12523 for TLSv1.2 and earlier, please check the "ciphers" keyword.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012524
Emeric Brunfd33a262012-10-11 16:28:27 +020012525crl-file <crlfile>
Emeric Brun1a073b42012-09-28 17:07:34 +020012526 This setting is only available when support for OpenSSL was built in. It
12527 designates a PEM file from which to load certificate revocation list used
12528 to verify client's certificate.
12529
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012530crt <cert>
Alex Davies0fbf0162013-03-02 16:04:50 +000012531 This setting is only available when support for OpenSSL was built in. It
12532 designates a PEM file containing both the required certificates and any
12533 associated private keys. This file can be built by concatenating multiple
12534 PEM files into one (e.g. cat cert.pem key.pem > combined.pem). If your CA
12535 requires an intermediate certificate, this can also be concatenated into this
Emmanuel Hocdet70df7bf2019-01-04 11:08:20 +010012536 file. Intermediate certificate can also be shared in a directory via
12537 "issuers-chain-path" directive.
Alex Davies0fbf0162013-03-02 16:04:50 +000012538
William Lallemand4c5adbf2020-02-24 14:23:22 +010012539 If the file does not contain a private key, HAProxy will try to load
12540 the key at the same path suffixed by a ".key".
12541
Alex Davies0fbf0162013-03-02 16:04:50 +000012542 If the OpenSSL used supports Diffie-Hellman, parameters present in this file
12543 are loaded.
12544
12545 If a directory name is used instead of a PEM file, then all files found in
William Lallemand3f25ae32020-02-24 16:30:12 +010012546 that directory will be loaded in alphabetic order unless their name ends
12547 with '.key', '.issuer', '.ocsp' or '.sctl' (reserved extensions). This
12548 directive may be specified multiple times in order to load certificates from
12549 multiple files or directories. The certificates will be presented to clients
12550 who provide a valid TLS Server Name Indication field matching one of their
12551 CN or alt subjects. Wildcards are supported, where a wildcard character '*'
12552 is used instead of the first hostname component (e.g. *.example.org matches
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010012553 www.example.org but not www.sub.example.org).
Alex Davies0fbf0162013-03-02 16:04:50 +000012554
12555 If no SNI is provided by the client or if the SSL library does not support
12556 TLS extensions, or if the client provides an SNI hostname which does not
12557 match any certificate, then the first loaded certificate will be presented.
12558 This means that when loading certificates from a directory, it is highly
Cyril Bonté3180f7b2015-01-25 00:16:08 +010012559 recommended to load the default one first as a file or to ensure that it will
12560 always be the first one in the directory.
Alex Davies0fbf0162013-03-02 16:04:50 +000012561
Emeric Brune032bfa2012-09-28 13:01:45 +020012562 Note that the same cert may be loaded multiple times without side effects.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012563
Davor Ocelice9ed2812017-12-25 17:49:28 +010012564 Some CAs (such as GoDaddy) offer a drop down list of server types that do not
Alex Davies0fbf0162013-03-02 16:04:50 +000012565 include HAProxy when obtaining a certificate. If this happens be sure to
Davor Ocelice9ed2812017-12-25 17:49:28 +010012566 choose a web server that the CA believes requires an intermediate CA (for
12567 GoDaddy, selection Apache Tomcat will get the correct bundle, but many
Alex Davies0fbf0162013-03-02 16:04:50 +000012568 others, e.g. nginx, result in a wrong bundle that will not work for some
12569 clients).
12570
Emeric Brun4147b2e2014-06-16 18:36:30 +020012571 For each PEM file, haproxy checks for the presence of file at the same path
12572 suffixed by ".ocsp". If such file is found, support for the TLS Certificate
12573 Status Request extension (also known as "OCSP stapling") is automatically
12574 enabled. The content of this file is optional. If not empty, it must contain
12575 a valid OCSP Response in DER format. In order to be valid an OCSP Response
12576 must comply with the following rules: it has to indicate a good status,
12577 it has to be a single response for the certificate of the PEM file, and it
12578 has to be valid at the moment of addition. If these rules are not respected
12579 the OCSP Response is ignored and a warning is emitted. In order to identify
12580 which certificate an OCSP Response applies to, the issuer's certificate is
12581 necessary. If the issuer's certificate is not found in the PEM file, it will
12582 be loaded from a file at the same path as the PEM file suffixed by ".issuer"
12583 if it exists otherwise it will fail with an error.
12584
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010012585 For each PEM file, haproxy also checks for the presence of file at the same
12586 path suffixed by ".sctl". If such file is found, support for Certificate
12587 Transparency (RFC6962) TLS extension is enabled. The file must contain a
12588 valid Signed Certificate Timestamp List, as described in RFC. File is parsed
12589 to check basic syntax, but no signatures are verified.
12590
yanbzhu6c25e9e2016-01-05 12:52:02 -050012591 There are cases where it is desirable to support multiple key types, e.g. RSA
12592 and ECDSA in the cipher suites offered to the clients. This allows clients
12593 that support EC certificates to be able to use EC ciphers, while
12594 simultaneously supporting older, RSA only clients.
yanbzhud19630c2015-12-14 15:10:25 -050012595
William Lallemandf9ff3ec2020-10-02 17:57:44 +020012596 To achieve this, OpenSSL 1.1.1 is required, you can configure this behavior
12597 by providing one crt entry per certificate type, or by configuring a "cert
12598 bundle" like it was required before HAProxy 1.8. See "ssl-load-extra-files".
yanbzhud19630c2015-12-14 15:10:25 -050012599
Emeric Brunb6dc9342012-09-28 17:55:37 +020012600crt-ignore-err <errors>
Alex Davies0fbf0162013-03-02 16:04:50 +000012601 This setting is only available when support for OpenSSL was built in. Sets a
Davor Ocelice9ed2812017-12-25 17:49:28 +010012602 comma separated list of errorIDs to ignore during verify at depth == 0. If
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012603 set to 'all', all errors are ignored. SSL handshake is not aborted if an error
Alex Davies0fbf0162013-03-02 16:04:50 +000012604 is ignored.
Emeric Brunb6dc9342012-09-28 17:55:37 +020012605
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010012606crt-list <file>
12607 This setting is only available when support for OpenSSL was built in. It
Emmanuel Hocdet98263292016-12-29 18:26:15 +010012608 designates a list of PEM file with an optional ssl configuration and a SNI
12609 filter per certificate, with the following format for each line :
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010012610
Emmanuel Hocdet98263292016-12-29 18:26:15 +010012611 <crtfile> [\[<sslbindconf> ...\]] [[!]<snifilter> ...]
12612
William Lallemand5d036392020-06-30 16:11:36 +020012613 sslbindconf supports "allow-0rtt", "alpn", "ca-file", "ca-verify-file",
12614 "ciphers", "ciphersuites", "crl-file", "curves", "ecdhe", "no-ca-names",
12615 "npn", "verify" configuration. With BoringSSL and Openssl >= 1.1.1
12616 "ssl-min-ver" and "ssl-max-ver" are also supported. It overrides the
12617 configuration set in bind line for the certificate.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010012618
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +020012619 Wildcards are supported in the SNI filter. Negative filter are also supported,
12620 only useful in combination with a wildcard filter to exclude a particular SNI.
12621 The certificates will be presented to clients who provide a valid TLS Server
12622 Name Indication field matching one of the SNI filters. If no SNI filter is
12623 specified, the CN and alt subjects are used. This directive may be specified
12624 multiple times. See the "crt" option for more information. The default
12625 certificate is still needed to meet OpenSSL expectations. If it is not used,
12626 the 'strict-sni' option may be used.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010012627
William Lallemandf9ff3ec2020-10-02 17:57:44 +020012628 Multi-cert bundling (see "ssl-load-extra-files") is supported with crt-list,
12629 as long as only the base name is given in the crt-list. SNI filter will do
12630 the same work on all bundled certificates.
yanbzhud19630c2015-12-14 15:10:25 -050012631
William Lallemand7c26ed72020-06-03 17:34:48 +020012632 Empty lines as well as lines beginning with a hash ('#') will be ignored.
12633
Emmanuel Hocdet98263292016-12-29 18:26:15 +010012634 crt-list file example:
12635 cert1.pem
William Lallemand7c26ed72020-06-03 17:34:48 +020012636 # comment
Emmanuel Hocdet05942112017-02-20 16:11:50 +010012637 cert2.pem [alpn h2,http/1.1]
Emmanuel Hocdet98263292016-12-29 18:26:15 +010012638 certW.pem *.domain.tld !secure.domain.tld
Emmanuel Hocdet05942112017-02-20 16:11:50 +010012639 certS.pem [curves X25519:P-256 ciphers ECDHE-ECDSA-AES256-GCM-SHA384] secure.domain.tld
Emmanuel Hocdet98263292016-12-29 18:26:15 +010012640
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012641defer-accept
12642 Is an optional keyword which is supported only on certain Linux kernels. It
12643 states that a connection will only be accepted once some data arrive on it,
12644 or at worst after the first retransmit. This should be used only on protocols
Davor Ocelice9ed2812017-12-25 17:49:28 +010012645 for which the client talks first (e.g. HTTP). It can slightly improve
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012646 performance by ensuring that most of the request is already available when
12647 the connection is accepted. On the other hand, it will not be able to detect
12648 connections which don't talk. It is important to note that this option is
12649 broken in all kernels up to 2.6.31, as the connection is never accepted until
12650 the client talks. This can cause issues with front firewalls which would see
12651 an established connection while the proxy will only see it in SYN_RECV. This
12652 option is only supported on TCPv4/TCPv6 sockets and ignored by other ones.
12653
William Lallemandf6975e92017-05-26 17:42:10 +020012654expose-fd listeners
12655 This option is only usable with the stats socket. It gives your stats socket
12656 the capability to pass listeners FD to another HAProxy process.
William Lallemande202b1e2017-06-01 17:38:56 +020012657 During a reload with the master-worker mode, the process is automatically
12658 reexecuted adding -x and one of the stats socket with this option.
Davor Ocelice9ed2812017-12-25 17:49:28 +010012659 See also "-x" in the management guide.
William Lallemandf6975e92017-05-26 17:42:10 +020012660
Emeric Brun2cb7ae52012-10-05 14:14:21 +020012661force-sslv3
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012662 This option enforces use of SSLv3 only on SSL connections instantiated from
Emeric Brun2cb7ae52012-10-05 14:14:21 +020012663 this listener. SSLv3 is generally less expensive than the TLS counterparts
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012664 for high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012665 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020012666
12667force-tlsv10
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012668 This option enforces use of TLSv1.0 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012669 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012670 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020012671
12672force-tlsv11
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012673 This option enforces use of TLSv1.1 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012674 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012675 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020012676
12677force-tlsv12
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012678 This option enforces use of TLSv1.2 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012679 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012680 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020012681
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020012682force-tlsv13
12683 This option enforces use of TLSv1.3 only on SSL connections instantiated from
12684 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012685 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020012686
Christopher Faulet31af49d2015-06-09 17:29:50 +020012687generate-certificates
12688 This setting is only available when support for OpenSSL was built in. It
12689 enables the dynamic SSL certificates generation. A CA certificate and its
12690 private key are necessary (see 'ca-sign-file'). When HAProxy is configured as
12691 a transparent forward proxy, SSL requests generate errors because of a common
12692 name mismatch on the certificate presented to the client. With this option
12693 enabled, HAProxy will try to forge a certificate using the SNI hostname
12694 indicated by the client. This is done only if no certificate matches the SNI
12695 hostname (see 'crt-list'). If an error occurs, the default certificate is
12696 used, else the 'strict-sni' option is set.
12697 It can also be used when HAProxy is configured as a reverse proxy to ease the
12698 deployment of an architecture with many backends.
12699
12700 Creating a SSL certificate is an expensive operation, so a LRU cache is used
12701 to store forged certificates (see 'tune.ssl.ssl-ctx-cache-size'). It
Davor Ocelice9ed2812017-12-25 17:49:28 +010012702 increases the HAProxy's memory footprint to reduce latency when the same
Christopher Faulet31af49d2015-06-09 17:29:50 +020012703 certificate is used many times.
12704
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012705gid <gid>
12706 Sets the group of the UNIX sockets to the designated system gid. It can also
12707 be set by default in the global section's "unix-bind" statement. Note that
12708 some platforms simply ignore this. This setting is equivalent to the "group"
12709 setting except that the group ID is used instead of its name. This setting is
12710 ignored by non UNIX sockets.
12711
12712group <group>
12713 Sets the group of the UNIX sockets to the designated system group. It can
12714 also be set by default in the global section's "unix-bind" statement. Note
12715 that some platforms simply ignore this. This setting is equivalent to the
12716 "gid" setting except that the group name is used instead of its gid. This
12717 setting is ignored by non UNIX sockets.
12718
12719id <id>
12720 Fixes the socket ID. By default, socket IDs are automatically assigned, but
12721 sometimes it is more convenient to fix them to ease monitoring. This value
12722 must be strictly positive and unique within the listener/frontend. This
12723 option can only be used when defining only a single socket.
12724
12725interface <interface>
Lukas Tribusfce2e962013-02-12 22:13:19 +010012726 Restricts the socket to a specific interface. When specified, only packets
12727 received from that particular interface are processed by the socket. This is
12728 currently only supported on Linux. The interface must be a primary system
12729 interface, not an aliased interface. It is also possible to bind multiple
12730 frontends to the same address if they are bound to different interfaces. Note
12731 that binding to a network interface requires root privileges. This parameter
Jérôme Magnin61275192018-02-07 11:39:58 +010012732 is only compatible with TCPv4/TCPv6 sockets. When specified, return traffic
12733 uses the same interface as inbound traffic, and its associated routing table,
12734 even if there are explicit routes through different interfaces configured.
12735 This can prove useful to address asymmetric routing issues when the same
12736 client IP addresses need to be able to reach frontends hosted on different
12737 interfaces.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012738
Willy Tarreauabb175f2012-09-24 12:43:26 +020012739level <level>
12740 This setting is used with the stats sockets only to restrict the nature of
12741 the commands that can be issued on the socket. It is ignored by other
12742 sockets. <level> can be one of :
Davor Ocelice9ed2812017-12-25 17:49:28 +010012743 - "user" is the least privileged level; only non-sensitive stats can be
Willy Tarreauabb175f2012-09-24 12:43:26 +020012744 read, and no change is allowed. It would make sense on systems where it
12745 is not easy to restrict access to the socket.
12746 - "operator" is the default level and fits most common uses. All data can
Davor Ocelice9ed2812017-12-25 17:49:28 +010012747 be read, and only non-sensitive changes are permitted (e.g. clear max
Willy Tarreauabb175f2012-09-24 12:43:26 +020012748 counters).
Davor Ocelice9ed2812017-12-25 17:49:28 +010012749 - "admin" should be used with care, as everything is permitted (e.g. clear
Willy Tarreauabb175f2012-09-24 12:43:26 +020012750 all counters).
12751
Andjelko Iharosc4df59e2017-07-20 11:59:48 +020012752severity-output <format>
12753 This setting is used with the stats sockets only to configure severity
12754 level output prepended to informational feedback messages. Severity
12755 level of messages can range between 0 and 7, conforming to syslog
12756 rfc5424. Valid and successful socket commands requesting data
12757 (i.e. "show map", "get acl foo" etc.) will never have a severity level
12758 prepended. It is ignored by other sockets. <format> can be one of :
12759 - "none" (default) no severity level is prepended to feedback messages.
12760 - "number" severity level is prepended as a number.
12761 - "string" severity level is prepended as a string following the
12762 rfc5424 convention.
12763
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012764maxconn <maxconn>
12765 Limits the sockets to this number of concurrent connections. Extraneous
12766 connections will remain in the system's backlog until a connection is
12767 released. If unspecified, the limit will be the same as the frontend's
12768 maxconn. Note that in case of port ranges or multiple addresses, the same
12769 value will be applied to each socket. This setting enables different
12770 limitations on expensive sockets, for instance SSL entries which may easily
12771 eat all memory.
12772
12773mode <mode>
12774 Sets the octal mode used to define access permissions on the UNIX socket. It
12775 can also be set by default in the global section's "unix-bind" statement.
12776 Note that some platforms simply ignore this. This setting is ignored by non
12777 UNIX sockets.
12778
12779mss <maxseg>
12780 Sets the TCP Maximum Segment Size (MSS) value to be advertised on incoming
12781 connections. This can be used to force a lower MSS for certain specific
12782 ports, for instance for connections passing through a VPN. Note that this
12783 relies on a kernel feature which is theoretically supported under Linux but
12784 was buggy in all versions prior to 2.6.28. It may or may not work on other
12785 operating systems. It may also not change the advertised value but change the
12786 effective size of outgoing segments. The commonly advertised value for TCPv4
12787 over Ethernet networks is 1460 = 1500(MTU) - 40(IP+TCP). If this value is
12788 positive, it will be used as the advertised MSS. If it is negative, it will
12789 indicate by how much to reduce the incoming connection's advertised MSS for
12790 outgoing segments. This parameter is only compatible with TCP v4/v6 sockets.
12791
12792name <name>
12793 Sets an optional name for these sockets, which will be reported on the stats
12794 page.
12795
Willy Tarreaud72f0f32015-10-13 14:50:22 +020012796namespace <name>
12797 On Linux, it is possible to specify which network namespace a socket will
12798 belong to. This directive makes it possible to explicitly bind a listener to
12799 a namespace different from the default one. Please refer to your operating
12800 system's documentation to find more details about network namespaces.
12801
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012802nice <nice>
12803 Sets the 'niceness' of connections initiated from the socket. Value must be
12804 in the range -1024..1024 inclusive, and defaults to zero. Positive values
12805 means that such connections are more friendly to others and easily offer
12806 their place in the scheduler. On the opposite, negative values mean that
12807 connections want to run with a higher priority than others. The difference
12808 only happens under high loads when the system is close to saturation.
12809 Negative values are appropriate for low-latency or administration services,
12810 and high values are generally recommended for CPU intensive tasks such as SSL
12811 processing or bulk transfers which are less sensible to latency. For example,
12812 it may make sense to use a positive value for an SMTP socket and a negative
12813 one for an RDP socket.
12814
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020012815no-ca-names
12816 This setting is only available when support for OpenSSL was built in. It
12817 prevents from send CA names in server hello message when ca-file is used.
Emmanuel Hocdet842e94e2019-12-16 16:39:17 +010012818 Use "ca-verify-file" instead of "ca-file" with "no-ca-names".
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020012819
Emeric Brun9b3009b2012-10-05 11:55:06 +020012820no-sslv3
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012821 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012822 disables support for SSLv3 on any sockets instantiated from the listener when
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012823 SSL is supported. Note that SSLv2 is forced disabled in the code and cannot
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012824 be enabled using any configuration option. This option is also available on
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012825 global statement "ssl-default-bind-options". Use "ssl-min-ver" and
12826 "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012827
Emeric Brun90ad8722012-10-02 14:00:59 +020012828no-tls-tickets
12829 This setting is only available when support for OpenSSL was built in. It
12830 disables the stateless session resumption (RFC 5077 TLS Ticket
12831 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012832 session resumption is more expensive in CPU usage. This option is also
12833 available on global statement "ssl-default-bind-options".
Lukas Tribusbdb386d2020-03-10 00:56:09 +010012834 The TLS ticket mechanism is only used up to TLS 1.2.
12835 Forward Secrecy is compromised with TLS tickets, unless ticket keys
12836 are periodically rotated (via reload or by using "tls-ticket-keys").
Emeric Brun90ad8722012-10-02 14:00:59 +020012837
Emeric Brun9b3009b2012-10-05 11:55:06 +020012838no-tlsv10
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012839 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012840 disables support for TLSv1.0 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020012841 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012842 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012843 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
12844 and "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012845
Emeric Brun9b3009b2012-10-05 11:55:06 +020012846no-tlsv11
Emeric Brunf5da4932012-09-28 19:42:54 +020012847 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012848 disables support for TLSv1.1 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020012849 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012850 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012851 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
12852 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020012853
Emeric Brun9b3009b2012-10-05 11:55:06 +020012854no-tlsv12
Emeric Brunf5da4932012-09-28 19:42:54 +020012855 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012856 disables support for TLSv1.2 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020012857 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012858 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012859 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
12860 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020012861
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020012862no-tlsv13
12863 This setting is only available when support for OpenSSL was built in. It
12864 disables support for TLSv1.3 on any sockets instantiated from the listener
12865 when SSL is supported. Note that SSLv2 is forced disabled in the code and
12866 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012867 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
12868 and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020012869
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020012870npn <protocols>
12871 This enables the NPN TLS extension and advertises the specified protocol list
12872 as supported on top of NPN. The protocol list consists in a comma-delimited
12873 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
John Roeslerfb2fce12019-07-10 15:45:51 -050012874 This requires that the SSL library is built with support for TLS extensions
Willy Tarreauab861d32013-04-02 02:30:41 +020012875 enabled (check with haproxy -vv). Note that the NPN extension has been
Willy Tarreau95c4e142017-11-26 12:18:55 +010012876 replaced with the ALPN extension (see the "alpn" keyword), though this one is
12877 only available starting with OpenSSL 1.0.2. If HTTP/2 is desired on an older
12878 version of OpenSSL, NPN might still be used as most clients still support it
12879 at the time of writing this. It is possible to enable both NPN and ALPN
12880 though it probably doesn't make any sense out of testing.
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020012881
Lukas Tribus53ae85c2017-05-04 15:45:40 +000012882prefer-client-ciphers
12883 Use the client's preference when selecting the cipher suite, by default
12884 the server's preference is enforced. This option is also available on
12885 global statement "ssl-default-bind-options".
Lukas Tribus926594f2018-05-18 17:55:57 +020012886 Note that with OpenSSL >= 1.1.1 ChaCha20-Poly1305 is reprioritized anyway
12887 (without setting this option), if a ChaCha20-Poly1305 cipher is at the top of
12888 the client cipher list.
Lukas Tribus53ae85c2017-05-04 15:45:40 +000012889
Christopher Fauletc644fa92017-11-23 22:44:11 +010012890process <process-set>[/<thread-set>]
Willy Tarreaua36b3242019-02-02 13:14:34 +010012891 This restricts the list of processes or threads on which this listener is
Christopher Fauletc644fa92017-11-23 22:44:11 +010012892 allowed to run. It does not enforce any process but eliminates those which do
Davor Ocelice9ed2812017-12-25 17:49:28 +010012893 not match. If the frontend uses a "bind-process" setting, the intersection
Christopher Fauletc644fa92017-11-23 22:44:11 +010012894 between the two is applied. If in the end the listener is not allowed to run
12895 on any remaining process, a warning is emitted, and the listener will either
12896 run on the first process of the listener if a single process was specified,
12897 or on all of its processes if multiple processes were specified. If a thread
Davor Ocelice9ed2812017-12-25 17:49:28 +010012898 set is specified, it limits the threads allowed to process incoming
Willy Tarreaua36b3242019-02-02 13:14:34 +010012899 connections for this listener, for the the process set. If multiple processes
12900 and threads are configured, a warning is emitted, as it either results from a
12901 configuration error or a misunderstanding of these models. For the unlikely
12902 case where several ranges are needed, this directive may be repeated.
12903 <process-set> and <thread-set> must use the format
Christopher Fauletc644fa92017-11-23 22:44:11 +010012904
12905 all | odd | even | number[-[number]]
12906
12907 Ranges can be partially defined. The higher bound can be omitted. In such
12908 case, it is replaced by the corresponding maximum value. The main purpose of
12909 this directive is to be used with the stats sockets and have one different
12910 socket per process. The second purpose is to have multiple bind lines sharing
12911 the same IP:port but not the same process in a listener, so that the system
12912 can distribute the incoming connections into multiple queues and allow a
12913 smoother inter-process load balancing. Currently Linux 3.9 and above is known
12914 for supporting this. See also "bind-process" and "nbproc".
Willy Tarreau6ae1ba62014-05-07 19:01:58 +020012915
Christopher Fauleta717b992018-04-10 14:43:00 +020012916proto <name>
12917 Forces the multiplexer's protocol to use for the incoming connections. It
12918 must be compatible with the mode of the frontend (TCP or HTTP). It must also
12919 be usable on the frontend side. The list of available protocols is reported
12920 in haproxy -vv.
Daniel Corbett67a82712020-07-06 23:01:19 -040012921 Idea behind this option is to bypass the selection of the best multiplexer's
Christopher Fauleta717b992018-04-10 14:43:00 +020012922 protocol for all connections instantiated from this listening socket. For
Joseph Herlant71b4b152018-11-13 16:55:16 -080012923 instance, it is possible to force the http/2 on clear TCP by specifying "proto
Christopher Fauleta717b992018-04-10 14:43:00 +020012924 h2" on the bind line.
12925
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012926ssl
12927 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012928 enables SSL deciphering on connections instantiated from this listener. A
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012929 certificate is necessary (see "crt" above). All contents in the buffers will
12930 appear in clear text, so that ACLs and HTTP processing will only have access
Emmanuel Hocdetbd695fe2017-05-15 15:53:41 +020012931 to deciphered contents. SSLv3 is disabled per default, use "ssl-min-ver SSLv3"
12932 to enable it.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012933
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012934ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
12935 This option enforces use of <version> or lower on SSL connections instantiated
William Lallemand50df1cb2020-06-02 10:52:24 +020012936 from this listener. Using this setting without "ssl-min-ver" can be
12937 ambiguous because the default ssl-min-ver value could change in future HAProxy
12938 versions. 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".
12940
12941ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
William Lallemand50df1cb2020-06-02 10:52:24 +020012942 This option enforces use of <version> or upper on SSL connections
12943 instantiated from this listener. The default value is "TLSv1.2". This option
12944 is also available on global statement "ssl-default-bind-options".
12945 See also "ssl-max-ver".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012946
Emmanuel Hocdet65623372013-01-24 17:17:15 +010012947strict-sni
12948 This setting is only available when support for OpenSSL was built in. The
12949 SSL/TLS negotiation is allow only if the client provided an SNI which match
12950 a certificate. The default certificate is not used.
12951 See the "crt" option for more information.
12952
Willy Tarreau2af207a2015-02-04 00:45:58 +010012953tcp-ut <delay>
Tim Düsterhus4896c442016-11-29 02:15:19 +010012954 Sets the TCP User Timeout for all incoming connections instantiated from this
Willy Tarreau2af207a2015-02-04 00:45:58 +010012955 listening socket. This option is available on Linux since version 2.6.37. It
12956 allows haproxy to configure a timeout for sockets which contain data not
Davor Ocelice9ed2812017-12-25 17:49:28 +010012957 receiving an acknowledgment for the configured delay. This is especially
Willy Tarreau2af207a2015-02-04 00:45:58 +010012958 useful on long-lived connections experiencing long idle periods such as
12959 remote terminals or database connection pools, where the client and server
12960 timeouts must remain high to allow a long period of idle, but where it is
12961 important to detect that the client has disappeared in order to release all
12962 resources associated with its connection (and the server's session). The
12963 argument is a delay expressed in milliseconds by default. This only works
12964 for regular TCP connections, and is ignored for other protocols.
12965
Willy Tarreau1c862c52012-10-05 16:21:00 +020012966tfo
Lukas Tribus0defb902013-02-13 23:35:39 +010012967 Is an optional keyword which is supported only on Linux kernels >= 3.7. It
Willy Tarreau1c862c52012-10-05 16:21:00 +020012968 enables TCP Fast Open on the listening socket, which means that clients which
12969 support this feature will be able to send a request and receive a response
12970 during the 3-way handshake starting from second connection, thus saving one
12971 round-trip after the first connection. This only makes sense with protocols
12972 that use high connection rates and where each round trip matters. This can
12973 possibly cause issues with many firewalls which do not accept data on SYN
12974 packets, so this option should only be enabled once well tested. This option
Lukas Tribus0999f762013-04-02 16:43:24 +020012975 is only supported on TCPv4/TCPv6 sockets and ignored by other ones. You may
12976 need to build HAProxy with USE_TFO=1 if your libc doesn't define
12977 TCP_FASTOPEN.
Willy Tarreau1c862c52012-10-05 16:21:00 +020012978
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010012979tls-ticket-keys <keyfile>
12980 Sets the TLS ticket keys file to load the keys from. The keys need to be 48
Emeric Brun9e754772019-01-10 17:51:55 +010012981 or 80 bytes long, depending if aes128 or aes256 is used, encoded with base64
12982 with one line per key (ex. openssl rand 80 | openssl base64 -A | xargs echo).
12983 The first key determines the key length used for next keys: you can't mix
12984 aes128 and aes256 keys. Number of keys is specified by the TLS_TICKETS_NO
12985 build option (default 3) and at least as many keys need to be present in
12986 the file. Last TLS_TICKETS_NO keys will be used for decryption and the
12987 penultimate one for encryption. This enables easy key rotation by just
12988 appending new key to the file and reloading the process. Keys must be
12989 periodically rotated (ex. every 12h) or Perfect Forward Secrecy is
12990 compromised. It is also a good idea to keep the keys off any permanent
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010012991 storage such as hard drives (hint: use tmpfs and don't swap those files).
12992 Lifetime hint can be changed using tune.ssl.timeout.
12993
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012994transparent
12995 Is an optional keyword which is supported only on certain Linux kernels. It
12996 indicates that the addresses will be bound even if they do not belong to the
12997 local machine, and that packets targeting any of these addresses will be
12998 intercepted just as if the addresses were locally configured. This normally
12999 requires that IP forwarding is enabled. Caution! do not use this with the
13000 default address '*', as it would redirect any traffic for the specified port.
13001 This keyword is available only when HAProxy is built with USE_LINUX_TPROXY=1.
13002 This parameter is only compatible with TCPv4 and TCPv6 sockets, depending on
13003 kernel version. Some distribution kernels include backports of the feature,
13004 so check for support with your vendor.
13005
Willy Tarreau77e3af92012-11-24 15:07:23 +010013006v4v6
13007 Is an optional keyword which is supported only on most recent systems
13008 including Linux kernels >= 2.4.21. It is used to bind a socket to both IPv4
13009 and IPv6 when it uses the default address. Doing so is sometimes necessary
13010 on systems which bind to IPv6 only by default. It has no effect on non-IPv6
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013011 sockets, and is overridden by the "v6only" option.
Willy Tarreau77e3af92012-11-24 15:07:23 +010013012
Willy Tarreau9b6700f2012-11-24 11:55:28 +010013013v6only
13014 Is an optional keyword which is supported only on most recent systems
13015 including Linux kernels >= 2.4.21. It is used to bind a socket to IPv6 only
13016 when it uses the default address. Doing so is sometimes preferred to doing it
Willy Tarreau77e3af92012-11-24 15:07:23 +010013017 system-wide as it is per-listener. It has no effect on non-IPv6 sockets and
13018 has precedence over the "v4v6" option.
Willy Tarreau9b6700f2012-11-24 11:55:28 +010013019
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013020uid <uid>
13021 Sets the owner of the UNIX sockets to the designated system uid. It can also
13022 be set by default in the global section's "unix-bind" statement. Note that
13023 some platforms simply ignore this. This setting is equivalent to the "user"
13024 setting except that the user numeric ID is used instead of its name. This
13025 setting is ignored by non UNIX sockets.
13026
13027user <user>
13028 Sets the owner of the UNIX sockets to the designated system user. It can also
13029 be set by default in the global section's "unix-bind" statement. Note that
13030 some platforms simply ignore this. This setting is equivalent to the "uid"
13031 setting except that the user name is used instead of its uid. This setting is
13032 ignored by non UNIX sockets.
13033
Emeric Brun1a073b42012-09-28 17:07:34 +020013034verify [none|optional|required]
13035 This setting is only available when support for OpenSSL was built in. If set
13036 to 'none', client certificate is not requested. This is the default. In other
13037 cases, a client certificate is requested. If the client does not provide a
13038 certificate after the request and if 'verify' is set to 'required', then the
13039 handshake is aborted, while it would have succeeded if set to 'optional'. The
Emeric Brunfd33a262012-10-11 16:28:27 +020013040 certificate provided by the client is always verified using CAs from
13041 'ca-file' and optional CRLs from 'crl-file'. On verify failure the handshake
13042 is aborted, regardless of the 'verify' option, unless the error code exactly
13043 matches one of those listed with 'ca-ignore-err' or 'crt-ignore-err'.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020013044
Willy Tarreaub6205fd2012-09-24 12:27:33 +0200130455.2. Server and default-server options
Cyril Bontéf0c60612010-02-06 14:44:47 +010013046------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020013047
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010013048The "server" and "default-server" keywords support a certain number of settings
13049which are all passed as arguments on the server line. The order in which those
13050arguments appear does not count, and they are all optional. Some of those
13051settings are single words (booleans) while others expect one or several values
13052after them. In this case, the values must immediately follow the setting name.
13053Except default-server, all those settings must be specified after the server's
13054address if they are used:
Willy Tarreau6a06a402007-07-15 20:15:28 +020013055
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013056 server <name> <address>[:port] [settings ...]
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010013057 default-server [settings ...]
Willy Tarreau6a06a402007-07-15 20:15:28 +020013058
Frédéric Lécailled2376272017-03-21 18:52:12 +010013059Note that all these settings are supported both by "server" and "default-server"
13060keywords, except "id" which is only supported by "server".
13061
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013062The currently supported settings are the following ones.
Willy Tarreau0ba27502007-12-24 16:55:16 +010013063
Willy Tarreauceb4ac92012-04-28 00:41:46 +020013064addr <ipv4|ipv6>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013065 Using the "addr" parameter, it becomes possible to use a different IP address
Baptiste Assmann13f83532016-03-06 23:14:36 +010013066 to send health-checks or to probe the agent-check. On some servers, it may be
13067 desirable to dedicate an IP address to specific component able to perform
13068 complex tests which are more suitable to health-checks than the application.
13069 This parameter is ignored if the "check" parameter is not set. See also the
13070 "port" parameter.
Willy Tarreau6a06a402007-07-15 20:15:28 +020013071
Simon Hormand60d6912013-11-25 10:46:36 +090013072agent-check
13073 Enable an auxiliary agent check which is run independently of a regular
Willy Tarreau81f5d942013-12-09 20:51:51 +010013074 health check. An agent health check is performed by making a TCP connection
Willy Tarreau7a0139e2018-12-16 08:42:56 +010013075 to the port set by the "agent-port" parameter and reading an ASCII string
13076 terminated by the first '\r' or '\n' met. The string is made of a series of
13077 words delimited by spaces, tabs or commas in any order, each consisting of :
Simon Hormand60d6912013-11-25 10:46:36 +090013078
Willy Tarreau81f5d942013-12-09 20:51:51 +010013079 - An ASCII representation of a positive integer percentage, e.g. "75%".
Simon Hormand60d6912013-11-25 10:46:36 +090013080 Values in this format will set the weight proportional to the initial
Willy Tarreauc5af3a62014-10-07 15:27:33 +020013081 weight of a server as configured when haproxy starts. Note that a zero
13082 weight is reported on the stats page as "DRAIN" since it has the same
13083 effect on the server (it's removed from the LB farm).
Simon Hormand60d6912013-11-25 10:46:36 +090013084
Davor Ocelice9ed2812017-12-25 17:49:28 +010013085 - The string "maxconn:" followed by an integer (no space between). Values
13086 in this format will set the maxconn of a server. The maximum number of
13087 connections advertised needs to be multiplied by the number of load
13088 balancers and different backends that use this health check to get the
13089 total number of connections the server might receive. Example: maxconn:30
Nenad Merdanovic174dd372016-04-24 23:10:06 +020013090
Willy Tarreau81f5d942013-12-09 20:51:51 +010013091 - The word "ready". This will turn the server's administrative state to the
Davor Ocelice9ed2812017-12-25 17:49:28 +010013092 READY mode, thus canceling any DRAIN or MAINT state
Simon Hormand60d6912013-11-25 10:46:36 +090013093
Willy Tarreau81f5d942013-12-09 20:51:51 +010013094 - The word "drain". This will turn the server's administrative state to the
13095 DRAIN mode, thus it will not accept any new connections other than those
13096 that are accepted via persistence.
Simon Hormand60d6912013-11-25 10:46:36 +090013097
Willy Tarreau81f5d942013-12-09 20:51:51 +010013098 - The word "maint". This will turn the server's administrative state to the
13099 MAINT mode, thus it will not accept any new connections at all, and health
13100 checks will be stopped.
Simon Hormand60d6912013-11-25 10:46:36 +090013101
William Dauchyf8e795c2020-09-26 13:35:51 +020013102 - The words "down", "fail", or "stopped", optionally followed by a
Willy Tarreau81f5d942013-12-09 20:51:51 +010013103 description string after a sharp ('#'). All of these mark the server's
13104 operating state as DOWN, but since the word itself is reported on the stats
13105 page, the difference allows an administrator to know if the situation was
13106 expected or not : the service may intentionally be stopped, may appear up
Davor Ocelice9ed2812017-12-25 17:49:28 +010013107 but fail some validity tests, or may be seen as down (e.g. missing process,
Willy Tarreau81f5d942013-12-09 20:51:51 +010013108 or port not responding).
Simon Hormand60d6912013-11-25 10:46:36 +090013109
Willy Tarreau81f5d942013-12-09 20:51:51 +010013110 - The word "up" sets back the server's operating state as UP if health checks
13111 also report that the service is accessible.
Simon Hormand60d6912013-11-25 10:46:36 +090013112
Willy Tarreau81f5d942013-12-09 20:51:51 +010013113 Parameters which are not advertised by the agent are not changed. For
13114 example, an agent might be designed to monitor CPU usage and only report a
13115 relative weight and never interact with the operating status. Similarly, an
13116 agent could be designed as an end-user interface with 3 radio buttons
13117 allowing an administrator to change only the administrative state. However,
13118 it is important to consider that only the agent may revert its own actions,
13119 so if a server is set to DRAIN mode or to DOWN state using the agent, the
13120 agent must implement the other equivalent actions to bring the service into
13121 operations again.
Simon Hormand60d6912013-11-25 10:46:36 +090013122
Simon Horman2f1f9552013-11-25 10:46:37 +090013123 Failure to connect to the agent is not considered an error as connectivity
13124 is tested by the regular health check which is enabled by the "check"
Willy Tarreau81f5d942013-12-09 20:51:51 +010013125 parameter. Warning though, it is not a good idea to stop an agent after it
13126 reports "down", since only an agent reporting "up" will be able to turn the
13127 server up again. Note that the CLI on the Unix stats socket is also able to
Willy Tarreau989222a2016-01-15 10:26:26 +010013128 force an agent's result in order to work around a bogus agent if needed.
Simon Horman2f1f9552013-11-25 10:46:37 +090013129
Willy Tarreau81f5d942013-12-09 20:51:51 +010013130 Requires the "agent-port" parameter to be set. See also the "agent-inter"
Frédéric Lécailled2376272017-03-21 18:52:12 +010013131 and "no-agent-check" parameters.
Simon Hormand60d6912013-11-25 10:46:36 +090013132
James Brown55f9ff12015-10-21 18:19:05 -070013133agent-send <string>
13134 If this option is specified, haproxy will send the given string (verbatim)
13135 to the agent server upon connection. You could, for example, encode
13136 the backend name into this string, which would enable your agent to send
13137 different responses based on the backend. Make sure to include a '\n' if
13138 you want to terminate your request with a newline.
13139
Simon Hormand60d6912013-11-25 10:46:36 +090013140agent-inter <delay>
13141 The "agent-inter" parameter sets the interval between two agent checks
13142 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
13143
13144 Just as with every other time-based parameter, it may be entered in any
13145 other explicit unit among { us, ms, s, m, h, d }. The "agent-inter"
13146 parameter also serves as a timeout for agent checks "timeout check" is
13147 not set. In order to reduce "resonance" effects when multiple servers are
13148 hosted on the same hardware, the agent and health checks of all servers
13149 are started with a small time offset between them. It is also possible to
13150 add some random noise in the agent and health checks interval using the
13151 global "spread-checks" keyword. This makes sense for instance when a lot
13152 of backends use the same servers.
13153
13154 See also the "agent-check" and "agent-port" parameters.
13155
Misiek768d8602017-01-09 09:52:43 +010013156agent-addr <addr>
13157 The "agent-addr" parameter sets address for agent check.
13158
13159 You can offload agent-check to another target, so you can make single place
13160 managing status and weights of servers defined in haproxy in case you can't
13161 make self-aware and self-managing services. You can specify both IP or
13162 hostname, it will be resolved.
13163
Simon Hormand60d6912013-11-25 10:46:36 +090013164agent-port <port>
13165 The "agent-port" parameter sets the TCP port used for agent checks.
13166
13167 See also the "agent-check" and "agent-inter" parameters.
13168
Olivier Houchard8cb2d2e2019-05-06 18:58:48 +020013169allow-0rtt
13170 Allow sending early data to the server when using TLS 1.3.
Olivier Houchard22c9b442019-05-06 19:01:04 +020013171 Note that early data will be sent only if the client used early data, or
13172 if the backend uses "retry-on" with the "0rtt-rejected" keyword.
Olivier Houchard8cb2d2e2019-05-06 18:58:48 +020013173
Olivier Houchardc7566002018-11-20 23:33:50 +010013174alpn <protocols>
13175 This enables the TLS ALPN extension and advertises the specified protocol
13176 list as supported on top of ALPN. The protocol list consists in a comma-
13177 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
John Roeslerfb2fce12019-07-10 15:45:51 -050013178 quotes). This requires that the SSL library is built with support for TLS
Olivier Houchardc7566002018-11-20 23:33:50 +010013179 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
13180 initial NPN extension. ALPN is required to connect to HTTP/2 servers.
13181 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
13182 now obsolete NPN extension.
13183 If both HTTP/2 and HTTP/1.1 are expected to be supported, both versions can
13184 be advertised, in order of preference, like below :
13185
13186 server 127.0.0.1:443 ssl crt pub.pem alpn h2,http/1.1
13187
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013188backup
13189 When "backup" is present on a server line, the server is only used in load
13190 balancing when all other non-backup servers are unavailable. Requests coming
13191 with a persistence cookie referencing the server will always be served
13192 though. By default, only the first operational backup server is used, unless
Frédéric Lécailled2376272017-03-21 18:52:12 +010013193 the "allbackups" option is set in the backend. See also the "no-backup" and
13194 "allbackups" options.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013195
Emeric Brunef42d922012-10-11 16:11:36 +020013196ca-file <cafile>
13197 This setting is only available when support for OpenSSL was built in. It
13198 designates a PEM file from which to load CA certificates used to verify
13199 server's certificate.
13200
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013201check
Jerome Magnin90702bc2020-04-26 14:23:04 +020013202 This option enables health checks on a server:
13203 - when not set, no health checking is performed, and the server is always
13204 considered available.
13205 - when set and no other check method is configured, the server is considered
13206 available when a connection can be established at the highest configured
13207 transport layer. This means TCP by default, or SSL/TLS when "ssl" or
13208 "check-ssl" are set, both possibly combined with connection prefixes such
13209 as a PROXY protocol header when "send-proxy" or "check-send-proxy" are
13210 set.
13211 - when set and an application-level health check is defined, the
13212 application-level exchanges are performed on top of the configured
13213 transport layer and the server is considered available if all of the
13214 exchanges succeed.
13215
13216 By default, health checks are performed on the same address and port as
13217 configured on the server, using the same encapsulation parameters (SSL/TLS,
13218 proxy-protocol header, etc... ). It is possible to change the destination
13219 address using "addr" and the port using "port". When done, it is assumed the
13220 server isn't checked on the service port, and configured encapsulation
Ilya Shipitsin4329a9a2020-05-05 21:17:10 +050013221 parameters are not reused. One must explicitly set "check-send-proxy" to send
Jerome Magnin90702bc2020-04-26 14:23:04 +020013222 connection headers, "check-ssl" to use SSL/TLS.
13223
13224 When "sni" or "alpn" are set on the server line, their value is not used for
13225 health checks and one must use "check-sni" or "check-alpn".
13226
13227 The default source address for health check traffic is the same as the one
13228 defined in the backend. It can be changed with the "source" keyword.
13229
13230 The interval between checks can be set using the "inter" keyword, and the
13231 "rise" and "fall" keywords can be used to define how many successful or
13232 failed health checks are required to flag a server available or not
13233 available.
13234
13235 Optional application-level health checks can be configured with "option
13236 httpchk", "option mysql-check" "option smtpchk", "option pgsql-check",
13237 "option ldap-check", or "option redis-check".
13238
13239 Example:
13240 # simple tcp check
13241 backend foo
13242 server s1 192.168.0.1:80 check
13243 # this does a tcp connect + tls handshake
13244 backend foo
13245 server s1 192.168.0.1:443 ssl check
13246 # simple tcp check is enough for check success
13247 backend foo
13248 option tcp-check
13249 tcp-check connect
13250 server s1 192.168.0.1:443 ssl check
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013251
Willy Tarreau6c16adc2012-10-05 00:04:16 +020013252check-send-proxy
13253 This option forces emission of a PROXY protocol line with outgoing health
13254 checks, regardless of whether the server uses send-proxy or not for the
13255 normal traffic. By default, the PROXY protocol is enabled for health checks
13256 if it is already enabled for normal traffic and if no "port" nor "addr"
13257 directive is present. However, if such a directive is present, the
13258 "check-send-proxy" option needs to be used to force the use of the
13259 protocol. See also the "send-proxy" option for more information.
13260
Olivier Houchard92150142018-12-21 19:47:01 +010013261check-alpn <protocols>
13262 Defines which protocols to advertise with ALPN. The protocol list consists in
13263 a comma-delimited list of protocol names, for instance: "http/1.1,http/1.0"
13264 (without quotes). If it is not set, the server ALPN is used.
13265
Christopher Fauletedc6ed92020-04-23 16:27:59 +020013266check-proto <name>
13267 Forces the multiplexer's protocol to use for the server's health-check
13268 connections. It must be compatible with the health-check type (TCP or
13269 HTTP). It must also be usable on the backend side. The list of available
13270 protocols is reported in haproxy -vv.
Daniel Corbett67a82712020-07-06 23:01:19 -040013271 Idea behind this option is to bypass the selection of the best multiplexer's
Christopher Fauletedc6ed92020-04-23 16:27:59 +020013272 protocol for health-check connections established to this server.
13273 If not defined, the server one will be used, if set.
13274
Jérôme Magninae9bb762018-12-09 16:08:26 +010013275check-sni <sni>
Olivier Houchard9130a962017-10-17 17:33:43 +020013276 This option allows you to specify the SNI to be used when doing health checks
Jérôme Magninae9bb762018-12-09 16:08:26 +010013277 over SSL. It is only possible to use a string to set <sni>. If you want to
13278 set a SNI for proxied traffic, see "sni".
Olivier Houchard9130a962017-10-17 17:33:43 +020013279
Willy Tarreau763a95b2012-10-04 23:15:39 +020013280check-ssl
13281 This option forces encryption of all health checks over SSL, regardless of
13282 whether the server uses SSL or not for the normal traffic. This is generally
13283 used when an explicit "port" or "addr" directive is specified and SSL health
13284 checks are not inherited. It is important to understand that this option
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013285 inserts an SSL transport layer below the checks, so that a simple TCP connect
Willy Tarreau763a95b2012-10-04 23:15:39 +020013286 check becomes an SSL connect, which replaces the old ssl-hello-chk. The most
13287 common use is to send HTTPS checks by combining "httpchk" with SSL checks.
Davor Ocelice9ed2812017-12-25 17:49:28 +010013288 All SSL settings are common to health checks and traffic (e.g. ciphers).
Frédéric Lécailled2376272017-03-21 18:52:12 +010013289 See the "ssl" option for more information and "no-check-ssl" to disable
13290 this option.
Willy Tarreau763a95b2012-10-04 23:15:39 +020013291
Alexander Liu2a54bb72019-05-22 19:44:48 +080013292check-via-socks4
John Roeslerfb2fce12019-07-10 15:45:51 -050013293 This option enables outgoing health checks using upstream socks4 proxy. By
Alexander Liu2a54bb72019-05-22 19:44:48 +080013294 default, the health checks won't go through socks tunnel even it was enabled
13295 for normal traffic.
13296
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020013297ciphers <ciphers>
Dirkjan Bussink415150f2018-09-14 11:14:21 +020013298 This setting is only available when support for OpenSSL was built in. This
13299 option sets the string describing the list of cipher algorithms that is
13300 negotiated during the SSL/TLS handshake with the server. The format of the
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000013301 string is defined in "man 1 ciphers" from OpenSSL man pages. For background
13302 information and recommendations see e.g.
13303 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
13304 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
13305 cipher configuration, please check the "ciphersuites" keyword.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020013306
Dirkjan Bussink415150f2018-09-14 11:14:21 +020013307ciphersuites <ciphersuites>
13308 This setting is only available when support for OpenSSL was built in and
13309 OpenSSL 1.1.1 or later was used to build HAProxy. This option sets the string
13310 describing the list of cipher algorithms that is negotiated during the TLS
13311 1.3 handshake with the server. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000013312 "man 1 ciphers" from OpenSSL man pages under the "ciphersuites" section.
13313 For cipher configuration for TLSv1.2 and earlier, please check the "ciphers"
13314 keyword.
Dirkjan Bussink415150f2018-09-14 11:14:21 +020013315
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013316cookie <value>
13317 The "cookie" parameter sets the cookie value assigned to the server to
13318 <value>. This value will be checked in incoming requests, and the first
13319 operational server possessing the same value will be selected. In return, in
13320 cookie insertion or rewrite modes, this value will be assigned to the cookie
13321 sent to the client. There is nothing wrong in having several servers sharing
13322 the same cookie value, and it is in fact somewhat common between normal and
13323 backup servers. See also the "cookie" keyword in backend section.
13324
Emeric Brunef42d922012-10-11 16:11:36 +020013325crl-file <crlfile>
13326 This setting is only available when support for OpenSSL was built in. It
13327 designates a PEM file from which to load certificate revocation list used
13328 to verify server's certificate.
13329
Emeric Bruna7aa3092012-10-26 12:58:00 +020013330crt <cert>
13331 This setting is only available when support for OpenSSL was built in.
13332 It designates a PEM file from which to load both a certificate and the
13333 associated private key. This file can be built by concatenating both PEM
13334 files into one. This certificate will be sent if the server send a client
13335 certificate request.
13336
Willy Tarreau96839092010-03-29 10:02:24 +020013337disabled
13338 The "disabled" keyword starts the server in the "disabled" state. That means
13339 that it is marked down in maintenance mode, and no connection other than the
13340 ones allowed by persist mode will reach it. It is very well suited to setup
13341 new servers, because normal traffic will never reach them, while it is still
13342 possible to test the service by making use of the force-persist mechanism.
Frédéric Lécailled2376272017-03-21 18:52:12 +010013343 See also "enabled" setting.
Willy Tarreau96839092010-03-29 10:02:24 +020013344
Frédéric Lécailled2376272017-03-21 18:52:12 +010013345enabled
13346 This option may be used as 'server' setting to reset any 'disabled'
13347 setting which would have been inherited from 'default-server' directive as
13348 default value.
13349 It may also be used as 'default-server' setting to reset any previous
13350 'default-server' 'disabled' setting.
Willy Tarreau96839092010-03-29 10:02:24 +020013351
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013352error-limit <count>
Willy Tarreau983e01e2010-01-11 18:42:06 +010013353 If health observing is enabled, the "error-limit" parameter specifies the
13354 number of consecutive errors that triggers event selected by the "on-error"
13355 option. By default it is set to 10 consecutive errors.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010013356
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013357 See also the "check", "error-limit" and "on-error".
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010013358
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013359fall <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013360 The "fall" parameter states that a server will be considered as dead after
13361 <count> consecutive unsuccessful health checks. This value defaults to 3 if
13362 unspecified. See also the "check", "inter" and "rise" parameters.
13363
Emeric Brun8694b9a2012-10-05 14:39:07 +020013364force-sslv3
13365 This option enforces use of SSLv3 only when SSL is used to communicate with
13366 the server. SSLv3 is generally less expensive than the TLS counterparts for
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013367 high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013368 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020013369
13370force-tlsv10
13371 This option enforces use of TLSv1.0 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013372 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013373 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020013374
13375force-tlsv11
13376 This option enforces use of TLSv1.1 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013377 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013378 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020013379
13380force-tlsv12
13381 This option enforces use of TLSv1.2 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013382 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013383 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020013384
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020013385force-tlsv13
13386 This option enforces use of TLSv1.3 only when SSL is used to communicate with
13387 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013388 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020013389
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013390id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +020013391 Set a persistent ID for the server. This ID must be positive and unique for
13392 the proxy. An unused ID will automatically be assigned if unset. The first
13393 assigned value will be 1. This ID is currently only returned in statistics.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013394
Willy Tarreau6a031d12016-11-07 19:42:35 +010013395init-addr {last | libc | none | <ip>},[...]*
13396 Indicate in what order the server's address should be resolved upon startup
13397 if it uses an FQDN. Attempts are made to resolve the address by applying in
Davor Ocelice9ed2812017-12-25 17:49:28 +010013398 turn each of the methods mentioned in the comma-delimited list. The first
Willy Tarreau6a031d12016-11-07 19:42:35 +010013399 method which succeeds is used. If the end of the list is reached without
13400 finding a working method, an error is thrown. Method "last" suggests to pick
13401 the address which appears in the state file (see "server-state-file"). Method
13402 "libc" uses the libc's internal resolver (gethostbyname() or getaddrinfo()
13403 depending on the operating system and build options). Method "none"
13404 specifically indicates that the server should start without any valid IP
13405 address in a down state. It can be useful to ignore some DNS issues upon
13406 startup, waiting for the situation to get fixed later. Finally, an IP address
13407 (IPv4 or IPv6) may be provided. It can be the currently known address of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010013408 server (e.g. filled by a configuration generator), or the address of a dummy
Willy Tarreau6a031d12016-11-07 19:42:35 +010013409 server used to catch old sessions and present them with a decent error
13410 message for example. When the "first" load balancing algorithm is used, this
13411 IP address could point to a fake server used to trigger the creation of new
13412 instances on the fly. This option defaults to "last,libc" indicating that the
13413 previous address found in the state file (if any) is used first, otherwise
13414 the libc's resolver is used. This ensures continued compatibility with the
Davor Ocelice9ed2812017-12-25 17:49:28 +010013415 historic behavior.
Willy Tarreau6a031d12016-11-07 19:42:35 +010013416
13417 Example:
13418 defaults
13419 # never fail on address resolution
13420 default-server init-addr last,libc,none
13421
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013422inter <delay>
13423fastinter <delay>
13424downinter <delay>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013425 The "inter" parameter sets the interval between two consecutive health checks
13426 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
13427 It is also possible to use "fastinter" and "downinter" to optimize delays
13428 between checks depending on the server state :
13429
Pieter Baauw44fc9df2015-09-17 21:30:46 +020013430 Server state | Interval used
13431 ----------------------------------------+----------------------------------
13432 UP 100% (non-transitional) | "inter"
13433 ----------------------------------------+----------------------------------
13434 Transitionally UP (going down "fall"), | "fastinter" if set,
13435 Transitionally DOWN (going up "rise"), | "inter" otherwise.
13436 or yet unchecked. |
13437 ----------------------------------------+----------------------------------
13438 DOWN 100% (non-transitional) | "downinter" if set,
13439 | "inter" otherwise.
13440 ----------------------------------------+----------------------------------
Willy Tarreaud72758d2010-01-12 10:42:19 +010013441
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013442 Just as with every other time-based parameter, they can be entered in any
13443 other explicit unit among { us, ms, s, m, h, d }. The "inter" parameter also
13444 serves as a timeout for health checks sent to servers if "timeout check" is
13445 not set. In order to reduce "resonance" effects when multiple servers are
Simon Hormand60d6912013-11-25 10:46:36 +090013446 hosted on the same hardware, the agent and health checks of all servers
13447 are started with a small time offset between them. It is also possible to
13448 add some random noise in the agent and health checks interval using the
13449 global "spread-checks" keyword. This makes sense for instance when a lot
13450 of backends use the same servers.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013451
Emeric Brun97556472020-05-30 01:42:45 +020013452log-proto <logproto>
13453 The "log-proto" specifies the protocol used to forward event messages to
13454 a server configured in a ring section. Possible values are "legacy"
13455 and "octet-count" corresponding respectively to "Non-transparent-framing"
13456 and "Octet counting" in rfc6587. "legacy" is the default.
13457
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013458maxconn <maxconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013459 The "maxconn" parameter specifies the maximal number of concurrent
13460 connections that will be sent to this server. If the number of incoming
Tim Duesterhuscefbbd92019-11-27 22:35:27 +010013461 concurrent connections goes higher than this value, they will be queued,
13462 waiting for a slot to be released. This parameter is very important as it can
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013463 save fragile servers from going down under extreme loads. If a "minconn"
13464 parameter is specified, the limit becomes dynamic. The default value is "0"
13465 which means unlimited. See also the "minconn" and "maxqueue" parameters, and
13466 the backend's "fullconn" keyword.
13467
Tim Duesterhuscefbbd92019-11-27 22:35:27 +010013468 In HTTP mode this parameter limits the number of concurrent requests instead
13469 of the number of connections. Multiple requests might be multiplexed over a
13470 single TCP connection to the server. As an example if you specify a maxconn
13471 of 50 you might see between 1 and 50 actual server connections, but no more
13472 than 50 concurrent requests.
13473
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013474maxqueue <maxqueue>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013475 The "maxqueue" parameter specifies the maximal number of connections which
13476 will wait in the queue for this server. If this limit is reached, next
13477 requests will be redispatched to other servers instead of indefinitely
13478 waiting to be served. This will break persistence but may allow people to
13479 quickly re-log in when the server they try to connect to is dying. The
13480 default value is "0" which means the queue is unlimited. See also the
13481 "maxconn" and "minconn" parameters.
13482
Willy Tarreau9c538e02019-01-23 10:21:49 +010013483max-reuse <count>
13484 The "max-reuse" argument indicates the HTTP connection processors that they
13485 should not reuse a server connection more than this number of times to send
13486 new requests. Permitted values are -1 (the default), which disables this
13487 limit, or any positive value. Value zero will effectively disable keep-alive.
13488 This is only used to work around certain server bugs which cause them to leak
13489 resources over time. The argument is not necessarily respected by the lower
13490 layers as there might be technical limitations making it impossible to
13491 enforce. At least HTTP/2 connections to servers will respect it.
13492
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013493minconn <minconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013494 When the "minconn" parameter is set, the maxconn limit becomes a dynamic
13495 limit following the backend's load. The server will always accept at least
13496 <minconn> connections, never more than <maxconn>, and the limit will be on
13497 the ramp between both values when the backend has less than <fullconn>
13498 concurrent connections. This makes it possible to limit the load on the
13499 server during normal loads, but push it further for important loads without
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010013500 overloading the server during exceptional loads. See also the "maxconn"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013501 and "maxqueue" parameters, as well as the "fullconn" backend keyword.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010013502
Willy Tarreaud72f0f32015-10-13 14:50:22 +020013503namespace <name>
13504 On Linux, it is possible to specify which network namespace a socket will
13505 belong to. This directive makes it possible to explicitly bind a server to
13506 a namespace different from the default one. Please refer to your operating
13507 system's documentation to find more details about network namespaces.
13508
Frédéric Lécailled2376272017-03-21 18:52:12 +010013509no-agent-check
13510 This option may be used as "server" setting to reset any "agent-check"
13511 setting which would have been inherited from "default-server" directive as
13512 default value.
13513 It may also be used as "default-server" setting to reset any previous
13514 "default-server" "agent-check" setting.
13515
13516no-backup
13517 This option may be used as "server" setting to reset any "backup"
13518 setting which would have been inherited from "default-server" directive as
13519 default value.
13520 It may also be used as "default-server" setting to reset any previous
13521 "default-server" "backup" setting.
13522
13523no-check
13524 This option may be used as "server" setting to reset any "check"
13525 setting which would have been inherited from "default-server" directive as
13526 default value.
13527 It may also be used as "default-server" setting to reset any previous
13528 "default-server" "check" setting.
13529
13530no-check-ssl
13531 This option may be used as "server" setting to reset any "check-ssl"
13532 setting which would have been inherited from "default-server" directive as
13533 default value.
13534 It may also be used as "default-server" setting to reset any previous
13535 "default-server" "check-ssl" setting.
13536
Frédéric Lécailled2376272017-03-21 18:52:12 +010013537no-send-proxy
13538 This option may be used as "server" setting to reset any "send-proxy"
13539 setting which would have been inherited from "default-server" directive as
13540 default value.
13541 It may also be used as "default-server" setting to reset any previous
13542 "default-server" "send-proxy" setting.
13543
13544no-send-proxy-v2
13545 This option may be used as "server" setting to reset any "send-proxy-v2"
13546 setting which would have been inherited from "default-server" directive as
13547 default value.
13548 It may also be used as "default-server" setting to reset any previous
13549 "default-server" "send-proxy-v2" setting.
13550
13551no-send-proxy-v2-ssl
13552 This option may be used as "server" setting to reset any "send-proxy-v2-ssl"
13553 setting which would have been inherited from "default-server" directive as
13554 default value.
13555 It may also be used as "default-server" setting to reset any previous
13556 "default-server" "send-proxy-v2-ssl" setting.
13557
13558no-send-proxy-v2-ssl-cn
13559 This option may be used as "server" setting to reset any "send-proxy-v2-ssl-cn"
13560 setting which would have been inherited from "default-server" directive as
13561 default value.
13562 It may also be used as "default-server" setting to reset any previous
13563 "default-server" "send-proxy-v2-ssl-cn" setting.
13564
13565no-ssl
13566 This option may be used as "server" setting to reset any "ssl"
13567 setting which would have been inherited from "default-server" directive as
13568 default value.
13569 It may also be used as "default-server" setting to reset any previous
13570 "default-server" "ssl" setting.
13571
Willy Tarreau2a3fb1c2015-02-05 16:47:07 +010013572no-ssl-reuse
13573 This option disables SSL session reuse when SSL is used to communicate with
13574 the server. It will force the server to perform a full handshake for every
13575 new connection. It's probably only useful for benchmarking, troubleshooting,
13576 and for paranoid users.
13577
Emeric Brun9b3009b2012-10-05 11:55:06 +020013578no-sslv3
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020013579 This option disables support for SSLv3 when SSL is used to communicate with
13580 the server. Note that SSLv2 is disabled in the code and cannot be enabled
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013581 using any configuration option. Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020013582
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020013583 Supported in default-server: No
13584
Emeric Brunf9c5c472012-10-11 15:28:34 +020013585no-tls-tickets
13586 This setting is only available when support for OpenSSL was built in. It
13587 disables the stateless session resumption (RFC 5077 TLS Ticket
13588 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013589 session resumption is more expensive in CPU usage for servers. This option
13590 is also available on global statement "ssl-default-server-options".
Lukas Tribusbdb386d2020-03-10 00:56:09 +010013591 The TLS ticket mechanism is only used up to TLS 1.2.
13592 Forward Secrecy is compromised with TLS tickets, unless ticket keys
13593 are periodically rotated (via reload or by using "tls-ticket-keys").
Frédéric Lécailled2376272017-03-21 18:52:12 +010013594 See also "tls-tickets".
Emeric Brunf9c5c472012-10-11 15:28:34 +020013595
Emeric Brun9b3009b2012-10-05 11:55:06 +020013596no-tlsv10
Emeric Brun8694b9a2012-10-05 14:39:07 +020013597 This option disables support for TLSv1.0 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020013598 the server. Note that SSLv2 is disabled in the code and cannot be enabled
13599 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013600 often makes sense to disable it when communicating with local servers. This
13601 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013602 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020013603
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020013604 Supported in default-server: No
13605
Emeric Brun9b3009b2012-10-05 11:55:06 +020013606no-tlsv11
Emeric Brun8694b9a2012-10-05 14:39:07 +020013607 This option disables support for TLSv1.1 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020013608 the server. Note that SSLv2 is disabled in the code and cannot be enabled
13609 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013610 often makes sense to disable it when communicating with local servers. This
13611 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013612 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020013613
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020013614 Supported in default-server: No
13615
Emeric Brun9b3009b2012-10-05 11:55:06 +020013616no-tlsv12
Emeric Brun8694b9a2012-10-05 14:39:07 +020013617 This option disables support for TLSv1.2 when SSL is used to communicate with
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020013618 the server. Note that SSLv2 is disabled in the code and cannot be enabled
13619 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013620 often makes sense to disable it when communicating with local servers. This
13621 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013622 Use "ssl-min-ver" and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020013623
13624 Supported in default-server: No
13625
13626no-tlsv13
13627 This option disables support for TLSv1.3 when SSL is used to communicate with
13628 the server. Note that SSLv2 is disabled in the code and cannot be enabled
13629 using any configuration option. TLSv1 is more expensive than SSLv3 so it
13630 often makes sense to disable it when communicating with local servers. This
13631 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013632 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020013633
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020013634 Supported in default-server: No
13635
Frédéric Lécailled2376272017-03-21 18:52:12 +010013636no-verifyhost
13637 This option may be used as "server" setting to reset any "verifyhost"
13638 setting which would have been inherited from "default-server" directive as
13639 default value.
13640 It may also be used as "default-server" setting to reset any previous
13641 "default-server" "verifyhost" setting.
Willy Tarreau763a95b2012-10-04 23:15:39 +020013642
Frédéric Lécaille1b9423d2019-07-04 14:19:06 +020013643no-tfo
13644 This option may be used as "server" setting to reset any "tfo"
13645 setting which would have been inherited from "default-server" directive as
13646 default value.
13647 It may also be used as "default-server" setting to reset any previous
13648 "default-server" "tfo" setting.
13649
Simon Hormanfa461682011-06-25 09:39:49 +090013650non-stick
13651 Never add connections allocated to this sever to a stick-table.
13652 This may be used in conjunction with backup to ensure that
13653 stick-table persistence is disabled for backup servers.
13654
Olivier Houchardc7566002018-11-20 23:33:50 +010013655npn <protocols>
13656 This enables the NPN TLS extension and advertises the specified protocol list
13657 as supported on top of NPN. The protocol list consists in a comma-delimited
13658 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
John Roeslerfb2fce12019-07-10 15:45:51 -050013659 This requires that the SSL library is built with support for TLS extensions
Olivier Houchardc7566002018-11-20 23:33:50 +010013660 enabled (check with haproxy -vv). Note that the NPN extension has been
13661 replaced with the ALPN extension (see the "alpn" keyword), though this one is
13662 only available starting with OpenSSL 1.0.2.
13663
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010013664observe <mode>
13665 This option enables health adjusting based on observing communication with
13666 the server. By default this functionality is disabled and enabling it also
13667 requires to enable health checks. There are two supported modes: "layer4" and
13668 "layer7". In layer4 mode, only successful/unsuccessful tcp connections are
13669 significant. In layer7, which is only allowed for http proxies, responses
13670 received from server are verified, like valid/wrong http code, unparsable
Willy Tarreau150d1462012-03-10 08:19:02 +010013671 headers, a timeout, etc. Valid status codes include 100 to 499, 501 and 505.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010013672
13673 See also the "check", "on-error" and "error-limit".
13674
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013675on-error <mode>
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010013676 Select what should happen when enough consecutive errors are detected.
13677 Currently, four modes are available:
13678 - fastinter: force fastinter
13679 - fail-check: simulate a failed check, also forces fastinter (default)
13680 - sudden-death: simulate a pre-fatal failed health check, one more failed
13681 check will mark a server down, forces fastinter
13682 - mark-down: mark the server immediately down and force fastinter
13683
13684 See also the "check", "observe" and "error-limit".
13685
Simon Hormane0d1bfb2011-06-21 14:34:58 +090013686on-marked-down <action>
13687 Modify what occurs when a server is marked down.
13688 Currently one action is available:
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070013689 - shutdown-sessions: Shutdown peer sessions. When this setting is enabled,
13690 all connections to the server are immediately terminated when the server
13691 goes down. It might be used if the health check detects more complex cases
13692 than a simple connection status, and long timeouts would cause the service
13693 to remain unresponsive for too long a time. For instance, a health check
13694 might detect that a database is stuck and that there's no chance to reuse
13695 existing connections anymore. Connections killed this way are logged with
13696 a 'D' termination code (for "Down").
Simon Hormane0d1bfb2011-06-21 14:34:58 +090013697
13698 Actions are disabled by default
13699
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070013700on-marked-up <action>
13701 Modify what occurs when a server is marked up.
13702 Currently one action is available:
13703 - shutdown-backup-sessions: Shutdown sessions on all backup servers. This is
13704 done only if the server is not in backup state and if it is not disabled
13705 (it must have an effective weight > 0). This can be used sometimes to force
13706 an active server to take all the traffic back after recovery when dealing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013707 with long sessions (e.g. LDAP, SQL, ...). Doing this can cause more trouble
13708 than it tries to solve (e.g. incomplete transactions), so use this feature
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070013709 with extreme care. Sessions killed because a server comes up are logged
13710 with an 'U' termination code (for "Up").
13711
13712 Actions are disabled by default
13713
Willy Tarreau2f3f4d32020-07-01 07:43:51 +020013714pool-low-conn <max>
13715 Set a low threshold on the number of idling connections for a server, below
13716 which a thread will not try to steal a connection from another thread. This
13717 can be useful to improve CPU usage patterns in scenarios involving many very
13718 fast servers, in order to ensure all threads will keep a few idle connections
13719 all the time instead of letting them accumulate over one thread and migrating
13720 them from thread to thread. Typical values of twice the number of threads
13721 seem to show very good performance already with sub-millisecond response
13722 times. The default is zero, indicating that any idle connection can be used
13723 at any time. It is the recommended setting for normal use. This only applies
13724 to connections that can be shared according to the same principles as those
13725 applying to "http-reuse".
13726
Olivier Houchard006e3102018-12-10 18:30:32 +010013727pool-max-conn <max>
13728 Set the maximum number of idling connections for a server. -1 means unlimited
13729 connections, 0 means no idle connections. The default is -1. When idle
13730 connections are enabled, orphaned idle connections which do not belong to any
13731 client session anymore are moved to a dedicated pool so that they remain
13732 usable by future clients. This only applies to connections that can be shared
13733 according to the same principles as those applying to "http-reuse".
13734
Olivier Houchardb7b3faa2018-12-14 18:15:36 +010013735pool-purge-delay <delay>
13736 Sets the delay to start purging idle connections. Each <delay> interval, half
Olivier Houcharda56eebf2019-03-19 16:44:02 +010013737 of the idle connections are closed. 0 means we don't keep any idle connection.
Willy Tarreaufb553652019-06-04 14:06:31 +020013738 The default is 5s.
Olivier Houchardb7b3faa2018-12-14 18:15:36 +010013739
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013740port <port>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013741 Using the "port" parameter, it becomes possible to use a different port to
13742 send health-checks. On some servers, it may be desirable to dedicate a port
13743 to a specific component able to perform complex tests which are more suitable
13744 to health-checks than the application. It is common to run a simple script in
13745 inetd for instance. This parameter is ignored if the "check" parameter is not
13746 set. See also the "addr" parameter.
13747
Christopher Faulet8ed0a3e2018-04-10 14:45:45 +020013748proto <name>
Christopher Faulet8ed0a3e2018-04-10 14:45:45 +020013749 Forces the multiplexer's protocol to use for the outgoing connections to this
13750 server. It must be compatible with the mode of the backend (TCP or HTTP). It
13751 must also be usable on the backend side. The list of available protocols is
13752 reported in haproxy -vv.
Daniel Corbett67a82712020-07-06 23:01:19 -040013753 Idea behind this option is to bypass the selection of the best multiplexer's
Christopher Faulet8ed0a3e2018-04-10 14:45:45 +020013754 protocol for all connections established to this server.
13755
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013756redir <prefix>
13757 The "redir" parameter enables the redirection mode for all GET and HEAD
13758 requests addressing this server. This means that instead of having HAProxy
13759 forward the request to the server, it will send an "HTTP 302" response with
13760 the "Location" header composed of this prefix immediately followed by the
13761 requested URI beginning at the leading '/' of the path component. That means
13762 that no trailing slash should be used after <prefix>. All invalid requests
13763 will be rejected, and all non-GET or HEAD requests will be normally served by
13764 the server. Note that since the response is completely forged, no header
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010013765 mangling nor cookie insertion is possible in the response. However, cookies in
Davor Ocelice9ed2812017-12-25 17:49:28 +010013766 requests are still analyzed, making this solution completely usable to direct
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013767 users to a remote location in case of local disaster. Main use consists in
13768 increasing bandwidth for static servers by having the clients directly
13769 connect to them. Note: never use a relative location here, it would cause a
13770 loop between the client and HAProxy!
13771
13772 Example : server srv1 192.168.1.1:80 redir http://image1.mydomain.com check
13773
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013774rise <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013775 The "rise" parameter states that a server will be considered as operational
13776 after <count> consecutive successful health checks. This value defaults to 2
13777 if unspecified. See also the "check", "inter" and "fall" parameters.
13778
Baptiste Assmann8e2d9432018-06-22 15:04:43 +020013779resolve-opts <option>,<option>,...
13780 Comma separated list of options to apply to DNS resolution linked to this
13781 server.
13782
13783 Available options:
13784
13785 * allow-dup-ip
13786 By default, HAProxy prevents IP address duplication in a backend when DNS
13787 resolution at runtime is in operation.
13788 That said, for some cases, it makes sense that two servers (in the same
13789 backend, being resolved by the same FQDN) have the same IP address.
13790 For such case, simply enable this option.
13791 This is the opposite of prevent-dup-ip.
13792
Daniel Corbettf8716912019-11-17 09:48:56 -050013793 * ignore-weight
13794 Ignore any weight that is set within an SRV record. This is useful when
13795 you would like to control the weights using an alternate method, such as
13796 using an "agent-check" or through the runtime api.
13797
Baptiste Assmann8e2d9432018-06-22 15:04:43 +020013798 * prevent-dup-ip
13799 Ensure HAProxy's default behavior is enforced on a server: prevent re-using
13800 an IP address already set to a server in the same backend and sharing the
13801 same fqdn.
13802 This is the opposite of allow-dup-ip.
13803
13804 Example:
13805 backend b_myapp
13806 default-server init-addr none resolvers dns
13807 server s1 myapp.example.com:80 check resolve-opts allow-dup-ip
13808 server s2 myapp.example.com:81 check resolve-opts allow-dup-ip
13809
13810 With the option allow-dup-ip set:
13811 * if the nameserver returns a single IP address, then both servers will use
13812 it
13813 * If the nameserver returns 2 IP addresses, then each server will pick up a
13814 different address
13815
13816 Default value: not set
13817
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013818resolve-prefer <family>
13819 When DNS resolution is enabled for a server and multiple IP addresses from
13820 different families are returned, HAProxy will prefer using an IP address
13821 from the family mentioned in the "resolve-prefer" parameter.
13822 Available families: "ipv4" and "ipv6"
13823
Baptiste Assmannc4aabae2015-08-04 22:43:06 +020013824 Default value: ipv6
13825
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020013826 Example:
13827
13828 server s1 app1.domain.com:80 resolvers mydns resolve-prefer ipv6
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013829
Thierry Fournierac88cfe2016-02-17 22:05:30 +010013830resolve-net <network>[,<network[,...]]
John Roeslerfb2fce12019-07-10 15:45:51 -050013831 This option prioritizes the choice of an ip address matching a network. This is
Thierry Fournierac88cfe2016-02-17 22:05:30 +010013832 useful with clouds to prefer a local ip. In some cases, a cloud high
Tim Düsterhus4896c442016-11-29 02:15:19 +010013833 availability service can be announced with many ip addresses on many
Davor Ocelice9ed2812017-12-25 17:49:28 +010013834 different datacenters. The latency between datacenter is not negligible, so
13835 this patch permits to prefer a local datacenter. If no address matches the
Thierry Fournierac88cfe2016-02-17 22:05:30 +010013836 configured network, another address is selected.
13837
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020013838 Example:
13839
13840 server s1 app1.domain.com:80 resolvers mydns resolve-net 10.0.0.0/8
Thierry Fournierac88cfe2016-02-17 22:05:30 +010013841
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013842resolvers <id>
13843 Points to an existing "resolvers" section to resolve current server's
13844 hostname.
13845
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020013846 Example:
13847
13848 server s1 app1.domain.com:80 check resolvers mydns
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013849
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020013850 See also section 5.3
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013851
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010013852send-proxy
13853 The "send-proxy" parameter enforces use of the PROXY protocol over any
13854 connection established to this server. The PROXY protocol informs the other
13855 end about the layer 3/4 addresses of the incoming connection, so that it can
13856 know the client's address or the public address it accessed to, whatever the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010013857 upper layer protocol. For connections accepted by an "accept-proxy" or
13858 "accept-netscaler-cip" listener, the advertised address will be used. Only
13859 TCPv4 and TCPv6 address families are supported. Other families such as
13860 Unix sockets, will report an UNKNOWN family. Servers using this option can
13861 fully be chained to another instance of haproxy listening with an
13862 "accept-proxy" setting. This setting must not be used if the server isn't
13863 aware of the protocol. When health checks are sent to the server, the PROXY
13864 protocol is automatically used when this option is set, unless there is an
13865 explicit "port" or "addr" directive, in which case an explicit
13866 "check-send-proxy" directive would also be needed to use the PROXY protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010013867 See also the "no-send-proxy" option of this section and "accept-proxy" and
13868 "accept-netscaler-cip" option of the "bind" keyword.
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010013869
David Safb76832014-05-08 23:42:08 -040013870send-proxy-v2
13871 The "send-proxy-v2" parameter enforces use of the PROXY protocol version 2
13872 over any connection established to this server. The PROXY protocol informs
13873 the other end about the layer 3/4 addresses of the incoming connection, so
13874 that it can know the client's address or the public address it accessed to,
Emmanuel Hocdet404d9782017-10-24 10:55:14 +020013875 whatever the upper layer protocol. It also send ALPN information if an alpn
13876 have been negotiated. This setting must not be used if the server isn't aware
13877 of this version of the protocol. See also the "no-send-proxy-v2" option of
13878 this section and send-proxy" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040013879
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010013880proxy-v2-options <option>[,<option>]*
Tim Duesterhuscf6e0c82020-03-13 12:34:24 +010013881 The "proxy-v2-options" parameter add options to send in PROXY protocol
13882 version 2 when "send-proxy-v2" is used. Options available are:
13883
13884 - ssl : See also "send-proxy-v2-ssl".
13885 - cert-cn : See also "send-proxy-v2-ssl-cn".
13886 - ssl-cipher: Name of the used cipher.
13887 - cert-sig : Signature algorithm of the used certificate.
13888 - cert-key : Key algorithm of the used certificate
13889 - authority : Host name value passed by the client (only SNI from a TLS
13890 connection is supported).
13891 - crc32c : Checksum of the PROXYv2 header.
13892 - unique-id : Send a unique ID generated using the frontend's
13893 "unique-id-format" within the PROXYv2 header.
13894 This unique-id is primarily meant for "mode tcp". It can
13895 lead to unexpected results in "mode http", because the
13896 generated unique ID is also used for the first HTTP request
13897 within a Keep-Alive connection.
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010013898
David Safb76832014-05-08 23:42:08 -040013899send-proxy-v2-ssl
13900 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
13901 2 over any connection established to this server. The PROXY protocol informs
13902 the other end about the layer 3/4 addresses of the incoming connection, so
13903 that it can know the client's address or the public address it accessed to,
13904 whatever the upper layer protocol. In addition, the SSL information extension
13905 of the PROXY protocol is added to the PROXY protocol header. This setting
13906 must not be used if the server isn't aware of this version of the protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010013907 See also the "no-send-proxy-v2-ssl" option of this section and the
13908 "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040013909
13910send-proxy-v2-ssl-cn
13911 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
13912 2 over any connection established to this server. The PROXY protocol informs
13913 the other end about the layer 3/4 addresses of the incoming connection, so
13914 that it can know the client's address or the public address it accessed to,
13915 whatever the upper layer protocol. In addition, the SSL information extension
13916 of the PROXY protocol, along along with the Common Name from the subject of
13917 the client certificate (if any), is added to the PROXY protocol header. This
13918 setting must not be used if the server isn't aware of this version of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010013919 protocol. See also the "no-send-proxy-v2-ssl-cn" option of this section and
13920 the "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040013921
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013922slowstart <start_time_in_ms>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013923 The "slowstart" parameter for a server accepts a value in milliseconds which
13924 indicates after how long a server which has just come back up will run at
13925 full speed. Just as with every other time-based parameter, it can be entered
13926 in any other explicit unit among { us, ms, s, m, h, d }. The speed grows
13927 linearly from 0 to 100% during this time. The limitation applies to two
13928 parameters :
13929
13930 - maxconn: the number of connections accepted by the server will grow from 1
13931 to 100% of the usual dynamic limit defined by (minconn,maxconn,fullconn).
13932
13933 - weight: when the backend uses a dynamic weighted algorithm, the weight
13934 grows linearly from 1 to 100%. In this case, the weight is updated at every
13935 health-check. For this reason, it is important that the "inter" parameter
13936 is smaller than the "slowstart", in order to maximize the number of steps.
13937
13938 The slowstart never applies when haproxy starts, otherwise it would cause
13939 trouble to running servers. It only applies when a server has been previously
13940 seen as failed.
13941
Willy Tarreau732eac42015-07-09 11:40:25 +020013942sni <expression>
13943 The "sni" parameter evaluates the sample fetch expression, converts it to a
13944 string and uses the result as the host name sent in the SNI TLS extension to
13945 the server. A typical use case is to send the SNI received from the client in
13946 a bridged HTTPS scenario, using the "ssl_fc_sni" sample fetch for the
Willy Tarreau2ab88672017-07-05 18:23:03 +020013947 expression, though alternatives such as req.hdr(host) can also make sense. If
13948 "verify required" is set (which is the recommended setting), the resulting
Willy Tarreauad92a9a2017-07-28 11:38:41 +020013949 name will also be matched against the server certificate's names. See the
Jérôme Magninb36a6d22018-12-09 16:03:40 +010013950 "verify" directive for more details. If you want to set a SNI for health
13951 checks, see the "check-sni" directive for more details.
Willy Tarreau732eac42015-07-09 11:40:25 +020013952
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020013953source <addr>[:<pl>[-<ph>]] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +020013954source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020013955source <addr>[:<pl>[-<ph>]] [interface <name>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013956 The "source" parameter sets the source address which will be used when
13957 connecting to the server. It follows the exact same parameters and principle
13958 as the backend "source" keyword, except that it only applies to the server
13959 referencing it. Please consult the "source" keyword for details.
13960
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020013961 Additionally, the "source" statement on a server line allows one to specify a
13962 source port range by indicating the lower and higher bounds delimited by a
13963 dash ('-'). Some operating systems might require a valid IP address when a
13964 source port range is specified. It is permitted to have the same IP/range for
13965 several servers. Doing so makes it possible to bypass the maximum of 64k
13966 total concurrent connections. The limit will then reach 64k connections per
13967 server.
13968
Lukas Tribus7d56c6d2016-09-13 09:51:15 +000013969 Since Linux 4.2/libc 2.23 IP_BIND_ADDRESS_NO_PORT is set for connections
13970 specifying the source address without port(s).
13971
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020013972ssl
Willy Tarreau44f65392013-06-25 07:56:20 +020013973 This option enables SSL ciphering on outgoing connections to the server. It
13974 is critical to verify server certificates using "verify" when using SSL to
13975 connect to servers, otherwise the communication is prone to trivial man in
13976 the-middle attacks rendering SSL useless. When this option is used, health
13977 checks are automatically sent in SSL too unless there is a "port" or an
13978 "addr" directive indicating the check should be sent to a different location.
Frédéric Lécailled2376272017-03-21 18:52:12 +010013979 See the "no-ssl" to disable "ssl" option and "check-ssl" option to force
13980 SSL health checks.
Willy Tarreau763a95b2012-10-04 23:15:39 +020013981
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013982ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
13983 This option enforces use of <version> or lower when SSL is used to communicate
13984 with the server. This option is also available on global statement
13985 "ssl-default-server-options". See also "ssl-min-ver".
13986
13987ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
13988 This option enforces use of <version> or upper when SSL is used to communicate
13989 with the server. This option is also available on global statement
13990 "ssl-default-server-options". See also "ssl-max-ver".
13991
Frédéric Lécailled2376272017-03-21 18:52:12 +010013992ssl-reuse
13993 This option may be used as "server" setting to reset any "no-ssl-reuse"
13994 setting which would have been inherited from "default-server" directive as
13995 default value.
13996 It may also be used as "default-server" setting to reset any previous
13997 "default-server" "no-ssl-reuse" setting.
13998
13999stick
14000 This option may be used as "server" setting to reset any "non-stick"
14001 setting which would have been inherited from "default-server" directive as
14002 default value.
14003 It may also be used as "default-server" setting to reset any previous
14004 "default-server" "non-stick" setting.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020014005
Alexander Liu2a54bb72019-05-22 19:44:48 +080014006socks4 <addr>:<port>
John Roeslerfb2fce12019-07-10 15:45:51 -050014007 This option enables upstream socks4 tunnel for outgoing connections to the
Alexander Liu2a54bb72019-05-22 19:44:48 +080014008 server. Using this option won't force the health check to go via socks4 by
14009 default. You will have to use the keyword "check-via-socks4" to enable it.
14010
Willy Tarreau163d4622015-10-13 16:16:41 +020014011tcp-ut <delay>
14012 Sets the TCP User Timeout for all outgoing connections to this server. This
14013 option is available on Linux since version 2.6.37. It allows haproxy to
14014 configure a timeout for sockets which contain data not receiving an
Davor Ocelice9ed2812017-12-25 17:49:28 +010014015 acknowledgment for the configured delay. This is especially useful on
Willy Tarreau163d4622015-10-13 16:16:41 +020014016 long-lived connections experiencing long idle periods such as remote
14017 terminals or database connection pools, where the client and server timeouts
14018 must remain high to allow a long period of idle, but where it is important to
14019 detect that the server has disappeared in order to release all resources
14020 associated with its connection (and the client's session). One typical use
14021 case is also to force dead server connections to die when health checks are
14022 too slow or during a soft reload since health checks are then disabled. The
14023 argument is a delay expressed in milliseconds by default. This only works for
14024 regular TCP connections, and is ignored for other protocols.
14025
Willy Tarreau034c88c2017-01-23 23:36:45 +010014026tfo
14027 This option enables using TCP fast open when connecting to servers, on
14028 systems that support it (currently only the Linux kernel >= 4.11).
14029 See the "tfo" bind option for more information about TCP fast open.
14030 Please note that when using tfo, you should also use the "conn-failure",
14031 "empty-response" and "response-timeout" keywords for "retry-on", or haproxy
Frédéric Lécaille1b9423d2019-07-04 14:19:06 +020014032 won't be able to retry the connection on failure. See also "no-tfo".
Willy Tarreau034c88c2017-01-23 23:36:45 +010014033
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014034track [<proxy>/]<server>
Willy Tarreau32091232014-05-16 13:52:00 +020014035 This option enables ability to set the current state of the server by tracking
14036 another one. It is possible to track a server which itself tracks another
14037 server, provided that at the end of the chain, a server has health checks
14038 enabled. If <proxy> is omitted the current one is used. If disable-on-404 is
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014039 used, it has to be enabled on both proxies.
14040
Frédéric Lécailled2376272017-03-21 18:52:12 +010014041tls-tickets
14042 This option may be used as "server" setting to reset any "no-tls-tickets"
14043 setting which would have been inherited from "default-server" directive as
14044 default value.
Lukas Tribusbdb386d2020-03-10 00:56:09 +010014045 The TLS ticket mechanism is only used up to TLS 1.2.
14046 Forward Secrecy is compromised with TLS tickets, unless ticket keys
14047 are periodically rotated (via reload or by using "tls-ticket-keys").
Frédéric Lécailled2376272017-03-21 18:52:12 +010014048 It may also be used as "default-server" setting to reset any previous
Bjoern Jacke5ab7eb62020-02-13 14:16:16 +010014049 "default-server" "no-tls-tickets" setting.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014050
Emeric Brunef42d922012-10-11 16:11:36 +020014051verify [none|required]
14052 This setting is only available when support for OpenSSL was built in. If set
Emeric Brun850efd52014-01-29 12:24:34 +010014053 to 'none', server certificate is not verified. In the other case, The
Willy Tarreauad92a9a2017-07-28 11:38:41 +020014054 certificate provided by the server is verified using CAs from 'ca-file' and
14055 optional CRLs from 'crl-file' after having checked that the names provided in
Davor Ocelice9ed2812017-12-25 17:49:28 +010014056 the certificate's subject and subjectAlternateNames attributes match either
Willy Tarreauad92a9a2017-07-28 11:38:41 +020014057 the name passed using the "sni" directive, or if not provided, the static
14058 host name passed using the "verifyhost" directive. When no name is found, the
14059 certificate's names are ignored. For this reason, without SNI it's important
14060 to use "verifyhost". On verification failure the handshake is aborted. It is
14061 critically important to verify server certificates when using SSL to connect
14062 to servers, otherwise the communication is prone to trivial man-in-the-middle
14063 attacks rendering SSL totally useless. Unless "ssl_server_verify" appears in
14064 the global section, "verify" is set to "required" by default.
Emeric Brunef42d922012-10-11 16:11:36 +020014065
Evan Broderbe554312013-06-27 00:05:25 -070014066verifyhost <hostname>
14067 This setting is only available when support for OpenSSL was built in, and
Willy Tarreauad92a9a2017-07-28 11:38:41 +020014068 only takes effect if 'verify required' is also specified. This directive sets
14069 a default static hostname to check the server's certificate against when no
14070 SNI was used to connect to the server. If SNI is not used, this is the only
14071 way to enable hostname verification. This static hostname, when set, will
14072 also be used for health checks (which cannot provide an SNI value). If none
14073 of the hostnames in the certificate match the specified hostname, the
14074 handshake is aborted. The hostnames in the server-provided certificate may
14075 include wildcards. See also "verify", "sni" and "no-verifyhost" options.
Evan Broderbe554312013-06-27 00:05:25 -070014076
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014077weight <weight>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014078 The "weight" parameter is used to adjust the server's weight relative to
14079 other servers. All servers will receive a load proportional to their weight
14080 relative to the sum of all weights, so the higher the weight, the higher the
Willy Tarreau6704d672009-06-15 10:56:05 +020014081 load. The default weight is 1, and the maximal value is 256. A value of 0
14082 means the server will not participate in load-balancing but will still accept
14083 persistent connections. If this parameter is used to distribute the load
14084 according to server's capacity, it is recommended to start with values which
14085 can both grow and shrink, for instance between 10 and 100 to leave enough
14086 room above and below for later adjustments.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014087
14088
Cyril Bonté46175dd2015-07-02 22:45:32 +0200140895.3. Server IP address resolution using DNS
14090-------------------------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014091
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014092HAProxy allows using a host name on the server line to retrieve its IP address
14093using name servers. By default, HAProxy resolves the name when parsing the
14094configuration file, at startup and cache the result for the process' life.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014095This is not sufficient in some cases, such as in Amazon where a server's IP
14096can change after a reboot or an ELB Virtual IP can change based on current
14097workload.
14098This chapter describes how HAProxy can be configured to process server's name
14099resolution at run time.
14100Whether run time server name resolution has been enable or not, HAProxy will
14101carry on doing the first resolution when parsing the configuration.
14102
14103
Cyril Bonté46175dd2015-07-02 22:45:32 +0200141045.3.1. Global overview
14105----------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014106
14107As we've seen in introduction, name resolution in HAProxy occurs at two
14108different steps of the process life:
14109
14110 1. when starting up, HAProxy parses the server line definition and matches a
14111 host name. It uses libc functions to get the host name resolved. This
14112 resolution relies on /etc/resolv.conf file.
14113
Christopher Faulet67957bd2017-09-27 11:00:59 +020014114 2. at run time, HAProxy performs periodically name resolutions for servers
14115 requiring DNS resolutions.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014116
14117A few other events can trigger a name resolution at run time:
14118 - when a server's health check ends up in a connection timeout: this may be
14119 because the server has a new IP address. So we need to trigger a name
14120 resolution to know this new IP.
14121
Christopher Faulet67957bd2017-09-27 11:00:59 +020014122When using resolvers, the server name can either be a hostname, or a SRV label.
Davor Ocelice9ed2812017-12-25 17:49:28 +010014123HAProxy considers anything that starts with an underscore as a SRV label. If a
Christopher Faulet67957bd2017-09-27 11:00:59 +020014124SRV label is specified, then the corresponding SRV records will be retrieved
14125from the DNS server, and the provided hostnames will be used. The SRV label
14126will be checked periodically, and if any server are added or removed, haproxy
14127will automatically do the same.
Olivier Houchardecfa18d2017-08-07 17:30:03 +020014128
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014129A few things important to notice:
John Roeslerfb2fce12019-07-10 15:45:51 -050014130 - all the name servers are queried in the meantime. HAProxy will process the
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014131 first valid response.
14132
14133 - a resolution is considered as invalid (NX, timeout, refused), when all the
14134 servers return an error.
14135
14136
Cyril Bonté46175dd2015-07-02 22:45:32 +0200141375.3.2. The resolvers section
14138----------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014139
14140This section is dedicated to host information related to name resolution in
Christopher Faulet67957bd2017-09-27 11:00:59 +020014141HAProxy. There can be as many as resolvers section as needed. Each section can
14142contain many name servers.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014143
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014144When multiple name servers are configured in a resolvers section, then HAProxy
14145uses the first valid response. In case of invalid responses, only the last one
14146is treated. Purpose is to give the chance to a slow server to deliver a valid
14147answer after a fast faulty or outdated server.
14148
14149When each server returns a different error type, then only the last error is
Christopher Faulet67957bd2017-09-27 11:00:59 +020014150used by HAProxy. The following processing is applied on this error:
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014151
Christopher Faulet67957bd2017-09-27 11:00:59 +020014152 1. HAProxy retries the same DNS query with a new query type. The A queries are
14153 switch to AAAA or the opposite. SRV queries are not concerned here. Timeout
14154 errors are also excluded.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014155
Christopher Faulet67957bd2017-09-27 11:00:59 +020014156 2. When the fallback on the query type was done (or not applicable), HAProxy
14157 retries the original DNS query, with the preferred query type.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014158
Christopher Faulet67957bd2017-09-27 11:00:59 +020014159 3. HAProxy retries previous steps <resolve_retires> times. If no valid
14160 response is received after that, it stops the DNS resolution and reports
14161 the error.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014162
Christopher Faulet67957bd2017-09-27 11:00:59 +020014163For example, with 2 name servers configured in a resolvers section, the
14164following scenarios are possible:
14165
14166 - First response is valid and is applied directly, second response is
14167 ignored
14168
14169 - First response is invalid and second one is valid, then second response is
14170 applied
14171
14172 - First response is a NX domain and second one a truncated response, then
14173 HAProxy retries the query with a new type
14174
14175 - First response is a NX domain and second one is a timeout, then HAProxy
14176 retries the query with a new type
14177
14178 - Query timed out for both name servers, then HAProxy retries it with the
14179 same query type
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014180
Olivier Houcharda8c6db82017-07-06 18:46:47 +020014181As a DNS server may not answer all the IPs in one DNS request, haproxy keeps
14182a cache of previous answers, an answer will be considered obsolete after
Christopher Faulet67957bd2017-09-27 11:00:59 +020014183<hold obsolete> seconds without the IP returned.
Olivier Houcharda8c6db82017-07-06 18:46:47 +020014184
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014185
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014186resolvers <resolvers id>
Davor Ocelice9ed2812017-12-25 17:49:28 +010014187 Creates a new name server list labeled <resolvers id>
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014188
14189A resolvers section accept the following parameters:
14190
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020014191accepted_payload_size <nb>
Davor Ocelice9ed2812017-12-25 17:49:28 +010014192 Defines the maximum payload size accepted by HAProxy and announced to all the
Christopher Faulet67957bd2017-09-27 11:00:59 +020014193 name servers configured in this resolvers section.
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020014194 <nb> is in bytes. If not set, HAProxy announces 512. (minimal value defined
14195 by RFC 6891)
14196
Baptiste Assmann9d8dbbc2017-08-18 23:35:08 +020014197 Note: the maximum allowed value is 8192.
14198
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014199nameserver <id> <ip>:<port>
14200 DNS server description:
14201 <id> : label of the server, should be unique
14202 <ip> : IP address of the server
14203 <port> : port where the DNS service actually runs
14204
Ben Draut44e609b2018-05-29 15:40:08 -060014205parse-resolv-conf
14206 Adds all nameservers found in /etc/resolv.conf to this resolvers nameservers
14207 list. Ordered as if each nameserver in /etc/resolv.conf was individually
14208 placed in the resolvers section in place of this directive.
14209
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014210hold <status> <period>
14211 Defines <period> during which the last name resolution should be kept based
14212 on last resolution <status>
Baptiste Assmann987e16d2016-11-02 22:23:31 +010014213 <status> : last name resolution status. Acceptable values are "nx",
Olivier Houcharda8c6db82017-07-06 18:46:47 +020014214 "other", "refused", "timeout", "valid", "obsolete".
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014215 <period> : interval between two successive name resolution when the last
14216 answer was in <status>. It follows the HAProxy time format.
14217 <period> is in milliseconds by default.
14218
Baptiste Assmann686408b2017-08-18 10:15:42 +020014219 Default value is 10s for "valid", 0s for "obsolete" and 30s for others.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014220
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014221resolve_retries <nb>
14222 Defines the number <nb> of queries to send to resolve a server name before
14223 giving up.
14224 Default value: 3
14225
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014226 A retry occurs on name server timeout or when the full sequence of DNS query
14227 type failover is over and we need to start up from the default ANY query
14228 type.
14229
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014230timeout <event> <time>
14231 Defines timeouts related to name resolution
14232 <event> : the event on which the <time> timeout period applies to.
14233 events available are:
Frédéric Lécaille93d33162019-03-06 09:35:59 +010014234 - resolve : default time to trigger name resolutions when no
14235 other time applied.
Christopher Faulet67957bd2017-09-27 11:00:59 +020014236 Default value: 1s
14237 - retry : time between two DNS queries, when no valid response
Frédéric Lécaille93d33162019-03-06 09:35:59 +010014238 have been received.
Christopher Faulet67957bd2017-09-27 11:00:59 +020014239 Default value: 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014240 <time> : time related to the event. It follows the HAProxy time format.
14241 <time> is expressed in milliseconds.
14242
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020014243 Example:
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014244
14245 resolvers mydns
14246 nameserver dns1 10.0.0.1:53
14247 nameserver dns2 10.0.0.2:53
Ben Draut44e609b2018-05-29 15:40:08 -060014248 parse-resolv-conf
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014249 resolve_retries 3
Christopher Faulet67957bd2017-09-27 11:00:59 +020014250 timeout resolve 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014251 timeout retry 1s
Baptiste Assmann987e16d2016-11-02 22:23:31 +010014252 hold other 30s
14253 hold refused 30s
14254 hold nx 30s
14255 hold timeout 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014256 hold valid 10s
Olivier Houcharda8c6db82017-07-06 18:46:47 +020014257 hold obsolete 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014258
14259
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200142606. Cache
14261---------
14262
14263HAProxy provides a cache, which was designed to perform cache on small objects
14264(favicon, css...). This is a minimalist low-maintenance cache which runs in
14265RAM.
14266
14267The cache is based on a memory which is shared between processes and threads,
14268this memory is split in blocks of 1k.
14269
14270If an object is not used anymore, it can be deleted to store a new object
14271independently of its expiration date. The oldest objects are deleted first
14272when we try to allocate a new one.
14273
14274The cache uses a hash of the host header and the URI as the key.
14275
14276It's possible to view the status of a cache using the Unix socket command
14277"show cache" consult section 9.3 "Unix Socket commands" of Management Guide
14278for more details.
14279
14280When an object is delivered from the cache, the server name in the log is
14281replaced by "<CACHE>".
14282
14283
142846.1. Limitation
14285----------------
14286
14287The cache won't store and won't deliver objects in these cases:
14288
14289- If the response is not a 200
14290- If the response contains a Vary header
14291- If the Content-Length + the headers size is greater than "max-object-size"
14292- If the response is not cacheable
14293
14294- If the request is not a GET
14295- If the HTTP version of the request is smaller than 1.1
14296- If the request contains an Authorization header
14297
14298
142996.2. Setup
14300-----------
14301
14302To setup a cache, you must define a cache section and use it in a proxy with
14303the corresponding http-request and response actions.
14304
14305
143066.2.1. Cache section
14307---------------------
14308
14309cache <name>
14310 Declare a cache section, allocate a shared cache memory named <name>, the
14311 size of cache is mandatory.
14312
14313total-max-size <megabytes>
14314 Define the size in RAM of the cache in megabytes. This size is split in
14315 blocks of 1kB which are used by the cache entries. Its maximum value is 4095.
14316
14317max-object-size <bytes>
14318 Define the maximum size of the objects to be cached. Must not be greater than
14319 an half of "total-max-size". If not set, it equals to a 256th of the cache size.
14320 All objects with sizes larger than "max-object-size" will not be cached.
14321
14322max-age <seconds>
14323 Define the maximum expiration duration. The expiration is set has the lowest
14324 value between the s-maxage or max-age (in this order) directive in the
14325 Cache-Control response header and this value. The default value is 60
14326 seconds, which means that you can't cache an object more than 60 seconds by
14327 default.
14328
14329
143306.2.2. Proxy section
14331---------------------
14332
14333http-request cache-use <name> [ { if | unless } <condition> ]
14334 Try to deliver a cached object from the cache <name>. This directive is also
14335 mandatory to store the cache as it calculates the cache hash. If you want to
14336 use a condition for both storage and delivering that's a good idea to put it
14337 after this one.
14338
14339http-response cache-store <name> [ { if | unless } <condition> ]
14340 Store an http-response within the cache. The storage of the response headers
14341 is done at this step, which means you can use others http-response actions
14342 to modify headers before or after the storage of the response. This action
14343 is responsible for the setup of the cache storage filter.
14344
14345
14346Example:
14347
14348 backend bck1
14349 mode http
14350
14351 http-request cache-use foobar
14352 http-response cache-store foobar
14353 server srv1 127.0.0.1:80
14354
14355 cache foobar
14356 total-max-size 4
14357 max-age 240
14358
14359
Willy Tarreau74ca5042013-06-11 23:12:07 +0200143607. Using ACLs and fetching samples
14361----------------------------------
14362
Davor Ocelice9ed2812017-12-25 17:49:28 +010014363HAProxy is capable of extracting data from request or response streams, from
Willy Tarreau74ca5042013-06-11 23:12:07 +020014364client or server information, from tables, environmental information etc...
14365The action of extracting such data is called fetching a sample. Once retrieved,
14366these samples may be used for various purposes such as a key to a stick-table,
14367but most common usages consist in matching them against predefined constant
14368data called patterns.
14369
14370
143717.1. ACL basics
14372---------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014373
14374The use of Access Control Lists (ACL) provides a flexible solution to perform
14375content switching and generally to take decisions based on content extracted
14376from the request, the response or any environmental status. The principle is
14377simple :
14378
Willy Tarreau74ca5042013-06-11 23:12:07 +020014379 - extract a data sample from a stream, table or the environment
Willy Tarreaue6b11e42013-11-26 19:02:32 +010014380 - optionally apply some format conversion to the extracted sample
Willy Tarreau74ca5042013-06-11 23:12:07 +020014381 - apply one or multiple pattern matching methods on this sample
14382 - perform actions only when a pattern matches the sample
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014383
Willy Tarreau74ca5042013-06-11 23:12:07 +020014384The actions generally consist in blocking a request, selecting a backend, or
14385adding a header.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014386
14387In order to define a test, the "acl" keyword is used. The syntax is :
14388
Willy Tarreau74ca5042013-06-11 23:12:07 +020014389 acl <aclname> <criterion> [flags] [operator] [<value>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014390
14391This creates a new ACL <aclname> or completes an existing one with new tests.
14392Those tests apply to the portion of request/response specified in <criterion>
14393and may be adjusted with optional flags [flags]. Some criteria also support
Willy Tarreaue6b11e42013-11-26 19:02:32 +010014394an operator which may be specified before the set of values. Optionally some
14395conversion operators may be applied to the sample, and they will be specified
14396as a comma-delimited list of keywords just after the first keyword. The values
14397are of the type supported by the criterion, and are separated by spaces.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014398
14399ACL names must be formed from upper and lower case letters, digits, '-' (dash),
14400'_' (underscore) , '.' (dot) and ':' (colon). ACL names are case-sensitive,
14401which means that "my_acl" and "My_Acl" are two different ACLs.
14402
14403There is no enforced limit to the number of ACLs. The unused ones do not affect
14404performance, they just consume a small amount of memory.
14405
Willy Tarreau74ca5042013-06-11 23:12:07 +020014406The criterion generally is the name of a sample fetch method, or one of its ACL
14407specific declinations. The default test method is implied by the output type of
14408this sample fetch method. The ACL declinations can describe alternate matching
Willy Tarreaue6b11e42013-11-26 19:02:32 +010014409methods of a same sample fetch method. The sample fetch methods are the only
14410ones supporting a conversion.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014411
14412Sample fetch methods return data which can be of the following types :
14413 - boolean
14414 - integer (signed or unsigned)
14415 - IPv4 or IPv6 address
14416 - string
14417 - data block
14418
Willy Tarreaue6b11e42013-11-26 19:02:32 +010014419Converters transform any of these data into any of these. For example, some
14420converters might convert a string to a lower-case string while other ones
14421would turn a string to an IPv4 address, or apply a netmask to an IP address.
14422The resulting sample is of the type of the last converter applied to the list,
14423which defaults to the type of the sample fetch method.
14424
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020014425Each sample or converter returns data of a specific type, specified with its
14426keyword in this documentation. When an ACL is declared using a standard sample
14427fetch method, certain types automatically involved a default matching method
14428which are summarized in the table below :
14429
14430 +---------------------+-----------------+
14431 | Sample or converter | Default |
14432 | output type | matching method |
14433 +---------------------+-----------------+
14434 | boolean | bool |
14435 +---------------------+-----------------+
14436 | integer | int |
14437 +---------------------+-----------------+
14438 | ip | ip |
14439 +---------------------+-----------------+
14440 | string | str |
14441 +---------------------+-----------------+
14442 | binary | none, use "-m" |
14443 +---------------------+-----------------+
14444
14445Note that in order to match a binary samples, it is mandatory to specify a
14446matching method, see below.
14447
Willy Tarreau74ca5042013-06-11 23:12:07 +020014448The ACL engine can match these types against patterns of the following types :
14449 - boolean
14450 - integer or integer range
14451 - IP address / network
14452 - string (exact, substring, suffix, prefix, subdir, domain)
14453 - regular expression
14454 - hex block
14455
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014456The following ACL flags are currently supported :
14457
Willy Tarreau2b5285d2010-05-09 23:45:24 +020014458 -i : ignore case during matching of all subsequent patterns.
14459 -f : load patterns from a file.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014460 -m : use a specific pattern matching method
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010014461 -n : forbid the DNS resolutions
Thierry FOURNIER9860c412014-01-29 14:23:29 +010014462 -M : load the file pointed by -f like a map file.
Thierry FOURNIER3534d882014-01-20 17:01:44 +010014463 -u : force the unique id of the ACL
Willy Tarreau6a06a402007-07-15 20:15:28 +020014464 -- : force end of flags. Useful when a string looks like one of the flags.
14465
Willy Tarreau74ca5042013-06-11 23:12:07 +020014466The "-f" flag is followed by the name of a file from which all lines will be
14467read as individual values. It is even possible to pass multiple "-f" arguments
14468if the patterns are to be loaded from multiple files. Empty lines as well as
14469lines beginning with a sharp ('#') will be ignored. All leading spaces and tabs
14470will be stripped. If it is absolutely necessary to insert a valid pattern
14471beginning with a sharp, just prefix it with a space so that it is not taken for
14472a comment. Depending on the data type and match method, haproxy may load the
14473lines into a binary tree, allowing very fast lookups. This is true for IPv4 and
14474exact string matching. In this case, duplicates will automatically be removed.
14475
Thierry FOURNIER9860c412014-01-29 14:23:29 +010014476The "-M" flag allows an ACL to use a map file. If this flag is set, the file is
14477parsed as two column file. The first column contains the patterns used by the
14478ACL, and the second column contain the samples. The sample can be used later by
14479a map. This can be useful in some rare cases where an ACL would just be used to
14480check for the existence of a pattern in a map before a mapping is applied.
14481
Thierry FOURNIER3534d882014-01-20 17:01:44 +010014482The "-u" flag forces the unique id of the ACL. This unique id is used with the
14483socket interface to identify ACL and dynamically change its values. Note that a
14484file is always identified by its name even if an id is set.
14485
Willy Tarreau74ca5042013-06-11 23:12:07 +020014486Also, note that the "-i" flag applies to subsequent entries and not to entries
14487loaded from files preceding it. For instance :
14488
14489 acl valid-ua hdr(user-agent) -f exact-ua.lst -i -f generic-ua.lst test
14490
14491In this example, each line of "exact-ua.lst" will be exactly matched against
14492the "user-agent" header of the request. Then each line of "generic-ua" will be
14493case-insensitively matched. Then the word "test" will be insensitively matched
14494as well.
14495
14496The "-m" flag is used to select a specific pattern matching method on the input
14497sample. All ACL-specific criteria imply a pattern matching method and generally
14498do not need this flag. However, this flag is useful with generic sample fetch
14499methods to describe how they're going to be matched against the patterns. This
14500is required for sample fetches which return data type for which there is no
Davor Ocelice9ed2812017-12-25 17:49:28 +010014501obvious matching method (e.g. string or binary). When "-m" is specified and
Willy Tarreau74ca5042013-06-11 23:12:07 +020014502followed by a pattern matching method name, this method is used instead of the
14503default one for the criterion. This makes it possible to match contents in ways
14504that were not initially planned, or with sample fetch methods which return a
14505string. The matching method also affects the way the patterns are parsed.
14506
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010014507The "-n" flag forbids the dns resolutions. It is used with the load of ip files.
14508By default, if the parser cannot parse ip address it considers that the parsed
14509string is maybe a domain name and try dns resolution. The flag "-n" disable this
14510resolution. It is useful for detecting malformed ip lists. Note that if the DNS
14511server is not reachable, the haproxy configuration parsing may last many minutes
John Roeslerfb2fce12019-07-10 15:45:51 -050014512waiting for the timeout. During this time no error messages are displayed. The
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010014513flag "-n" disable this behavior. Note also that during the runtime, this
14514function is disabled for the dynamic acl modifications.
14515
Willy Tarreau74ca5042013-06-11 23:12:07 +020014516There are some restrictions however. Not all methods can be used with all
14517sample fetch methods. Also, if "-m" is used in conjunction with "-f", it must
14518be placed first. The pattern matching method must be one of the following :
Willy Tarreau5adeda12013-03-31 22:13:34 +020014519
14520 - "found" : only check if the requested sample could be found in the stream,
14521 but do not compare it against any pattern. It is recommended not
Willy Tarreau74ca5042013-06-11 23:12:07 +020014522 to pass any pattern to avoid confusion. This matching method is
14523 particularly useful to detect presence of certain contents such
14524 as headers, cookies, etc... even if they are empty and without
14525 comparing them to anything nor counting them.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014526
14527 - "bool" : check the value as a boolean. It can only be applied to fetches
14528 which return a boolean or integer value, and takes no pattern.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014529 Value zero or false does not match, all other values do match.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014530
14531 - "int" : match the value as an integer. It can be used with integer and
Willy Tarreau74ca5042013-06-11 23:12:07 +020014532 boolean samples. Boolean false is integer 0, true is integer 1.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014533
14534 - "ip" : match the value as an IPv4 or IPv6 address. It is compatible
Willy Tarreau74ca5042013-06-11 23:12:07 +020014535 with IP address samples only, so it is implied and never needed.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014536
Davor Ocelice9ed2812017-12-25 17:49:28 +010014537 - "bin" : match the contents against a hexadecimal string representing a
Willy Tarreau5adeda12013-03-31 22:13:34 +020014538 binary sequence. This may be used with binary or string samples.
14539
14540 - "len" : match the sample's length as an integer. This may be used with
14541 binary or string samples.
14542
Willy Tarreau74ca5042013-06-11 23:12:07 +020014543 - "str" : exact match : match the contents against a string. This may be
14544 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014545
Willy Tarreau74ca5042013-06-11 23:12:07 +020014546 - "sub" : substring match : check that the contents contain at least one of
14547 the provided string patterns. This may be used with binary or
14548 string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014549
Willy Tarreau74ca5042013-06-11 23:12:07 +020014550 - "reg" : regex match : match the contents against a list of regular
14551 expressions. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014552
Willy Tarreau74ca5042013-06-11 23:12:07 +020014553 - "beg" : prefix match : check that the contents begin like the provided
14554 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014555
Willy Tarreau74ca5042013-06-11 23:12:07 +020014556 - "end" : suffix match : check that the contents end like the provided
14557 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014558
Willy Tarreau74ca5042013-06-11 23:12:07 +020014559 - "dir" : subdir match : check that a slash-delimited portion of the
14560 contents exactly matches one of the provided string patterns.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014561 This may be used with binary or string samples.
14562
Willy Tarreau74ca5042013-06-11 23:12:07 +020014563 - "dom" : domain match : check that a dot-delimited portion of the contents
14564 exactly match one of the provided string patterns. This may be
14565 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014566
14567For example, to quickly detect the presence of cookie "JSESSIONID" in an HTTP
14568request, it is possible to do :
14569
14570 acl jsess_present cook(JSESSIONID) -m found
14571
14572In order to apply a regular expression on the 500 first bytes of data in the
14573buffer, one would use the following acl :
14574
14575 acl script_tag payload(0,500) -m reg -i <script>
14576
Willy Tarreaue6b11e42013-11-26 19:02:32 +010014577On systems where the regex library is much slower when using "-i", it is
14578possible to convert the sample to lowercase before matching, like this :
14579
14580 acl script_tag payload(0,500),lower -m reg <script>
14581
Willy Tarreau74ca5042013-06-11 23:12:07 +020014582All ACL-specific criteria imply a default matching method. Most often, these
14583criteria are composed by concatenating the name of the original sample fetch
14584method and the matching method. For example, "hdr_beg" applies the "beg" match
14585to samples retrieved using the "hdr" fetch method. Since all ACL-specific
14586criteria rely on a sample fetch method, it is always possible instead to use
14587the original sample fetch method and the explicit matching method using "-m".
Willy Tarreau2b5285d2010-05-09 23:45:24 +020014588
Willy Tarreau74ca5042013-06-11 23:12:07 +020014589If an alternate match is specified using "-m" on an ACL-specific criterion,
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014590the matching method is simply applied to the underlying sample fetch method.
14591For example, all ACLs below are exact equivalent :
Willy Tarreau2b5285d2010-05-09 23:45:24 +020014592
Willy Tarreau74ca5042013-06-11 23:12:07 +020014593 acl short_form hdr_beg(host) www.
14594 acl alternate1 hdr_beg(host) -m beg www.
14595 acl alternate2 hdr_dom(host) -m beg www.
14596 acl alternate3 hdr(host) -m beg www.
Willy Tarreau2b5285d2010-05-09 23:45:24 +020014597
Willy Tarreau2b5285d2010-05-09 23:45:24 +020014598
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020014599The table below summarizes the compatibility matrix between sample or converter
14600types and the pattern types to fetch against. It indicates for each compatible
14601combination the name of the matching method to be used, surrounded with angle
14602brackets ">" and "<" when the method is the default one and will work by
14603default without "-m".
Willy Tarreau0ba27502007-12-24 16:55:16 +010014604
Willy Tarreau74ca5042013-06-11 23:12:07 +020014605 +-------------------------------------------------+
14606 | Input sample type |
14607 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020014608 | pattern type | boolean | integer | ip | string | binary |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014609 +----------------------+---------+---------+---------+---------+---------+
14610 | none (presence only) | found | found | found | found | found |
14611 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020014612 | none (boolean value) |> bool <| bool | | bool | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014613 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020014614 | integer (value) | int |> int <| int | int | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014615 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010014616 | integer (length) | len | len | len | len | len |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014617 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020014618 | IP address | | |> ip <| ip | ip |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014619 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020014620 | exact string | str | str | str |> str <| str |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014621 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010014622 | prefix | beg | beg | beg | beg | beg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014623 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010014624 | suffix | end | end | end | end | end |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014625 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010014626 | substring | sub | sub | sub | sub | sub |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014627 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010014628 | subdir | dir | dir | dir | dir | dir |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014629 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010014630 | domain | dom | dom | dom | dom | dom |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014631 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010014632 | regex | reg | reg | reg | reg | reg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014633 +----------------------+---------+---------+---------+---------+---------+
14634 | hex block | | | | bin | bin |
14635 +----------------------+---------+---------+---------+---------+---------+
Willy Tarreau6a06a402007-07-15 20:15:28 +020014636
14637
Willy Tarreau74ca5042013-06-11 23:12:07 +0200146387.1.1. Matching booleans
14639------------------------
14640
14641In order to match a boolean, no value is needed and all values are ignored.
14642Boolean matching is used by default for all fetch methods of type "boolean".
14643When boolean matching is used, the fetched value is returned as-is, which means
14644that a boolean "true" will always match and a boolean "false" will never match.
14645
14646Boolean matching may also be enforced using "-m bool" on fetch methods which
14647return an integer value. Then, integer value 0 is converted to the boolean
14648"false" and all other values are converted to "true".
14649
Willy Tarreau6a06a402007-07-15 20:15:28 +020014650
Willy Tarreau74ca5042013-06-11 23:12:07 +0200146517.1.2. Matching integers
14652------------------------
14653
14654Integer matching applies by default to integer fetch methods. It can also be
14655enforced on boolean fetches using "-m int". In this case, "false" is converted
14656to the integer 0, and "true" is converted to the integer 1.
14657
14658Integer matching also supports integer ranges and operators. Note that integer
14659matching only applies to positive values. A range is a value expressed with a
14660lower and an upper bound separated with a colon, both of which may be omitted.
Willy Tarreau6a06a402007-07-15 20:15:28 +020014661
14662For instance, "1024:65535" is a valid range to represent a range of
14663unprivileged ports, and "1024:" would also work. "0:1023" is a valid
14664representation of privileged ports, and ":1023" would also work.
14665
Willy Tarreau62644772008-07-16 18:36:06 +020014666As a special case, some ACL functions support decimal numbers which are in fact
14667two integers separated by a dot. This is used with some version checks for
14668instance. All integer properties apply to those decimal numbers, including
14669ranges and operators.
14670
Willy Tarreau6a06a402007-07-15 20:15:28 +020014671For an easier usage, comparison operators are also supported. Note that using
Willy Tarreau0ba27502007-12-24 16:55:16 +010014672operators with ranges does not make much sense and is strongly discouraged.
14673Similarly, it does not make much sense to perform order comparisons with a set
14674of values.
Willy Tarreau6a06a402007-07-15 20:15:28 +020014675
Willy Tarreau0ba27502007-12-24 16:55:16 +010014676Available operators for integer matching are :
Willy Tarreau6a06a402007-07-15 20:15:28 +020014677
14678 eq : true if the tested value equals at least one value
14679 ge : true if the tested value is greater than or equal to at least one value
14680 gt : true if the tested value is greater than at least one value
14681 le : true if the tested value is less than or equal to at least one value
14682 lt : true if the tested value is less than at least one value
14683
Willy Tarreau0ba27502007-12-24 16:55:16 +010014684For instance, the following ACL matches any negative Content-Length header :
Willy Tarreau6a06a402007-07-15 20:15:28 +020014685
14686 acl negative-length hdr_val(content-length) lt 0
14687
Willy Tarreau62644772008-07-16 18:36:06 +020014688This one matches SSL versions between 3.0 and 3.1 (inclusive) :
14689
14690 acl sslv3 req_ssl_ver 3:3.1
14691
Willy Tarreau6a06a402007-07-15 20:15:28 +020014692
Willy Tarreau74ca5042013-06-11 23:12:07 +0200146937.1.3. Matching strings
14694-----------------------
14695
14696String matching applies to string or binary fetch methods, and exists in 6
14697different forms :
14698
14699 - exact match (-m str) : the extracted string must exactly match the
Davor Ocelice9ed2812017-12-25 17:49:28 +010014700 patterns;
Willy Tarreau74ca5042013-06-11 23:12:07 +020014701
14702 - substring match (-m sub) : the patterns are looked up inside the
Davor Ocelice9ed2812017-12-25 17:49:28 +010014703 extracted string, and the ACL matches if any of them is found inside;
Willy Tarreau74ca5042013-06-11 23:12:07 +020014704
14705 - prefix match (-m beg) : the patterns are compared with the beginning of
14706 the extracted string, and the ACL matches if any of them matches.
14707
14708 - suffix match (-m end) : the patterns are compared with the end of the
14709 extracted string, and the ACL matches if any of them matches.
14710
Baptiste Assmann33db6002016-03-06 23:32:10 +010014711 - subdir match (-m dir) : the patterns are looked up inside the extracted
Willy Tarreau74ca5042013-06-11 23:12:07 +020014712 string, delimited with slashes ("/"), and the ACL matches if any of them
14713 matches.
14714
14715 - domain match (-m dom) : the patterns are looked up inside the extracted
14716 string, delimited with dots ("."), and the ACL matches if any of them
14717 matches.
Willy Tarreau6a06a402007-07-15 20:15:28 +020014718
14719String matching applies to verbatim strings as they are passed, with the
14720exception of the backslash ("\") which makes it possible to escape some
14721characters such as the space. If the "-i" flag is passed before the first
14722string, then the matching will be performed ignoring the case. In order
14723to match the string "-i", either set it second, or pass the "--" flag
Willy Tarreau0ba27502007-12-24 16:55:16 +010014724before the first string. Same applies of course to match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020014725
Mathias Weiersmuellercb250fc2019-12-02 09:43:40 +010014726Do not use string matches for binary fetches which might contain null bytes
14727(0x00), as the comparison stops at the occurrence of the first null byte.
14728Instead, convert the binary fetch to a hex string with the hex converter first.
14729
14730Example:
14731 # matches if the string <tag> is present in the binary sample
14732 acl tag_found req.payload(0,0),hex -m sub 3C7461673E
14733
Willy Tarreau6a06a402007-07-15 20:15:28 +020014734
Willy Tarreau74ca5042013-06-11 23:12:07 +0200147357.1.4. Matching regular expressions (regexes)
14736---------------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020014737
14738Just like with string matching, regex matching applies to verbatim strings as
14739they are passed, with the exception of the backslash ("\") which makes it
14740possible to escape some characters such as the space. If the "-i" flag is
14741passed before the first regex, then the matching will be performed ignoring
14742the case. In order to match the string "-i", either set it second, or pass
Willy Tarreau0ba27502007-12-24 16:55:16 +010014743the "--" flag before the first string. Same principle applies of course to
14744match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020014745
14746
Willy Tarreau74ca5042013-06-11 23:12:07 +0200147477.1.5. Matching arbitrary data blocks
14748-------------------------------------
14749
14750It is possible to match some extracted samples against a binary block which may
14751not safely be represented as a string. For this, the patterns must be passed as
14752a series of hexadecimal digits in an even number, when the match method is set
14753to binary. Each sequence of two digits will represent a byte. The hexadecimal
14754digits may be used upper or lower case.
14755
14756Example :
14757 # match "Hello\n" in the input stream (\x48 \x65 \x6c \x6c \x6f \x0a)
14758 acl hello payload(0,6) -m bin 48656c6c6f0a
14759
14760
147617.1.6. Matching IPv4 and IPv6 addresses
14762---------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020014763
14764IPv4 addresses values can be specified either as plain addresses or with a
14765netmask appended, in which case the IPv4 address matches whenever it is
14766within the network. Plain addresses may also be replaced with a resolvable
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010014767host name, but this practice is generally discouraged as it makes it more
Willy Tarreau0ba27502007-12-24 16:55:16 +010014768difficult to read and debug configurations. If hostnames are used, you should
14769at least ensure that they are present in /etc/hosts so that the configuration
14770does not depend on any random DNS match at the moment the configuration is
14771parsed.
Willy Tarreau6a06a402007-07-15 20:15:28 +020014772
Daniel Schnellereba56342016-04-13 00:26:52 +020014773The dotted IPv4 address notation is supported in both regular as well as the
14774abbreviated form with all-0-octets omitted:
14775
14776 +------------------+------------------+------------------+
14777 | Example 1 | Example 2 | Example 3 |
14778 +------------------+------------------+------------------+
14779 | 192.168.0.1 | 10.0.0.12 | 127.0.0.1 |
14780 | 192.168.1 | 10.12 | 127.1 |
14781 | 192.168.0.1/22 | 10.0.0.12/8 | 127.0.0.1/8 |
14782 | 192.168.1/22 | 10.12/8 | 127.1/8 |
14783 +------------------+------------------+------------------+
14784
14785Notice that this is different from RFC 4632 CIDR address notation in which
14786192.168.42/24 would be equivalent to 192.168.42.0/24.
14787
Willy Tarreauceb4ac92012-04-28 00:41:46 +020014788IPv6 may be entered in their usual form, with or without a netmask appended.
14789Only bit counts are accepted for IPv6 netmasks. In order to avoid any risk of
14790trouble with randomly resolved IP addresses, host names are never allowed in
14791IPv6 patterns.
14792
14793HAProxy is also able to match IPv4 addresses with IPv6 addresses in the
14794following situations :
14795 - tested address is IPv4, pattern address is IPv4, the match applies
14796 in IPv4 using the supplied mask if any.
14797 - tested address is IPv6, pattern address is IPv6, the match applies
14798 in IPv6 using the supplied mask if any.
14799 - tested address is IPv6, pattern address is IPv4, the match applies in IPv4
14800 using the pattern's mask if the IPv6 address matches with 2002:IPV4::,
14801 ::IPV4 or ::ffff:IPV4, otherwise it fails.
14802 - tested address is IPv4, pattern address is IPv6, the IPv4 address is first
14803 converted to IPv6 by prefixing ::ffff: in front of it, then the match is
14804 applied in IPv6 using the supplied IPv6 mask.
14805
Willy Tarreau74ca5042013-06-11 23:12:07 +020014806
148077.2. Using ACLs to form conditions
14808----------------------------------
14809
14810Some actions are only performed upon a valid condition. A condition is a
14811combination of ACLs with operators. 3 operators are supported :
14812
14813 - AND (implicit)
14814 - OR (explicit with the "or" keyword or the "||" operator)
14815 - Negation with the exclamation mark ("!")
Willy Tarreau6a06a402007-07-15 20:15:28 +020014816
Willy Tarreau74ca5042013-06-11 23:12:07 +020014817A condition is formed as a disjunctive form:
Willy Tarreau6a06a402007-07-15 20:15:28 +020014818
Willy Tarreau74ca5042013-06-11 23:12:07 +020014819 [!]acl1 [!]acl2 ... [!]acln { or [!]acl1 [!]acl2 ... [!]acln } ...
Willy Tarreaubef91e72013-03-31 23:14:46 +020014820
Willy Tarreau74ca5042013-06-11 23:12:07 +020014821Such conditions are generally used after an "if" or "unless" statement,
14822indicating when the condition will trigger the action.
Willy Tarreaubef91e72013-03-31 23:14:46 +020014823
Willy Tarreau74ca5042013-06-11 23:12:07 +020014824For instance, to block HTTP requests to the "*" URL with methods other than
14825"OPTIONS", as well as POST requests without content-length, and GET or HEAD
14826requests with a content-length greater than 0, and finally every request which
14827is not either GET/HEAD/POST/OPTIONS !
14828
14829 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030014830 http-request deny if HTTP_URL_STAR !METH_OPTIONS || METH_POST missing_cl
14831 http-request deny if METH_GET HTTP_CONTENT
14832 http-request deny unless METH_GET or METH_POST or METH_OPTIONS
Willy Tarreau74ca5042013-06-11 23:12:07 +020014833
14834To select a different backend for requests to static contents on the "www" site
14835and to every request on the "img", "video", "download" and "ftp" hosts :
14836
14837 acl url_static path_beg /static /images /img /css
14838 acl url_static path_end .gif .png .jpg .css .js
14839 acl host_www hdr_beg(host) -i www
14840 acl host_static hdr_beg(host) -i img. video. download. ftp.
14841
Davor Ocelice9ed2812017-12-25 17:49:28 +010014842 # now use backend "static" for all static-only hosts, and for static URLs
Willy Tarreau74ca5042013-06-11 23:12:07 +020014843 # of host "www". Use backend "www" for the rest.
14844 use_backend static if host_static or host_www url_static
14845 use_backend www if host_www
14846
14847It is also possible to form rules using "anonymous ACLs". Those are unnamed ACL
14848expressions that are built on the fly without needing to be declared. They must
14849be enclosed between braces, with a space before and after each brace (because
14850the braces must be seen as independent words). Example :
14851
14852 The following rule :
14853
14854 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030014855 http-request deny if METH_POST missing_cl
Willy Tarreau74ca5042013-06-11 23:12:07 +020014856
14857 Can also be written that way :
14858
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030014859 http-request deny if METH_POST { hdr_cnt(Content-length) eq 0 }
Willy Tarreau74ca5042013-06-11 23:12:07 +020014860
14861It is generally not recommended to use this construct because it's a lot easier
14862to leave errors in the configuration when written that way. However, for very
14863simple rules matching only one source IP address for instance, it can make more
14864sense to use them than to declare ACLs with random names. Another example of
14865good use is the following :
14866
14867 With named ACLs :
14868
14869 acl site_dead nbsrv(dynamic) lt 2
14870 acl site_dead nbsrv(static) lt 2
14871 monitor fail if site_dead
14872
14873 With anonymous ACLs :
14874
14875 monitor fail if { nbsrv(dynamic) lt 2 } || { nbsrv(static) lt 2 }
14876
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030014877See section 4.2 for detailed help on the "http-request deny" and "use_backend"
14878keywords.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014879
14880
148817.3. Fetching samples
14882---------------------
14883
14884Historically, sample fetch methods were only used to retrieve data to match
14885against patterns using ACLs. With the arrival of stick-tables, a new class of
14886sample fetch methods was created, most often sharing the same syntax as their
14887ACL counterpart. These sample fetch methods are also known as "fetches". As
14888of now, ACLs and fetches have converged. All ACL fetch methods have been made
14889available as fetch methods, and ACLs may use any sample fetch method as well.
14890
14891This section details all available sample fetch methods and their output type.
14892Some sample fetch methods have deprecated aliases that are used to maintain
14893compatibility with existing configurations. They are then explicitly marked as
14894deprecated and should not be used in new setups.
14895
14896The ACL derivatives are also indicated when available, with their respective
14897matching methods. These ones all have a well defined default pattern matching
14898method, so it is never necessary (though allowed) to pass the "-m" option to
14899indicate how the sample will be matched using ACLs.
14900
14901As indicated in the sample type versus matching compatibility matrix above,
14902when using a generic sample fetch method in an ACL, the "-m" option is
14903mandatory unless the sample type is one of boolean, integer, IPv4 or IPv6. When
14904the same keyword exists as an ACL keyword and as a standard fetch method, the
14905ACL engine will automatically pick the ACL-only one by default.
14906
14907Some of these keywords support one or multiple mandatory arguments, and one or
14908multiple optional arguments. These arguments are strongly typed and are checked
14909when the configuration is parsed so that there is no risk of running with an
Davor Ocelice9ed2812017-12-25 17:49:28 +010014910incorrect argument (e.g. an unresolved backend name). Fetch function arguments
14911are passed between parenthesis and are delimited by commas. When an argument
Willy Tarreau74ca5042013-06-11 23:12:07 +020014912is optional, it will be indicated below between square brackets ('[ ]'). When
14913all arguments are optional, the parenthesis may be omitted.
14914
14915Thus, the syntax of a standard sample fetch method is one of the following :
14916 - name
14917 - name(arg1)
14918 - name(arg1,arg2)
14919
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014920
149217.3.1. Converters
14922-----------------
14923
Willy Tarreaue6b11e42013-11-26 19:02:32 +010014924Sample fetch methods may be combined with transformations to be applied on top
14925of the fetched sample (also called "converters"). These combinations form what
14926is called "sample expressions" and the result is a "sample". Initially this
14927was only supported by "stick on" and "stick store-request" directives but this
Davor Ocelice9ed2812017-12-25 17:49:28 +010014928has now be extended to all places where samples may be used (ACLs, log-format,
Willy Tarreaue6b11e42013-11-26 19:02:32 +010014929unique-id-format, add-header, ...).
14930
14931These transformations are enumerated as a series of specific keywords after the
14932sample fetch method. These keywords may equally be appended immediately after
14933the fetch keyword's argument, delimited by a comma. These keywords can also
Davor Ocelice9ed2812017-12-25 17:49:28 +010014934support some arguments (e.g. a netmask) which must be passed in parenthesis.
Willy Tarreau0ba27502007-12-24 16:55:16 +010014935
Willy Tarreau97707872015-01-27 15:12:13 +010014936A certain category of converters are bitwise and arithmetic operators which
14937support performing basic operations on integers. Some bitwise operations are
14938supported (and, or, xor, cpl) and some arithmetic operations are supported
14939(add, sub, mul, div, mod, neg). Some comparators are provided (odd, even, not,
14940bool) which make it possible to report a match without having to write an ACL.
14941
Willy Tarreau74ca5042013-06-11 23:12:07 +020014942The currently available list of transformation keywords include :
Willy Tarreau0ba27502007-12-24 16:55:16 +010014943
Ben Shillitof25e8e52016-12-02 14:25:37 +00001494451d.single(<prop>[,<prop>*])
14945 Returns values for the properties requested as a string, where values are
14946 separated by the delimiter specified with "51degrees-property-separator".
14947 The device is identified using the User-Agent header passed to the
14948 converter. The function can be passed up to five property names, and if a
14949 property name can't be found, the value "NoData" is returned.
14950
14951 Example :
Davor Ocelice9ed2812017-12-25 17:49:28 +010014952 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request,
14953 # containing values for the three properties requested by using the
Ben Shillitof25e8e52016-12-02 14:25:37 +000014954 # User-Agent passed to the converter.
14955 frontend http-in
14956 bind *:8081
14957 default_backend servers
14958 http-request set-header X-51D-DeviceTypeMobileTablet \
14959 %[req.fhdr(User-Agent),51d.single(DeviceType,IsMobile,IsTablet)]
14960
Willy Tarreau97707872015-01-27 15:12:13 +010014961add(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014962 Adds <value> to the input value of type signed integer, and returns the
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014963 result as a signed integer. <value> can be a numeric value or a variable
Daniel Schneller0b547052016-03-21 20:46:57 +010014964 name. The name of the variable starts with an indication about its scope. The
14965 scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010014966 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010014967 "sess" : the variable is shared with the whole session
14968 "txn" : the variable is shared with the transaction (request and response)
14969 "req" : the variable is shared only during request processing
14970 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010014971 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014972 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010014973
Nenad Merdanovicc31499d2019-03-23 11:00:32 +010014974aes_gcm_dec(<bits>,<nonce>,<key>,<aead_tag>)
14975 Decrypts the raw byte input using the AES128-GCM, AES192-GCM or
14976 AES256-GCM algorithm, depending on the <bits> parameter. All other parameters
14977 need to be base64 encoded and the returned result is in raw byte format.
14978 If the <aead_tag> validation fails, the converter doesn't return any data.
14979 The <nonce>, <key> and <aead_tag> can either be strings or variables. This
14980 converter requires at least OpenSSL 1.0.1.
14981
14982 Example:
14983 http-response set-header X-Decrypted-Text %[var(txn.enc),\
14984 aes_gcm_dec(128,txn.nonce,Zm9vb2Zvb29mb29wZm9vbw==,txn.aead_tag)]
14985
Willy Tarreau97707872015-01-27 15:12:13 +010014986and(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014987 Performs a bitwise "AND" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014988 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010014989 numeric value or a variable name. The name of the variable starts with an
14990 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010014991 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010014992 "sess" : the variable is shared with the whole session
14993 "txn" : the variable is shared with the transaction (request and response)
14994 "req" : the variable is shared only during request processing
14995 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010014996 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014997 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010014998
Holger Just1bfc24b2017-05-06 00:56:53 +020014999b64dec
15000 Converts (decodes) a base64 encoded input string to its binary
15001 representation. It performs the inverse operation of base64().
15002
Emeric Brun53d1a982014-04-30 18:21:37 +020015003base64
15004 Converts a binary input sample to a base64 string. It is used to log or
Davor Ocelice9ed2812017-12-25 17:49:28 +010015005 transfer binary content in a way that can be reliably transferred (e.g.
Emeric Brun53d1a982014-04-30 18:21:37 +020015006 an SSL ID can be copied in a header).
15007
Willy Tarreau97707872015-01-27 15:12:13 +010015008bool
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015009 Returns a boolean TRUE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010015010 non-null, otherwise returns FALSE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010015011 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010015012 presence of a flag).
15013
Emeric Brun54c4ac82014-11-03 15:32:43 +010015014bytes(<offset>[,<length>])
15015 Extracts some bytes from an input binary sample. The result is a binary
15016 sample starting at an offset (in bytes) of the original sample and
Tim Düsterhus4896c442016-11-29 02:15:19 +010015017 optionally truncated at the given length.
Emeric Brun54c4ac82014-11-03 15:32:43 +010015018
Willy Tarreau280f42b2018-02-19 15:34:12 +010015019concat([<start>],[<var>],[<end>])
15020 Concatenates up to 3 fields after the current sample which is then turned to
15021 a string. The first one, <start>, is a constant string, that will be appended
15022 immediately after the existing sample. It may be omitted if not used. The
15023 second one, <var>, is a variable name. The variable will be looked up, its
15024 contents converted to a string, and it will be appended immediately after the
15025 <first> part. If the variable is not found, nothing is appended. It may be
15026 omitted as well. The third field, <end> is a constant string that will be
15027 appended after the variable. It may also be omitted. Together, these elements
15028 allow to concatenate variables with delimiters to an existing set of
15029 variables. This can be used to build new variables made of a succession of
Willy Tarreauef21fac2020-02-14 13:37:20 +010015030 other variables, such as colon-delimited values. If commas or closing
Daniel Corbett67a82712020-07-06 23:01:19 -040015031 parenthesis are needed as delimiters, they must be protected by quotes or
Willy Tarreauef21fac2020-02-14 13:37:20 +010015032 backslashes, themselves protected so that they are not stripped by the first
15033 level parser. See examples below.
Willy Tarreau280f42b2018-02-19 15:34:12 +010015034
15035 Example:
15036 tcp-request session set-var(sess.src) src
15037 tcp-request session set-var(sess.dn) ssl_c_s_dn
15038 tcp-request session set-var(txn.sig) str(),concat(<ip=,sess.ip,>),concat(<dn=,sess.dn,>)
Willy Tarreauef21fac2020-02-14 13:37:20 +010015039 tcp-request session set-var(txn.ipport) "str(),concat('addr=(',sess.ip),concat(',',sess.port,')')"
Willy Tarreau280f42b2018-02-19 15:34:12 +010015040 http-request set-header x-hap-sig %[var(txn.sig)]
15041
Willy Tarreau97707872015-01-27 15:12:13 +010015042cpl
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015043 Takes the input value of type signed integer, applies a ones-complement
15044 (flips all bits) and returns the result as an signed integer.
Willy Tarreau97707872015-01-27 15:12:13 +010015045
Willy Tarreau80599772015-01-20 19:35:24 +010015046crc32([<avalanche>])
15047 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32
15048 hash function. Optionally, it is possible to apply a full avalanche hash
15049 function to the output if the optional <avalanche> argument equals 1. This
15050 converter uses the same functions as used by the various hash-based load
15051 balancing algorithms, so it will provide exactly the same results. It is
15052 provided for compatibility with other software which want a CRC32 to be
15053 computed on some input keys, so it follows the most common implementation as
15054 found in Ethernet, Gzip, PNG, etc... It is slower than the other algorithms
15055 but may provide a better or at least less predictable distribution. It must
15056 not be used for security purposes as a 32-bit hash is trivial to break. See
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010015057 also "djb2", "sdbm", "wt6", "crc32c" and the "hash-type" directive.
15058
15059crc32c([<avalanche>])
15060 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32C
15061 hash function. Optionally, it is possible to apply a full avalanche hash
15062 function to the output if the optional <avalanche> argument equals 1. This
15063 converter uses the same functions as described in RFC4960, Appendix B [8].
15064 It is provided for compatibility with other software which want a CRC32C to be
15065 computed on some input keys. It is slower than the other algorithms and it must
15066 not be used for security purposes as a 32-bit hash is trivial to break. See
15067 also "djb2", "sdbm", "wt6", "crc32" and the "hash-type" directive.
Willy Tarreau80599772015-01-20 19:35:24 +010015068
Christopher Fauletea159d62020-04-01 16:21:44 +020015069cut_crlf
15070 Cuts the string representation of the input sample on the first carriage
15071 return ('\r') or newline ('\n') character found. Only the string length is
15072 updated.
15073
David Carlier29b3ca32015-09-25 14:09:21 +010015074da-csv-conv(<prop>[,<prop>*])
David Carlier4542b102015-06-01 13:54:29 +020015075 Asks the DeviceAtlas converter to identify the User Agent string passed on
15076 input, and to emit a string made of the concatenation of the properties
15077 enumerated in argument, delimited by the separator defined by the global
15078 keyword "deviceatlas-property-separator", or by default the pipe character
David Carlier840b0242016-03-16 10:09:55 +000015079 ('|'). There's a limit of 12 different properties imposed by the haproxy
David Carlier4542b102015-06-01 13:54:29 +020015080 configuration language.
15081
15082 Example:
15083 frontend www
Cyril Bonté307ee1e2015-09-28 23:16:06 +020015084 bind *:8881
15085 default_backend servers
David Carlier840b0242016-03-16 10:09:55 +000015086 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 +020015087
Willy Tarreau0851fd52019-12-17 10:07:25 +010015088debug([<prefix][,<destination>])
15089 This converter is used as debug tool. It takes a capture of the input sample
15090 and sends it to event sink <destination>, which may designate a ring buffer
15091 such as "buf0", as well as "stdout", or "stderr". Available sinks may be
15092 checked at run time by issuing "show events" on the CLI. When not specified,
15093 the output will be "buf0", which may be consulted via the CLI's "show events"
15094 command. An optional prefix <prefix> may be passed to help distinguish
15095 outputs from multiple expressions. It will then appear before the colon in
15096 the output message. The input sample is passed as-is on the output, so that
15097 it is safe to insert the debug converter anywhere in a chain, even with non-
15098 printable sample types.
15099
15100 Example:
15101 tcp-request connection track-sc0 src,debug(track-sc)
Thierry FOURNIER9687c772015-05-07 15:46:29 +020015102
Patrick Gansterer8e366512020-04-22 16:47:57 +020015103digest(<algorithm>)
15104 Converts a binary input sample to a message digest. The result is a binary
15105 sample. The <algorithm> must be an OpenSSL message digest name (e.g. sha256).
15106
15107 Please note that this converter is only available when haproxy has been
15108 compiled with USE_OPENSSL.
15109
Willy Tarreau97707872015-01-27 15:12:13 +010015110div(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015111 Divides the input value of type signed integer by <value>, and returns the
15112 result as an signed integer. If <value> is null, the largest unsigned
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015113 integer is returned (typically 2^63-1). <value> can be a numeric value or a
Daniel Schneller0b547052016-03-21 20:46:57 +010015114 variable name. The name of the variable starts with an indication about its
15115 scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015116 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015117 "sess" : the variable is shared with the whole session
15118 "txn" : the variable is shared with the transaction (request and response)
15119 "req" : the variable is shared only during request processing
15120 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010015121 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015122 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010015123
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015124djb2([<avalanche>])
15125 Hashes a binary input sample into an unsigned 32-bit quantity using the DJB2
15126 hash function. Optionally, it is possible to apply a full avalanche hash
15127 function to the output if the optional <avalanche> argument equals 1. This
15128 converter uses the same functions as used by the various hash-based load
15129 balancing algorithms, so it will provide exactly the same results. It is
15130 mostly intended for debugging, but can be used as a stick-table entry to
15131 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010015132 32-bit hash is trivial to break. See also "crc32", "sdbm", "wt6", "crc32c",
15133 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015134
Willy Tarreau97707872015-01-27 15:12:13 +010015135even
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015136 Returns a boolean TRUE if the input value of type signed integer is even
Willy Tarreau97707872015-01-27 15:12:13 +010015137 otherwise returns FALSE. It is functionally equivalent to "not,and(1),bool".
15138
Marcin Deranek9631a282018-04-16 14:30:46 +020015139field(<index>,<delimiters>[,<count>])
15140 Extracts the substring at the given index counting from the beginning
15141 (positive index) or from the end (negative index) considering given delimiters
15142 from an input string. Indexes start at 1 or -1 and delimiters are a string
15143 formatted list of chars. Optionally you can specify <count> of fields to
15144 extract (default: 1). Value of 0 indicates extraction of all remaining
15145 fields.
15146
15147 Example :
15148 str(f1_f2_f3__f5),field(5,_) # f5
15149 str(f1_f2_f3__f5),field(2,_,0) # f2_f3__f5
15150 str(f1_f2_f3__f5),field(2,_,2) # f2_f3
15151 str(f1_f2_f3__f5),field(-2,_,3) # f2_f3_
15152 str(f1_f2_f3__f5),field(-3,_,0) # f1_f2_f3
Emeric Brunf399b0d2014-11-03 17:07:03 +010015153
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015154hex
Davor Ocelice9ed2812017-12-25 17:49:28 +010015155 Converts a binary input sample to a hex string containing two hex digits per
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015156 input byte. It is used to log or transfer hex dumps of some binary input data
Davor Ocelice9ed2812017-12-25 17:49:28 +010015157 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 +020015158 header).
Thierry FOURNIER2f49d6d2014-03-12 15:01:52 +010015159
Dragan Dosen3f957b22017-10-24 09:27:34 +020015160hex2i
15161 Converts a hex string containing two hex digits per input byte to an
John Roeslerfb2fce12019-07-10 15:45:51 -050015162 integer. If the input value cannot be converted, then zero is returned.
Dragan Dosen3f957b22017-10-24 09:27:34 +020015163
Christopher Faulet4ccc12f2020-04-01 09:08:32 +020015164htonl
15165 Converts the input integer value to its 32-bit binary representation in the
15166 network byte order. Because sample fetches own signed 64-bit integer, when
15167 this converter is used, the input integer value is first casted to an
15168 unsigned 32-bit integer.
15169
Patrick Gansterer8e366512020-04-22 16:47:57 +020015170hmac(<algorithm>, <key>)
15171 Converts a binary input sample to a message authentication code with the given
15172 key. The result is a binary sample. The <algorithm> must be one of the
15173 registered OpenSSL message digest names (e.g. sha256). The <key> parameter must
15174 be base64 encoded and can either be a string or a variable.
15175
15176 Please note that this converter is only available when haproxy has been
15177 compiled with USE_OPENSSL.
15178
Cyril Bonté6bcd1822019-11-05 23:13:59 +010015179http_date([<offset],[<unit>])
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015180 Converts an integer supposed to contain a date since epoch to a string
15181 representing this date in a format suitable for use in HTTP header fields. If
Damien Claisseae6f1252019-10-30 15:57:28 +000015182 an offset value is specified, then it is added to the date before the
15183 conversion is operated. This is particularly useful to emit Date header fields,
15184 Expires values in responses when combined with a positive offset, or
15185 Last-Modified values when the offset is negative.
15186 If a unit value is specified, then consider the timestamp as either
15187 "s" for seconds (default behavior), "ms" for milliseconds, or "us" for
15188 microseconds since epoch. Offset is assumed to have the same unit as
15189 input timestamp.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015190
Tim Duesterhus3943e4f2020-09-11 14:25:23 +020015191iif(<true>,<false>)
15192 Returns the <true> string if the input value is true. Returns the <false>
15193 string otherwise.
15194
15195 Example:
Tim Duesterhus870713b2020-09-11 17:13:12 +020015196 http-request set-header x-forwarded-proto %[ssl_fc,iif(https,http)]
Tim Duesterhus3943e4f2020-09-11 14:25:23 +020015197
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015198in_table(<table>)
15199 Uses the string representation of the input sample to perform a look up in
15200 the specified table. If the key is not found in the table, a boolean false
15201 is returned. Otherwise a boolean true is returned. This can be used to verify
Davor Ocelice9ed2812017-12-25 17:49:28 +010015202 the presence of a certain key in a table tracking some elements (e.g. whether
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015203 or not a source IP address or an Authorization header was already seen).
15204
Tim Duesterhus1478aa72018-01-25 16:24:51 +010015205ipmask(<mask4>, [<mask6>])
15206 Apply a mask to an IP address, and use the result for lookups and storage.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020015207 This can be used to make all hosts within a certain mask to share the same
Tim Duesterhus1478aa72018-01-25 16:24:51 +010015208 table entries and as such use the same server. The mask4 can be passed in
15209 dotted form (e.g. 255.255.255.0) or in CIDR form (e.g. 24). The mask6 can
15210 be passed in quadruplet form (e.g. ffff:ffff::) or in CIDR form (e.g. 64).
15211 If no mask6 is given IPv6 addresses will fail to convert for backwards
15212 compatibility reasons.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020015213
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015214json([<input-code>])
Davor Ocelice9ed2812017-12-25 17:49:28 +010015215 Escapes the input string and produces an ASCII output string ready to use as a
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015216 JSON string. The converter tries to decode the input string according to the
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020015217 <input-code> parameter. It can be "ascii", "utf8", "utf8s", "utf8p" or
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015218 "utf8ps". The "ascii" decoder never fails. The "utf8" decoder detects 3 types
15219 of errors:
15220 - bad UTF-8 sequence (lone continuation byte, bad number of continuation
15221 bytes, ...)
15222 - invalid range (the decoded value is within a UTF-8 prohibited range),
15223 - code overlong (the value is encoded with more bytes than necessary).
15224
15225 The UTF-8 JSON encoding can produce a "too long value" error when the UTF-8
15226 character is greater than 0xffff because the JSON string escape specification
15227 only authorizes 4 hex digits for the value encoding. The UTF-8 decoder exists
15228 in 4 variants designated by a combination of two suffix letters : "p" for
15229 "permissive" and "s" for "silently ignore". The behaviors of the decoders
15230 are :
Davor Ocelice9ed2812017-12-25 17:49:28 +010015231 - "ascii" : never fails;
15232 - "utf8" : fails on any detected errors;
15233 - "utf8s" : never fails, but removes characters corresponding to errors;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015234 - "utf8p" : accepts and fixes the overlong errors, but fails on any other
Davor Ocelice9ed2812017-12-25 17:49:28 +010015235 error;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015236 - "utf8ps" : never fails, accepts and fixes the overlong errors, but removes
15237 characters corresponding to the other errors.
15238
15239 This converter is particularly useful for building properly escaped JSON for
Davor Ocelice9ed2812017-12-25 17:49:28 +010015240 logging to servers which consume JSON-formatted traffic logs.
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015241
15242 Example:
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015243 capture request header Host len 15
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020015244 capture request header user-agent len 150
15245 log-format '{"ip":"%[src]","user-agent":"%[capture.req.hdr(1),json(utf8s)]"}'
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015246
15247 Input request from client 127.0.0.1:
15248 GET / HTTP/1.0
15249 User-Agent: Very "Ugly" UA 1/2
15250
15251 Output log:
15252 {"ip":"127.0.0.1","user-agent":"Very \"Ugly\" UA 1\/2"}
15253
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015254language(<value>[,<default>])
15255 Returns the value with the highest q-factor from a list as extracted from the
15256 "accept-language" header using "req.fhdr". Values with no q-factor have a
15257 q-factor of 1. Values with a q-factor of 0 are dropped. Only values which
15258 belong to the list of semi-colon delimited <values> will be considered. The
15259 argument <value> syntax is "lang[;lang[;lang[;...]]]". If no value matches the
15260 given list and a default value is provided, it is returned. Note that language
15261 names may have a variant after a dash ('-'). If this variant is present in the
15262 list, it will be matched, but if it is not, only the base language is checked.
15263 The match is case-sensitive, and the output string is always one of those
Davor Ocelice9ed2812017-12-25 17:49:28 +010015264 provided in arguments. The ordering of arguments is meaningless, only the
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015265 ordering of the values in the request counts, as the first value among
15266 multiple sharing the same q-factor is used.
Thierry FOURNIERad903512014-04-11 17:51:01 +020015267
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015268 Example :
Thierry FOURNIERad903512014-04-11 17:51:01 +020015269
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015270 # this configuration switches to the backend matching a
15271 # given language based on the request :
Thierry FOURNIERad903512014-04-11 17:51:01 +020015272
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015273 acl es req.fhdr(accept-language),language(es;fr;en) -m str es
15274 acl fr req.fhdr(accept-language),language(es;fr;en) -m str fr
15275 acl en req.fhdr(accept-language),language(es;fr;en) -m str en
15276 use_backend spanish if es
15277 use_backend french if fr
15278 use_backend english if en
15279 default_backend choose_your_language
Thierry FOURNIERad903512014-04-11 17:51:01 +020015280
Willy Tarreau60a2ee72017-12-15 07:13:48 +010015281length
Etienne Carriereed0d24e2017-12-13 13:41:34 +010015282 Get the length of the string. This can only be placed after a string
15283 sample fetch function or after a transformation keyword returning a string
15284 type. The result is of type integer.
15285
Willy Tarreauffcb2e42014-07-10 16:29:08 +020015286lower
15287 Convert a string sample to lower case. This can only be placed after a string
15288 sample fetch function or after a transformation keyword returning a string
15289 type. The result is of type string.
15290
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020015291ltime(<format>[,<offset>])
15292 Converts an integer supposed to contain a date since epoch to a string
15293 representing this date in local time using a format defined by the <format>
15294 string using strftime(3). The purpose is to allow any date format to be used
15295 in logs. An optional <offset> in seconds may be applied to the input date
15296 (positive or negative). See the strftime() man page for the format supported
15297 by your operating system. See also the utime converter.
15298
15299 Example :
15300
15301 # Emit two colons, one with the local time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010015302 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020015303 log-format %[date,ltime(%Y%m%d%H%M%S)]\ %ci:%cp
15304
Christopher Faulet51fc9d12020-04-01 17:24:41 +020015305ltrim(<chars>)
15306 Skips any characters from <chars> from the beginning of the string
15307 representation of the input sample.
15308
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015309map(<map_file>[,<default_value>])
15310map_<match_type>(<map_file>[,<default_value>])
15311map_<match_type>_<output_type>(<map_file>[,<default_value>])
15312 Search the input value from <map_file> using the <match_type> matching method,
15313 and return the associated value converted to the type <output_type>. If the
15314 input value cannot be found in the <map_file>, the converter returns the
15315 <default_value>. If the <default_value> is not set, the converter fails and
15316 acts as if no input value could be fetched. If the <match_type> is not set, it
15317 defaults to "str". Likewise, if the <output_type> is not set, it defaults to
15318 "str". For convenience, the "map" keyword is an alias for "map_str" and maps a
15319 string to another string.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010015320
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015321 It is important to avoid overlapping between the keys : IP addresses and
15322 strings are stored in trees, so the first of the finest match will be used.
15323 Other keys are stored in lists, so the first matching occurrence will be used.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010015324
Tim Düsterhus4896c442016-11-29 02:15:19 +010015325 The following array contains the list of all map functions available sorted by
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015326 input type, match type and output type.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010015327
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015328 input type | match method | output type str | output type int | output type ip
15329 -----------+--------------+-----------------+-----------------+---------------
15330 str | str | map_str | map_str_int | map_str_ip
15331 -----------+--------------+-----------------+-----------------+---------------
Willy Tarreau787a4c02014-05-10 07:55:30 +020015332 str | beg | map_beg | map_beg_int | map_end_ip
15333 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015334 str | sub | map_sub | map_sub_int | map_sub_ip
15335 -----------+--------------+-----------------+-----------------+---------------
15336 str | dir | map_dir | map_dir_int | map_dir_ip
15337 -----------+--------------+-----------------+-----------------+---------------
15338 str | dom | map_dom | map_dom_int | map_dom_ip
15339 -----------+--------------+-----------------+-----------------+---------------
15340 str | end | map_end | map_end_int | map_end_ip
15341 -----------+--------------+-----------------+-----------------+---------------
Ruoshan Huang3c5e3742016-12-02 16:25:31 +080015342 str | reg | map_reg | map_reg_int | map_reg_ip
15343 -----------+--------------+-----------------+-----------------+---------------
15344 str | reg | map_regm | map_reg_int | map_reg_ip
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015345 -----------+--------------+-----------------+-----------------+---------------
15346 int | int | map_int | map_int_int | map_int_ip
15347 -----------+--------------+-----------------+-----------------+---------------
15348 ip | ip | map_ip | map_ip_int | map_ip_ip
15349 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010015350
Thierry Fournier8feaa662016-02-10 22:55:20 +010015351 The special map called "map_regm" expect matching zone in the regular
15352 expression and modify the output replacing back reference (like "\1") by
15353 the corresponding match text.
15354
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015355 The file contains one key + value per line. Lines which start with '#' are
15356 ignored, just like empty lines. Leading tabs and spaces are stripped. The key
15357 is then the first "word" (series of non-space/tabs characters), and the value
15358 is what follows this series of space/tab till the end of the line excluding
15359 trailing spaces/tabs.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010015360
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015361 Example :
15362
15363 # this is a comment and is ignored
15364 2.22.246.0/23 United Kingdom \n
15365 <-><-----------><--><------------><---->
15366 | | | | `- trailing spaces ignored
15367 | | | `---------- value
15368 | | `-------------------- middle spaces ignored
15369 | `---------------------------- key
15370 `------------------------------------ leading spaces ignored
15371
Willy Tarreau97707872015-01-27 15:12:13 +010015372mod(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015373 Divides the input value of type signed integer by <value>, and returns the
15374 remainder as an signed integer. If <value> is null, then zero is returned.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015375 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010015376 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015377 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015378 "sess" : the variable is shared with the whole session
15379 "txn" : the variable is shared with the transaction (request and response)
15380 "req" : the variable is shared only during request processing
15381 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010015382 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015383 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010015384
15385mul(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015386 Multiplies the input value of type signed integer by <value>, and returns
Thierry FOURNIER00c005c2015-07-08 01:10:21 +020015387 the product as an signed integer. In case of overflow, the largest possible
15388 value for the sign is returned so that the operation doesn't wrap around.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015389 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010015390 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015391 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015392 "sess" : the variable is shared with the whole session
15393 "txn" : the variable is shared with the transaction (request and response)
15394 "req" : the variable is shared only during request processing
15395 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010015396 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015397 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010015398
Nenad Merdanovicb7e7c472017-03-12 21:56:55 +010015399nbsrv
15400 Takes an input value of type string, interprets it as a backend name and
15401 returns the number of usable servers in that backend. Can be used in places
15402 where we want to look up a backend from a dynamic name, like a result of a
15403 map lookup.
15404
Willy Tarreau97707872015-01-27 15:12:13 +010015405neg
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015406 Takes the input value of type signed integer, computes the opposite value,
15407 and returns the remainder as an signed integer. 0 is identity. This operator
15408 is provided for reversed subtracts : in order to subtract the input from a
15409 constant, simply perform a "neg,add(value)".
Willy Tarreau97707872015-01-27 15:12:13 +010015410
15411not
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015412 Returns a boolean FALSE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010015413 non-null, otherwise returns TRUE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010015414 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010015415 absence of a flag).
15416
15417odd
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015418 Returns a boolean TRUE if the input value of type signed integer is odd
Willy Tarreau97707872015-01-27 15:12:13 +010015419 otherwise returns FALSE. It is functionally equivalent to "and(1),bool".
15420
15421or(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015422 Performs a bitwise "OR" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015423 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010015424 numeric value or a variable name. The name of the variable starts with an
15425 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015426 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015427 "sess" : the variable is shared with the whole session
15428 "txn" : the variable is shared with the transaction (request and response)
15429 "req" : the variable is shared only during request processing
15430 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010015431 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015432 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010015433
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010015434protobuf(<field_number>,[<field_type>])
15435 This extracts the protocol buffers message field in raw mode of an input binary
15436 sample representation of a protocol buffer message with <field_number> as field
15437 number (dotted notation) if <field_type> is not present, or as an integer sample
15438 if this field is present (see also "ungrpc" below).
15439 The list of the authorized types is the following one: "int32", "int64", "uint32",
15440 "uint64", "sint32", "sint64", "bool", "enum" for the "varint" wire type 0
15441 "fixed64", "sfixed64", "double" for the 64bit wire type 1, "fixed32", "sfixed32",
15442 "float" for the wire type 5. Note that "string" is considered as a length-delimited
15443 type, so it does not require any <field_type> argument to be extracted.
15444 More information may be found here about the protocol buffers message field types:
15445 https://developers.google.com/protocol-buffers/docs/encoding
15446
Willy Tarreauc4dc3502015-01-23 20:39:28 +010015447regsub(<regex>,<subst>[,<flags>])
Willy Tarreau7eda8492015-01-20 19:47:06 +010015448 Applies a regex-based substitution to the input string. It does the same
15449 operation as the well-known "sed" utility with "s/<regex>/<subst>/". By
15450 default it will replace in the input string the first occurrence of the
15451 largest part matching the regular expression <regex> with the substitution
15452 string <subst>. It is possible to replace all occurrences instead by adding
15453 the flag "g" in the third argument <flags>. It is also possible to make the
15454 regex case insensitive by adding the flag "i" in <flags>. Since <flags> is a
15455 string, it is made up from the concatenation of all desired flags. Thus if
15456 both "i" and "g" are desired, using "gi" or "ig" will have the same effect.
Willy Tarreauef21fac2020-02-14 13:37:20 +010015457 The first use of this converter is to replace certain characters or sequence
15458 of characters with other ones.
15459
15460 It is highly recommended to enclose the regex part using protected quotes to
15461 improve clarity and never have a closing parenthesis from the regex mixed up
15462 with the parenthesis from the function. Just like in Bourne shell, the first
15463 level of quotes is processed when delimiting word groups on the line, a
15464 second level is usable for argument. It is recommended to use single quotes
15465 outside since these ones do not try to resolve backslashes nor dollar signs.
Willy Tarreau7eda8492015-01-20 19:47:06 +010015466
Willy Tarreaucd0d2ed2020-02-14 17:33:06 +010015467 Examples:
Willy Tarreau7eda8492015-01-20 19:47:06 +010015468
15469 # de-duplicate "/" in header "x-path".
15470 # input: x-path: /////a///b/c/xzxyz/
15471 # output: x-path: /a/b/c/xzxyz/
Willy Tarreauef21fac2020-02-14 13:37:20 +010015472 http-request set-header x-path "%[hdr(x-path),regsub('/+','/','g')]"
Willy Tarreau7eda8492015-01-20 19:47:06 +010015473
Willy Tarreaucd0d2ed2020-02-14 17:33:06 +010015474 # copy query string to x-query and drop all leading '?', ';' and '&'
15475 http-request set-header x-query "%[query,regsub([?;&]*,'')]"
15476
Jerome Magnin07e1e3c2020-02-16 19:20:19 +010015477 # capture groups and backreferences
15478 # both lines do the same.
Willy Tarreau465dc7d2020-10-08 18:05:56 +020015479 http-request redirect location %[url,'regsub("(foo|bar)([0-9]+)?","\2\1",i)']
Jerome Magnin07e1e3c2020-02-16 19:20:19 +010015480 http-request redirect location %[url,regsub(\"(foo|bar)([0-9]+)?\",\"\2\1\",i)]
15481
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020015482capture-req(<id>)
15483 Capture the string entry in the request slot <id> and returns the entry as
15484 is. If the slot doesn't exist, the capture fails silently.
15485
15486 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020015487 "http-response capture", "capture.req.hdr" and
15488 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020015489
15490capture-res(<id>)
15491 Capture the string entry in the response slot <id> and returns the entry as
15492 is. If the slot doesn't exist, the capture fails silently.
15493
15494 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020015495 "http-response capture", "capture.req.hdr" and
15496 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020015497
Christopher Faulet568415a2020-04-01 17:24:47 +020015498rtrim(<chars>)
15499 Skips any characters from <chars> from the end of the string representation
15500 of the input sample.
15501
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015502sdbm([<avalanche>])
15503 Hashes a binary input sample into an unsigned 32-bit quantity using the SDBM
15504 hash function. Optionally, it is possible to apply a full avalanche hash
15505 function to the output if the optional <avalanche> argument equals 1. This
15506 converter uses the same functions as used by the various hash-based load
15507 balancing algorithms, so it will provide exactly the same results. It is
15508 mostly intended for debugging, but can be used as a stick-table entry to
15509 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010015510 32-bit hash is trivial to break. See also "crc32", "djb2", "wt6", "crc32c",
15511 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015512
Tim Duesterhusf38175c2020-06-09 11:48:42 +020015513secure_memcmp(<var>)
15514 Compares the contents of <var> with the input value. Both values are treated
15515 as a binary string. Returns a boolean indicating whether both binary strings
15516 match.
15517
15518 If both binary strings have the same length then the comparison will be
15519 performed in constant time.
15520
15521 Please note that this converter is only available when haproxy has been
15522 compiled with USE_OPENSSL.
15523
15524 Example :
15525
15526 http-request set-var(txn.token) hdr(token)
15527 # Check whether the token sent by the client matches the secret token
15528 # value, without leaking the contents using a timing attack.
15529 acl token_given str(my_secret_token),secure_memcmp(txn.token)
15530
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020015531set-var(<var name>)
Davor Ocelice9ed2812017-12-25 17:49:28 +010015532 Sets a variable with the input content and returns the content on the output
15533 as-is. The variable keeps the value and the associated input type. The name of
15534 the variable starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015535 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015536 "sess" : the variable is shared with the whole session
15537 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020015538 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010015539 "req" : the variable is shared only during request processing,
15540 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020015541 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015542 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020015543
Dragan Dosen6e5a9ca2017-10-24 09:18:23 +020015544sha1
Tim Duesterhusd4376302019-06-17 12:41:44 +020015545 Converts a binary input sample to a SHA-1 digest. The result is a binary
Dragan Dosen6e5a9ca2017-10-24 09:18:23 +020015546 sample with length of 20 bytes.
15547
Tim Duesterhusd4376302019-06-17 12:41:44 +020015548sha2([<bits>])
15549 Converts a binary input sample to a digest in the SHA-2 family. The result
15550 is a binary sample with length of <bits>/8 bytes.
15551
15552 Valid values for <bits> are 224, 256, 384, 512, each corresponding to
15553 SHA-<bits>. The default value is 256.
15554
15555 Please note that this converter is only available when haproxy has been
15556 compiled with USE_OPENSSL.
15557
Nenad Merdanovic177adc92019-08-27 01:58:13 +020015558srv_queue
15559 Takes an input value of type string, either a server name or <backend>/<server>
15560 format and returns the number of queued sessions on that server. Can be used
15561 in places where we want to look up queued sessions from a dynamic name, like a
15562 cookie value (e.g. req.cook(SRVID),srv_queue) and then make a decision to break
15563 persistence or direct a request elsewhere.
15564
Tim Duesterhusca097c12018-04-27 21:18:45 +020015565strcmp(<var>)
15566 Compares the contents of <var> with the input value of type string. Returns
15567 the result as a signed integer compatible with strcmp(3): 0 if both strings
15568 are identical. A value less than 0 if the left string is lexicographically
15569 smaller than the right string or if the left string is shorter. A value greater
15570 than 0 otherwise (right string greater than left string or the right string is
15571 shorter).
15572
Tim Duesterhusf38175c2020-06-09 11:48:42 +020015573 See also the secure_memcmp converter if you need to compare two binary
15574 strings in constant time.
15575
Tim Duesterhusca097c12018-04-27 21:18:45 +020015576 Example :
15577
15578 http-request set-var(txn.host) hdr(host)
15579 # Check whether the client is attempting domain fronting.
15580 acl ssl_sni_http_host_match ssl_fc_sni,strcmp(txn.host) eq 0
15581
15582
Willy Tarreau97707872015-01-27 15:12:13 +010015583sub(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015584 Subtracts <value> from the input value of type signed integer, and returns
15585 the result as an signed integer. Note: in order to subtract the input from
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015586 a constant, simply perform a "neg,add(value)". <value> can be a numeric value
Daniel Schneller0b547052016-03-21 20:46:57 +010015587 or a variable name. The name of the variable starts with an indication about
15588 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015589 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015590 "sess" : the variable is shared with the whole session
15591 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015592 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010015593 "req" : the variable is shared only during request processing,
15594 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015595 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015596 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010015597
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015598table_bytes_in_rate(<table>)
15599 Uses the string representation of the input sample to perform a look up in
15600 the specified table. If the key is not found in the table, integer value zero
15601 is returned. Otherwise the converter returns the average client-to-server
15602 bytes rate associated with the input sample in the designated table, measured
15603 in amount of bytes over the period configured in the table. See also the
15604 sc_bytes_in_rate sample fetch keyword.
15605
15606
15607table_bytes_out_rate(<table>)
15608 Uses the string representation of the input sample to perform a look up in
15609 the specified table. If the key is not found in the table, integer value zero
15610 is returned. Otherwise the converter returns the average server-to-client
15611 bytes rate associated with the input sample in the designated table, measured
15612 in amount of bytes over the period configured in the table. See also the
15613 sc_bytes_out_rate sample fetch keyword.
15614
15615table_conn_cnt(<table>)
15616 Uses the string representation of the input sample to perform a look up in
15617 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010015618 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015619 connections associated with the input sample in the designated table. See
15620 also the sc_conn_cnt sample fetch keyword.
15621
15622table_conn_cur(<table>)
15623 Uses the string representation of the input sample to perform a look up in
15624 the specified table. If the key is not found in the table, integer value zero
15625 is returned. Otherwise the converter returns the current amount of concurrent
15626 tracked connections associated with the input sample in the designated table.
15627 See also the sc_conn_cur sample fetch keyword.
15628
15629table_conn_rate(<table>)
15630 Uses the string representation of the input sample to perform a look up in
15631 the specified table. If the key is not found in the table, integer value zero
15632 is returned. Otherwise the converter returns the average incoming connection
15633 rate associated with the input sample in the designated table. See also the
15634 sc_conn_rate sample fetch keyword.
15635
Thierry FOURNIER236657b2015-08-19 08:25:14 +020015636table_gpt0(<table>)
15637 Uses the string representation of the input sample to perform a look up in
15638 the specified table. If the key is not found in the table, boolean value zero
15639 is returned. Otherwise the converter returns the current value of the first
15640 general purpose tag associated with the input sample in the designated table.
15641 See also the sc_get_gpt0 sample fetch keyword.
15642
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015643table_gpc0(<table>)
15644 Uses the string representation of the input sample to perform a look up in
15645 the specified table. If the key is not found in the table, integer value zero
15646 is returned. Otherwise the converter returns the current value of the first
15647 general purpose counter associated with the input sample in the designated
15648 table. See also the sc_get_gpc0 sample fetch keyword.
15649
15650table_gpc0_rate(<table>)
15651 Uses the string representation of the input sample to perform a look up in
15652 the specified table. If the key is not found in the table, integer value zero
15653 is returned. Otherwise the converter returns the frequency which the gpc0
15654 counter was incremented over the configured period in the table, associated
15655 with the input sample in the designated table. See also the sc_get_gpc0_rate
15656 sample fetch keyword.
15657
Frédéric Lécaille6778b272018-01-29 15:22:53 +010015658table_gpc1(<table>)
15659 Uses the string representation of the input sample to perform a look up in
15660 the specified table. If the key is not found in the table, integer value zero
15661 is returned. Otherwise the converter returns the current value of the second
15662 general purpose counter associated with the input sample in the designated
15663 table. See also the sc_get_gpc1 sample fetch keyword.
15664
15665table_gpc1_rate(<table>)
15666 Uses the string representation of the input sample to perform a look up in
15667 the specified table. If the key is not found in the table, integer value zero
15668 is returned. Otherwise the converter returns the frequency which the gpc1
15669 counter was incremented over the configured period in the table, associated
15670 with the input sample in the designated table. See also the sc_get_gpc1_rate
15671 sample fetch keyword.
15672
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015673table_http_err_cnt(<table>)
15674 Uses the string representation of the input sample to perform a look up in
15675 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010015676 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015677 errors associated with the input sample in the designated table. See also the
15678 sc_http_err_cnt sample fetch keyword.
15679
15680table_http_err_rate(<table>)
15681 Uses the string representation of the input sample to perform a look up in
15682 the specified table. If the key is not found in the table, integer value zero
15683 is returned. Otherwise the average rate of HTTP errors associated with the
15684 input sample in the designated table, measured in amount of errors over the
15685 period configured in the table. See also the sc_http_err_rate sample fetch
15686 keyword.
15687
15688table_http_req_cnt(<table>)
15689 Uses the string representation of the input sample to perform a look up in
15690 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010015691 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015692 requests associated with the input sample in the designated table. See also
15693 the sc_http_req_cnt sample fetch keyword.
15694
15695table_http_req_rate(<table>)
15696 Uses the string representation of the input sample to perform a look up in
15697 the specified table. If the key is not found in the table, integer value zero
15698 is returned. Otherwise the average rate of HTTP requests associated with the
15699 input sample in the designated table, measured in amount of requests over the
15700 period configured in the table. See also the sc_http_req_rate sample fetch
15701 keyword.
15702
15703table_kbytes_in(<table>)
15704 Uses the string representation of the input sample to perform a look up in
15705 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010015706 is returned. Otherwise the converter returns the cumulative number of client-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015707 to-server data associated with the input sample in the designated table,
15708 measured in kilobytes. The test is currently performed on 32-bit integers,
15709 which limits values to 4 terabytes. See also the sc_kbytes_in sample fetch
15710 keyword.
15711
15712table_kbytes_out(<table>)
15713 Uses the string representation of the input sample to perform a look up in
15714 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010015715 is returned. Otherwise the converter returns the cumulative number of server-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015716 to-client data associated with the input sample in the designated table,
15717 measured in kilobytes. The test is currently performed on 32-bit integers,
15718 which limits values to 4 terabytes. See also the sc_kbytes_out sample fetch
15719 keyword.
15720
15721table_server_id(<table>)
15722 Uses the string representation of the input sample to perform a look up in
15723 the specified table. If the key is not found in the table, integer value zero
15724 is returned. Otherwise the converter returns the server ID associated with
15725 the input sample in the designated table. A server ID is associated to a
15726 sample by a "stick" rule when a connection to a server succeeds. A server ID
15727 zero means that no server is associated with this key.
15728
15729table_sess_cnt(<table>)
15730 Uses the string representation of the input sample to perform a look up in
15731 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010015732 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015733 sessions associated with the input sample in the designated table. Note that
15734 a session here refers to an incoming connection being accepted by the
15735 "tcp-request connection" rulesets. See also the sc_sess_cnt sample fetch
15736 keyword.
15737
15738table_sess_rate(<table>)
15739 Uses the string representation of the input sample to perform a look up in
15740 the specified table. If the key is not found in the table, integer value zero
15741 is returned. Otherwise the converter returns the average incoming session
15742 rate associated with the input sample in the designated table. Note that a
15743 session here refers to an incoming connection being accepted by the
15744 "tcp-request connection" rulesets. See also the sc_sess_rate sample fetch
15745 keyword.
15746
15747table_trackers(<table>)
15748 Uses the string representation of the input sample to perform a look up in
15749 the specified table. If the key is not found in the table, integer value zero
15750 is returned. Otherwise the converter returns the current amount of concurrent
15751 connections tracking the same key as the input sample in the designated
15752 table. It differs from table_conn_cur in that it does not rely on any stored
15753 information but on the table's reference count (the "use" value which is
15754 returned by "show table" on the CLI). This may sometimes be more suited for
15755 layer7 tracking. It can be used to tell a server how many concurrent
15756 connections there are from a given address for example. See also the
15757 sc_trackers sample fetch keyword.
15758
Willy Tarreauffcb2e42014-07-10 16:29:08 +020015759upper
15760 Convert a string sample to upper case. This can only be placed after a string
15761 sample fetch function or after a transformation keyword returning a string
15762 type. The result is of type string.
15763
Willy Tarreau62ba9ba2020-04-23 17:54:47 +020015764url_dec([<in_form>])
15765 Takes an url-encoded string provided as input and returns the decoded version
15766 as output. The input and the output are of type string. If the <in_form>
15767 argument is set to a non-zero integer value, the input string is assumed to
15768 be part of a form or query string and the '+' character will be turned into a
15769 space (' '). Otherwise this will only happen after a question mark indicating
15770 a query string ('?').
Thierry FOURNIER82ff3c92015-05-07 15:46:20 +020015771
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010015772ungrpc(<field_number>,[<field_type>])
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010015773 This extracts the protocol buffers message field in raw mode of an input binary
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010015774 sample representation of a gRPC message with <field_number> as field number
15775 (dotted notation) if <field_type> is not present, or as an integer sample if this
15776 field is present.
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010015777 The list of the authorized types is the following one: "int32", "int64", "uint32",
15778 "uint64", "sint32", "sint64", "bool", "enum" for the "varint" wire type 0
15779 "fixed64", "sfixed64", "double" for the 64bit wire type 1, "fixed32", "sfixed32",
15780 "float" for the wire type 5. Note that "string" is considered as a length-delimited
Frédéric Lécaille93d33162019-03-06 09:35:59 +010015781 type, so it does not require any <field_type> argument to be extracted.
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010015782 More information may be found here about the protocol buffers message field types:
15783 https://developers.google.com/protocol-buffers/docs/encoding
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010015784
15785 Example:
15786 // with such a protocol buffer .proto file content adapted from
15787 // https://github.com/grpc/grpc/blob/master/examples/protos/route_guide.proto
15788
15789 message Point {
15790 int32 latitude = 1;
15791 int32 longitude = 2;
15792 }
15793
15794 message PPoint {
15795 Point point = 59;
15796 }
15797
15798 message Rectangle {
15799 // One corner of the rectangle.
15800 PPoint lo = 48;
15801 // The other corner of the rectangle.
15802 PPoint hi = 49;
15803 }
15804
Peter Gervaidf4c9d22020-06-11 18:05:11 +020015805 let's say a body request is made of a "Rectangle" object value (two PPoint
15806 protocol buffers messages), the four protocol buffers fields could be
15807 extracted with these "ungrpc" directives:
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010015808
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010015809 req.body,ungrpc(48.59.1,int32) # "latitude" of "lo" first PPoint
15810 req.body,ungrpc(48.59.2,int32) # "longitude" of "lo" first PPoint
John Roeslerfb2fce12019-07-10 15:45:51 -050015811 req.body,ungrpc(49.59.1,int32) # "latitude" of "hi" second PPoint
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010015812 req.body,ungrpc(49.59.2,int32) # "longitude" of "hi" second PPoint
15813
Peter Gervaidf4c9d22020-06-11 18:05:11 +020015814 We could also extract the intermediary 48.59 field as a binary sample as follows:
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010015815
Frédéric Lécaille93d33162019-03-06 09:35:59 +010015816 req.body,ungrpc(48.59)
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010015817
Peter Gervaidf4c9d22020-06-11 18:05:11 +020015818 As a gRPC message is always made of a gRPC header followed by protocol buffers
15819 messages, in the previous example the "latitude" of "lo" first PPoint
15820 could be extracted with these equivalent directives:
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010015821
15822 req.body,ungrpc(48.59),protobuf(1,int32)
15823 req.body,ungrpc(48),protobuf(59.1,int32)
15824 req.body,ungrpc(48),protobuf(59),protobuf(1,int32)
15825
Peter Gervaidf4c9d22020-06-11 18:05:11 +020015826 Note that the first convert must be "ungrpc", the remaining ones must be
15827 "protobuf" and only the last one may have or not a second argument to
15828 interpret the previous binary sample.
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010015829
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010015830
Christopher Faulet85d79c92016-11-09 16:54:56 +010015831unset-var(<var name>)
15832 Unsets a variable if the input content is defined. The name of the variable
15833 starts with an indication about its scope. The scopes allowed are:
15834 "proc" : the variable is shared with the whole process
15835 "sess" : the variable is shared with the whole session
15836 "txn" : the variable is shared with the transaction (request and
15837 response),
15838 "req" : the variable is shared only during request processing,
15839 "res" : the variable is shared only during response processing.
15840 This prefix is followed by a name. The separator is a '.'. The name may only
15841 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
15842
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020015843utime(<format>[,<offset>])
15844 Converts an integer supposed to contain a date since epoch to a string
15845 representing this date in UTC time using a format defined by the <format>
15846 string using strftime(3). The purpose is to allow any date format to be used
15847 in logs. An optional <offset> in seconds may be applied to the input date
15848 (positive or negative). See the strftime() man page for the format supported
15849 by your operating system. See also the ltime converter.
15850
15851 Example :
15852
15853 # Emit two colons, one with the UTC time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010015854 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020015855 log-format %[date,utime(%Y%m%d%H%M%S)]\ %ci:%cp
15856
Marcin Deranek9631a282018-04-16 14:30:46 +020015857word(<index>,<delimiters>[,<count>])
15858 Extracts the nth word counting from the beginning (positive index) or from
15859 the end (negative index) considering given delimiters from an input string.
15860 Indexes start at 1 or -1 and delimiters are a string formatted list of chars.
Jerome Magnin88209322020-01-28 13:33:44 +010015861 Delimiters at the beginning or end of the input string are ignored.
Marcin Deranek9631a282018-04-16 14:30:46 +020015862 Optionally you can specify <count> of words to extract (default: 1).
15863 Value of 0 indicates extraction of all remaining words.
15864
15865 Example :
15866 str(f1_f2_f3__f5),word(4,_) # f5
15867 str(f1_f2_f3__f5),word(2,_,0) # f2_f3__f5
15868 str(f1_f2_f3__f5),word(3,_,2) # f3__f5
15869 str(f1_f2_f3__f5),word(-2,_,3) # f1_f2_f3
15870 str(f1_f2_f3__f5),word(-3,_,0) # f1_f2
Jerome Magnin88209322020-01-28 13:33:44 +010015871 str(/f1/f2/f3/f4),word(1,/) # f1
Emeric Brunc9a0f6d2014-11-25 14:09:01 +010015872
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015873wt6([<avalanche>])
15874 Hashes a binary input sample into an unsigned 32-bit quantity using the WT6
15875 hash function. Optionally, it is possible to apply a full avalanche hash
15876 function to the output if the optional <avalanche> argument equals 1. This
15877 converter uses the same functions as used by the various hash-based load
15878 balancing algorithms, so it will provide exactly the same results. It is
15879 mostly intended for debugging, but can be used as a stick-table entry to
15880 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010015881 32-bit hash is trivial to break. See also "crc32", "djb2", "sdbm", "crc32c",
15882 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015883
Willy Tarreau97707872015-01-27 15:12:13 +010015884xor(<value>)
15885 Performs a bitwise "XOR" (exclusive OR) between <value> and the input value
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015886 of type signed integer, and returns the result as an signed integer.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015887 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010015888 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015889 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015890 "sess" : the variable is shared with the whole session
15891 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015892 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010015893 "req" : the variable is shared only during request processing,
15894 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015895 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015896 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010015897
Thierry FOURNIER01e09742016-12-26 11:46:11 +010015898xxh32([<seed>])
15899 Hashes a binary input sample into an unsigned 32-bit quantity using the 32-bit
15900 variant of the XXHash hash function. This hash supports a seed which defaults
15901 to zero but a different value maybe passed as the <seed> argument. This hash
15902 is known to be very good and very fast so it can be used to hash URLs and/or
15903 URL parameters for use as stick-table keys to collect statistics with a low
15904 collision rate, though care must be taken as the algorithm is not considered
15905 as cryptographically secure.
15906
15907xxh64([<seed>])
15908 Hashes a binary input sample into a signed 64-bit quantity using the 64-bit
15909 variant of the XXHash hash function. This hash supports a seed which defaults
15910 to zero but a different value maybe passed as the <seed> argument. This hash
15911 is known to be very good and very fast so it can be used to hash URLs and/or
15912 URL parameters for use as stick-table keys to collect statistics with a low
15913 collision rate, though care must be taken as the algorithm is not considered
15914 as cryptographically secure.
15915
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010015916
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200159177.3.2. Fetching samples from internal states
Willy Tarreau74ca5042013-06-11 23:12:07 +020015918--------------------------------------------
15919
15920A first set of sample fetch methods applies to internal information which does
15921not even relate to any client information. These ones are sometimes used with
15922"monitor-fail" directives to report an internal status to external watchers.
15923The sample fetch methods described in this section are usable anywhere.
15924
15925always_false : boolean
15926 Always returns the boolean "false" value. It may be used with ACLs as a
15927 temporary replacement for another one when adjusting configurations.
15928
15929always_true : boolean
15930 Always returns the boolean "true" value. It may be used with ACLs as a
15931 temporary replacement for another one when adjusting configurations.
15932
15933avg_queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010015934 Returns the total number of queued connections of the designated backend
Willy Tarreau74ca5042013-06-11 23:12:07 +020015935 divided by the number of active servers. The current backend is used if no
15936 backend is specified. This is very similar to "queue" except that the size of
15937 the farm is considered, in order to give a more accurate measurement of the
15938 time it may take for a new connection to be processed. The main usage is with
15939 ACL to return a sorry page to new users when it becomes certain they will get
15940 a degraded service, or to pass to the backend servers in a header so that
15941 they decide to work in degraded mode or to disable some functions to speed up
15942 the processing a bit. Note that in the event there would not be any active
15943 server anymore, twice the number of queued connections would be considered as
15944 the measured value. This is a fair estimate, as we expect one server to get
15945 back soon anyway, but we still prefer to send new traffic to another backend
15946 if in better shape. See also the "queue", "be_conn", and "be_sess_rate"
15947 sample fetches.
Krzysztof Piotr Oledzki346f76d2010-01-12 21:59:30 +010015948
Willy Tarreau74ca5042013-06-11 23:12:07 +020015949be_conn([<backend>]) : integer
Willy Tarreaua36af912009-10-10 12:02:45 +020015950 Applies to the number of currently established connections on the backend,
15951 possibly including the connection being evaluated. If no backend name is
15952 specified, the current one is used. But it is also possible to check another
15953 backend. It can be used to use a specific farm when the nominal one is full.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040015954 See also the "fe_conn", "queue", "be_conn_free", and "be_sess_rate" criteria.
15955
15956be_conn_free([<backend>]) : integer
15957 Returns an integer value corresponding to the number of available connections
15958 across available servers in the backend. Queue slots are not included. Backup
15959 servers are also not included, unless all other servers are down. If no
15960 backend name is specified, the current one is used. But it is also possible
15961 to check another backend. It can be used to use a specific farm when the
Patrick Hemmer155e93e2018-06-14 18:01:35 -040015962 nominal one is full. See also the "be_conn", "connslots", and "srv_conn_free"
15963 criteria.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040015964
15965 OTHER CAVEATS AND NOTES: if any of the server maxconn, or maxqueue is 0
15966 (meaning unlimited), then this fetch clearly does not make sense, in which
15967 case the value returned will be -1.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015968
Willy Tarreau74ca5042013-06-11 23:12:07 +020015969be_sess_rate([<backend>]) : integer
15970 Returns an integer value corresponding to the sessions creation rate on the
15971 backend, in number of new sessions per second. This is used with ACLs to
15972 switch to an alternate backend when an expensive or fragile one reaches too
Davor Ocelice9ed2812017-12-25 17:49:28 +010015973 high a session rate, or to limit abuse of service (e.g. prevent sucking of an
Willy Tarreau74ca5042013-06-11 23:12:07 +020015974 online dictionary). It can also be useful to add this element to logs using a
15975 log-format directive.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015976
15977 Example :
15978 # Redirect to an error page if the dictionary is requested too often
15979 backend dynamic
15980 mode http
15981 acl being_scanned be_sess_rate gt 100
15982 redirect location /denied.html if being_scanned
Willy Tarreau0ba27502007-12-24 16:55:16 +010015983
Davor Ocelice9ed2812017-12-25 17:49:28 +010015984bin(<hex>) : bin
Thierry FOURNIERcc103292015-06-06 19:30:17 +020015985 Returns a binary chain. The input is the hexadecimal representation
15986 of the string.
15987
15988bool(<bool>) : bool
15989 Returns a boolean value. <bool> can be 'true', 'false', '1' or '0'.
15990 'false' and '0' are the same. 'true' and '1' are the same.
15991
Willy Tarreau74ca5042013-06-11 23:12:07 +020015992connslots([<backend>]) : integer
15993 Returns an integer value corresponding to the number of connection slots
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015994 still available in the backend, by totaling the maximum amount of
Willy Tarreau74ca5042013-06-11 23:12:07 +020015995 connections on all servers and the maximum queue size. This is probably only
15996 used with ACLs.
Tait Clarridge7896d522012-12-05 21:39:31 -050015997
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080015998 The basic idea here is to be able to measure the number of connection "slots"
Willy Tarreau55165fe2009-05-10 12:02:55 +020015999 still available (connection + queue), so that anything beyond that (intended
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080016000 usage; see "use_backend" keyword) can be redirected to a different backend.
16001
Willy Tarreau55165fe2009-05-10 12:02:55 +020016002 'connslots' = number of available server connection slots, + number of
16003 available server queue slots.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080016004
Willy Tarreaua36af912009-10-10 12:02:45 +020016005 Note that while "fe_conn" may be used, "connslots" comes in especially
Willy Tarreau55165fe2009-05-10 12:02:55 +020016006 useful when you have a case of traffic going to one single ip, splitting into
Willy Tarreau74ca5042013-06-11 23:12:07 +020016007 multiple backends (perhaps using ACLs to do name-based load balancing) and
Willy Tarreau55165fe2009-05-10 12:02:55 +020016008 you want to be able to differentiate between different backends, and their
Davor Ocelice9ed2812017-12-25 17:49:28 +010016009 available "connslots". Also, whereas "nbsrv" only measures servers that are
Willy Tarreau74ca5042013-06-11 23:12:07 +020016010 actually *down*, this fetch is more fine-grained and looks into the number of
Willy Tarreaua36af912009-10-10 12:02:45 +020016011 available connection slots as well. See also "queue" and "avg_queue".
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080016012
Willy Tarreau55165fe2009-05-10 12:02:55 +020016013 OTHER CAVEATS AND NOTES: at this point in time, the code does not take care
16014 of dynamic connections. Also, if any of the server maxconn, or maxqueue is 0,
Willy Tarreau74ca5042013-06-11 23:12:07 +020016015 then this fetch clearly does not make sense, in which case the value returned
Willy Tarreau55165fe2009-05-10 12:02:55 +020016016 will be -1.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080016017
Willy Tarreau70fe9442018-11-22 16:07:39 +010016018cpu_calls : integer
16019 Returns the number of calls to the task processing the stream or current
16020 request since it was allocated. This number is reset for each new request on
16021 the same connections in case of HTTP keep-alive. This value should usually be
16022 low and stable (around 2 calls for a typically simple request) but may become
16023 high if some processing (compression, caching or analysis) is performed. This
16024 is purely for performance monitoring purposes.
16025
16026cpu_ns_avg : integer
16027 Returns the average number of nanoseconds spent in each call to the task
16028 processing the stream or current request. This number is reset for each new
16029 request on the same connections in case of HTTP keep-alive. This value
16030 indicates the overall cost of processing the request or the connection for
16031 each call. There is no good nor bad value but the time spent in a call
16032 automatically causes latency for other processing (see lat_ns_avg below),
16033 and may affect other connection's apparent response time. Certain operations
16034 like compression, complex regex matching or heavy Lua operations may directly
16035 affect this value, and having it in the logs will make it easier to spot the
16036 faulty processing that needs to be fixed to recover decent performance.
16037 Note: this value is exactly cpu_ns_tot divided by cpu_calls.
16038
16039cpu_ns_tot : integer
16040 Returns the total number of nanoseconds spent in each call to the task
16041 processing the stream or current request. This number is reset for each new
16042 request on the same connections in case of HTTP keep-alive. This value
16043 indicates the overall cost of processing the request or the connection for
16044 each call. There is no good nor bad value but the time spent in a call
16045 automatically causes latency for other processing (see lat_ns_avg below),
16046 induces CPU costs on the machine, and may affect other connection's apparent
16047 response time. Certain operations like compression, complex regex matching or
16048 heavy Lua operations may directly affect this value, and having it in the
16049 logs will make it easier to spot the faulty processing that needs to be fixed
16050 to recover decent performance. The value may be artificially high due to a
16051 high cpu_calls count, for example when processing many HTTP chunks, and for
16052 this reason it is often preferred to log cpu_ns_avg instead.
16053
Cyril Bonté6bcd1822019-11-05 23:13:59 +010016054date([<offset>],[<unit>]) : integer
Willy Tarreau6236d3a2013-07-25 14:28:25 +020016055 Returns the current date as the epoch (number of seconds since 01/01/1970).
Damien Claisseae6f1252019-10-30 15:57:28 +000016056
16057 If an offset value is specified, then it is added to the current date before
16058 returning the value. This is particularly useful to compute relative dates,
16059 as both positive and negative offsets are allowed.
Willy Tarreau276fae92013-07-25 14:36:01 +020016060 It is useful combined with the http_date converter.
16061
Damien Claisseae6f1252019-10-30 15:57:28 +000016062 <unit> is facultative, and can be set to "s" for seconds (default behavior),
16063 "ms" for milliseconds or "us" for microseconds.
16064 If unit is set, return value is an integer reflecting either seconds,
16065 milliseconds or microseconds since epoch, plus offset.
16066 It is useful when a time resolution of less than a second is needed.
16067
Willy Tarreau276fae92013-07-25 14:36:01 +020016068 Example :
16069
16070 # set an expires header to now+1 hour in every response
16071 http-response set-header Expires %[date(3600),http_date]
Willy Tarreau6236d3a2013-07-25 14:28:25 +020016072
Damien Claisseae6f1252019-10-30 15:57:28 +000016073 # set an expires header to now+1 hour in every response, with
16074 # millisecond granularity
16075 http-response set-header Expires %[date(3600000,ms),http_date(0,ms)]
16076
Etienne Carrierea792a0a2018-01-17 13:43:24 +010016077date_us : integer
16078 Return the microseconds part of the date (the "second" part is returned by
16079 date sample). This sample is coherent with the date sample as it is comes
16080 from the same timeval structure.
16081
Willy Tarreaud716f9b2017-10-13 11:03:15 +020016082distcc_body(<token>[,<occ>]) : binary
16083 Parses a distcc message and returns the body associated to occurrence #<occ>
16084 of the token <token>. Occurrences start at 1, and when unspecified, any may
16085 match though in practice only the first one is checked for now. This can be
16086 used to extract file names or arguments in files built using distcc through
16087 haproxy. Please refer to distcc's protocol documentation for the complete
16088 list of supported tokens.
16089
16090distcc_param(<token>[,<occ>]) : integer
16091 Parses a distcc message and returns the parameter associated to occurrence
16092 #<occ> of the token <token>. Occurrences start at 1, and when unspecified,
16093 any may match though in practice only the first one is checked for now. This
16094 can be used to extract certain information such as the protocol version, the
16095 file size or the argument in files built using distcc through haproxy.
16096 Another use case consists in waiting for the start of the preprocessed file
16097 contents before connecting to the server to avoid keeping idle connections.
16098 Please refer to distcc's protocol documentation for the complete list of
16099 supported tokens.
16100
16101 Example :
16102 # wait up to 20s for the pre-processed file to be uploaded
16103 tcp-request inspect-delay 20s
16104 tcp-request content accept if { distcc_param(DOTI) -m found }
16105 # send large files to the big farm
16106 use_backend big_farm if { distcc_param(DOTI) gt 1000000 }
16107
Willy Tarreau595ec542013-06-12 21:34:28 +020016108env(<name>) : string
16109 Returns a string containing the value of environment variable <name>. As a
16110 reminder, environment variables are per-process and are sampled when the
16111 process starts. This can be useful to pass some information to a next hop
16112 server, or with ACLs to take specific action when the process is started a
16113 certain way.
16114
16115 Examples :
16116 # Pass the Via header to next hop with the local hostname in it
16117 http-request add-header Via 1.1\ %[env(HOSTNAME)]
16118
16119 # reject cookie-less requests when the STOP environment variable is set
16120 http-request deny if !{ cook(SESSIONID) -m found } { env(STOP) -m found }
16121
Willy Tarreau74ca5042013-06-11 23:12:07 +020016122fe_conn([<frontend>]) : integer
16123 Returns the number of currently established connections on the frontend,
Willy Tarreaud63335a2010-02-26 12:56:52 +010016124 possibly including the connection being evaluated. If no frontend name is
16125 specified, the current one is used. But it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020016126 frontend. It can be used to return a sorry page before hard-blocking, or to
16127 use a specific backend to drain new requests when the farm is considered
Davor Ocelice9ed2812017-12-25 17:49:28 +010016128 full. This is mostly used with ACLs but can also be used to pass some
Willy Tarreau74ca5042013-06-11 23:12:07 +020016129 statistics to servers in HTTP headers. See also the "dst_conn", "be_conn",
16130 "fe_sess_rate" fetches.
Willy Tarreaua36af912009-10-10 12:02:45 +020016131
Nenad Merdanovicad9a7e92016-10-03 04:57:37 +020016132fe_req_rate([<frontend>]) : integer
16133 Returns an integer value corresponding to the number of HTTP requests per
16134 second sent to a frontend. This number can differ from "fe_sess_rate" in
16135 situations where client-side keep-alive is enabled.
16136
Willy Tarreau74ca5042013-06-11 23:12:07 +020016137fe_sess_rate([<frontend>]) : integer
16138 Returns an integer value corresponding to the sessions creation rate on the
16139 frontend, in number of new sessions per second. This is used with ACLs to
16140 limit the incoming session rate to an acceptable range in order to prevent
16141 abuse of service at the earliest moment, for example when combined with other
16142 layer 4 ACLs in order to force the clients to wait a bit for the rate to go
16143 down below the limit. It can also be useful to add this element to logs using
16144 a log-format directive. See also the "rate-limit sessions" directive for use
16145 in frontends.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010016146
16147 Example :
16148 # This frontend limits incoming mails to 10/s with a max of 100
16149 # concurrent connections. We accept any connection below 10/s, and
16150 # force excess clients to wait for 100 ms. Since clients are limited to
16151 # 100 max, there cannot be more than 10 incoming mails per second.
16152 frontend mail
16153 bind :25
16154 mode tcp
16155 maxconn 100
16156 acl too_fast fe_sess_rate ge 10
16157 tcp-request inspect-delay 100ms
16158 tcp-request content accept if ! too_fast
16159 tcp-request content accept if WAIT_END
Willy Tarreaud72758d2010-01-12 10:42:19 +010016160
Nenad Merdanovic807a6e72017-03-12 22:00:00 +010016161hostname : string
16162 Returns the system hostname.
16163
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016164int(<integer>) : signed integer
16165 Returns a signed integer.
16166
Thierry FOURNIERcc103292015-06-06 19:30:17 +020016167ipv4(<ipv4>) : ipv4
16168 Returns an ipv4.
16169
16170ipv6(<ipv6>) : ipv6
16171 Returns an ipv6.
16172
Willy Tarreau70fe9442018-11-22 16:07:39 +010016173lat_ns_avg : integer
16174 Returns the average number of nanoseconds spent between the moment the task
16175 handling the stream is woken up and the moment it is effectively called. This
16176 number is reset for each new request on the same connections in case of HTTP
16177 keep-alive. This value indicates the overall latency inflicted to the current
16178 request by all other requests being processed in parallel, and is a direct
16179 indicator of perceived performance due to noisy neighbours. In order to keep
16180 the value low, it is possible to reduce the scheduler's run queue depth using
16181 "tune.runqueue-depth", to reduce the number of concurrent events processed at
16182 once using "tune.maxpollevents", to decrease the stream's nice value using
Willy Tarreaue7723bd2020-06-24 11:11:02 +020016183 the "nice" option on the "bind" lines or in the frontend, to enable low
16184 latency scheduling using "tune.sched.low-latency", or to look for other heavy
16185 requests in logs (those exhibiting large values of "cpu_ns_avg"), whose
16186 processing needs to be adjusted or fixed. Compression of large buffers could
16187 be a culprit, like heavy regex or long lists of regex. Note: this value is
16188 exactly lat_ns_tot divided by cpu_calls.
Willy Tarreau70fe9442018-11-22 16:07:39 +010016189
16190lat_ns_tot : integer
16191 Returns the total number of nanoseconds spent between the moment the task
16192 handling the stream is woken up and the moment it is effectively called. This
16193 number is reset for each new request on the same connections in case of HTTP
16194 keep-alive. This value indicates the overall latency inflicted to the current
16195 request by all other requests being processed in parallel, and is a direct
16196 indicator of perceived performance due to noisy neighbours. In order to keep
16197 the value low, it is possible to reduce the scheduler's run queue depth using
16198 "tune.runqueue-depth", to reduce the number of concurrent events processed at
16199 once using "tune.maxpollevents", to decrease the stream's nice value using
Willy Tarreaue7723bd2020-06-24 11:11:02 +020016200 the "nice" option on the "bind" lines or in the frontend, to enable low
16201 latency scheduling using "tune.sched.low-latency", or to look for other heavy
16202 requests in logs (those exhibiting large values of "cpu_ns_avg"), whose
16203 processing needs to be adjusted or fixed. Compression of large buffers could
16204 be a culprit, like heavy regex or long lists of regex. Note: while it
Willy Tarreau70fe9442018-11-22 16:07:39 +010016205 may intuitively seem that the total latency adds to a transfer time, it is
16206 almost never true because while a task waits for the CPU, network buffers
16207 continue to fill up and the next call will process more at once. The value
16208 may be artificially high due to a high cpu_calls count, for example when
16209 processing many HTTP chunks, and for this reason it is often preferred to log
16210 lat_ns_avg instead, which is a more relevant performance indicator.
16211
Thierry FOURNIERcc103292015-06-06 19:30:17 +020016212meth(<method>) : method
16213 Returns a method.
16214
Willy Tarreau0f30d262014-11-24 16:02:05 +010016215nbproc : integer
16216 Returns an integer value corresponding to the number of processes that were
16217 started (it equals the global "nbproc" setting). This is useful for logging
16218 and debugging purposes.
16219
Willy Tarreau74ca5042013-06-11 23:12:07 +020016220nbsrv([<backend>]) : integer
16221 Returns an integer value corresponding to the number of usable servers of
16222 either the current backend or the named backend. This is mostly used with
16223 ACLs but can also be useful when added to logs. This is normally used to
Willy Tarreaud63335a2010-02-26 12:56:52 +010016224 switch to an alternate backend when the number of servers is too low to
16225 to handle some load. It is useful to report a failure when combined with
16226 "monitor fail".
Willy Tarreau079ff0a2009-03-05 21:34:28 +010016227
Patrick Hemmerfabb24f2018-08-13 14:07:57 -040016228prio_class : integer
16229 Returns the priority class of the current session for http mode or connection
16230 for tcp mode. The value will be that set by the last call to "http-request
16231 set-priority-class" or "tcp-request content set-priority-class".
16232
16233prio_offset : integer
16234 Returns the priority offset of the current session for http mode or
16235 connection for tcp mode. The value will be that set by the last call to
16236 "http-request set-priority-offset" or "tcp-request content
16237 set-priority-offset".
16238
Willy Tarreau0f30d262014-11-24 16:02:05 +010016239proc : integer
16240 Returns an integer value corresponding to the position of the process calling
16241 the function, between 1 and global.nbproc. This is useful for logging and
16242 debugging purposes.
16243
Willy Tarreau74ca5042013-06-11 23:12:07 +020016244queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010016245 Returns the total number of queued connections of the designated backend,
16246 including all the connections in server queues. If no backend name is
16247 specified, the current one is used, but it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020016248 one. This is useful with ACLs or to pass statistics to backend servers. This
16249 can be used to take actions when queuing goes above a known level, generally
16250 indicating a surge of traffic or a massive slowdown on the servers. One
16251 possible action could be to reject new users but still accept old ones. See
16252 also the "avg_queue", "be_conn", and "be_sess_rate" fetches.
16253
Willy Tarreau84310e22014-02-14 11:59:04 +010016254rand([<range>]) : integer
16255 Returns a random integer value within a range of <range> possible values,
16256 starting at zero. If the range is not specified, it defaults to 2^32, which
16257 gives numbers between 0 and 4294967295. It can be useful to pass some values
16258 needed to take some routing decisions for example, or just for debugging
16259 purposes. This random must not be used for security purposes.
16260
Luca Schimweg8a694b82019-09-10 15:42:52 +020016261uuid([<version>]) : string
16262 Returns a UUID following the RFC4122 standard. If the version is not
16263 specified, a UUID version 4 (fully random) is returned.
16264 Currently, only version 4 is supported.
16265
Willy Tarreau74ca5042013-06-11 23:12:07 +020016266srv_conn([<backend>/]<server>) : integer
16267 Returns an integer value corresponding to the number of currently established
16268 connections on the designated server, possibly including the connection being
16269 evaluated. If <backend> is omitted, then the server is looked up in the
16270 current backend. It can be used to use a specific farm when one server is
16271 full, or to inform the server about our view of the number of active
Patrick Hemmer155e93e2018-06-14 18:01:35 -040016272 connections with it. See also the "fe_conn", "be_conn", "queue", and
16273 "srv_conn_free" fetch methods.
16274
16275srv_conn_free([<backend>/]<server>) : integer
16276 Returns an integer value corresponding to the number of available connections
16277 on the designated server, possibly including the connection being evaluated.
16278 The value does not include queue slots. If <backend> is omitted, then the
16279 server is looked up in the current backend. It can be used to use a specific
16280 farm when one server is full, or to inform the server about our view of the
16281 number of active connections with it. See also the "be_conn_free" and
16282 "srv_conn" fetch methods.
16283
16284 OTHER CAVEATS AND NOTES: If the server maxconn is 0, then this fetch clearly
16285 does not make sense, in which case the value returned will be -1.
Willy Tarreau74ca5042013-06-11 23:12:07 +020016286
16287srv_is_up([<backend>/]<server>) : boolean
16288 Returns true when the designated server is UP, and false when it is either
16289 DOWN or in maintenance mode. If <backend> is omitted, then the server is
16290 looked up in the current backend. It is mainly used to take action based on
Davor Ocelice9ed2812017-12-25 17:49:28 +010016291 an external status reported via a health check (e.g. a geographical site's
Willy Tarreau74ca5042013-06-11 23:12:07 +020016292 availability). Another possible use which is more of a hack consists in
16293 using dummy servers as boolean variables that can be enabled or disabled from
16294 the CLI, so that rules depending on those ACLs can be tweaked in realtime.
16295
Willy Tarreauff2b7af2017-10-13 11:46:26 +020016296srv_queue([<backend>/]<server>) : integer
16297 Returns an integer value corresponding to the number of connections currently
16298 pending in the designated server's queue. If <backend> is omitted, then the
16299 server is looked up in the current backend. It can sometimes be used together
16300 with the "use-server" directive to force to use a known faster server when it
16301 is not much loaded. See also the "srv_conn", "avg_queue" and "queue" sample
16302 fetch methods.
16303
Willy Tarreau74ca5042013-06-11 23:12:07 +020016304srv_sess_rate([<backend>/]<server>) : integer
16305 Returns an integer corresponding to the sessions creation rate on the
16306 designated server, in number of new sessions per second. If <backend> is
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016307 omitted, then the server is looked up in the current backend. This is mostly
Willy Tarreau74ca5042013-06-11 23:12:07 +020016308 used with ACLs but can make sense with logs too. This is used to switch to an
16309 alternate backend when an expensive or fragile one reaches too high a session
Davor Ocelice9ed2812017-12-25 17:49:28 +010016310 rate, or to limit abuse of service (e.g. prevent latent requests from
Willy Tarreau74ca5042013-06-11 23:12:07 +020016311 overloading servers).
16312
16313 Example :
16314 # Redirect to a separate back
16315 acl srv1_full srv_sess_rate(be1/srv1) gt 50
16316 acl srv2_full srv_sess_rate(be1/srv2) gt 50
16317 use_backend be2 if srv1_full or srv2_full
16318
Christopher Faulet1bea8652020-07-10 16:03:45 +020016319srv_iweight([<backend>/]<server>): integer
16320 Returns an integer corresponding to the server's initial weight. If <backend>
16321 is omitted, then the server is looked up in the current backend. See also
16322 "srv_weight" and "srv_uweight".
16323
16324srv_uweight([<backend>/]<server>): integer
16325 Returns an integer corresponding to the user visible server's weight. If
16326 <backend> is omitted, then the server is looked up in the current
16327 backend. See also "srv_weight" and "srv_iweight".
16328
16329srv_weight([<backend>/]<server>): integer
16330 Returns an integer corresponding to the current (or effective) server's
16331 weight. If <backend> is omitted, then the server is looked up in the current
16332 backend. See also "srv_iweight" and "srv_uweight".
16333
Willy Tarreau0f30d262014-11-24 16:02:05 +010016334stopping : boolean
16335 Returns TRUE if the process calling the function is currently stopping. This
16336 can be useful for logging, or for relaxing certain checks or helping close
16337 certain connections upon graceful shutdown.
16338
Thierry FOURNIERcc103292015-06-06 19:30:17 +020016339str(<string>) : string
16340 Returns a string.
16341
Willy Tarreau74ca5042013-06-11 23:12:07 +020016342table_avl([<table>]) : integer
16343 Returns the total number of available entries in the current proxy's
16344 stick-table or in the designated stick-table. See also table_cnt.
16345
16346table_cnt([<table>]) : integer
16347 Returns the total number of entries currently in use in the current proxy's
16348 stick-table or in the designated stick-table. See also src_conn_cnt and
16349 table_avl for other entry counting methods.
16350
Christopher Faulet34adb2a2017-11-21 21:45:38 +010016351thread : integer
16352 Returns an integer value corresponding to the position of the thread calling
16353 the function, between 0 and (global.nbthread-1). This is useful for logging
16354 and debugging purposes.
16355
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020016356var(<var-name>) : undefined
16357 Returns a variable with the stored type. If the variable is not set, the
Daniel Schneller0b547052016-03-21 20:46:57 +010016358 sample fetch fails. The name of the variable starts with an indication
16359 about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010016360 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010016361 "sess" : the variable is shared with the whole session
16362 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020016363 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010016364 "req" : the variable is shared only during request processing,
16365 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020016366 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010016367 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020016368
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200163697.3.3. Fetching samples at Layer 4
Willy Tarreau74ca5042013-06-11 23:12:07 +020016370----------------------------------
16371
16372The layer 4 usually describes just the transport layer which in haproxy is
16373closest to the connection, where no content is yet made available. The fetch
16374methods described here are usable as low as the "tcp-request connection" rule
16375sets unless they require some future information. Those generally include
16376TCP/IP addresses and ports, as well as elements from stick-tables related to
Willy Tarreau4d4149c2013-07-23 19:33:46 +020016377the incoming connection. For retrieving a value from a sticky counters, the
16378counter number can be explicitly set as 0, 1, or 2 using the pre-defined
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020016379"sc0_", "sc1_", or "sc2_" prefix. These three pre-defined prefixes can only be
16380used if MAX_SESS_STKCTR value does not exceed 3, otherwise the counter number
16381can be specified as the first integer argument when using the "sc_" prefix.
16382Starting from "sc_0" to "sc_N" where N is (MAX_SESS_STKCTR-1). An optional
16383table may be specified with the "sc*" form, in which case the currently
16384tracked key will be looked up into this alternate table instead of the table
16385currently being tracked.
Willy Tarreau74ca5042013-06-11 23:12:07 +020016386
Jérôme Magnin35e53a62019-01-16 14:38:37 +010016387bc_http_major : integer
Jérôme Magnin86577422018-12-07 09:03:11 +010016388 Returns the backend connection's HTTP major version encoding, which may be 1
16389 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
16390 encoding and not the version present in the request header.
16391
Willy Tarreau74ca5042013-06-11 23:12:07 +020016392be_id : integer
16393 Returns an integer containing the current backend's id. It can be used in
Christopher Fauletd1b44642020-04-30 09:51:15 +020016394 frontends with responses to check which backend processed the request. It can
16395 also be used in a tcp-check or an http-check ruleset.
Willy Tarreau74ca5042013-06-11 23:12:07 +020016396
Marcin Deranekd2471c22016-12-12 14:08:05 +010016397be_name : string
16398 Returns a string containing the current backend's name. It can be used in
Christopher Fauletd1b44642020-04-30 09:51:15 +020016399 frontends with responses to check which backend processed the request. It can
16400 also be used in a tcp-check or an http-check ruleset.
Marcin Deranekd2471c22016-12-12 14:08:05 +010016401
Willy Tarreau74ca5042013-06-11 23:12:07 +020016402dst : ip
16403 This is the destination IPv4 address of the connection on the client side,
16404 which is the address the client connected to. It can be useful when running
16405 in transparent mode. It is of type IP and works on both IPv4 and IPv6 tables.
16406 On IPv6 tables, IPv4 address is mapped to its IPv6 equivalent, according to
Willy Tarreau64ded3d2019-01-23 10:02:15 +010016407 RFC 4291. When the incoming connection passed through address translation or
16408 redirection involving connection tracking, the original destination address
16409 before the redirection will be reported. On Linux systems, the source and
16410 destination may seldom appear reversed if the nf_conntrack_tcp_loose sysctl
16411 is set, because a late response may reopen a timed out connection and switch
16412 what is believed to be the source and the destination.
Willy Tarreau74ca5042013-06-11 23:12:07 +020016413
16414dst_conn : integer
16415 Returns an integer value corresponding to the number of currently established
16416 connections on the same socket including the one being evaluated. It is
16417 normally used with ACLs but can as well be used to pass the information to
16418 servers in an HTTP header or in logs. It can be used to either return a sorry
16419 page before hard-blocking, or to use a specific backend to drain new requests
16420 when the socket is considered saturated. This offers the ability to assign
16421 different limits to different listening ports or addresses. See also the
16422 "fe_conn" and "be_conn" fetches.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016423
Willy Tarreau16e01562016-08-09 16:46:18 +020016424dst_is_local : boolean
16425 Returns true if the destination address of the incoming connection is local
16426 to the system, or false if the address doesn't exist on the system, meaning
16427 that it was intercepted in transparent mode. It can be useful to apply
16428 certain rules by default to forwarded traffic and other rules to the traffic
Davor Ocelice9ed2812017-12-25 17:49:28 +010016429 targeting the real address of the machine. For example the stats page could
Willy Tarreau16e01562016-08-09 16:46:18 +020016430 be delivered only on this address, or SSH access could be locally redirected.
16431 Please note that the check involves a few system calls, so it's better to do
16432 it only once per connection.
16433
Willy Tarreau74ca5042013-06-11 23:12:07 +020016434dst_port : integer
16435 Returns an integer value corresponding to the destination TCP port of the
16436 connection on the client side, which is the port the client connected to.
16437 This might be used when running in transparent mode, when assigning dynamic
16438 ports to some clients for a whole application session, to stick all users to
16439 a same server, or to pass the destination port information to a server using
16440 an HTTP header.
16441
Willy Tarreau60ca10a2017-08-18 15:26:54 +020016442fc_http_major : integer
16443 Reports the front connection's HTTP major version encoding, which may be 1
16444 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
16445 encoding and not on the version present in the request header.
16446
Geoff Simmons7185b782019-08-27 18:31:16 +020016447fc_pp_authority : string
16448 Returns the authority TLV sent by the client in the PROXY protocol header,
16449 if any.
16450
Tim Duesterhusd1b15b62020-03-13 12:34:23 +010016451fc_pp_unique_id : string
16452 Returns the unique ID TLV sent by the client in the PROXY protocol header,
16453 if any.
16454
Emeric Brun4f603012017-01-05 15:11:44 +010016455fc_rcvd_proxy : boolean
16456 Returns true if the client initiated the connection with a PROXY protocol
16457 header.
16458
Thierry Fournier / OZON.IO6310bef2016-07-24 20:16:50 +020016459fc_rtt(<unit>) : integer
16460 Returns the Round Trip Time (RTT) measured by the kernel for the client
16461 connection. <unit> is facultative, by default the unit is milliseconds. <unit>
16462 can be set to "ms" for milliseconds or "us" for microseconds. If the server
16463 connection is not established, if the connection is not TCP or if the
16464 operating system does not support TCP_INFO, for example Linux kernels before
16465 2.4, the sample fetch fails.
16466
16467fc_rttvar(<unit>) : integer
16468 Returns the Round Trip Time (RTT) variance measured by the kernel for the
16469 client connection. <unit> is facultative, by default the unit is milliseconds.
16470 <unit> can be set to "ms" for milliseconds or "us" for microseconds. If the
16471 server connection is not established, if the connection is not TCP or if the
16472 operating system does not support TCP_INFO, for example Linux kernels before
16473 2.4, the sample fetch fails.
16474
Christopher Fauletba0c53e2019-10-17 14:40:48 +020016475fc_unacked : integer
Joe Williams30fcd392016-08-10 07:06:44 -070016476 Returns the unacked counter measured by the kernel for the client connection.
16477 If the server connection is not established, if the connection is not TCP or
16478 if the operating system does not support TCP_INFO, for example Linux kernels
16479 before 2.4, the sample fetch fails.
16480
Christopher Fauletba0c53e2019-10-17 14:40:48 +020016481fc_sacked : integer
Joe Williams30fcd392016-08-10 07:06:44 -070016482 Returns the sacked counter measured by the kernel for the client connection.
16483 If the server connection is not established, if the connection is not TCP or
16484 if the operating system does not support TCP_INFO, for example Linux kernels
16485 before 2.4, the sample fetch fails.
16486
Christopher Fauletba0c53e2019-10-17 14:40:48 +020016487fc_retrans : integer
Joe Williams30fcd392016-08-10 07:06:44 -070016488 Returns the retransmits counter measured by the kernel for the client
16489 connection. If the server connection is not established, if the connection is
16490 not TCP or if the operating system does not support TCP_INFO, for example
16491 Linux kernels before 2.4, the sample fetch fails.
16492
Christopher Fauletba0c53e2019-10-17 14:40:48 +020016493fc_fackets : integer
Joe Williams30fcd392016-08-10 07:06:44 -070016494 Returns the fack counter measured by the kernel for the client
16495 connection. If the server connection is not established, if the connection is
16496 not TCP or if the operating system does not support TCP_INFO, for example
16497 Linux kernels before 2.4, the sample fetch fails.
16498
Christopher Fauletba0c53e2019-10-17 14:40:48 +020016499fc_lost : integer
Joe Williams30fcd392016-08-10 07:06:44 -070016500 Returns the lost counter measured by the kernel for the client
16501 connection. If the server connection is not established, if the connection is
16502 not TCP or if the operating system does not support TCP_INFO, for example
16503 Linux kernels before 2.4, the sample fetch fails.
16504
Christopher Fauletba0c53e2019-10-17 14:40:48 +020016505fc_reordering : integer
Joe Williams30fcd392016-08-10 07:06:44 -070016506 Returns the reordering counter measured by the kernel for the client
16507 connection. If the server connection is not established, if the connection is
16508 not TCP or if the operating system does not support TCP_INFO, for example
16509 Linux kernels before 2.4, the sample fetch fails.
16510
Marcin Deranek9a66dfb2018-04-13 14:37:50 +020016511fe_defbe : string
16512 Returns a string containing the frontend's default backend name. It can be
16513 used in frontends to check which backend will handle requests by default.
16514
Willy Tarreau74ca5042013-06-11 23:12:07 +020016515fe_id : integer
16516 Returns an integer containing the current frontend's id. It can be used in
Marcin Deranek6e413ed2016-12-13 12:40:01 +010016517 backends to check from which frontend it was called, or to stick all users
Willy Tarreau74ca5042013-06-11 23:12:07 +020016518 coming via a same frontend to the same server.
16519
Marcin Deranekd2471c22016-12-12 14:08:05 +010016520fe_name : string
16521 Returns a string containing the current frontend's name. It can be used in
16522 backends to check from which frontend it was called, or to stick all users
16523 coming via a same frontend to the same server.
16524
Cyril Bonté62ba8702014-04-22 23:52:25 +020016525sc_bytes_in_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016526sc0_bytes_in_rate([<table>]) : integer
16527sc1_bytes_in_rate([<table>]) : integer
16528sc2_bytes_in_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020016529 Returns the average client-to-server bytes rate from the currently tracked
16530 counters, measured in amount of bytes over the period configured in the
16531 table. See also src_bytes_in_rate.
16532
Cyril Bonté62ba8702014-04-22 23:52:25 +020016533sc_bytes_out_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016534sc0_bytes_out_rate([<table>]) : integer
16535sc1_bytes_out_rate([<table>]) : integer
16536sc2_bytes_out_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020016537 Returns the average server-to-client bytes rate from the currently tracked
16538 counters, measured in amount of bytes over the period configured in the
16539 table. See also src_bytes_out_rate.
16540
Cyril Bonté62ba8702014-04-22 23:52:25 +020016541sc_clr_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016542sc0_clr_gpc0([<table>]) : integer
16543sc1_clr_gpc0([<table>]) : integer
16544sc2_clr_gpc0([<table>]) : integer
Willy Tarreauf73cd112011-08-13 01:45:16 +020016545 Clears the first General Purpose Counter associated to the currently tracked
16546 counters, and returns its previous value. Before the first invocation, the
Willy Tarreau869948b2013-01-04 14:14:57 +010016547 stored value is zero, so first invocation will always return zero. This is
16548 typically used as a second ACL in an expression in order to mark a connection
16549 when a first ACL was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020016550
Jarno Huuskonen676f6222017-03-30 09:19:45 +030016551 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020016552 # block if 5 consecutive requests continue to come faster than 10 sess
16553 # per second, and reset the counter as soon as the traffic slows down.
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020016554 acl abuse sc0_http_req_rate gt 10
16555 acl kill sc0_inc_gpc0 gt 5
16556 acl save sc0_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020016557 tcp-request connection accept if !abuse save
16558 tcp-request connection reject if abuse kill
16559
Frédéric Lécaille6778b272018-01-29 15:22:53 +010016560sc_clr_gpc1(<ctr>[,<table>]) : integer
16561sc0_clr_gpc1([<table>]) : integer
16562sc1_clr_gpc1([<table>]) : integer
16563sc2_clr_gpc1([<table>]) : integer
16564 Clears the second General Purpose Counter associated to the currently tracked
16565 counters, and returns its previous value. Before the first invocation, the
16566 stored value is zero, so first invocation will always return zero. This is
16567 typically used as a second ACL in an expression in order to mark a connection
16568 when a first ACL was verified.
16569
Cyril Bonté62ba8702014-04-22 23:52:25 +020016570sc_conn_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016571sc0_conn_cnt([<table>]) : integer
16572sc1_conn_cnt([<table>]) : integer
16573sc2_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010016574 Returns the cumulative number of incoming connections from currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020016575 counters. See also src_conn_cnt.
16576
Cyril Bonté62ba8702014-04-22 23:52:25 +020016577sc_conn_cur(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016578sc0_conn_cur([<table>]) : integer
16579sc1_conn_cur([<table>]) : integer
16580sc2_conn_cur([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020016581 Returns the current amount of concurrent connections tracking the same
16582 tracked counters. This number is automatically incremented when tracking
16583 begins and decremented when tracking stops. See also src_conn_cur.
16584
Cyril Bonté62ba8702014-04-22 23:52:25 +020016585sc_conn_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016586sc0_conn_rate([<table>]) : integer
16587sc1_conn_rate([<table>]) : integer
16588sc2_conn_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020016589 Returns the average connection rate from the currently tracked counters,
16590 measured in amount of connections over the period configured in the table.
16591 See also src_conn_rate.
16592
Cyril Bonté62ba8702014-04-22 23:52:25 +020016593sc_get_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016594sc0_get_gpc0([<table>]) : integer
16595sc1_get_gpc0([<table>]) : integer
16596sc2_get_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020016597 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020016598 currently tracked counters. See also src_get_gpc0 and sc/sc0/sc1/sc2_inc_gpc0.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020016599
Frédéric Lécaille6778b272018-01-29 15:22:53 +010016600sc_get_gpc1(<ctr>[,<table>]) : integer
16601sc0_get_gpc1([<table>]) : integer
16602sc1_get_gpc1([<table>]) : integer
16603sc2_get_gpc1([<table>]) : integer
16604 Returns the value of the second General Purpose Counter associated to the
16605 currently tracked counters. See also src_get_gpc1 and sc/sc0/sc1/sc2_inc_gpc1.
16606
Thierry FOURNIER236657b2015-08-19 08:25:14 +020016607sc_get_gpt0(<ctr>[,<table>]) : integer
16608sc0_get_gpt0([<table>]) : integer
16609sc1_get_gpt0([<table>]) : integer
16610sc2_get_gpt0([<table>]) : integer
16611 Returns the value of the first General Purpose Tag associated to the
16612 currently tracked counters. See also src_get_gpt0.
16613
Cyril Bonté62ba8702014-04-22 23:52:25 +020016614sc_gpc0_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016615sc0_gpc0_rate([<table>]) : integer
16616sc1_gpc0_rate([<table>]) : integer
16617sc2_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020016618 Returns the average increment rate of the first General Purpose Counter
16619 associated to the currently tracked counters. It reports the frequency
16620 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020016621 src_gpc0_rate, sc/sc0/sc1/sc2_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
16622 that the "gpc0_rate" counter must be stored in the stick-table for a value to
16623 be returned, as "gpc0" only holds the event count.
Willy Tarreaue9656522010-08-17 15:40:09 +020016624
Frédéric Lécaille6778b272018-01-29 15:22:53 +010016625sc_gpc1_rate(<ctr>[,<table>]) : integer
16626sc0_gpc1_rate([<table>]) : integer
16627sc1_gpc1_rate([<table>]) : integer
16628sc2_gpc1_rate([<table>]) : integer
16629 Returns the average increment rate of the second General Purpose Counter
16630 associated to the currently tracked counters. It reports the frequency
16631 which the gpc1 counter was incremented over the configured period. See also
16632 src_gpcA_rate, sc/sc0/sc1/sc2_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
16633 that the "gpc1_rate" counter must be stored in the stick-table for a value to
16634 be returned, as "gpc1" only holds the event count.
16635
Cyril Bonté62ba8702014-04-22 23:52:25 +020016636sc_http_err_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016637sc0_http_err_cnt([<table>]) : integer
16638sc1_http_err_cnt([<table>]) : integer
16639sc2_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010016640 Returns the cumulative number of HTTP errors from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020016641 counters. This includes the both request errors and 4xx error responses.
16642 See also src_http_err_cnt.
16643
Cyril Bonté62ba8702014-04-22 23:52:25 +020016644sc_http_err_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016645sc0_http_err_rate([<table>]) : integer
16646sc1_http_err_rate([<table>]) : integer
16647sc2_http_err_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020016648 Returns the average rate of HTTP errors from the currently tracked counters,
16649 measured in amount of errors over the period configured in the table. This
16650 includes the both request errors and 4xx error responses. See also
16651 src_http_err_rate.
16652
Cyril Bonté62ba8702014-04-22 23:52:25 +020016653sc_http_req_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016654sc0_http_req_cnt([<table>]) : integer
16655sc1_http_req_cnt([<table>]) : integer
16656sc2_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010016657 Returns the cumulative number of HTTP requests from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020016658 counters. This includes every started request, valid or not. See also
16659 src_http_req_cnt.
16660
Cyril Bonté62ba8702014-04-22 23:52:25 +020016661sc_http_req_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016662sc0_http_req_rate([<table>]) : integer
16663sc1_http_req_rate([<table>]) : integer
16664sc2_http_req_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020016665 Returns the average rate of HTTP requests from the currently tracked
16666 counters, measured in amount of requests over the period configured in
16667 the table. This includes every started request, valid or not. See also
16668 src_http_req_rate.
16669
Cyril Bonté62ba8702014-04-22 23:52:25 +020016670sc_inc_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016671sc0_inc_gpc0([<table>]) : integer
16672sc1_inc_gpc0([<table>]) : integer
16673sc2_inc_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020016674 Increments the first General Purpose Counter associated to the currently
Willy Tarreau869948b2013-01-04 14:14:57 +010016675 tracked counters, and returns its new value. Before the first invocation,
16676 the stored value is zero, so first invocation will increase it to 1 and will
16677 return 1. This is typically used as a second ACL in an expression in order
16678 to mark a connection when a first ACL was verified :
Willy Tarreaue9656522010-08-17 15:40:09 +020016679
Jarno Huuskonen676f6222017-03-30 09:19:45 +030016680 Example:
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020016681 acl abuse sc0_http_req_rate gt 10
16682 acl kill sc0_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020016683 tcp-request connection reject if abuse kill
16684
Frédéric Lécaille6778b272018-01-29 15:22:53 +010016685sc_inc_gpc1(<ctr>[,<table>]) : integer
16686sc0_inc_gpc1([<table>]) : integer
16687sc1_inc_gpc1([<table>]) : integer
16688sc2_inc_gpc1([<table>]) : integer
16689 Increments the second General Purpose Counter associated to the currently
16690 tracked counters, and returns its new value. Before the first invocation,
16691 the stored value is zero, so first invocation will increase it to 1 and will
16692 return 1. This is typically used as a second ACL in an expression in order
16693 to mark a connection when a first ACL was verified.
16694
Cyril Bonté62ba8702014-04-22 23:52:25 +020016695sc_kbytes_in(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016696sc0_kbytes_in([<table>]) : integer
16697sc1_kbytes_in([<table>]) : integer
16698sc2_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020016699 Returns the total amount of client-to-server data from the currently tracked
16700 counters, measured in kilobytes. The test is currently performed on 32-bit
16701 integers, which limits values to 4 terabytes. See also src_kbytes_in.
Willy Tarreaue9656522010-08-17 15:40:09 +020016702
Cyril Bonté62ba8702014-04-22 23:52:25 +020016703sc_kbytes_out(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016704sc0_kbytes_out([<table>]) : integer
16705sc1_kbytes_out([<table>]) : integer
16706sc2_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020016707 Returns the total amount of server-to-client data from the currently tracked
16708 counters, measured in kilobytes. The test is currently performed on 32-bit
16709 integers, which limits values to 4 terabytes. See also src_kbytes_out.
Willy Tarreaue9656522010-08-17 15:40:09 +020016710
Cyril Bonté62ba8702014-04-22 23:52:25 +020016711sc_sess_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016712sc0_sess_cnt([<table>]) : integer
16713sc1_sess_cnt([<table>]) : integer
16714sc2_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010016715 Returns the cumulative number of incoming connections that were transformed
Willy Tarreaue9656522010-08-17 15:40:09 +020016716 into sessions, which means that they were accepted by a "tcp-request
16717 connection" rule, from the currently tracked counters. A backend may count
16718 more sessions than connections because each connection could result in many
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016719 backend sessions if some HTTP keep-alive is performed over the connection
Willy Tarreaue9656522010-08-17 15:40:09 +020016720 with the client. See also src_sess_cnt.
16721
Cyril Bonté62ba8702014-04-22 23:52:25 +020016722sc_sess_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016723sc0_sess_rate([<table>]) : integer
16724sc1_sess_rate([<table>]) : integer
16725sc2_sess_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020016726 Returns the average session rate from the currently tracked counters,
16727 measured in amount of sessions over the period configured in the table. A
16728 session is a connection that got past the early "tcp-request connection"
16729 rules. A backend may count more sessions than connections because each
16730 connection could result in many backend sessions if some HTTP keep-alive is
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016731 performed over the connection with the client. See also src_sess_rate.
Willy Tarreaue9656522010-08-17 15:40:09 +020016732
Cyril Bonté62ba8702014-04-22 23:52:25 +020016733sc_tracked(<ctr>[,<table>]) : boolean
Willy Tarreau0f791d42013-07-23 19:56:43 +020016734sc0_tracked([<table>]) : boolean
16735sc1_tracked([<table>]) : boolean
16736sc2_tracked([<table>]) : boolean
Willy Tarreau6f1615f2013-06-03 15:15:22 +020016737 Returns true if the designated session counter is currently being tracked by
16738 the current session. This can be useful when deciding whether or not we want
16739 to set some values in a header passed to the server.
16740
Cyril Bonté62ba8702014-04-22 23:52:25 +020016741sc_trackers(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020016742sc0_trackers([<table>]) : integer
16743sc1_trackers([<table>]) : integer
16744sc2_trackers([<table>]) : integer
Willy Tarreau2406db42012-12-09 12:16:43 +010016745 Returns the current amount of concurrent connections tracking the same
16746 tracked counters. This number is automatically incremented when tracking
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020016747 begins and decremented when tracking stops. It differs from sc0_conn_cur in
Willy Tarreau2406db42012-12-09 12:16:43 +010016748 that it does not rely on any stored information but on the table's reference
16749 count (the "use" value which is returned by "show table" on the CLI). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020016750 may sometimes be more suited for layer7 tracking. It can be used to tell a
16751 server how many concurrent connections there are from a given address for
16752 example.
Willy Tarreau2406db42012-12-09 12:16:43 +010016753
Willy Tarreau74ca5042013-06-11 23:12:07 +020016754so_id : integer
16755 Returns an integer containing the current listening socket's id. It is useful
16756 in frontends involving many "bind" lines, or to stick all users coming via a
16757 same socket to the same server.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016758
Jerome Magnineb421b22020-03-27 22:08:40 +010016759so_name : string
16760 Returns a string containing the current listening socket's name, as defined
16761 with name on a "bind" line. It can serve the same purposes as so_id but with
16762 strings instead of integers.
16763
Willy Tarreau74ca5042013-06-11 23:12:07 +020016764src : ip
Davor Ocelice9ed2812017-12-25 17:49:28 +010016765 This is the source IPv4 address of the client of the session. It is of type
Willy Tarreau74ca5042013-06-11 23:12:07 +020016766 IP and works on both IPv4 and IPv6 tables. On IPv6 tables, IPv4 addresses are
16767 mapped to their IPv6 equivalent, according to RFC 4291. Note that it is the
16768 TCP-level source address which is used, and not the address of a client
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016769 behind a proxy. However if the "accept-proxy" or "accept-netscaler-cip" bind
16770 directive is used, it can be the address of a client behind another
16771 PROXY-protocol compatible component for all rule sets except
Willy Tarreau64ded3d2019-01-23 10:02:15 +010016772 "tcp-request connection" which sees the real address. When the incoming
16773 connection passed through address translation or redirection involving
16774 connection tracking, the original destination address before the redirection
16775 will be reported. On Linux systems, the source and destination may seldom
16776 appear reversed if the nf_conntrack_tcp_loose sysctl is set, because a late
16777 response may reopen a timed out connection and switch what is believed to be
16778 the source and the destination.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016779
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010016780 Example:
16781 # add an HTTP header in requests with the originating address' country
16782 http-request set-header X-Country %[src,map_ip(geoip.lst)]
16783
Willy Tarreau74ca5042013-06-11 23:12:07 +020016784src_bytes_in_rate([<table>]) : integer
16785 Returns the average bytes rate from the incoming connection's source address
16786 in the current proxy's stick-table or in the designated stick-table, measured
16787 in amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020016788 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_in_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020016789
Willy Tarreau74ca5042013-06-11 23:12:07 +020016790src_bytes_out_rate([<table>]) : integer
16791 Returns the average bytes rate to the incoming connection's source address in
16792 the current proxy's stick-table or in the designated stick-table, measured in
Willy Tarreauc9705a12010-07-27 20:05:50 +020016793 amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020016794 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_out_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020016795
Willy Tarreau74ca5042013-06-11 23:12:07 +020016796src_clr_gpc0([<table>]) : integer
16797 Clears the first General Purpose Counter associated to the incoming
16798 connection's source address in the current proxy's stick-table or in the
16799 designated stick-table, and returns its previous value. If the address is not
16800 found, an entry is created and 0 is returned. This is typically used as a
16801 second ACL in an expression in order to mark a connection when a first ACL
16802 was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020016803
Jarno Huuskonen676f6222017-03-30 09:19:45 +030016804 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020016805 # block if 5 consecutive requests continue to come faster than 10 sess
16806 # per second, and reset the counter as soon as the traffic slows down.
16807 acl abuse src_http_req_rate gt 10
16808 acl kill src_inc_gpc0 gt 5
Willy Tarreau869948b2013-01-04 14:14:57 +010016809 acl save src_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020016810 tcp-request connection accept if !abuse save
16811 tcp-request connection reject if abuse kill
16812
Frédéric Lécaille6778b272018-01-29 15:22:53 +010016813src_clr_gpc1([<table>]) : integer
16814 Clears the second General Purpose Counter associated to the incoming
16815 connection's source address in the current proxy's stick-table or in the
16816 designated stick-table, and returns its previous value. If the address is not
16817 found, an entry is created and 0 is returned. This is typically used as a
16818 second ACL in an expression in order to mark a connection when a first ACL
16819 was verified.
16820
Willy Tarreau74ca5042013-06-11 23:12:07 +020016821src_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010016822 Returns the cumulative number of connections initiated from the current
Willy Tarreau74ca5042013-06-11 23:12:07 +020016823 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020016824 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020016825 See also sc/sc0/sc1/sc2_conn_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020016826
Willy Tarreau74ca5042013-06-11 23:12:07 +020016827src_conn_cur([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020016828 Returns the current amount of concurrent connections initiated from the
Willy Tarreau74ca5042013-06-11 23:12:07 +020016829 current incoming connection's source address in the current proxy's
16830 stick-table or in the designated stick-table. If the address is not found,
Willy Tarreau4d4149c2013-07-23 19:33:46 +020016831 zero is returned. See also sc/sc0/sc1/sc2_conn_cur.
Willy Tarreauc9705a12010-07-27 20:05:50 +020016832
Willy Tarreau74ca5042013-06-11 23:12:07 +020016833src_conn_rate([<table>]) : integer
16834 Returns the average connection rate from the incoming connection's source
16835 address in the current proxy's stick-table or in the designated stick-table,
16836 measured in amount of connections over the period configured in the table. If
Willy Tarreau4d4149c2013-07-23 19:33:46 +020016837 the address is not found, zero is returned. See also sc/sc0/sc1/sc2_conn_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020016838
Willy Tarreau74ca5042013-06-11 23:12:07 +020016839src_get_gpc0([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020016840 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau74ca5042013-06-11 23:12:07 +020016841 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020016842 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020016843 See also sc/sc0/sc1/sc2_get_gpc0 and src_inc_gpc0.
Willy Tarreauc9705a12010-07-27 20:05:50 +020016844
Frédéric Lécaille6778b272018-01-29 15:22:53 +010016845src_get_gpc1([<table>]) : integer
16846 Returns the value of the second General Purpose Counter associated to the
16847 incoming connection's source address in the current proxy's stick-table or in
16848 the designated stick-table. If the address is not found, zero is returned.
16849 See also sc/sc0/sc1/sc2_get_gpc1 and src_inc_gpc1.
16850
Thierry FOURNIER236657b2015-08-19 08:25:14 +020016851src_get_gpt0([<table>]) : integer
16852 Returns the value of the first General Purpose Tag associated to the
16853 incoming connection's source address in the current proxy's stick-table or in
16854 the designated stick-table. If the address is not found, zero is returned.
16855 See also sc/sc0/sc1/sc2_get_gpt0.
16856
Willy Tarreau74ca5042013-06-11 23:12:07 +020016857src_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020016858 Returns the average increment rate of the first General Purpose Counter
Willy Tarreau74ca5042013-06-11 23:12:07 +020016859 associated to the incoming connection's source address in the current proxy's
Willy Tarreauba2ffd12013-05-29 15:54:14 +020016860 stick-table or in the designated stick-table. It reports the frequency
16861 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020016862 sc/sc0/sc1/sc2_gpc0_rate, src_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
16863 that the "gpc0_rate" counter must be stored in the stick-table for a value to
16864 be returned, as "gpc0" only holds the event count.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020016865
Frédéric Lécaille6778b272018-01-29 15:22:53 +010016866src_gpc1_rate([<table>]) : integer
16867 Returns the average increment rate of the second General Purpose Counter
16868 associated to the incoming connection's source address in the current proxy's
16869 stick-table or in the designated stick-table. It reports the frequency
16870 which the gpc1 counter was incremented over the configured period. See also
16871 sc/sc0/sc1/sc2_gpc1_rate, src_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
16872 that the "gpc1_rate" counter must be stored in the stick-table for a value to
16873 be returned, as "gpc1" only holds the event count.
16874
Willy Tarreau74ca5042013-06-11 23:12:07 +020016875src_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010016876 Returns the cumulative number of HTTP errors from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020016877 source address in the current proxy's stick-table or in the designated
Willy Tarreauc9705a12010-07-27 20:05:50 +020016878 stick-table. This includes the both request errors and 4xx error responses.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020016879 See also sc/sc0/sc1/sc2_http_err_cnt. If the address is not found, zero is
Willy Tarreau74ca5042013-06-11 23:12:07 +020016880 returned.
Willy Tarreauc9705a12010-07-27 20:05:50 +020016881
Willy Tarreau74ca5042013-06-11 23:12:07 +020016882src_http_err_rate([<table>]) : integer
16883 Returns the average rate of HTTP errors from the incoming connection's source
16884 address in the current proxy's stick-table or in the designated stick-table,
16885 measured in amount of errors over the period configured in the table. This
16886 includes the both request errors and 4xx error responses. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020016887 not found, zero is returned. See also sc/sc0/sc1/sc2_http_err_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020016888
Willy Tarreau74ca5042013-06-11 23:12:07 +020016889src_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010016890 Returns the cumulative number of HTTP requests from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020016891 source address in the current proxy's stick-table or in the designated stick-
16892 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020016893 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020016894
Willy Tarreau74ca5042013-06-11 23:12:07 +020016895src_http_req_rate([<table>]) : integer
16896 Returns the average rate of HTTP requests from the incoming connection's
16897 source address in the current proxy's stick-table or in the designated stick-
16898 table, measured in amount of requests over the period configured in the
Willy Tarreauc9705a12010-07-27 20:05:50 +020016899 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020016900 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020016901
Willy Tarreau74ca5042013-06-11 23:12:07 +020016902src_inc_gpc0([<table>]) : integer
16903 Increments the first General Purpose Counter associated to the incoming
16904 connection's source address in the current proxy's stick-table or in the
16905 designated stick-table, and returns its new value. If the address is not
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020016906 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc0.
Willy Tarreau74ca5042013-06-11 23:12:07 +020016907 This is typically used as a second ACL in an expression in order to mark a
16908 connection when a first ACL was verified :
Willy Tarreauc9705a12010-07-27 20:05:50 +020016909
Jarno Huuskonen676f6222017-03-30 09:19:45 +030016910 Example:
Willy Tarreauc9705a12010-07-27 20:05:50 +020016911 acl abuse src_http_req_rate gt 10
Willy Tarreau869948b2013-01-04 14:14:57 +010016912 acl kill src_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020016913 tcp-request connection reject if abuse kill
Willy Tarreauc9705a12010-07-27 20:05:50 +020016914
Frédéric Lécaille6778b272018-01-29 15:22:53 +010016915src_inc_gpc1([<table>]) : integer
16916 Increments the second General Purpose Counter associated to the incoming
16917 connection's source address in the current proxy's stick-table or in the
16918 designated stick-table, and returns its new value. If the address is not
16919 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc1.
16920 This is typically used as a second ACL in an expression in order to mark a
16921 connection when a first ACL was verified.
16922
Willy Tarreau16e01562016-08-09 16:46:18 +020016923src_is_local : boolean
16924 Returns true if the source address of the incoming connection is local to the
16925 system, or false if the address doesn't exist on the system, meaning that it
16926 comes from a remote machine. Note that UNIX addresses are considered local.
16927 It can be useful to apply certain access restrictions based on where the
Davor Ocelice9ed2812017-12-25 17:49:28 +010016928 client comes from (e.g. require auth or https for remote machines). Please
Willy Tarreau16e01562016-08-09 16:46:18 +020016929 note that the check involves a few system calls, so it's better to do it only
16930 once per connection.
16931
Willy Tarreau74ca5042013-06-11 23:12:07 +020016932src_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020016933 Returns the total amount of data received from the incoming connection's
16934 source address in the current proxy's stick-table or in the designated
16935 stick-table, measured in kilobytes. If the address is not found, zero is
16936 returned. The test is currently performed on 32-bit integers, which limits
16937 values to 4 terabytes. See also sc/sc0/sc1/sc2_kbytes_in.
Willy Tarreauc9705a12010-07-27 20:05:50 +020016938
Willy Tarreau74ca5042013-06-11 23:12:07 +020016939src_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020016940 Returns the total amount of data sent to the incoming connection's source
16941 address in the current proxy's stick-table or in the designated stick-table,
16942 measured in kilobytes. If the address is not found, zero is returned. The
16943 test is currently performed on 32-bit integers, which limits values to 4
16944 terabytes. See also sc/sc0/sc1/sc2_kbytes_out.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020016945
Willy Tarreau74ca5042013-06-11 23:12:07 +020016946src_port : integer
16947 Returns an integer value corresponding to the TCP source port of the
16948 connection on the client side, which is the port the client connected from.
16949 Usage of this function is very limited as modern protocols do not care much
16950 about source ports nowadays.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010016951
Willy Tarreau74ca5042013-06-11 23:12:07 +020016952src_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010016953 Returns the cumulative number of connections initiated from the incoming
Willy Tarreauc9705a12010-07-27 20:05:50 +020016954 connection's source IPv4 address in the current proxy's stick-table or in the
16955 designated stick-table, that were transformed into sessions, which means that
16956 they were accepted by "tcp-request" rules. If the address is not found, zero
Willy Tarreau4d4149c2013-07-23 19:33:46 +020016957 is returned. See also sc/sc0/sc1/sc2_sess_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020016958
Willy Tarreau74ca5042013-06-11 23:12:07 +020016959src_sess_rate([<table>]) : integer
16960 Returns the average session rate from the incoming connection's source
16961 address in the current proxy's stick-table or in the designated stick-table,
16962 measured in amount of sessions over the period configured in the table. A
16963 session is a connection that went past the early "tcp-request" rules. If the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020016964 address is not found, zero is returned. See also sc/sc0/sc1/sc2_sess_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020016965
Willy Tarreau74ca5042013-06-11 23:12:07 +020016966src_updt_conn_cnt([<table>]) : integer
16967 Creates or updates the entry associated to the incoming connection's source
16968 address in the current proxy's stick-table or in the designated stick-table.
16969 This table must be configured to store the "conn_cnt" data type, otherwise
16970 the match will be ignored. The current count is incremented by one, and the
16971 expiration timer refreshed. The updated count is returned, so this match
16972 can't return zero. This was used to reject service abusers based on their
16973 source address. Note: it is recommended to use the more complete "track-sc*"
16974 actions in "tcp-request" rules instead.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020016975
16976 Example :
16977 # This frontend limits incoming SSH connections to 3 per 10 second for
16978 # each source address, and rejects excess connections until a 10 second
16979 # silence is observed. At most 20 addresses are tracked.
16980 listen ssh
16981 bind :22
16982 mode tcp
16983 maxconn 100
Willy Tarreauc9705a12010-07-27 20:05:50 +020016984 stick-table type ip size 20 expire 10s store conn_cnt
Willy Tarreau74ca5042013-06-11 23:12:07 +020016985 tcp-request content reject if { src_updt_conn_cnt gt 3 }
Willy Tarreaua975b8f2010-06-05 19:13:27 +020016986 server local 127.0.0.1:22
16987
Willy Tarreau74ca5042013-06-11 23:12:07 +020016988srv_id : integer
16989 Returns an integer containing the server's id when processing the response.
16990 While it's almost only used with ACLs, it may be used for logging or
Christopher Fauletd1b44642020-04-30 09:51:15 +020016991 debugging. It can also be used in a tcp-check or an http-check ruleset.
Hervé COMMOWICKdaa824e2011-08-05 12:09:44 +020016992
vkill1dfd1652019-10-30 16:58:14 +080016993srv_name : string
16994 Returns a string containing the server's name when processing the response.
16995 While it's almost only used with ACLs, it may be used for logging or
Christopher Fauletd1b44642020-04-30 09:51:15 +020016996 debugging. It can also be used in a tcp-check or an http-check ruleset.
vkill1dfd1652019-10-30 16:58:14 +080016997
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200169987.3.4. Fetching samples at Layer 5
Willy Tarreau74ca5042013-06-11 23:12:07 +020016999----------------------------------
Willy Tarreau0b1cd942010-05-16 22:18:27 +020017000
Willy Tarreau74ca5042013-06-11 23:12:07 +020017001The layer 5 usually describes just the session layer which in haproxy is
17002closest to the session once all the connection handshakes are finished, but
17003when no content is yet made available. The fetch methods described here are
17004usable as low as the "tcp-request content" rule sets unless they require some
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017005future information. Those generally include the results of SSL negotiations.
Willy Tarreauc735a072011-03-29 00:57:02 +020017006
Ben Shillitof25e8e52016-12-02 14:25:37 +00001700751d.all(<prop>[,<prop>*]) : string
17008 Returns values for the properties requested as a string, where values are
17009 separated by the delimiter specified with "51degrees-property-separator".
17010 The device is identified using all the important HTTP headers from the
17011 request. The function can be passed up to five property names, and if a
17012 property name can't be found, the value "NoData" is returned.
17013
17014 Example :
17015 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request
17016 # containing the three properties requested using all relevant headers from
17017 # the request.
17018 frontend http-in
17019 bind *:8081
17020 default_backend servers
17021 http-request set-header X-51D-DeviceTypeMobileTablet \
17022 %[51d.all(DeviceType,IsMobile,IsTablet)]
17023
Emeric Brun645ae792014-04-30 14:21:06 +020017024ssl_bc : boolean
17025 Returns true when the back connection was made via an SSL/TLS transport
17026 layer and is locally deciphered. This means the outgoing connection was made
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017027 other a server with the "ssl" option. It can be used in a tcp-check or an
17028 http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017029
17030ssl_bc_alg_keysize : integer
17031 Returns the symmetric cipher key size supported in bits when the outgoing
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017032 connection was made over an SSL/TLS transport layer. It can be used in a
17033 tcp-check or an http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017034
Olivier Houchard6b77f492018-11-22 18:18:29 +010017035ssl_bc_alpn : string
17036 This extracts the Application Layer Protocol Negotiation field from an
17037 outgoing connection made via a TLS transport layer.
Michael Prokop4438c602019-05-24 10:25:45 +020017038 The result is a string containing the protocol name negotiated with the
Olivier Houchard6b77f492018-11-22 18:18:29 +010017039 server. The SSL library must have been built with support for TLS
17040 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
17041 not advertised unless the "alpn" keyword on the "server" line specifies a
17042 protocol list. Also, nothing forces the server to pick a protocol from this
17043 list, any other one may be requested. The TLS ALPN extension is meant to
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017044 replace the TLS NPN extension. See also "ssl_bc_npn". It can be used in a
17045 tcp-check or an http-check ruleset.
Olivier Houchard6b77f492018-11-22 18:18:29 +010017046
Emeric Brun645ae792014-04-30 14:21:06 +020017047ssl_bc_cipher : string
17048 Returns the name of the used cipher when the outgoing connection was made
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017049 over an SSL/TLS transport layer. It can be used in a tcp-check or an
17050 http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017051
Patrick Hemmer65674662019-06-04 08:13:03 -040017052ssl_bc_client_random : binary
17053 Returns the client random of the back connection when the incoming connection
17054 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
17055 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017056 It can be used in a tcp-check or an http-check ruleset.
Patrick Hemmer65674662019-06-04 08:13:03 -040017057
Emeric Brun74f7ffa2018-02-19 16:14:12 +010017058ssl_bc_is_resumed : boolean
17059 Returns true when the back connection was made over an SSL/TLS transport
17060 layer and the newly created SSL session was resumed using a cached
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017061 session or a TLS ticket. It can be used in a tcp-check or an http-check
17062 ruleset.
Emeric Brun74f7ffa2018-02-19 16:14:12 +010017063
Olivier Houchard6b77f492018-11-22 18:18:29 +010017064ssl_bc_npn : string
17065 This extracts the Next Protocol Negotiation field from an outgoing connection
17066 made via a TLS transport layer. The result is a string containing the
Michael Prokop4438c602019-05-24 10:25:45 +020017067 protocol name negotiated with the server . The SSL library must have been
Olivier Houchard6b77f492018-11-22 18:18:29 +010017068 built with support for TLS extensions enabled (check haproxy -vv). Note that
17069 the TLS NPN extension is not advertised unless the "npn" keyword on the
17070 "server" line specifies a protocol list. Also, nothing forces the server to
17071 pick a protocol from this list, any other one may be used. Please note that
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017072 the TLS NPN extension was replaced with ALPN. It can be used in a tcp-check
17073 or an http-check ruleset.
Olivier Houchard6b77f492018-11-22 18:18:29 +010017074
Emeric Brun645ae792014-04-30 14:21:06 +020017075ssl_bc_protocol : string
17076 Returns the name of the used protocol when the outgoing connection was made
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017077 over an SSL/TLS transport layer. It can be used in a tcp-check or an
17078 http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017079
Emeric Brunb73a9b02014-04-30 18:49:19 +020017080ssl_bc_unique_id : binary
Emeric Brun645ae792014-04-30 14:21:06 +020017081 When the outgoing connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020017082 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017083 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64". It
17084 can be used in a tcp-check or an http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017085
Patrick Hemmer65674662019-06-04 08:13:03 -040017086ssl_bc_server_random : binary
17087 Returns the server random of the back connection when the incoming connection
17088 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
17089 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017090 It can be used in a tcp-check or an http-check ruleset.
Patrick Hemmer65674662019-06-04 08:13:03 -040017091
Emeric Brun645ae792014-04-30 14:21:06 +020017092ssl_bc_session_id : binary
17093 Returns the SSL ID of the back connection when the outgoing connection was
17094 made over an SSL/TLS transport layer. It is useful to log if we want to know
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017095 if session was reused or not. It can be used in a tcp-check or an http-check
17096 ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017097
Patrick Hemmere0275472018-04-28 19:15:51 -040017098ssl_bc_session_key : binary
17099 Returns the SSL session master key of the back connection when the outgoing
17100 connection was made over an SSL/TLS transport layer. It is useful to decrypt
17101 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017102 BoringSSL. It can be used in a tcp-check or an http-check ruleset.
Patrick Hemmere0275472018-04-28 19:15:51 -040017103
Emeric Brun645ae792014-04-30 14:21:06 +020017104ssl_bc_use_keysize : integer
17105 Returns the symmetric cipher key size used in bits when the outgoing
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017106 connection was made over an SSL/TLS transport layer. It can be used in a
17107 tcp-check or an http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017108
Willy Tarreau74ca5042013-06-11 23:12:07 +020017109ssl_c_ca_err : integer
17110 When the incoming connection was made over an SSL/TLS transport layer,
17111 returns the ID of the first error detected during verification of the client
17112 certificate at depth > 0, or 0 if no error was encountered during this
17113 verification process. Please refer to your SSL library's documentation to
17114 find the exhaustive list of error codes.
Willy Tarreauc735a072011-03-29 00:57:02 +020017115
Willy Tarreau74ca5042013-06-11 23:12:07 +020017116ssl_c_ca_err_depth : integer
17117 When the incoming connection was made over an SSL/TLS transport layer,
17118 returns the depth in the CA chain of the first error detected during the
17119 verification of the client certificate. If no error is encountered, 0 is
17120 returned.
Willy Tarreau0ba27502007-12-24 16:55:16 +010017121
Emeric Brun43e79582014-10-29 19:03:26 +010017122ssl_c_der : binary
17123 Returns the DER formatted certificate presented by the client when the
17124 incoming connection was made over an SSL/TLS transport layer. When used for
17125 an ACL, the value(s) to match against can be passed in hexadecimal form.
17126
William Dauchya598b502020-08-06 18:11:38 +020017127ssl_c_der_chain : binary
17128 Returns the DER formatted chain certificate presented by the client when the
17129 incoming connection was made over an SSL/TLS transport layer. When used for
17130 an ACL, the value(s) to match against can be passed in hexadecimal form. One
17131 can parse the result with any lib accepting ASN.1 DER data. It currentlly
17132 does not support resumed sessions.
17133
Willy Tarreau74ca5042013-06-11 23:12:07 +020017134ssl_c_err : integer
17135 When the incoming connection was made over an SSL/TLS transport layer,
17136 returns the ID of the first error detected during verification at depth 0, or
17137 0 if no error was encountered during this verification process. Please refer
17138 to your SSL library's documentation to find the exhaustive list of error
17139 codes.
Willy Tarreau62644772008-07-16 18:36:06 +020017140
Elliot Otchet71f82972020-01-15 08:12:14 -050017141ssl_c_i_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020017142 When the incoming connection was made over an SSL/TLS transport layer,
17143 returns the full distinguished name of the issuer of the certificate
17144 presented by the client when no <entry> is specified, or the value of the
17145 first given entry found from the beginning of the DN. If a positive/negative
17146 occurrence number is specified as the optional second argument, it returns
17147 the value of the nth given entry value from the beginning/end of the DN.
17148 For instance, "ssl_c_i_dn(OU,2)" the second organization unit, and
17149 "ssl_c_i_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050017150 The <format> parameter allows you to receive the DN suitable for
17151 consumption by different protocols. Currently supported is rfc2253 for
17152 LDAP v3.
17153 If you'd like to modify the format only you can specify an empty string
17154 and zero for the first two parameters. Example: ssl_c_i_dn(,0,rfc2253)
Willy Tarreau62644772008-07-16 18:36:06 +020017155
Willy Tarreau74ca5042013-06-11 23:12:07 +020017156ssl_c_key_alg : string
17157 Returns the name of the algorithm used to generate the key of the certificate
17158 presented by the client when the incoming connection was made over an SSL/TLS
17159 transport layer.
Willy Tarreau62644772008-07-16 18:36:06 +020017160
Willy Tarreau74ca5042013-06-11 23:12:07 +020017161ssl_c_notafter : string
17162 Returns the end date presented by the client as a formatted string
17163 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
17164 transport layer.
Emeric Brunbede3d02009-06-30 17:54:00 +020017165
Willy Tarreau74ca5042013-06-11 23:12:07 +020017166ssl_c_notbefore : string
17167 Returns the start date presented by the client as a formatted string
17168 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
17169 transport layer.
Willy Tarreaub6672b52011-12-12 17:23:41 +010017170
Elliot Otchet71f82972020-01-15 08:12:14 -050017171ssl_c_s_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020017172 When the incoming connection was made over an SSL/TLS transport layer,
17173 returns the full distinguished name of the subject of the certificate
17174 presented by the client when no <entry> is specified, or the value of the
17175 first given entry found from the beginning of the DN. If a positive/negative
17176 occurrence number is specified as the optional second argument, it returns
17177 the value of the nth given entry value from the beginning/end of the DN.
17178 For instance, "ssl_c_s_dn(OU,2)" the second organization unit, and
17179 "ssl_c_s_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050017180 The <format> parameter allows you to receive the DN suitable for
17181 consumption by different protocols. Currently supported is rfc2253 for
17182 LDAP v3.
17183 If you'd like to modify the format only you can specify an empty string
17184 and zero for the first two parameters. Example: ssl_c_s_dn(,0,rfc2253)
Willy Tarreaub6672b52011-12-12 17:23:41 +010017185
Willy Tarreau74ca5042013-06-11 23:12:07 +020017186ssl_c_serial : binary
17187 Returns the serial of the certificate presented by the client when the
17188 incoming connection was made over an SSL/TLS transport layer. When used for
17189 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun2525b6b2012-10-18 15:59:43 +020017190
Willy Tarreau74ca5042013-06-11 23:12:07 +020017191ssl_c_sha1 : binary
17192 Returns the SHA-1 fingerprint of the certificate presented by the client when
17193 the incoming connection was made over an SSL/TLS transport layer. This can be
17194 used to stick a client to a server, or to pass this information to a server.
Willy Tarreau2d0caa32014-07-02 19:01:22 +020017195 Note that the output is binary, so if you want to pass that signature to the
17196 server, you need to encode it in hex or base64, such as in the example below:
17197
Jarno Huuskonen676f6222017-03-30 09:19:45 +030017198 Example:
Willy Tarreau2d0caa32014-07-02 19:01:22 +020017199 http-request set-header X-SSL-Client-SHA1 %[ssl_c_sha1,hex]
Emeric Brun2525b6b2012-10-18 15:59:43 +020017200
Willy Tarreau74ca5042013-06-11 23:12:07 +020017201ssl_c_sig_alg : string
17202 Returns the name of the algorithm used to sign the certificate presented by
17203 the client when the incoming connection was made over an SSL/TLS transport
17204 layer.
Emeric Brun87855892012-10-17 17:39:35 +020017205
Willy Tarreau74ca5042013-06-11 23:12:07 +020017206ssl_c_used : boolean
17207 Returns true if current SSL session uses a client certificate even if current
17208 connection uses SSL session resumption. See also "ssl_fc_has_crt".
Emeric Brun7f56e742012-10-19 18:15:40 +020017209
Willy Tarreau74ca5042013-06-11 23:12:07 +020017210ssl_c_verify : integer
17211 Returns the verify result error ID when the incoming connection was made over
17212 an SSL/TLS transport layer, otherwise zero if no error is encountered. Please
17213 refer to your SSL library's documentation for an exhaustive list of error
17214 codes.
Emeric Brunce5ad802012-10-22 14:11:22 +020017215
Willy Tarreau74ca5042013-06-11 23:12:07 +020017216ssl_c_version : integer
17217 Returns the version of the certificate presented by the client when the
17218 incoming connection was made over an SSL/TLS transport layer.
Emeric Brunce5ad802012-10-22 14:11:22 +020017219
Emeric Brun43e79582014-10-29 19:03:26 +010017220ssl_f_der : binary
17221 Returns the DER formatted certificate presented by the frontend when the
17222 incoming connection was made over an SSL/TLS transport layer. When used for
17223 an ACL, the value(s) to match against can be passed in hexadecimal form.
17224
Elliot Otchet71f82972020-01-15 08:12:14 -050017225ssl_f_i_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020017226 When the incoming connection was made over an SSL/TLS transport layer,
17227 returns the full distinguished name of the issuer of the certificate
17228 presented by the frontend when no <entry> is specified, or the value of the
17229 first given entry found from the beginning of the DN. If a positive/negative
Emeric Brun87855892012-10-17 17:39:35 +020017230 occurrence number is specified as the optional second argument, it returns
Willy Tarreau74ca5042013-06-11 23:12:07 +020017231 the value of the nth given entry value from the beginning/end of the DN.
17232 For instance, "ssl_f_i_dn(OU,2)" the second organization unit, and
17233 "ssl_f_i_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050017234 The <format> parameter allows you to receive the DN suitable for
17235 consumption by different protocols. Currently supported is rfc2253 for
17236 LDAP v3.
17237 If you'd like to modify the format only you can specify an empty string
17238 and zero for the first two parameters. Example: ssl_f_i_dn(,0,rfc2253)
Emeric Brun87855892012-10-17 17:39:35 +020017239
Willy Tarreau74ca5042013-06-11 23:12:07 +020017240ssl_f_key_alg : string
17241 Returns the name of the algorithm used to generate the key of the certificate
17242 presented by the frontend when the incoming connection was made over an
17243 SSL/TLS transport layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020017244
Willy Tarreau74ca5042013-06-11 23:12:07 +020017245ssl_f_notafter : string
17246 Returns the end date presented by the frontend as a formatted string
17247 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
17248 transport layer.
Emeric Brun2525b6b2012-10-18 15:59:43 +020017249
Willy Tarreau74ca5042013-06-11 23:12:07 +020017250ssl_f_notbefore : string
17251 Returns the start date presented by the frontend as a formatted string
17252 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
17253 transport layer.
Emeric Brun87855892012-10-17 17:39:35 +020017254
Elliot Otchet71f82972020-01-15 08:12:14 -050017255ssl_f_s_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020017256 When the incoming connection was made over an SSL/TLS transport layer,
17257 returns the full distinguished name of the subject of the certificate
17258 presented by the frontend when no <entry> is specified, or the value of the
17259 first given entry found from the beginning of the DN. If a positive/negative
17260 occurrence number is specified as the optional second argument, it returns
17261 the value of the nth given entry value from the beginning/end of the DN.
17262 For instance, "ssl_f_s_dn(OU,2)" the second organization unit, and
17263 "ssl_f_s_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050017264 The <format> parameter allows you to receive the DN suitable for
17265 consumption by different protocols. Currently supported is rfc2253 for
17266 LDAP v3.
17267 If you'd like to modify the format only you can specify an empty string
17268 and zero for the first two parameters. Example: ssl_f_s_dn(,0,rfc2253)
Emeric Brunce5ad802012-10-22 14:11:22 +020017269
Willy Tarreau74ca5042013-06-11 23:12:07 +020017270ssl_f_serial : binary
17271 Returns the serial of the certificate presented by the frontend when the
17272 incoming connection was made over an SSL/TLS transport layer. When used for
17273 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun87855892012-10-17 17:39:35 +020017274
Emeric Brun55f4fa82014-04-30 17:11:25 +020017275ssl_f_sha1 : binary
17276 Returns the SHA-1 fingerprint of the certificate presented by the frontend
17277 when the incoming connection was made over an SSL/TLS transport layer. This
17278 can be used to know which certificate was chosen using SNI.
17279
Willy Tarreau74ca5042013-06-11 23:12:07 +020017280ssl_f_sig_alg : string
17281 Returns the name of the algorithm used to sign the certificate presented by
17282 the frontend when the incoming connection was made over an SSL/TLS transport
17283 layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020017284
Willy Tarreau74ca5042013-06-11 23:12:07 +020017285ssl_f_version : integer
17286 Returns the version of the certificate presented by the frontend when the
17287 incoming connection was made over an SSL/TLS transport layer.
17288
17289ssl_fc : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020017290 Returns true when the front connection was made via an SSL/TLS transport
17291 layer and is locally deciphered. This means it has matched a socket declared
17292 with a "bind" line having the "ssl" option.
17293
Willy Tarreau74ca5042013-06-11 23:12:07 +020017294 Example :
17295 # This passes "X-Proto: https" to servers when client connects over SSL
17296 listen http-https
17297 bind :80
17298 bind :443 ssl crt /etc/haproxy.pem
17299 http-request add-header X-Proto https if { ssl_fc }
17300
17301ssl_fc_alg_keysize : integer
17302 Returns the symmetric cipher key size supported in bits when the incoming
17303 connection was made over an SSL/TLS transport layer.
17304
17305ssl_fc_alpn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017306 This extracts the Application Layer Protocol Negotiation field from an
Willy Tarreau74ca5042013-06-11 23:12:07 +020017307 incoming connection made via a TLS transport layer and locally deciphered by
17308 haproxy. The result is a string containing the protocol name advertised by
17309 the client. The SSL library must have been built with support for TLS
17310 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
17311 not advertised unless the "alpn" keyword on the "bind" line specifies a
17312 protocol list. Also, nothing forces the client to pick a protocol from this
17313 list, any other one may be requested. The TLS ALPN extension is meant to
17314 replace the TLS NPN extension. See also "ssl_fc_npn".
17315
Willy Tarreau74ca5042013-06-11 23:12:07 +020017316ssl_fc_cipher : string
17317 Returns the name of the used cipher when the incoming connection was made
17318 over an SSL/TLS transport layer.
Willy Tarreauab861d32013-04-02 02:30:41 +020017319
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010017320ssl_fc_cipherlist_bin : binary
17321 Returns the binary form of the client hello cipher list. The maximum returned
17322 value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010017323 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010017324
17325ssl_fc_cipherlist_hex : string
17326 Returns the binary form of the client hello cipher list encoded as
17327 hexadecimal. The maximum returned value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010017328 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010017329
17330ssl_fc_cipherlist_str : string
17331 Returns the decoded text form of the client hello cipher list. The maximum
17332 number of ciphers returned is according with the value of
17333 "tune.ssl.capture-cipherlist-size". Note that this sample-fetch is only
Davor Ocelice9ed2812017-12-25 17:49:28 +010017334 available with OpenSSL >= 1.0.2. If the function is not enabled, this
Emmanuel Hocdetddcde192017-09-01 17:32:08 +020017335 sample-fetch returns the hash like "ssl_fc_cipherlist_xxh".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010017336
17337ssl_fc_cipherlist_xxh : integer
17338 Returns a xxh64 of the cipher list. This hash can be return only is the value
17339 "tune.ssl.capture-cipherlist-size" is set greater than 0, however the hash
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010017340 take in account all the data of the cipher list.
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010017341
Patrick Hemmer65674662019-06-04 08:13:03 -040017342ssl_fc_client_random : binary
17343 Returns the client random of the front connection when the incoming connection
17344 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
17345 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
17346
William Lallemand7d42ef52020-07-06 11:41:30 +020017347ssl_fc_client_early_traffic_secret : string
17348 Return the CLIENT_EARLY_TRAFFIC_SECRET as an hexadecimal string for the
17349 front connection when the incoming connection was made over a TLS 1.3
17350 transport layer.
17351 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
17352 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
17353 activated with "tune.ssl.keylog on" in the global section. See also
17354 "tune.ssl.keylog"
17355
17356ssl_fc_client_handshake_traffic_secret : string
17357 Return the CLIENT_HANDSHAKE_TRAFFIC_SECRET as an hexadecimal string for the
17358 front connection when the incoming connection was made over a TLS 1.3
17359 transport layer.
17360 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
17361 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
17362 activated with "tune.ssl.keylog on" in the global section. See also
17363 "tune.ssl.keylog"
17364
17365ssl_fc_client_traffic_secret_0 : string
17366 Return the CLIENT_TRAFFIC_SECRET_0 as an hexadecimal string for the
17367 front connection when the incoming connection was made over a TLS 1.3
17368 transport layer.
17369 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
17370 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
17371 activated with "tune.ssl.keylog on" in the global section. See also
17372 "tune.ssl.keylog"
17373
17374ssl_fc_exporter_secret : string
17375 Return the EXPORTER_SECRET as an hexadecimal string for the
17376 front connection when the incoming connection was made over a TLS 1.3
17377 transport layer.
17378 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
17379 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
17380 activated with "tune.ssl.keylog on" in the global section. See also
17381 "tune.ssl.keylog"
17382
17383ssl_fc_early_exporter_secret : string
17384 Return the EARLY_EXPORTER_SECRET as an hexadecimal string for the
17385 front connection when the incoming connection was made over an TLS 1.3
17386 transport layer.
17387 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
17388 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
17389 activated with "tune.ssl.keylog on" in the global section. See also
17390 "tune.ssl.keylog"
17391
Willy Tarreau74ca5042013-06-11 23:12:07 +020017392ssl_fc_has_crt : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020017393 Returns true if a client certificate is present in an incoming connection over
17394 SSL/TLS transport layer. Useful if 'verify' statement is set to 'optional'.
Emeric Brun9143d372012-12-20 15:44:16 +010017395 Note: on SSL session resumption with Session ID or TLS ticket, client
17396 certificate is not present in the current connection but may be retrieved
17397 from the cache or the ticket. So prefer "ssl_c_used" if you want to check if
17398 current SSL session uses a client certificate.
Emeric Brun2525b6b2012-10-18 15:59:43 +020017399
Olivier Houchardccaa7de2017-10-02 11:51:03 +020017400ssl_fc_has_early : boolean
17401 Returns true if early data were sent, and the handshake didn't happen yet. As
17402 it has security implications, it is useful to be able to refuse those, or
17403 wait until the handshake happened.
17404
Willy Tarreau74ca5042013-06-11 23:12:07 +020017405ssl_fc_has_sni : boolean
17406 This checks for the presence of a Server Name Indication TLS extension (SNI)
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020017407 in an incoming connection was made over an SSL/TLS transport layer. Returns
17408 true when the incoming connection presents a TLS SNI field. This requires
John Roeslerfb2fce12019-07-10 15:45:51 -050017409 that the SSL library is built with support for TLS extensions enabled (check
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020017410 haproxy -vv).
Willy Tarreau7875d092012-09-10 08:20:03 +020017411
Nenad Merdanovic1516fe32016-05-17 03:31:21 +020017412ssl_fc_is_resumed : boolean
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020017413 Returns true if the SSL/TLS session has been resumed through the use of
Jérôme Magnin4a326cb2018-01-15 14:01:17 +010017414 SSL session cache or TLS tickets on an incoming connection over an SSL/TLS
17415 transport layer.
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020017416
Willy Tarreau74ca5042013-06-11 23:12:07 +020017417ssl_fc_npn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017418 This extracts the Next Protocol Negotiation field from an incoming connection
Willy Tarreau74ca5042013-06-11 23:12:07 +020017419 made via a TLS transport layer and locally deciphered by haproxy. The result
17420 is a string containing the protocol name advertised by the client. The SSL
17421 library must have been built with support for TLS extensions enabled (check
17422 haproxy -vv). Note that the TLS NPN extension is not advertised unless the
17423 "npn" keyword on the "bind" line specifies a protocol list. Also, nothing
17424 forces the client to pick a protocol from this list, any other one may be
17425 requested. Please note that the TLS NPN extension was replaced with ALPN.
Willy Tarreaua33c6542012-10-15 13:19:06 +020017426
Willy Tarreau74ca5042013-06-11 23:12:07 +020017427ssl_fc_protocol : string
17428 Returns the name of the used protocol when the incoming connection was made
17429 over an SSL/TLS transport layer.
Willy Tarreau7875d092012-09-10 08:20:03 +020017430
Emeric Brunb73a9b02014-04-30 18:49:19 +020017431ssl_fc_unique_id : binary
David Sc1ad52e2014-04-08 18:48:47 -040017432 When the incoming connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020017433 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
17434 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
David Sc1ad52e2014-04-08 18:48:47 -040017435
William Lallemand7d42ef52020-07-06 11:41:30 +020017436ssl_fc_server_handshake_traffic_secret : string
17437 Return the SERVER_HANDSHAKE_TRAFFIC_SECRET as an hexadecimal string for the
17438 front connection when the incoming connection was made over a TLS 1.3
17439 transport layer.
17440 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
17441 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
17442 activated with "tune.ssl.keylog on" in the global section. See also
17443 "tune.ssl.keylog"
17444
17445ssl_fc_server_traffic_secret_0 : string
17446 Return the SERVER_TRAFFIC_SECRET_0 as an hexadecimal string for the
17447 front connection when the incoming connection was made over an TLS 1.3
17448 transport layer.
17449 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
17450 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
17451 activated with "tune.ssl.keylog on" in the global section. See also
17452 "tune.ssl.keylog"
17453
Patrick Hemmer65674662019-06-04 08:13:03 -040017454ssl_fc_server_random : binary
17455 Returns the server random of the front connection when the incoming connection
17456 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
17457 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
17458
Willy Tarreau74ca5042013-06-11 23:12:07 +020017459ssl_fc_session_id : binary
17460 Returns the SSL ID of the front connection when the incoming connection was
17461 made over an SSL/TLS transport layer. It is useful to stick a given client to
17462 a server. It is important to note that some browsers refresh their session ID
17463 every few minutes.
Willy Tarreau7875d092012-09-10 08:20:03 +020017464
Patrick Hemmere0275472018-04-28 19:15:51 -040017465ssl_fc_session_key : binary
17466 Returns the SSL session master key of the front connection when the incoming
17467 connection was made over an SSL/TLS transport layer. It is useful to decrypt
17468 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
17469 BoringSSL.
17470
17471
Willy Tarreau74ca5042013-06-11 23:12:07 +020017472ssl_fc_sni : string
17473 This extracts the Server Name Indication TLS extension (SNI) field from an
17474 incoming connection made via an SSL/TLS transport layer and locally
17475 deciphered by haproxy. The result (when present) typically is a string
17476 matching the HTTPS host name (253 chars or less). The SSL library must have
17477 been built with support for TLS extensions enabled (check haproxy -vv).
17478
17479 This fetch is different from "req_ssl_sni" above in that it applies to the
17480 connection being deciphered by haproxy and not to SSL contents being blindly
17481 forwarded. See also "ssl_fc_sni_end" and "ssl_fc_sni_reg" below. This
John Roeslerfb2fce12019-07-10 15:45:51 -050017482 requires that the SSL library is built with support for TLS extensions
Cyril Bonté9c1eb1e2012-10-09 22:45:34 +020017483 enabled (check haproxy -vv).
Willy Tarreau62644772008-07-16 18:36:06 +020017484
Willy Tarreau74ca5042013-06-11 23:12:07 +020017485 ACL derivatives :
Willy Tarreau74ca5042013-06-11 23:12:07 +020017486 ssl_fc_sni_end : suffix match
17487 ssl_fc_sni_reg : regex match
Emeric Brun589fcad2012-10-16 14:13:26 +020017488
Willy Tarreau74ca5042013-06-11 23:12:07 +020017489ssl_fc_use_keysize : integer
17490 Returns the symmetric cipher key size used in bits when the incoming
17491 connection was made over an SSL/TLS transport layer.
Willy Tarreaub6fb4202008-07-20 11:18:28 +020017492
William Lallemandbfa3e812020-06-25 20:07:18 +020017493ssl_s_der : binary
17494 Returns the DER formatted certificate presented by the server when the
17495 outgoing connection was made over an SSL/TLS transport layer. When used for
17496 an ACL, the value(s) to match against can be passed in hexadecimal form.
17497
William Dauchya598b502020-08-06 18:11:38 +020017498ssl_s_chain_der : binary
17499 Returns the DER formatted chain certificate presented by the server when the
17500 outgoing connection was made over an SSL/TLS transport layer. When used for
17501 an ACL, the value(s) to match against can be passed in hexadecimal form. One
17502 can parse the result with any lib accepting ASN.1 DER data. It currentlly
17503 does not support resumed sessions.
17504
William Lallemandbfa3e812020-06-25 20:07:18 +020017505ssl_s_key_alg : string
17506 Returns the name of the algorithm used to generate the key of the certificate
17507 presented by the server when the outgoing connection was made over an
17508 SSL/TLS transport layer.
17509
17510ssl_s_notafter : string
17511 Returns the end date presented by the server as a formatted string
17512 YYMMDDhhmmss[Z] when the outgoing connection was made over an SSL/TLS
17513 transport layer.
17514
17515ssl_s_notbefore : string
17516 Returns the start date presented by the server as a formatted string
17517 YYMMDDhhmmss[Z] when the outgoing connection was made over an SSL/TLS
17518 transport layer.
17519
17520ssl_s_i_dn([<entry>[,<occ>[,<format>]]]) : string
17521 When the outgoing connection was made over an SSL/TLS transport layer,
17522 returns the full distinguished name of the issuer of the certificate
17523 presented by the server when no <entry> is specified, or the value of the
17524 first given entry found from the beginning of the DN. If a positive/negative
17525 occurrence number is specified as the optional second argument, it returns
17526 the value of the nth given entry value from the beginning/end of the DN.
William Lallemand8f600c82020-06-26 09:55:06 +020017527 For instance, "ssl_s_i_dn(OU,2)" the second organization unit, and
17528 "ssl_s_i_dn(CN)" retrieves the common name.
William Lallemandbfa3e812020-06-25 20:07:18 +020017529 The <format> parameter allows you to receive the DN suitable for
17530 consumption by different protocols. Currently supported is rfc2253 for
17531 LDAP v3.
17532 If you'd like to modify the format only you can specify an empty string
17533 and zero for the first two parameters. Example: ssl_s_i_dn(,0,rfc2253)
17534
17535ssl_s_s_dn([<entry>[,<occ>[,<format>]]]) : string
17536 When the outgoing connection was made over an SSL/TLS transport layer,
17537 returns the full distinguished name of the subject of the certificate
17538 presented by the server when no <entry> is specified, or the value of the
17539 first given entry found from the beginning of the DN. If a positive/negative
17540 occurrence number is specified as the optional second argument, it returns
17541 the value of the nth given entry value from the beginning/end of the DN.
William Lallemand8f600c82020-06-26 09:55:06 +020017542 For instance, "ssl_s_s_dn(OU,2)" the second organization unit, and
17543 "ssl_s_s_dn(CN)" retrieves the common name.
William Lallemandbfa3e812020-06-25 20:07:18 +020017544 The <format> parameter allows you to receive the DN suitable for
17545 consumption by different protocols. Currently supported is rfc2253 for
17546 LDAP v3.
17547 If you'd like to modify the format only you can specify an empty string
17548 and zero for the first two parameters. Example: ssl_s_s_dn(,0,rfc2253)
17549
17550ssl_s_serial : binary
17551 Returns the serial of the certificate presented by the server when the
17552 outgoing connection was made over an SSL/TLS transport layer. When used for
17553 an ACL, the value(s) to match against can be passed in hexadecimal form.
17554
17555ssl_s_sha1 : binary
17556 Returns the SHA-1 fingerprint of the certificate presented by the server
17557 when the outgoing connection was made over an SSL/TLS transport layer. This
17558 can be used to know which certificate was chosen using SNI.
17559
17560ssl_s_sig_alg : string
17561 Returns the name of the algorithm used to sign the certificate presented by
17562 the server when the outgoing connection was made over an SSL/TLS transport
17563 layer.
17564
17565ssl_s_version : integer
17566 Returns the version of the certificate presented by the server when the
17567 outgoing connection was made over an SSL/TLS transport layer.
Willy Tarreaub6fb4202008-07-20 11:18:28 +020017568
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200175697.3.5. Fetching samples from buffer contents (Layer 6)
Willy Tarreau74ca5042013-06-11 23:12:07 +020017570------------------------------------------------------
Willy Tarreaub6fb4202008-07-20 11:18:28 +020017571
Willy Tarreau74ca5042013-06-11 23:12:07 +020017572Fetching samples from buffer contents is a bit different from the previous
17573sample fetches above because the sampled data are ephemeral. These data can
17574only be used when they're available and will be lost when they're forwarded.
17575For this reason, samples fetched from buffer contents during a request cannot
17576be used in a response for example. Even while the data are being fetched, they
17577can change. Sometimes it is necessary to set some delays or combine multiple
17578sample fetch methods to ensure that the expected data are complete and usable,
17579for example through TCP request content inspection. Please see the "tcp-request
17580content" keyword for more detailed information on the subject.
Willy Tarreau62644772008-07-16 18:36:06 +020017581
Willy Tarreau74ca5042013-06-11 23:12:07 +020017582payload(<offset>,<length>) : binary (deprecated)
Davor Ocelice9ed2812017-12-25 17:49:28 +010017583 This is an alias for "req.payload" when used in the context of a request (e.g.
Willy Tarreau74ca5042013-06-11 23:12:07 +020017584 "stick on", "stick match"), and for "res.payload" when used in the context of
17585 a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010017586
Willy Tarreau74ca5042013-06-11 23:12:07 +020017587payload_lv(<offset1>,<length>[,<offset2>]) : binary (deprecated)
17588 This is an alias for "req.payload_lv" when used in the context of a request
Davor Ocelice9ed2812017-12-25 17:49:28 +010017589 (e.g. "stick on", "stick match"), and for "res.payload_lv" when used in the
Willy Tarreau74ca5042013-06-11 23:12:07 +020017590 context of a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010017591
Thierry FOURNIERd7d88812017-04-19 15:15:14 +020017592req.hdrs : string
17593 Returns the current request headers as string including the last empty line
17594 separating headers from the request body. The last empty line can be used to
17595 detect a truncated header block. This sample fetch is useful for some SPOE
17596 headers analyzers and for advanced logging.
17597
Thierry FOURNIER5617dce2017-04-09 05:38:19 +020017598req.hdrs_bin : binary
17599 Returns the current request headers contained in preparsed binary form. This
17600 is useful for offloading some processing with SPOE. Each string is described
17601 by a length followed by the number of bytes indicated in the length. The
17602 length is represented using the variable integer encoding detailed in the
17603 SPOE documentation. The end of the list is marked by a couple of empty header
17604 names and values (length of 0 for both).
17605
17606 *(<str:header-name><str:header-value>)<empty string><empty string>
17607
17608 int: refer to the SPOE documentation for the encoding
17609 str: <int:length><bytes>
17610
Willy Tarreau74ca5042013-06-11 23:12:07 +020017611req.len : integer
17612req_len : integer (deprecated)
17613 Returns an integer value corresponding to the number of bytes present in the
17614 request buffer. This is mostly used in ACL. It is important to understand
17615 that this test does not return false as long as the buffer is changing. This
17616 means that a check with equality to zero will almost always immediately match
17617 at the beginning of the session, while a test for more data will wait for
17618 that data to come in and return false only when haproxy is certain that no
17619 more data will come in. This test was designed to be used with TCP request
17620 content inspection.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020017621
Willy Tarreau74ca5042013-06-11 23:12:07 +020017622req.payload(<offset>,<length>) : binary
17623 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020017624 in the request buffer. As a special case, if the <length> argument is zero,
17625 the the whole buffer from <offset> to the end is extracted. This can be used
17626 with ACLs in order to check for the presence of some content in a buffer at
17627 any location.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020017628
Willy Tarreau74ca5042013-06-11 23:12:07 +020017629 ACL alternatives :
17630 payload(<offset>,<length>) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020017631
Willy Tarreau74ca5042013-06-11 23:12:07 +020017632req.payload_lv(<offset1>,<length>[,<offset2>]) : binary
17633 This extracts a binary block whose size is specified at <offset1> for <length>
17634 bytes, and which starts at <offset2> if specified or just after the length in
17635 the request buffer. The <offset2> parameter also supports relative offsets if
17636 prepended with a '+' or '-' sign.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020017637
Willy Tarreau74ca5042013-06-11 23:12:07 +020017638 ACL alternatives :
17639 payload_lv(<offset1>,<length>[,<offset2>]) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020017640
Willy Tarreau74ca5042013-06-11 23:12:07 +020017641 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020017642
Willy Tarreau74ca5042013-06-11 23:12:07 +020017643req.proto_http : boolean
17644req_proto_http : boolean (deprecated)
17645 Returns true when data in the request buffer look like HTTP and correctly
17646 parses as such. It is the same parser as the common HTTP request parser which
17647 is used so there should be no surprises. The test does not match until the
17648 request is complete, failed or timed out. This test may be used to report the
17649 protocol in TCP logs, but the biggest use is to block TCP request analysis
17650 until a complete HTTP request is present in the buffer, for example to track
17651 a header.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020017652
Willy Tarreau74ca5042013-06-11 23:12:07 +020017653 Example:
17654 # track request counts per "base" (concatenation of Host+URL)
17655 tcp-request inspect-delay 10s
17656 tcp-request content reject if !HTTP
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020017657 tcp-request content track-sc0 base table req-rate
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020017658
Willy Tarreau74ca5042013-06-11 23:12:07 +020017659req.rdp_cookie([<name>]) : string
17660rdp_cookie([<name>]) : string (deprecated)
17661 When the request buffer looks like the RDP protocol, extracts the RDP cookie
17662 <name>, or any cookie if unspecified. The parser only checks for the first
17663 cookie, as illustrated in the RDP protocol specification. The cookie name is
17664 case insensitive. Generally the "MSTS" cookie name will be used, as it can
17665 contain the user name of the client connecting to the server if properly
17666 configured on the client. The "MSTSHASH" cookie is often used as well for
17667 session stickiness to servers.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020017668
Willy Tarreau74ca5042013-06-11 23:12:07 +020017669 This differs from "balance rdp-cookie" in that any balancing algorithm may be
17670 used and thus the distribution of clients to backend servers is not linked to
17671 a hash of the RDP cookie. It is envisaged that using a balancing algorithm
17672 such as "balance roundrobin" or "balance leastconn" will lead to a more even
17673 distribution of clients to backend servers than the hash used by "balance
17674 rdp-cookie".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020017675
Willy Tarreau74ca5042013-06-11 23:12:07 +020017676 ACL derivatives :
17677 req_rdp_cookie([<name>]) : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020017678
Willy Tarreau74ca5042013-06-11 23:12:07 +020017679 Example :
17680 listen tse-farm
17681 bind 0.0.0.0:3389
17682 # wait up to 5s for an RDP cookie in the request
17683 tcp-request inspect-delay 5s
17684 tcp-request content accept if RDP_COOKIE
17685 # apply RDP cookie persistence
17686 persist rdp-cookie
17687 # Persist based on the mstshash cookie
17688 # This is only useful makes sense if
17689 # balance rdp-cookie is not used
17690 stick-table type string size 204800
17691 stick on req.rdp_cookie(mstshash)
17692 server srv1 1.1.1.1:3389
17693 server srv1 1.1.1.2:3389
Willy Tarreau04aa6a92012-04-06 18:57:55 +020017694
Willy Tarreau74ca5042013-06-11 23:12:07 +020017695 See also : "balance rdp-cookie", "persist rdp-cookie", "tcp-request" and the
17696 "req_rdp_cookie" ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020017697
Willy Tarreau74ca5042013-06-11 23:12:07 +020017698req.rdp_cookie_cnt([name]) : integer
17699rdp_cookie_cnt([name]) : integer (deprecated)
17700 Tries to parse the request buffer as RDP protocol, then returns an integer
17701 corresponding to the number of RDP cookies found. If an optional cookie name
17702 is passed, only cookies matching this name are considered. This is mostly
17703 used in ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020017704
Willy Tarreau74ca5042013-06-11 23:12:07 +020017705 ACL derivatives :
17706 req_rdp_cookie_cnt([<name>]) : integer match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020017707
Alex Zorin4afdd132018-12-30 13:56:28 +110017708req.ssl_alpn : string
17709 Returns a string containing the values of the Application-Layer Protocol
17710 Negotiation (ALPN) TLS extension (RFC7301), sent by the client within the SSL
17711 ClientHello message. Note that this only applies to raw contents found in the
17712 request buffer and not to the contents deciphered via an SSL data layer, so
17713 this will not work with "bind" lines having the "ssl" option. This is useful
17714 in ACL to make a routing decision based upon the ALPN preferences of a TLS
Jarno Huuskonene504f812019-01-03 07:56:49 +020017715 client, like in the example below. See also "ssl_fc_alpn".
Alex Zorin4afdd132018-12-30 13:56:28 +110017716
17717 Examples :
17718 # Wait for a client hello for at most 5 seconds
17719 tcp-request inspect-delay 5s
17720 tcp-request content accept if { req_ssl_hello_type 1 }
Jarno Huuskonene504f812019-01-03 07:56:49 +020017721 use_backend bk_acme if { req.ssl_alpn acme-tls/1 }
Alex Zorin4afdd132018-12-30 13:56:28 +110017722 default_backend bk_default
17723
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020017724req.ssl_ec_ext : boolean
17725 Returns a boolean identifying if client sent the Supported Elliptic Curves
17726 Extension as defined in RFC4492, section 5.1. within the SSL ClientHello
Cyril Bonté307ee1e2015-09-28 23:16:06 +020017727 message. This can be used to present ECC compatible clients with EC
17728 certificate and to use RSA for all others, on the same IP address. Note that
17729 this only applies to raw contents found in the request buffer and not to
17730 contents deciphered via an SSL data layer, so this will not work with "bind"
17731 lines having the "ssl" option.
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020017732
Willy Tarreau74ca5042013-06-11 23:12:07 +020017733req.ssl_hello_type : integer
17734req_ssl_hello_type : integer (deprecated)
17735 Returns an integer value containing the type of the SSL hello message found
17736 in the request buffer if the buffer contains data that parse as a complete
17737 SSL (v3 or superior) client hello message. Note that this only applies to raw
17738 contents found in the request buffer and not to contents deciphered via an
17739 SSL data layer, so this will not work with "bind" lines having the "ssl"
17740 option. This is mostly used in ACL to detect presence of an SSL hello message
17741 that is supposed to contain an SSL session ID usable for stickiness.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020017742
Willy Tarreau74ca5042013-06-11 23:12:07 +020017743req.ssl_sni : string
17744req_ssl_sni : string (deprecated)
17745 Returns a string containing the value of the Server Name TLS extension sent
17746 by a client in a TLS stream passing through the request buffer if the buffer
17747 contains data that parse as a complete SSL (v3 or superior) client hello
17748 message. Note that this only applies to raw contents found in the request
17749 buffer and not to contents deciphered via an SSL data layer, so this will not
Lukas Tribusa267b5d2020-07-19 00:25:06 +020017750 work with "bind" lines having the "ssl" option. This will only work for actual
17751 implicit TLS based protocols like HTTPS (443), IMAPS (993), SMTPS (465),
17752 however it will not work for explicit TLS based protocols, like SMTP (25/587)
17753 or IMAP (143). SNI normally contains the name of the host the client tries to
17754 connect to (for recent browsers). SNI is useful for allowing or denying access
17755 to certain hosts when SSL/TLS is used by the client. This test was designed to
17756 be used with TCP request content inspection. If content switching is needed,
17757 it is recommended to first wait for a complete client hello (type 1), like in
17758 the example below. See also "ssl_fc_sni".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020017759
Willy Tarreau74ca5042013-06-11 23:12:07 +020017760 ACL derivatives :
17761 req_ssl_sni : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020017762
Willy Tarreau74ca5042013-06-11 23:12:07 +020017763 Examples :
17764 # Wait for a client hello for at most 5 seconds
17765 tcp-request inspect-delay 5s
17766 tcp-request content accept if { req_ssl_hello_type 1 }
17767 use_backend bk_allow if { req_ssl_sni -f allowed_sites }
17768 default_backend bk_sorry_page
Willy Tarreau04aa6a92012-04-06 18:57:55 +020017769
Pradeep Jindalbb2acf52015-09-29 10:12:57 +053017770req.ssl_st_ext : integer
17771 Returns 0 if the client didn't send a SessionTicket TLS Extension (RFC5077)
17772 Returns 1 if the client sent SessionTicket TLS Extension
17773 Returns 2 if the client also sent non-zero length TLS SessionTicket
17774 Note that this only applies to raw contents found in the request buffer and
17775 not to contents deciphered via an SSL data layer, so this will not work with
17776 "bind" lines having the "ssl" option. This can for example be used to detect
17777 whether the client sent a SessionTicket or not and stick it accordingly, if
17778 no SessionTicket then stick on SessionID or don't stick as there's no server
17779 side state is there when SessionTickets are in use.
17780
Willy Tarreau74ca5042013-06-11 23:12:07 +020017781req.ssl_ver : integer
17782req_ssl_ver : integer (deprecated)
17783 Returns an integer value containing the version of the SSL/TLS protocol of a
17784 stream present in the request buffer. Both SSLv2 hello messages and SSLv3
17785 messages are supported. TLSv1 is announced as SSL version 3.1. The value is
17786 composed of the major version multiplied by 65536, added to the minor
17787 version. Note that this only applies to raw contents found in the request
17788 buffer and not to contents deciphered via an SSL data layer, so this will not
17789 work with "bind" lines having the "ssl" option. The ACL version of the test
Davor Ocelice9ed2812017-12-25 17:49:28 +010017790 matches against a decimal notation in the form MAJOR.MINOR (e.g. 3.1). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020017791 fetch is mostly used in ACL.
Willy Tarreaud63335a2010-02-26 12:56:52 +010017792
Willy Tarreau74ca5042013-06-11 23:12:07 +020017793 ACL derivatives :
17794 req_ssl_ver : decimal match
Willy Tarreaud63335a2010-02-26 12:56:52 +010017795
Willy Tarreau47e8eba2013-09-11 23:28:46 +020017796res.len : integer
17797 Returns an integer value corresponding to the number of bytes present in the
17798 response buffer. This is mostly used in ACL. It is important to understand
17799 that this test does not return false as long as the buffer is changing. This
17800 means that a check with equality to zero will almost always immediately match
17801 at the beginning of the session, while a test for more data will wait for
17802 that data to come in and return false only when haproxy is certain that no
17803 more data will come in. This test was designed to be used with TCP response
Christopher Faulete596d182020-05-05 17:46:34 +020017804 content inspection. But it may also be used in tcp-check based expect rules.
Willy Tarreau47e8eba2013-09-11 23:28:46 +020017805
Willy Tarreau74ca5042013-06-11 23:12:07 +020017806res.payload(<offset>,<length>) : binary
17807 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020017808 in the response buffer. As a special case, if the <length> argument is zero,
Christopher Faulete596d182020-05-05 17:46:34 +020017809 the whole buffer from <offset> to the end is extracted. This can be used
Willy Tarreau00f00842013-08-02 11:07:32 +020017810 with ACLs in order to check for the presence of some content in a buffer at
Christopher Faulete596d182020-05-05 17:46:34 +020017811 any location. It may also be used in tcp-check based expect rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +010017812
Willy Tarreau74ca5042013-06-11 23:12:07 +020017813res.payload_lv(<offset1>,<length>[,<offset2>]) : binary
17814 This extracts a binary block whose size is specified at <offset1> for <length>
17815 bytes, and which starts at <offset2> if specified or just after the length in
17816 the response buffer. The <offset2> parameter also supports relative offsets
Christopher Faulete596d182020-05-05 17:46:34 +020017817 if prepended with a '+' or '-' sign. It may also be used in tcp-check based
17818 expect rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +010017819
Willy Tarreau74ca5042013-06-11 23:12:07 +020017820 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaud63335a2010-02-26 12:56:52 +010017821
Willy Tarreau971f7b62015-09-29 14:06:59 +020017822res.ssl_hello_type : integer
17823rep_ssl_hello_type : integer (deprecated)
17824 Returns an integer value containing the type of the SSL hello message found
17825 in the response buffer if the buffer contains data that parses as a complete
17826 SSL (v3 or superior) hello message. Note that this only applies to raw
17827 contents found in the response buffer and not to contents deciphered via an
17828 SSL data layer, so this will not work with "server" lines having the "ssl"
17829 option. This is mostly used in ACL to detect presence of an SSL hello message
17830 that is supposed to contain an SSL session ID usable for stickiness.
17831
Willy Tarreau74ca5042013-06-11 23:12:07 +020017832wait_end : boolean
17833 This fetch either returns true when the inspection period is over, or does
17834 not fetch. It is only used in ACLs, in conjunction with content analysis to
Davor Ocelice9ed2812017-12-25 17:49:28 +010017835 avoid returning a wrong verdict early. It may also be used to delay some
Willy Tarreau74ca5042013-06-11 23:12:07 +020017836 actions, such as a delayed reject for some special addresses. Since it either
17837 stops the rules evaluation or immediately returns true, it is recommended to
Davor Ocelice9ed2812017-12-25 17:49:28 +010017838 use this acl as the last one in a rule. Please note that the default ACL
Willy Tarreau74ca5042013-06-11 23:12:07 +020017839 "WAIT_END" is always usable without prior declaration. This test was designed
17840 to be used with TCP request content inspection.
Willy Tarreaud63335a2010-02-26 12:56:52 +010017841
Willy Tarreau74ca5042013-06-11 23:12:07 +020017842 Examples :
17843 # delay every incoming request by 2 seconds
17844 tcp-request inspect-delay 2s
17845 tcp-request content accept if WAIT_END
Willy Tarreaud63335a2010-02-26 12:56:52 +010017846
Willy Tarreau74ca5042013-06-11 23:12:07 +020017847 # don't immediately tell bad guys they are rejected
17848 tcp-request inspect-delay 10s
17849 acl goodguys src 10.0.0.0/24
17850 acl badguys src 10.0.1.0/24
17851 tcp-request content accept if goodguys
17852 tcp-request content reject if badguys WAIT_END
17853 tcp-request content reject
17854
17855
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200178567.3.6. Fetching HTTP samples (Layer 7)
Willy Tarreau74ca5042013-06-11 23:12:07 +020017857--------------------------------------
17858
17859It is possible to fetch samples from HTTP contents, requests and responses.
17860This application layer is also called layer 7. It is only possible to fetch the
17861data in this section when a full HTTP request or response has been parsed from
17862its respective request or response buffer. This is always the case with all
17863HTTP specific rules and for sections running with "mode http". When using TCP
17864content inspection, it may be necessary to support an inspection delay in order
17865to let the request or response come in first. These fetches may require a bit
17866more CPU resources than the layer 4 ones, but not much since the request and
17867response are indexed.
17868
17869base : string
17870 This returns the concatenation of the first Host header and the path part of
17871 the request, which starts at the first slash and ends before the question
17872 mark. It can be useful in virtual hosted environments to detect URL abuses as
17873 well as to improve shared caches efficiency. Using this with a limited size
17874 stick table also allows one to collect statistics about most commonly
17875 requested objects by host/path. With ACLs it can allow simple content
17876 switching rules involving the host and the path at the same time, such as
17877 "www.example.com/favicon.ico". See also "path" and "uri".
17878
17879 ACL derivatives :
17880 base : exact string match
17881 base_beg : prefix match
17882 base_dir : subdir match
17883 base_dom : domain match
17884 base_end : suffix match
17885 base_len : length match
17886 base_reg : regex match
17887 base_sub : substring match
17888
17889base32 : integer
17890 This returns a 32-bit hash of the value returned by the "base" fetch method
17891 above. This is useful to track per-URL activity on high traffic sites without
17892 having to store all URLs. Instead a shorter hash is stored, saving a lot of
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020017893 memory. The output type is an unsigned integer. The hash function used is
17894 SDBM with full avalanche on the output. Technically, base32 is exactly equal
17895 to "base,sdbm(1)".
Willy Tarreau74ca5042013-06-11 23:12:07 +020017896
17897base32+src : binary
17898 This returns the concatenation of the base32 fetch above and the src fetch
17899 below. The resulting type is of type binary, with a size of 8 or 20 bytes
17900 depending on the source address family. This can be used to track per-IP,
17901 per-URL counters.
17902
William Lallemand65ad6e12014-01-31 15:08:02 +010017903capture.req.hdr(<idx>) : string
17904 This extracts the content of the header captured by the "capture request
17905 header", idx is the position of the capture keyword in the configuration.
17906 The first entry is an index of 0. See also: "capture request header".
17907
17908capture.req.method : string
17909 This extracts the METHOD of an HTTP request. It can be used in both request
17910 and response. Unlike "method", it can be used in both request and response
17911 because it's allocated.
17912
17913capture.req.uri : string
17914 This extracts the request's URI, which starts at the first slash and ends
17915 before the first space in the request (without the host part). Unlike "path"
17916 and "url", it can be used in both request and response because it's
17917 allocated.
17918
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020017919capture.req.ver : string
17920 This extracts the request's HTTP version and returns either "HTTP/1.0" or
17921 "HTTP/1.1". Unlike "req.ver", it can be used in both request, response, and
17922 logs because it relies on a persistent flag.
17923
William Lallemand65ad6e12014-01-31 15:08:02 +010017924capture.res.hdr(<idx>) : string
17925 This extracts the content of the header captured by the "capture response
17926 header", idx is the position of the capture keyword in the configuration.
17927 The first entry is an index of 0.
17928 See also: "capture response header"
17929
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020017930capture.res.ver : string
17931 This extracts the response's HTTP version and returns either "HTTP/1.0" or
17932 "HTTP/1.1". Unlike "res.ver", it can be used in logs because it relies on a
17933 persistent flag.
17934
Willy Tarreaua5910cc2015-05-02 00:46:08 +020017935req.body : binary
Christopher Fauletaf4dc4c2020-05-05 17:33:25 +020017936 This returns the HTTP request's available body as a block of data. It is
17937 recommended to use "option http-buffer-request" to be sure to wait, as much
17938 as possible, for the request's body.
Willy Tarreaua5910cc2015-05-02 00:46:08 +020017939
Thierry FOURNIER9826c772015-05-20 15:50:54 +020017940req.body_param([<name>) : string
17941 This fetch assumes that the body of the POST request is url-encoded. The user
17942 can check if the "content-type" contains the value
17943 "application/x-www-form-urlencoded". This extracts the first occurrence of the
17944 parameter <name> in the body, which ends before '&'. The parameter name is
17945 case-sensitive. If no name is given, any parameter will match, and the first
17946 one will be returned. The result is a string corresponding to the value of the
17947 parameter <name> as presented in the request body (no URL decoding is
17948 performed). Note that the ACL version of this fetch iterates over multiple
17949 parameters and will iteratively report all parameters values if no name is
17950 given.
17951
Willy Tarreaua5910cc2015-05-02 00:46:08 +020017952req.body_len : integer
17953 This returns the length of the HTTP request's available body in bytes. It may
17954 be lower than the advertised length if the body is larger than the buffer. It
Christopher Fauletaf4dc4c2020-05-05 17:33:25 +020017955 is recommended to use "option http-buffer-request" to be sure to wait, as
17956 much as possible, for the request's body.
Willy Tarreaua5910cc2015-05-02 00:46:08 +020017957
17958req.body_size : integer
17959 This returns the advertised length of the HTTP request's body in bytes. It
Christopher Fauletaf4dc4c2020-05-05 17:33:25 +020017960 will represent the advertised Content-Length header, or the size of the
17961 available data in case of chunked encoding.
Willy Tarreaua5910cc2015-05-02 00:46:08 +020017962
Willy Tarreau74ca5042013-06-11 23:12:07 +020017963req.cook([<name>]) : string
17964cook([<name>]) : string (deprecated)
17965 This extracts the last occurrence of the cookie name <name> on a "Cookie"
17966 header line from the request, and returns its value as string. If no name is
17967 specified, the first cookie value is returned. When used with ACLs, all
17968 matching cookies are evaluated. Spaces around the name and the value are
17969 ignored as requested by the Cookie header specification (RFC6265). The cookie
17970 name is case-sensitive. Empty cookies are valid, so an empty cookie may very
17971 well return an empty value if it is present. Use the "found" match to detect
17972 presence. Use the res.cook() variant for response cookies sent by the server.
17973
17974 ACL derivatives :
17975 cook([<name>]) : exact string match
17976 cook_beg([<name>]) : prefix match
17977 cook_dir([<name>]) : subdir match
17978 cook_dom([<name>]) : domain match
17979 cook_end([<name>]) : suffix match
17980 cook_len([<name>]) : length match
17981 cook_reg([<name>]) : regex match
17982 cook_sub([<name>]) : substring match
Willy Tarreaud63335a2010-02-26 12:56:52 +010017983
Willy Tarreau74ca5042013-06-11 23:12:07 +020017984req.cook_cnt([<name>]) : integer
17985cook_cnt([<name>]) : integer (deprecated)
17986 Returns an integer value representing the number of occurrences of the cookie
17987 <name> in the request, or all cookies if <name> is not specified.
Willy Tarreaud63335a2010-02-26 12:56:52 +010017988
Willy Tarreau74ca5042013-06-11 23:12:07 +020017989req.cook_val([<name>]) : integer
17990cook_val([<name>]) : integer (deprecated)
17991 This extracts the last occurrence of the cookie name <name> on a "Cookie"
17992 header line from the request, and converts its value to an integer which is
17993 returned. If no name is specified, the first cookie value is returned. When
17994 used in ACLs, all matching names are iterated over until a value matches.
Willy Tarreau0e698542011-09-16 08:32:32 +020017995
Willy Tarreau74ca5042013-06-11 23:12:07 +020017996cookie([<name>]) : string (deprecated)
17997 This extracts the last occurrence of the cookie name <name> on a "Cookie"
17998 header line from the request, or a "Set-Cookie" header from the response, and
17999 returns its value as a string. A typical use is to get multiple clients
18000 sharing a same profile use the same server. This can be similar to what
Willy Tarreau294d0f02015-08-10 19:40:12 +020018001 "appsession" did with the "request-learn" statement, but with support for
Willy Tarreau74ca5042013-06-11 23:12:07 +020018002 multi-peer synchronization and state keeping across restarts. If no name is
18003 specified, the first cookie value is returned. This fetch should not be used
18004 anymore and should be replaced by req.cook() or res.cook() instead as it
18005 ambiguously uses the direction based on the context where it is used.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018006
Willy Tarreau74ca5042013-06-11 23:12:07 +020018007hdr([<name>[,<occ>]]) : string
18008 This is equivalent to req.hdr() when used on requests, and to res.hdr() when
18009 used on responses. Please refer to these respective fetches for more details.
18010 In case of doubt about the fetch direction, please use the explicit ones.
18011 Note that contrary to the hdr() sample fetch method, the hdr_* ACL keywords
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030018012 unambiguously apply to the request headers.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018013
Willy Tarreau74ca5042013-06-11 23:12:07 +020018014req.fhdr(<name>[,<occ>]) : string
18015 This extracts the last occurrence of header <name> in an HTTP request. When
18016 used from an ACL, all occurrences are iterated over until a match is found.
18017 Optionally, a specific occurrence might be specified as a position number.
18018 Positive values indicate a position from the first occurrence, with 1 being
18019 the first one. Negative values indicate positions relative to the last one,
18020 with -1 being the last one. It differs from req.hdr() in that any commas
18021 present in the value are returned and are not used as delimiters. This is
18022 sometimes useful with headers such as User-Agent.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018023
Willy Tarreau74ca5042013-06-11 23:12:07 +020018024req.fhdr_cnt([<name>]) : integer
18025 Returns an integer value representing the number of occurrences of request
18026 header field name <name>, or the total number of header fields if <name> is
18027 not specified. Contrary to its req.hdr_cnt() cousin, this function returns
18028 the number of full line headers and does not stop on commas.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018029
Willy Tarreau74ca5042013-06-11 23:12:07 +020018030req.hdr([<name>[,<occ>]]) : string
18031 This extracts the last occurrence of header <name> in an HTTP request. When
18032 used from an ACL, all occurrences are iterated over until a match is found.
18033 Optionally, a specific occurrence might be specified as a position number.
18034 Positive values indicate a position from the first occurrence, with 1 being
18035 the first one. Negative values indicate positions relative to the last one,
18036 with -1 being the last one. A typical use is with the X-Forwarded-For header
18037 once converted to IP, associated with an IP stick-table. The function
18038 considers any comma as a delimiter for distinct values. If full-line headers
Lukas Tribus23953682017-04-28 13:24:30 +000018039 are desired instead, use req.fhdr(). Please carefully check RFC7231 to know
Willy Tarreau74ca5042013-06-11 23:12:07 +020018040 how certain headers are supposed to be parsed. Also, some of them are case
Davor Ocelice9ed2812017-12-25 17:49:28 +010018041 insensitive (e.g. Connection).
Willy Tarreaud63335a2010-02-26 12:56:52 +010018042
Willy Tarreau74ca5042013-06-11 23:12:07 +020018043 ACL derivatives :
18044 hdr([<name>[,<occ>]]) : exact string match
18045 hdr_beg([<name>[,<occ>]]) : prefix match
18046 hdr_dir([<name>[,<occ>]]) : subdir match
18047 hdr_dom([<name>[,<occ>]]) : domain match
18048 hdr_end([<name>[,<occ>]]) : suffix match
18049 hdr_len([<name>[,<occ>]]) : length match
18050 hdr_reg([<name>[,<occ>]]) : regex match
18051 hdr_sub([<name>[,<occ>]]) : substring match
18052
18053req.hdr_cnt([<name>]) : integer
18054hdr_cnt([<header>]) : integer (deprecated)
18055 Returns an integer value representing the number of occurrences of request
18056 header field name <name>, or the total number of header field values if
18057 <name> is not specified. It is important to remember that one header line may
18058 count as several headers if it has several values. The function considers any
18059 comma as a delimiter for distinct values. If full-line headers are desired
18060 instead, req.fhdr_cnt() should be used instead. With ACLs, it can be used to
18061 detect presence, absence or abuse of a specific header, as well as to block
18062 request smuggling attacks by rejecting requests which contain more than one
18063 of certain headers. See "req.hdr" for more information on header matching.
18064
18065req.hdr_ip([<name>[,<occ>]]) : ip
18066hdr_ip([<name>[,<occ>]]) : ip (deprecated)
18067 This extracts the last occurrence of header <name> in an HTTP request,
18068 converts it to an IPv4 or IPv6 address and returns this address. When used
18069 with ACLs, all occurrences are checked, and if <name> is omitted, every value
18070 of every header is checked. Optionally, a specific occurrence might be
18071 specified as a position number. Positive values indicate a position from the
Davor Ocelice9ed2812017-12-25 17:49:28 +010018072 first occurrence, with 1 being the first one. Negative values indicate
Willy Tarreau74ca5042013-06-11 23:12:07 +020018073 positions relative to the last one, with -1 being the last one. A typical use
18074 is with the X-Forwarded-For and X-Client-IP headers.
18075
18076req.hdr_val([<name>[,<occ>]]) : integer
18077hdr_val([<name>[,<occ>]]) : integer (deprecated)
18078 This extracts the last occurrence of header <name> in an HTTP request, and
18079 converts it to an integer value. When used with ACLs, all occurrences are
18080 checked, and if <name> is omitted, every value of every header is checked.
18081 Optionally, a specific occurrence might be specified as a position number.
18082 Positive values indicate a position from the first occurrence, with 1 being
18083 the first one. Negative values indicate positions relative to the last one,
18084 with -1 being the last one. A typical use is with the X-Forwarded-For header.
18085
Frédéric Lécailleec891192019-02-26 15:02:35 +010018086
18087
Willy Tarreau74ca5042013-06-11 23:12:07 +020018088http_auth(<userlist>) : boolean
18089 Returns a boolean indicating whether the authentication data received from
18090 the client match a username & password stored in the specified userlist. This
18091 fetch function is not really useful outside of ACLs. Currently only http
18092 basic auth is supported.
18093
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010018094http_auth_group(<userlist>) : string
18095 Returns a string corresponding to the user name found in the authentication
18096 data received from the client if both the user name and password are valid
18097 according to the specified userlist. The main purpose is to use it in ACLs
18098 where it is then checked whether the user belongs to any group within a list.
Willy Tarreau74ca5042013-06-11 23:12:07 +020018099 This fetch function is not really useful outside of ACLs. Currently only http
18100 basic auth is supported.
18101
18102 ACL derivatives :
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010018103 http_auth_group(<userlist>) : group ...
18104 Returns true when the user extracted from the request and whose password is
18105 valid according to the specified userlist belongs to at least one of the
18106 groups.
Willy Tarreau74ca5042013-06-11 23:12:07 +020018107
Christopher Fauleta4063562019-08-02 11:51:37 +020018108http_auth_pass : string
Willy Tarreauc9c6cdb2020-03-05 16:03:58 +010018109 Returns the user's password found in the authentication data received from
18110 the client, as supplied in the Authorization header. Not checks are
18111 performed by this sample fetch. Only Basic authentication is supported.
Christopher Fauleta4063562019-08-02 11:51:37 +020018112
18113http_auth_type : string
Willy Tarreauc9c6cdb2020-03-05 16:03:58 +010018114 Returns the authentication method found in the authentication data received from
18115 the client, as supplied in the Authorization header. Not checks are
18116 performed by this sample fetch. Only Basic authentication is supported.
Christopher Fauleta4063562019-08-02 11:51:37 +020018117
18118http_auth_user : string
Willy Tarreauc9c6cdb2020-03-05 16:03:58 +010018119 Returns the user name found in the authentication data received from the
18120 client, as supplied in the Authorization header. Not checks are performed by
18121 this sample fetch. Only Basic authentication is supported.
Christopher Fauleta4063562019-08-02 11:51:37 +020018122
Willy Tarreau74ca5042013-06-11 23:12:07 +020018123http_first_req : boolean
Willy Tarreau7f18e522010-10-22 20:04:13 +020018124 Returns true when the request being processed is the first one of the
18125 connection. This can be used to add or remove headers that may be missing
Willy Tarreau74ca5042013-06-11 23:12:07 +020018126 from some requests when a request is not the first one, or to help grouping
18127 requests in the logs.
Willy Tarreau7f18e522010-10-22 20:04:13 +020018128
Willy Tarreau74ca5042013-06-11 23:12:07 +020018129method : integer + string
18130 Returns an integer value corresponding to the method in the HTTP request. For
18131 example, "GET" equals 1 (check sources to establish the matching). Value 9
18132 means "other method" and may be converted to a string extracted from the
18133 stream. This should not be used directly as a sample, this is only meant to
18134 be used from ACLs, which transparently convert methods from patterns to these
18135 integer + string values. Some predefined ACL already check for most common
18136 methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020018137
Willy Tarreau74ca5042013-06-11 23:12:07 +020018138 ACL derivatives :
18139 method : case insensitive method match
Willy Tarreau6a06a402007-07-15 20:15:28 +020018140
Willy Tarreau74ca5042013-06-11 23:12:07 +020018141 Example :
18142 # only accept GET and HEAD requests
18143 acl valid_method method GET HEAD
18144 http-request deny if ! valid_method
Willy Tarreau6a06a402007-07-15 20:15:28 +020018145
Willy Tarreau74ca5042013-06-11 23:12:07 +020018146path : string
18147 This extracts the request's URL path, which starts at the first slash and
18148 ends before the question mark (without the host part). A typical use is with
18149 prefetch-capable caches, and with portals which need to aggregate multiple
18150 information from databases and keep them in caches. Note that with outgoing
18151 caches, it would be wiser to use "url" instead. With ACLs, it's typically
Davor Ocelice9ed2812017-12-25 17:49:28 +010018152 used to match exact file names (e.g. "/login.php"), or directory parts using
Willy Tarreau74ca5042013-06-11 23:12:07 +020018153 the derivative forms. See also the "url" and "base" fetch methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020018154
Willy Tarreau74ca5042013-06-11 23:12:07 +020018155 ACL derivatives :
18156 path : exact string match
18157 path_beg : prefix match
18158 path_dir : subdir match
18159 path_dom : domain match
18160 path_end : suffix match
18161 path_len : length match
18162 path_reg : regex match
18163 path_sub : substring match
Willy Tarreau6a06a402007-07-15 20:15:28 +020018164
Christopher Faulete720c322020-09-02 17:25:18 +020018165pathq : string
18166 This extracts the request's URL path with the query-string, which starts at
18167 the first slash. This sample fetch is pretty handy to always retrieve a
18168 relative URI, excluding the scheme and the authority part, if any. Indeed,
18169 while it is the common representation for an HTTP/1.1 request target, in
18170 HTTP/2, an absolute URI is often used. This sample fetch will return the same
18171 result in both cases.
18172
Willy Tarreau49ad95c2015-01-19 15:06:26 +010018173query : string
18174 This extracts the request's query string, which starts after the first
18175 question mark. If no question mark is present, this fetch returns nothing. If
18176 a question mark is present but nothing follows, it returns an empty string.
18177 This means it's possible to easily know whether a query string is present
Tim Düsterhus4896c442016-11-29 02:15:19 +010018178 using the "found" matching method. This fetch is the complement of "path"
Willy Tarreau49ad95c2015-01-19 15:06:26 +010018179 which stops before the question mark.
18180
Willy Tarreaueb27ec72015-02-20 13:55:29 +010018181req.hdr_names([<delim>]) : string
18182 This builds a string made from the concatenation of all header names as they
18183 appear in the request when the rule is evaluated. The default delimiter is
18184 the comma (',') but it may be overridden as an optional argument <delim>. In
18185 this case, only the first character of <delim> is considered.
18186
Willy Tarreau74ca5042013-06-11 23:12:07 +020018187req.ver : string
18188req_ver : string (deprecated)
18189 Returns the version string from the HTTP request, for example "1.1". This can
18190 be useful for logs, but is mostly there for ACL. Some predefined ACL already
18191 check for versions 1.0 and 1.1.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018192
Willy Tarreau74ca5042013-06-11 23:12:07 +020018193 ACL derivatives :
18194 req_ver : exact string match
Willy Tarreau0e698542011-09-16 08:32:32 +020018195
Christopher Faulete596d182020-05-05 17:46:34 +020018196res.body : binary
18197 This returns the HTTP response's available body as a block of data. Unlike
18198 the request side, there is no directive to wait for the response's body. This
18199 sample fetch is really useful (and usable) in the health-check context. It
18200 may be used in tcp-check based expect rules.
18201
18202res.body_len : integer
18203 This returns the length of the HTTP response available body in bytes. Unlike
18204 the request side, there is no directive to wait for the response's body. This
18205 sample fetch is really useful (and usable) in the health-check context. It
18206 may be used in tcp-check based expect rules.
18207
18208res.body_size : integer
18209 This returns the advertised length of the HTTP response body in bytes. It
18210 will represent the advertised Content-Length header, or the size of the
18211 available data in case of chunked encoding. Unlike the request side, there is
18212 no directive to wait for the response body. This sample fetch is really
18213 useful (and usable) in the health-check context. It may be used in tcp-check
18214 based expect rules.
18215
Willy Tarreau74ca5042013-06-11 23:12:07 +020018216res.comp : boolean
18217 Returns the boolean "true" value if the response has been compressed by
18218 HAProxy, otherwise returns boolean "false". This may be used to add
18219 information in the logs.
Willy Tarreau6a06a402007-07-15 20:15:28 +020018220
Willy Tarreau74ca5042013-06-11 23:12:07 +020018221res.comp_algo : string
18222 Returns a string containing the name of the algorithm used if the response
18223 was compressed by HAProxy, for example : "deflate". This may be used to add
18224 some information in the logs.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018225
Willy Tarreau74ca5042013-06-11 23:12:07 +020018226res.cook([<name>]) : string
18227scook([<name>]) : string (deprecated)
18228 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
18229 header line from the response, and returns its value as string. If no name is
Christopher Faulete596d182020-05-05 17:46:34 +020018230 specified, the first cookie value is returned. It may be used in tcp-check
18231 based expect rules.
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020018232
Willy Tarreau74ca5042013-06-11 23:12:07 +020018233 ACL derivatives :
18234 scook([<name>] : exact string match
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020018235
Willy Tarreau74ca5042013-06-11 23:12:07 +020018236res.cook_cnt([<name>]) : integer
18237scook_cnt([<name>]) : integer (deprecated)
18238 Returns an integer value representing the number of occurrences of the cookie
18239 <name> in the response, or all cookies if <name> is not specified. This is
Christopher Faulete596d182020-05-05 17:46:34 +020018240 mostly useful when combined with ACLs to detect suspicious responses. It may
18241 be used in tcp-check based expect rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018242
Willy Tarreau74ca5042013-06-11 23:12:07 +020018243res.cook_val([<name>]) : integer
18244scook_val([<name>]) : integer (deprecated)
18245 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
18246 header line from the response, and converts its value to an integer which is
Christopher Faulete596d182020-05-05 17:46:34 +020018247 returned. If no name is specified, the first cookie value is returned. It may
18248 be used in tcp-check based expect rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018249
Willy Tarreau74ca5042013-06-11 23:12:07 +020018250res.fhdr([<name>[,<occ>]]) : string
18251 This extracts the last occurrence of header <name> in an HTTP response, or of
18252 the last header if no <name> is specified. Optionally, a specific occurrence
18253 might be specified as a position number. Positive values indicate a position
18254 from the first occurrence, with 1 being the first one. Negative values
18255 indicate positions relative to the last one, with -1 being the last one. It
18256 differs from res.hdr() in that any commas present in the value are returned
18257 and are not used as delimiters. If this is not desired, the res.hdr() fetch
18258 should be used instead. This is sometimes useful with headers such as Date or
Christopher Faulete596d182020-05-05 17:46:34 +020018259 Expires. It may be used in tcp-check based expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020018260
Willy Tarreau74ca5042013-06-11 23:12:07 +020018261res.fhdr_cnt([<name>]) : integer
18262 Returns an integer value representing the number of occurrences of response
18263 header field name <name>, or the total number of header fields if <name> is
18264 not specified. Contrary to its res.hdr_cnt() cousin, this function returns
18265 the number of full line headers and does not stop on commas. If this is not
Christopher Faulete596d182020-05-05 17:46:34 +020018266 desired, the res.hdr_cnt() fetch should be used instead. It may be used in
18267 tcp-check based expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020018268
Willy Tarreau74ca5042013-06-11 23:12:07 +020018269res.hdr([<name>[,<occ>]]) : string
18270shdr([<name>[,<occ>]]) : string (deprecated)
18271 This extracts the last occurrence of header <name> in an HTTP response, or of
18272 the last header if no <name> is specified. Optionally, a specific occurrence
18273 might be specified as a position number. Positive values indicate a position
18274 from the first occurrence, with 1 being the first one. Negative values
18275 indicate positions relative to the last one, with -1 being the last one. This
18276 can be useful to learn some data into a stick-table. The function considers
18277 any comma as a delimiter for distinct values. If this is not desired, the
Christopher Faulete596d182020-05-05 17:46:34 +020018278 res.fhdr() fetch should be used instead. It may be used in tcp-check based
18279 expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020018280
Willy Tarreau74ca5042013-06-11 23:12:07 +020018281 ACL derivatives :
18282 shdr([<name>[,<occ>]]) : exact string match
18283 shdr_beg([<name>[,<occ>]]) : prefix match
18284 shdr_dir([<name>[,<occ>]]) : subdir match
18285 shdr_dom([<name>[,<occ>]]) : domain match
18286 shdr_end([<name>[,<occ>]]) : suffix match
18287 shdr_len([<name>[,<occ>]]) : length match
18288 shdr_reg([<name>[,<occ>]]) : regex match
18289 shdr_sub([<name>[,<occ>]]) : substring match
18290
18291res.hdr_cnt([<name>]) : integer
18292shdr_cnt([<name>]) : integer (deprecated)
18293 Returns an integer value representing the number of occurrences of response
18294 header field name <name>, or the total number of header fields if <name> is
18295 not specified. The function considers any comma as a delimiter for distinct
18296 values. If this is not desired, the res.fhdr_cnt() fetch should be used
Christopher Faulete596d182020-05-05 17:46:34 +020018297 instead. It may be used in tcp-check based expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020018298
Willy Tarreau74ca5042013-06-11 23:12:07 +020018299res.hdr_ip([<name>[,<occ>]]) : ip
18300shdr_ip([<name>[,<occ>]]) : ip (deprecated)
18301 This extracts the last occurrence of header <name> in an HTTP response,
18302 convert it to an IPv4 or IPv6 address and returns this address. Optionally, a
18303 specific occurrence might be specified as a position number. Positive values
18304 indicate a position from the first occurrence, with 1 being the first one.
18305 Negative values indicate positions relative to the last one, with -1 being
Christopher Faulete596d182020-05-05 17:46:34 +020018306 the last one. This can be useful to learn some data into a stick table. It
18307 may be used in tcp-check based expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020018308
Willy Tarreaueb27ec72015-02-20 13:55:29 +010018309res.hdr_names([<delim>]) : string
18310 This builds a string made from the concatenation of all header names as they
18311 appear in the response when the rule is evaluated. The default delimiter is
18312 the comma (',') but it may be overridden as an optional argument <delim>. In
Christopher Faulete596d182020-05-05 17:46:34 +020018313 this case, only the first character of <delim> is considered. It may be used
18314 in tcp-check based expect rules.
Willy Tarreaueb27ec72015-02-20 13:55:29 +010018315
Willy Tarreau74ca5042013-06-11 23:12:07 +020018316res.hdr_val([<name>[,<occ>]]) : integer
18317shdr_val([<name>[,<occ>]]) : integer (deprecated)
18318 This extracts the last occurrence of header <name> in an HTTP response, and
18319 converts it to an integer value. Optionally, a specific occurrence might be
18320 specified as a position number. Positive values indicate a position from the
18321 first occurrence, with 1 being the first one. Negative values indicate
18322 positions relative to the last one, with -1 being the last one. This can be
Christopher Faulete596d182020-05-05 17:46:34 +020018323 useful to learn some data into a stick table. It may be used in tcp-check
18324 based expect rules.
18325
18326res.hdrs : string
18327 Returns the current response headers as string including the last empty line
18328 separating headers from the request body. The last empty line can be used to
18329 detect a truncated header block. This sample fetch is useful for some SPOE
18330 headers analyzers and for advanced logging. It may also be used in tcp-check
18331 based expect rules.
18332
18333res.hdrs_bin : binary
18334 Returns the current response headers contained in preparsed binary form. This
18335 is useful for offloading some processing with SPOE. It may be used in
18336 tcp-check based expect rules. Each string is described by a length followed
18337 by the number of bytes indicated in the length. The length is represented
18338 using the variable integer encoding detailed in the SPOE documentation. The
18339 end of the list is marked by a couple of empty header names and values
18340 (length of 0 for both).
18341
18342 *(<str:header-name><str:header-value>)<empty string><empty string>
18343
18344 int: refer to the SPOE documentation for the encoding
18345 str: <int:length><bytes>
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010018346
Willy Tarreau74ca5042013-06-11 23:12:07 +020018347res.ver : string
18348resp_ver : string (deprecated)
18349 Returns the version string from the HTTP response, for example "1.1". This
Christopher Faulete596d182020-05-05 17:46:34 +020018350 can be useful for logs, but is mostly there for ACL. It may be used in
18351 tcp-check based expect rules.
Willy Tarreau0e698542011-09-16 08:32:32 +020018352
Willy Tarreau74ca5042013-06-11 23:12:07 +020018353 ACL derivatives :
18354 resp_ver : exact string match
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010018355
Willy Tarreau74ca5042013-06-11 23:12:07 +020018356set-cookie([<name>]) : string (deprecated)
18357 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
18358 header line from the response and uses the corresponding value to match. This
Willy Tarreau294d0f02015-08-10 19:40:12 +020018359 can be comparable to what "appsession" did with default options, but with
Willy Tarreau74ca5042013-06-11 23:12:07 +020018360 support for multi-peer synchronization and state keeping across restarts.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010018361
Willy Tarreau74ca5042013-06-11 23:12:07 +020018362 This fetch function is deprecated and has been superseded by the "res.cook"
18363 fetch. This keyword will disappear soon.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010018364
Willy Tarreau74ca5042013-06-11 23:12:07 +020018365status : integer
18366 Returns an integer containing the HTTP status code in the HTTP response, for
18367 example, 302. It is mostly used within ACLs and integer ranges, for example,
Christopher Faulete596d182020-05-05 17:46:34 +020018368 to remove any Location header if the response is not a 3xx. It may be used in
18369 tcp-check based expect rules.
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018370
Thierry Fournier0e00dca2016-04-07 15:47:40 +020018371unique-id : string
18372 Returns the unique-id attached to the request. The directive
18373 "unique-id-format" must be set. If it is not set, the unique-id sample fetch
18374 fails. Note that the unique-id is usually used with HTTP requests, however this
18375 sample fetch can be used with other protocols. Obviously, if it is used with
18376 other protocols than HTTP, the unique-id-format directive must not contain
18377 HTTP parts. See: unique-id-format and unique-id-header
18378
Willy Tarreau74ca5042013-06-11 23:12:07 +020018379url : string
18380 This extracts the request's URL as presented in the request. A typical use is
18381 with prefetch-capable caches, and with portals which need to aggregate
18382 multiple information from databases and keep them in caches. With ACLs, using
18383 "path" is preferred over using "url", because clients may send a full URL as
18384 is normally done with proxies. The only real use is to match "*" which does
18385 not match in "path", and for which there is already a predefined ACL. See
18386 also "path" and "base".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018387
Willy Tarreau74ca5042013-06-11 23:12:07 +020018388 ACL derivatives :
18389 url : exact string match
18390 url_beg : prefix match
18391 url_dir : subdir match
18392 url_dom : domain match
18393 url_end : suffix match
18394 url_len : length match
18395 url_reg : regex match
18396 url_sub : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018397
Willy Tarreau74ca5042013-06-11 23:12:07 +020018398url_ip : ip
18399 This extracts the IP address from the request's URL when the host part is
18400 presented as an IP address. Its use is very limited. For instance, a
18401 monitoring system might use this field as an alternative for the source IP in
18402 order to test what path a given source address would follow, or to force an
18403 entry in a table for a given source address. With ACLs it can be used to
18404 restrict access to certain systems through a proxy, for example when combined
18405 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018406
Willy Tarreau74ca5042013-06-11 23:12:07 +020018407url_port : integer
18408 This extracts the port part from the request's URL. Note that if the port is
18409 not specified in the request, port 80 is assumed. With ACLs it can be used to
18410 restrict access to certain systems through a proxy, for example when combined
18411 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018412
Willy Tarreau1ede1da2015-05-07 16:06:18 +020018413urlp([<name>[,<delim>]]) : string
18414url_param([<name>[,<delim>]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020018415 This extracts the first occurrence of the parameter <name> in the query
18416 string, which begins after either '?' or <delim>, and which ends before '&',
Willy Tarreau1ede1da2015-05-07 16:06:18 +020018417 ';' or <delim>. The parameter name is case-sensitive. If no name is given,
18418 any parameter will match, and the first one will be returned. The result is
18419 a string corresponding to the value of the parameter <name> as presented in
18420 the request (no URL decoding is performed). This can be used for session
Willy Tarreau74ca5042013-06-11 23:12:07 +020018421 stickiness based on a client ID, to extract an application cookie passed as a
18422 URL parameter, or in ACLs to apply some checks. Note that the ACL version of
Willy Tarreau1ede1da2015-05-07 16:06:18 +020018423 this fetch iterates over multiple parameters and will iteratively report all
18424 parameters values if no name is given
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018425
Willy Tarreau74ca5042013-06-11 23:12:07 +020018426 ACL derivatives :
18427 urlp(<name>[,<delim>]) : exact string match
18428 urlp_beg(<name>[,<delim>]) : prefix match
18429 urlp_dir(<name>[,<delim>]) : subdir match
18430 urlp_dom(<name>[,<delim>]) : domain match
18431 urlp_end(<name>[,<delim>]) : suffix match
18432 urlp_len(<name>[,<delim>]) : length match
18433 urlp_reg(<name>[,<delim>]) : regex match
18434 urlp_sub(<name>[,<delim>]) : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018435
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018436
Willy Tarreau74ca5042013-06-11 23:12:07 +020018437 Example :
18438 # match http://example.com/foo?PHPSESSIONID=some_id
18439 stick on urlp(PHPSESSIONID)
18440 # match http://example.com/foo;JSESSIONID=some_id
18441 stick on urlp(JSESSIONID,;)
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018442
Jarno Huuskonen676f6222017-03-30 09:19:45 +030018443urlp_val([<name>[,<delim>]]) : integer
Willy Tarreau74ca5042013-06-11 23:12:07 +020018444 See "urlp" above. This one extracts the URL parameter <name> in the request
18445 and converts it to an integer value. This can be used for session stickiness
18446 based on a user ID for example, or with ACLs to match a page number or price.
Willy Tarreaua9fddca2012-07-31 07:51:48 +020018447
Dragan Dosen0070cd52016-06-16 12:19:49 +020018448url32 : integer
18449 This returns a 32-bit hash of the value obtained by concatenating the first
18450 Host header and the whole URL including parameters (not only the path part of
18451 the request, as in the "base32" fetch above). This is useful to track per-URL
18452 activity. A shorter hash is stored, saving a lot of memory. The output type
18453 is an unsigned integer.
18454
18455url32+src : binary
18456 This returns the concatenation of the "url32" fetch and the "src" fetch. The
18457 resulting type is of type binary, with a size of 8 or 20 bytes depending on
18458 the source address family. This can be used to track per-IP, per-URL counters.
18459
Christopher Faulet16032ab2020-04-30 11:30:00 +020018460
Christopher Faulete596d182020-05-05 17:46:34 +0200184617.3.7. Fetching samples for developers
Christopher Fauletd47941d2020-01-08 14:40:19 +010018462---------------------------------------
18463
18464This set of sample fetch methods is reserved to developers and must never be
18465used on a production environment, except on developer demand, for debugging
18466purposes. Moreover, no special care will be taken on backwards compatibility.
18467There is no warranty the following sample fetches will never change, be renamed
18468or simply removed. So be really careful if you should use one of them. To avoid
18469any ambiguity, these sample fetches are placed in the dedicated scope "internal",
18470for instance "internal.strm.is_htx".
18471
18472internal.htx.data : integer
18473 Returns the size in bytes used by data in the HTX message associated to a
18474 channel. The channel is chosen depending on the sample direction.
18475
18476internal.htx.free : integer
18477 Returns the free space (size - used) in bytes in the HTX message associated
18478 to a channel. The channel is chosen depending on the sample direction.
18479
18480internal.htx.free_data : integer
18481 Returns the free space for the data in bytes in the HTX message associated to
18482 a channel. The channel is chosen depending on the sample direction.
18483
18484internal.htx.has_eom : boolean
18485 Returns true if the HTX message associated to a channel contains an
18486 end-of-message block (EOM). Otherwise, it returns false. The channel is
18487 chosen depending on the sample direction.
18488
18489internal.htx.nbblks : integer
18490 Returns the number of blocks present in the HTX message associated to a
18491 channel. The channel is chosen depending on the sample direction.
18492
18493internal.htx.size : integer
18494 Returns the total size in bytes of the HTX message associated to a
18495 channel. The channel is chosen depending on the sample direction.
18496
18497internal.htx.used : integer
18498 Returns the total size used in bytes (data + metadata) in the HTX message
18499 associated to a channel. The channel is chosen depending on the sample
18500 direction.
18501
18502internal.htx_blk.size(<idx>) : integer
18503 Returns the size of the block at the position <idx> in the HTX message
18504 associated to a channel or 0 if it does not exist. The channel is chosen
18505 depending on the sample direction. <idx> may be any positive integer or one
18506 of the special value :
18507 * head : The oldest inserted block
18508 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050018509 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010018510
18511internal.htx_blk.type(<idx>) : string
18512 Returns the type of the block at the position <idx> in the HTX message
18513 associated to a channel or "HTX_BLK_UNUSED" if it does not exist. The channel
18514 is chosen depending on the sample direction. <idx> may be any positive
18515 integer or one of the special value :
18516 * head : The oldest inserted block
18517 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050018518 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010018519
18520internal.htx_blk.data(<idx>) : binary
18521 Returns the value of the DATA block at the position <idx> in the HTX message
18522 associated to a channel or an empty string if it does not exist or if it is
18523 not a DATA block. The channel is chosen depending on the sample direction.
18524 <idx> may be any positive integer or one of the special value :
18525
18526 * head : The oldest inserted block
18527 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050018528 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010018529
18530internal.htx_blk.hdrname(<idx>) : string
18531 Returns the header name of the HEADER block at the position <idx> in the HTX
18532 message associated to a channel or an empty string if it does not exist or if
18533 it is not an HEADER block. The channel is chosen depending on the sample
18534 direction. <idx> may be any positive integer or one of the special value :
18535
18536 * head : The oldest inserted block
18537 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050018538 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010018539
18540internal.htx_blk.hdrval(<idx>) : string
18541 Returns the header value of the HEADER block at the position <idx> in the HTX
18542 message associated to a channel or an empty string if it does not exist or if
18543 it is not an HEADER block. The channel is chosen depending on the sample
18544 direction. <idx> may be any positive integer or one of the special value :
18545
18546 * head : The oldest inserted block
18547 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050018548 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010018549
18550internal.htx_blk.start_line(<idx>) : string
18551 Returns the value of the REQ_SL or RES_SL block at the position <idx> in the
18552 HTX message associated to a channel or an empty string if it does not exist
18553 or if it is not a SL block. The channel is chosen depending on the sample
18554 direction. <idx> may be any positive integer or one of the special value :
18555
18556 * head : The oldest inserted block
18557 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050018558 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010018559
18560internal.strm.is_htx : boolean
18561 Returns true if the current stream is an HTX stream. It means the data in the
18562 channels buffers are stored using the internal HTX representation. Otherwise,
18563 it returns false.
18564
18565
Willy Tarreau74ca5042013-06-11 23:12:07 +0200185667.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018567---------------------
Willy Tarreauced27012008-01-17 20:35:34 +010018568
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018569Some predefined ACLs are hard-coded so that they do not have to be declared in
18570every frontend which needs them. They all have their names in upper case in
Patrick Mézard2382ad62010-05-09 10:43:32 +020018571order to avoid confusion. Their equivalence is provided below.
Willy Tarreauced27012008-01-17 20:35:34 +010018572
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018573ACL name Equivalent to Usage
18574---------------+-----------------------------+---------------------------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018575FALSE always_false never match
Willy Tarreau2492d5b2009-07-11 00:06:00 +020018576HTTP req_proto_http match if protocol is valid HTTP
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018577HTTP_1.0 req_ver 1.0 match HTTP version 1.0
18578HTTP_1.1 req_ver 1.1 match HTTP version 1.1
Willy Tarreaud63335a2010-02-26 12:56:52 +010018579HTTP_CONTENT hdr_val(content-length) gt 0 match an existing content-length
18580HTTP_URL_ABS url_reg ^[^/:]*:// match absolute URL with scheme
18581HTTP_URL_SLASH url_beg / match URL beginning with "/"
18582HTTP_URL_STAR url * match URL equal to "*"
18583LOCALHOST src 127.0.0.1/8 match connection from local host
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018584METH_CONNECT method CONNECT match HTTP CONNECT method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020018585METH_DELETE method DELETE match HTTP DELETE method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018586METH_GET method GET HEAD match HTTP GET or HEAD method
18587METH_HEAD method HEAD match HTTP HEAD method
18588METH_OPTIONS method OPTIONS match HTTP OPTIONS method
18589METH_POST method POST match HTTP POST method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020018590METH_PUT method PUT match HTTP PUT method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018591METH_TRACE method TRACE match HTTP TRACE method
Emeric Brunbede3d02009-06-30 17:54:00 +020018592RDP_COOKIE req_rdp_cookie_cnt gt 0 match presence of an RDP cookie
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018593REQ_CONTENT req_len gt 0 match data in the request buffer
Willy Tarreaud63335a2010-02-26 12:56:52 +010018594TRUE always_true always match
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018595WAIT_END wait_end wait for end of content analysis
18596---------------+-----------------------------+---------------------------------
Willy Tarreauced27012008-01-17 20:35:34 +010018597
Willy Tarreaub937b7e2010-01-12 15:27:54 +010018598
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200185998. Logging
18600----------
Willy Tarreau844e3c52008-01-11 16:28:18 +010018601
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018602One of HAProxy's strong points certainly lies is its precise logs. It probably
18603provides the finest level of information available for such a product, which is
18604very important for troubleshooting complex environments. Standard information
18605provided in logs include client ports, TCP/HTTP state timers, precise session
18606state at termination and precise termination cause, information about decisions
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010018607to direct traffic to a server, and of course the ability to capture arbitrary
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018608headers.
18609
18610In order to improve administrators reactivity, it offers a great transparency
18611about encountered problems, both internal and external, and it is possible to
18612send logs to different sources at the same time with different level filters :
18613
18614 - global process-level logs (system errors, start/stop, etc..)
18615 - per-instance system and internal errors (lack of resource, bugs, ...)
18616 - per-instance external troubles (servers up/down, max connections)
18617 - per-instance activity (client connections), either at the establishment or
18618 at the termination.
Davor Ocelice9ed2812017-12-25 17:49:28 +010018619 - per-request control of log-level, e.g.
Jim Freeman9e8714b2015-05-26 09:16:34 -060018620 http-request set-log-level silent if sensitive_request
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018621
18622The ability to distribute different levels of logs to different log servers
18623allow several production teams to interact and to fix their problems as soon
18624as possible. For example, the system team might monitor system-wide errors,
18625while the application team might be monitoring the up/down for their servers in
18626real time, and the security team might analyze the activity logs with one hour
18627delay.
18628
18629
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200186308.1. Log levels
18631---------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018632
Simon Hormandf791f52011-05-29 15:01:10 +090018633TCP and HTTP connections can be logged with information such as the date, time,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018634source IP address, destination address, connection duration, response times,
Simon Hormandf791f52011-05-29 15:01:10 +090018635HTTP request, HTTP return code, number of bytes transmitted, conditions
18636in which the session ended, and even exchanged cookies values. For example
18637track a particular user's problems. All messages may be sent to up to two
18638syslog servers. Check the "log" keyword in section 4.2 for more information
18639about log facilities.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018640
18641
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200186428.2. Log formats
18643----------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018644
William Lallemand48940402012-01-30 16:47:22 +010018645HAProxy supports 5 log formats. Several fields are common between these formats
Simon Hormandf791f52011-05-29 15:01:10 +090018646and will be detailed in the following sections. A few of them may vary
18647slightly with the configuration, due to indicators specific to certain
18648options. The supported formats are as follows :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018649
18650 - the default format, which is very basic and very rarely used. It only
18651 provides very basic information about the incoming connection at the moment
18652 it is accepted : source IP:port, destination IP:port, and frontend-name.
18653 This mode will eventually disappear so it will not be described to great
18654 extents.
18655
18656 - the TCP format, which is more advanced. This format is enabled when "option
18657 tcplog" is set on the frontend. HAProxy will then usually wait for the
18658 connection to terminate before logging. This format provides much richer
18659 information, such as timers, connection counts, queue size, etc... This
18660 format is recommended for pure TCP proxies.
18661
18662 - the HTTP format, which is the most advanced for HTTP proxying. This format
18663 is enabled when "option httplog" is set on the frontend. It provides the
18664 same information as the TCP format with some HTTP-specific fields such as
18665 the request, the status code, and captures of headers and cookies. This
18666 format is recommended for HTTP proxies.
18667
Emeric Brun3a058f32009-06-30 18:26:00 +020018668 - the CLF HTTP format, which is equivalent to the HTTP format, but with the
18669 fields arranged in the same order as the CLF format. In this mode, all
18670 timers, captures, flags, etc... appear one per field after the end of the
18671 common fields, in the same order they appear in the standard HTTP format.
18672
William Lallemand48940402012-01-30 16:47:22 +010018673 - the custom log format, allows you to make your own log line.
18674
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018675Next sections will go deeper into details for each of these formats. Format
18676specification will be performed on a "field" basis. Unless stated otherwise, a
18677field is a portion of text delimited by any number of spaces. Since syslog
18678servers are susceptible of inserting fields at the beginning of a line, it is
18679always assumed that the first field is the one containing the process name and
18680identifier.
18681
18682Note : Since log lines may be quite long, the log examples in sections below
18683 might be broken into multiple lines. The example log lines will be
18684 prefixed with 3 closing angle brackets ('>>>') and each time a log is
18685 broken into multiple lines, each non-final line will end with a
18686 backslash ('\') and the next line will start indented by two characters.
18687
18688
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200186898.2.1. Default log format
18690-------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018691
18692This format is used when no specific option is set. The log is emitted as soon
18693as the connection is accepted. One should note that this currently is the only
18694format which logs the request's destination IP and ports.
18695
18696 Example :
18697 listen www
18698 mode http
18699 log global
18700 server srv1 127.0.0.1:8000
18701
18702 >>> Feb 6 12:12:09 localhost \
18703 haproxy[14385]: Connect from 10.0.1.2:33312 to 10.0.3.31:8012 \
18704 (www/HTTP)
18705
18706 Field Format Extract from the example above
18707 1 process_name '[' pid ']:' haproxy[14385]:
18708 2 'Connect from' Connect from
18709 3 source_ip ':' source_port 10.0.1.2:33312
18710 4 'to' to
18711 5 destination_ip ':' destination_port 10.0.3.31:8012
18712 6 '(' frontend_name '/' mode ')' (www/HTTP)
18713
18714Detailed fields description :
18715 - "source_ip" is the IP address of the client which initiated the connection.
18716 - "source_port" is the TCP port of the client which initiated the connection.
18717 - "destination_ip" is the IP address the client connected to.
18718 - "destination_port" is the TCP port the client connected to.
18719 - "frontend_name" is the name of the frontend (or listener) which received
18720 and processed the connection.
18721 - "mode is the mode the frontend is operating (TCP or HTTP).
18722
Willy Tarreauceb24bc2010-11-09 12:46:41 +010018723In case of a UNIX socket, the source and destination addresses are marked as
18724"unix:" and the ports reflect the internal ID of the socket which accepted the
18725connection (the same ID as reported in the stats).
18726
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018727It is advised not to use this deprecated format for newer installations as it
18728will eventually disappear.
18729
18730
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200187318.2.2. TCP log format
18732---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018733
18734The TCP format is used when "option tcplog" is specified in the frontend, and
18735is the recommended format for pure TCP proxies. It provides a lot of precious
18736information for troubleshooting. Since this format includes timers and byte
18737counts, the log is normally emitted at the end of the session. It can be
18738emitted earlier if "option logasap" is specified, which makes sense in most
18739environments with long sessions such as remote terminals. Sessions which match
18740the "monitor" rules are never logged. It is also possible not to emit logs for
18741sessions for which no data were exchanged between the client and the server, by
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020018742specifying "option dontlognull" in the frontend. Successful connections will
18743not be logged if "option dontlog-normal" is specified in the frontend. A few
18744fields may slightly vary depending on some configuration options, those are
18745marked with a star ('*') after the field name below.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018746
18747 Example :
18748 frontend fnt
18749 mode tcp
18750 option tcplog
18751 log global
18752 default_backend bck
18753
18754 backend bck
18755 server srv1 127.0.0.1:8000
18756
18757 >>> Feb 6 12:12:56 localhost \
18758 haproxy[14387]: 10.0.1.2:33313 [06/Feb/2009:12:12:51.443] fnt \
18759 bck/srv1 0/0/5007 212 -- 0/0/0/0/3 0/0
18760
18761 Field Format Extract from the example above
18762 1 process_name '[' pid ']:' haproxy[14387]:
18763 2 client_ip ':' client_port 10.0.1.2:33313
18764 3 '[' accept_date ']' [06/Feb/2009:12:12:51.443]
18765 4 frontend_name fnt
18766 5 backend_name '/' server_name bck/srv1
18767 6 Tw '/' Tc '/' Tt* 0/0/5007
18768 7 bytes_read* 212
18769 8 termination_state --
18770 9 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 0/0/0/0/3
18771 10 srv_queue '/' backend_queue 0/0
18772
18773Detailed fields description :
18774 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010018775 connection to haproxy. If the connection was accepted on a UNIX socket
18776 instead, the IP address would be replaced with the word "unix". Note that
18777 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010018778 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010018779 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010018780 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018781
18782 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010018783 If the connection was accepted on a UNIX socket instead, the port would be
18784 replaced with the ID of the accepting socket, which is also reported in the
18785 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018786
18787 - "accept_date" is the exact date when the connection was received by haproxy
18788 (which might be very slightly different from the date observed on the
18789 network if there was some queuing in the system's backlog). This is usually
Willy Tarreau590a0512018-09-05 11:56:48 +020018790 the same date which may appear in any upstream firewall's log. When used in
18791 HTTP mode, the accept_date field will be reset to the first moment the
18792 connection is ready to receive a new request (end of previous response for
18793 HTTP/1, immediately after previous request for HTTP/2).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018794
18795 - "frontend_name" is the name of the frontend (or listener) which received
18796 and processed the connection.
18797
18798 - "backend_name" is the name of the backend (or listener) which was selected
18799 to manage the connection to the server. This will be the same as the
18800 frontend if no switching rule has been applied, which is common for TCP
18801 applications.
18802
18803 - "server_name" is the name of the last server to which the connection was
18804 sent, which might differ from the first one if there were connection errors
18805 and a redispatch occurred. Note that this server belongs to the backend
18806 which processed the request. If the connection was aborted before reaching
18807 a server, "<NOSRV>" is indicated instead of a server name.
18808
18809 - "Tw" is the total time in milliseconds spent waiting in the various queues.
18810 It can be "-1" if the connection was aborted before reaching the queue.
18811 See "Timers" below for more details.
18812
18813 - "Tc" is the total time in milliseconds spent waiting for the connection to
18814 establish to the final server, including retries. It can be "-1" if the
18815 connection was aborted before a connection could be established. See
18816 "Timers" below for more details.
18817
18818 - "Tt" is the total time in milliseconds elapsed between the accept and the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030018819 last close. It covers all possible processing. There is one exception, if
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018820 "option logasap" was specified, then the time counting stops at the moment
18821 the log is emitted. In this case, a '+' sign is prepended before the value,
18822 indicating that the final one will be larger. See "Timers" below for more
18823 details.
18824
18825 - "bytes_read" is the total number of bytes transmitted from the server to
18826 the client when the log is emitted. If "option logasap" is specified, the
18827 this value will be prefixed with a '+' sign indicating that the final one
18828 may be larger. Please note that this value is a 64-bit counter, so log
18829 analysis tools must be able to handle it without overflowing.
18830
18831 - "termination_state" is the condition the session was in when the session
18832 ended. This indicates the session state, which side caused the end of
18833 session to happen, and for what reason (timeout, error, ...). The normal
18834 flags should be "--", indicating the session was closed by either end with
18835 no data remaining in buffers. See below "Session state at disconnection"
18836 for more details.
18837
18838 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040018839 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018840 limits have been reached. For instance, if actconn is close to 512 when
18841 multiple connection errors occur, chances are high that the system limits
18842 the process to use a maximum of 1024 file descriptors and that all of them
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018843 are used. See section 3 "Global parameters" to find how to tune the system.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018844
18845 - "feconn" is the total number of concurrent connections on the frontend when
18846 the session was logged. It is useful to estimate the amount of resource
18847 required to sustain high loads, and to detect when the frontend's "maxconn"
18848 has been reached. Most often when this value increases by huge jumps, it is
18849 because there is congestion on the backend servers, but sometimes it can be
18850 caused by a denial of service attack.
18851
18852 - "beconn" is the total number of concurrent connections handled by the
18853 backend when the session was logged. It includes the total number of
18854 concurrent connections active on servers as well as the number of
18855 connections pending in queues. It is useful to estimate the amount of
18856 additional servers needed to support high loads for a given application.
18857 Most often when this value increases by huge jumps, it is because there is
18858 congestion on the backend servers, but sometimes it can be caused by a
18859 denial of service attack.
18860
18861 - "srv_conn" is the total number of concurrent connections still active on
18862 the server when the session was logged. It can never exceed the server's
18863 configured "maxconn" parameter. If this value is very often close or equal
18864 to the server's "maxconn", it means that traffic regulation is involved a
18865 lot, meaning that either the server's maxconn value is too low, or that
18866 there aren't enough servers to process the load with an optimal response
18867 time. When only one of the server's "srv_conn" is high, it usually means
18868 that this server has some trouble causing the connections to take longer to
18869 be processed than on other servers.
18870
18871 - "retries" is the number of connection retries experienced by this session
18872 when trying to connect to the server. It must normally be zero, unless a
18873 server is being stopped at the same moment the connection was attempted.
18874 Frequent retries generally indicate either a network problem between
18875 haproxy and the server, or a misconfigured system backlog on the server
18876 preventing new connections from being queued. This field may optionally be
18877 prefixed with a '+' sign, indicating that the session has experienced a
18878 redispatch after the maximal retry count has been reached on the initial
18879 server. In this case, the server name appearing in the log is the one the
18880 connection was redispatched to, and not the first one, though both may
18881 sometimes be the same in case of hashing for instance. So as a general rule
18882 of thumb, when a '+' is present in front of the retry count, this count
18883 should not be attributed to the logged server.
18884
18885 - "srv_queue" is the total number of requests which were processed before
18886 this one in the server queue. It is zero when the request has not gone
18887 through the server queue. It makes it possible to estimate the approximate
18888 server's response time by dividing the time spent in queue by the number of
18889 requests in the queue. It is worth noting that if a session experiences a
18890 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010018891 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018892 backend queue unless a redispatch occurs.
18893
18894 - "backend_queue" is the total number of requests which were processed before
18895 this one in the backend's global queue. It is zero when the request has not
18896 gone through the global queue. It makes it possible to estimate the average
18897 queue length, which easily translates into a number of missing servers when
18898 divided by a server's "maxconn" parameter. It is worth noting that if a
18899 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010018900 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018901 through both the server queue and the backend queue unless a redispatch
18902 occurs.
18903
18904
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200189058.2.3. HTTP log format
18906----------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018907
18908The HTTP format is the most complete and the best suited for HTTP proxies. It
18909is enabled by when "option httplog" is specified in the frontend. It provides
18910the same level of information as the TCP format with additional features which
18911are specific to the HTTP protocol. Just like the TCP format, the log is usually
18912emitted at the end of the session, unless "option logasap" is specified, which
18913generally only makes sense for download sites. A session which matches the
18914"monitor" rules will never logged. It is also possible not to log sessions for
18915which no data were sent by the client by specifying "option dontlognull" in the
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020018916frontend. Successful connections will not be logged if "option dontlog-normal"
18917is specified in the frontend.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018918
18919Most fields are shared with the TCP log, some being different. A few fields may
18920slightly vary depending on some configuration options. Those ones are marked
18921with a star ('*') after the field name below.
18922
18923 Example :
18924 frontend http-in
18925 mode http
18926 option httplog
18927 log global
18928 default_backend bck
18929
18930 backend static
18931 server srv1 127.0.0.1:8000
18932
18933 >>> Feb 6 12:14:14 localhost \
18934 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
18935 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 +010018936 {} "GET /index.html HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018937
18938 Field Format Extract from the example above
18939 1 process_name '[' pid ']:' haproxy[14389]:
18940 2 client_ip ':' client_port 10.0.1.2:33317
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020018941 3 '[' request_date ']' [06/Feb/2009:12:14:14.655]
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018942 4 frontend_name http-in
18943 5 backend_name '/' server_name static/srv1
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020018944 6 TR '/' Tw '/' Tc '/' Tr '/' Ta* 10/0/30/69/109
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018945 7 status_code 200
18946 8 bytes_read* 2750
18947 9 captured_request_cookie -
18948 10 captured_response_cookie -
18949 11 termination_state ----
18950 12 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 1/1/1/1/0
18951 13 srv_queue '/' backend_queue 0/0
18952 14 '{' captured_request_headers* '}' {haproxy.1wt.eu}
18953 15 '{' captured_response_headers* '}' {}
18954 16 '"' http_request '"' "GET /index.html HTTP/1.1"
Willy Tarreaud72758d2010-01-12 10:42:19 +010018955
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018956Detailed fields description :
18957 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010018958 connection to haproxy. If the connection was accepted on a UNIX socket
18959 instead, the IP address would be replaced with the word "unix". Note that
18960 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010018961 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010018962 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010018963 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018964
18965 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010018966 If the connection was accepted on a UNIX socket instead, the port would be
18967 replaced with the ID of the accepting socket, which is also reported in the
18968 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018969
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020018970 - "request_date" is the exact date when the first byte of the HTTP request
18971 was received by haproxy (log field %tr).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018972
18973 - "frontend_name" is the name of the frontend (or listener) which received
18974 and processed the connection.
18975
18976 - "backend_name" is the name of the backend (or listener) which was selected
18977 to manage the connection to the server. This will be the same as the
18978 frontend if no switching rule has been applied.
18979
18980 - "server_name" is the name of the last server to which the connection was
18981 sent, which might differ from the first one if there were connection errors
18982 and a redispatch occurred. Note that this server belongs to the backend
18983 which processed the request. If the request was aborted before reaching a
18984 server, "<NOSRV>" is indicated instead of a server name. If the request was
18985 intercepted by the stats subsystem, "<STATS>" is indicated instead.
18986
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020018987 - "TR" is the total time in milliseconds spent waiting for a full HTTP
18988 request from the client (not counting body) after the first byte was
18989 received. It can be "-1" if the connection was aborted before a complete
John Roeslerfb2fce12019-07-10 15:45:51 -050018990 request could be received or a bad request was received. It should
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020018991 always be very small because a request generally fits in one single packet.
18992 Large times here generally indicate network issues between the client and
Willy Tarreau590a0512018-09-05 11:56:48 +020018993 haproxy or requests being typed by hand. See section 8.4 "Timing Events"
18994 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018995
18996 - "Tw" is the total time in milliseconds spent waiting in the various queues.
18997 It can be "-1" if the connection was aborted before reaching the queue.
Willy Tarreau590a0512018-09-05 11:56:48 +020018998 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018999
19000 - "Tc" is the total time in milliseconds spent waiting for the connection to
19001 establish to the final server, including retries. It can be "-1" if the
Willy Tarreau590a0512018-09-05 11:56:48 +020019002 request was aborted before a connection could be established. See section
19003 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019004
19005 - "Tr" is the total time in milliseconds spent waiting for the server to send
19006 a full HTTP response, not counting data. It can be "-1" if the request was
19007 aborted before a complete response could be received. It generally matches
19008 the server's processing time for the request, though it may be altered by
19009 the amount of data sent by the client to the server. Large times here on
Willy Tarreau590a0512018-09-05 11:56:48 +020019010 "GET" requests generally indicate an overloaded server. See section 8.4
19011 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019012
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019013 - "Ta" is the time the request remained active in haproxy, which is the total
19014 time in milliseconds elapsed between the first byte of the request was
19015 received and the last byte of response was sent. It covers all possible
19016 processing except the handshake (see Th) and idle time (see Ti). There is
19017 one exception, if "option logasap" was specified, then the time counting
19018 stops at the moment the log is emitted. In this case, a '+' sign is
19019 prepended before the value, indicating that the final one will be larger.
Willy Tarreau590a0512018-09-05 11:56:48 +020019020 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019021
19022 - "status_code" is the HTTP status code returned to the client. This status
19023 is generally set by the server, but it might also be set by haproxy when
19024 the server cannot be reached or when its response is blocked by haproxy.
19025
19026 - "bytes_read" is the total number of bytes transmitted to the client when
19027 the log is emitted. This does include HTTP headers. If "option logasap" is
John Roeslerfb2fce12019-07-10 15:45:51 -050019028 specified, this value will be prefixed with a '+' sign indicating that
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019029 the final one may be larger. Please note that this value is a 64-bit
19030 counter, so log analysis tools must be able to handle it without
19031 overflowing.
19032
19033 - "captured_request_cookie" is an optional "name=value" entry indicating that
19034 the client had this cookie in the request. The cookie name and its maximum
19035 length are defined by the "capture cookie" statement in the frontend
19036 configuration. The field is a single dash ('-') when the option is not
19037 set. Only one cookie may be captured, it is generally used to track session
19038 ID exchanges between a client and a server to detect session crossing
19039 between clients due to application bugs. For more details, please consult
19040 the section "Capturing HTTP headers and cookies" below.
19041
19042 - "captured_response_cookie" is an optional "name=value" entry indicating
19043 that the server has returned a cookie with its response. The cookie name
19044 and its maximum length are defined by the "capture cookie" statement in the
19045 frontend configuration. The field is a single dash ('-') when the option is
19046 not set. Only one cookie may be captured, it is generally used to track
19047 session ID exchanges between a client and a server to detect session
19048 crossing between clients due to application bugs. For more details, please
19049 consult the section "Capturing HTTP headers and cookies" below.
19050
19051 - "termination_state" is the condition the session was in when the session
19052 ended. This indicates the session state, which side caused the end of
19053 session to happen, for what reason (timeout, error, ...), just like in TCP
19054 logs, and information about persistence operations on cookies in the last
19055 two characters. The normal flags should begin with "--", indicating the
19056 session was closed by either end with no data remaining in buffers. See
19057 below "Session state at disconnection" for more details.
19058
19059 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040019060 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019061 limits have been reached. For instance, if actconn is close to 512 or 1024
19062 when multiple connection errors occur, chances are high that the system
19063 limits the process to use a maximum of 1024 file descriptors and that all
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019064 of them are used. See section 3 "Global parameters" to find how to tune the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019065 system.
19066
19067 - "feconn" is the total number of concurrent connections on the frontend when
19068 the session was logged. It is useful to estimate the amount of resource
19069 required to sustain high loads, and to detect when the frontend's "maxconn"
19070 has been reached. Most often when this value increases by huge jumps, it is
19071 because there is congestion on the backend servers, but sometimes it can be
19072 caused by a denial of service attack.
19073
19074 - "beconn" is the total number of concurrent connections handled by the
19075 backend when the session was logged. It includes the total number of
19076 concurrent connections active on servers as well as the number of
19077 connections pending in queues. It is useful to estimate the amount of
19078 additional servers needed to support high loads for a given application.
19079 Most often when this value increases by huge jumps, it is because there is
19080 congestion on the backend servers, but sometimes it can be caused by a
19081 denial of service attack.
19082
19083 - "srv_conn" is the total number of concurrent connections still active on
19084 the server when the session was logged. It can never exceed the server's
19085 configured "maxconn" parameter. If this value is very often close or equal
19086 to the server's "maxconn", it means that traffic regulation is involved a
19087 lot, meaning that either the server's maxconn value is too low, or that
19088 there aren't enough servers to process the load with an optimal response
19089 time. When only one of the server's "srv_conn" is high, it usually means
19090 that this server has some trouble causing the requests to take longer to be
19091 processed than on other servers.
19092
19093 - "retries" is the number of connection retries experienced by this session
19094 when trying to connect to the server. It must normally be zero, unless a
19095 server is being stopped at the same moment the connection was attempted.
19096 Frequent retries generally indicate either a network problem between
19097 haproxy and the server, or a misconfigured system backlog on the server
19098 preventing new connections from being queued. This field may optionally be
19099 prefixed with a '+' sign, indicating that the session has experienced a
19100 redispatch after the maximal retry count has been reached on the initial
19101 server. In this case, the server name appearing in the log is the one the
19102 connection was redispatched to, and not the first one, though both may
19103 sometimes be the same in case of hashing for instance. So as a general rule
19104 of thumb, when a '+' is present in front of the retry count, this count
19105 should not be attributed to the logged server.
19106
19107 - "srv_queue" is the total number of requests which were processed before
19108 this one in the server queue. It is zero when the request has not gone
19109 through the server queue. It makes it possible to estimate the approximate
19110 server's response time by dividing the time spent in queue by the number of
19111 requests in the queue. It is worth noting that if a session experiences a
19112 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010019113 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019114 backend queue unless a redispatch occurs.
19115
19116 - "backend_queue" is the total number of requests which were processed before
19117 this one in the backend's global queue. It is zero when the request has not
19118 gone through the global queue. It makes it possible to estimate the average
19119 queue length, which easily translates into a number of missing servers when
19120 divided by a server's "maxconn" parameter. It is worth noting that if a
19121 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010019122 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019123 through both the server queue and the backend queue unless a redispatch
19124 occurs.
19125
19126 - "captured_request_headers" is a list of headers captured in the request due
19127 to the presence of the "capture request header" statement in the frontend.
19128 Multiple headers can be captured, they will be delimited by a vertical bar
19129 ('|'). When no capture is enabled, the braces do not appear, causing a
19130 shift of remaining fields. It is important to note that this field may
19131 contain spaces, and that using it requires a smarter log parser than when
19132 it's not used. Please consult the section "Capturing HTTP headers and
19133 cookies" below for more details.
19134
19135 - "captured_response_headers" is a list of headers captured in the response
19136 due to the presence of the "capture response header" statement in the
19137 frontend. Multiple headers can be captured, they will be delimited by a
19138 vertical bar ('|'). When no capture is enabled, the braces do not appear,
19139 causing a shift of remaining fields. It is important to note that this
19140 field may contain spaces, and that using it requires a smarter log parser
19141 than when it's not used. Please consult the section "Capturing HTTP headers
19142 and cookies" below for more details.
19143
19144 - "http_request" is the complete HTTP request line, including the method,
19145 request and HTTP version string. Non-printable characters are encoded (see
19146 below the section "Non-printable characters"). This is always the last
19147 field, and it is always delimited by quotes and is the only one which can
19148 contain quotes. If new fields are added to the log format, they will be
19149 added before this field. This field might be truncated if the request is
19150 huge and does not fit in the standard syslog buffer (1024 characters). This
19151 is the reason why this field must always remain the last one.
19152
19153
Cyril Bontédc4d9032012-04-08 21:57:39 +0200191548.2.4. Custom log format
19155------------------------
William Lallemand48940402012-01-30 16:47:22 +010019156
Willy Tarreau2beef582012-12-20 17:22:52 +010019157The directive log-format allows you to customize the logs in http mode and tcp
William Lallemandbddd4fd2012-02-27 11:23:10 +010019158mode. It takes a string as argument.
William Lallemand48940402012-01-30 16:47:22 +010019159
Davor Ocelice9ed2812017-12-25 17:49:28 +010019160HAProxy understands some log format variables. % precedes log format variables.
William Lallemand48940402012-01-30 16:47:22 +010019161Variables can take arguments using braces ('{}'), and multiple arguments are
19162separated by commas within the braces. Flags may be added or removed by
19163prefixing them with a '+' or '-' sign.
19164
19165Special variable "%o" may be used to propagate its flags to all other
19166variables on the same format string. This is particularly handy with quoted
Dragan Dosen835b9212016-02-12 13:23:03 +010019167("Q") and escaped ("E") string formats.
William Lallemand48940402012-01-30 16:47:22 +010019168
Willy Tarreauc8368452012-12-21 00:09:23 +010019169If a variable is named between square brackets ('[' .. ']') then it is used
Willy Tarreaube722a22014-06-13 16:31:59 +020019170as a sample expression rule (see section 7.3). This it useful to add some
Willy Tarreauc8368452012-12-21 00:09:23 +010019171less common information such as the client's SSL certificate's DN, or to log
19172the key that would be used to store an entry into a stick table.
19173
Dragan Dosen1e3b16f2020-06-23 18:16:44 +020019174Note: spaces must be escaped. In configuration directives "log-format",
19175"log-format-sd" and "unique-id-format", spaces are considered as
19176delimiters and are merged. In order to emit a verbatim '%', it must be
19177preceded by another '%' resulting in '%%'.
William Lallemand48940402012-01-30 16:47:22 +010019178
Dragan Dosen835b9212016-02-12 13:23:03 +010019179Note: when using the RFC5424 syslog message format, the characters '"',
19180'\' and ']' inside PARAM-VALUE should be escaped with '\' as prefix (see
19181https://tools.ietf.org/html/rfc5424#section-6.3.3 for more details). In
19182such cases, the use of the flag "E" should be considered.
19183
William Lallemand48940402012-01-30 16:47:22 +010019184Flags are :
19185 * Q: quote a string
Jamie Gloudonaaa21002012-08-25 00:18:33 -040019186 * X: hexadecimal representation (IPs, Ports, %Ts, %rt, %pid)
Dragan Dosen835b9212016-02-12 13:23:03 +010019187 * E: escape characters '"', '\' and ']' in a string with '\' as prefix
19188 (intended purpose is for the RFC5424 structured-data log formats)
William Lallemand48940402012-01-30 16:47:22 +010019189
19190 Example:
19191
19192 log-format %T\ %t\ Some\ Text
19193 log-format %{+Q}o\ %t\ %s\ %{-Q}r
19194
Dragan Dosen835b9212016-02-12 13:23:03 +010019195 log-format-sd %{+Q,+E}o\ [exampleSDID@1234\ header=%[capture.req.hdr(0)]]
19196
William Lallemand48940402012-01-30 16:47:22 +010019197At the moment, the default HTTP format is defined this way :
19198
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019199 log-format "%ci:%cp [%tr] %ft %b/%s %TR/%Tw/%Tc/%Tr/%Ta %ST %B %CC \
19200 %CS %tsc %ac/%fc/%bc/%sc/%rc %sq/%bq %hr %hs %{+Q}r"
William Lallemand48940402012-01-30 16:47:22 +010019201
William Lallemandbddd4fd2012-02-27 11:23:10 +010019202the default CLF format is defined this way :
William Lallemand48940402012-01-30 16:47:22 +010019203
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019204 log-format "%{+Q}o %{-Q}ci - - [%trg] %r %ST %B \"\" \"\" %cp \
19205 %ms %ft %b %s %TR %Tw %Tc %Tr %Ta %tsc %ac %fc \
19206 %bc %sc %rc %sq %bq %CC %CS %hrl %hsl"
William Lallemand48940402012-01-30 16:47:22 +010019207
William Lallemandbddd4fd2012-02-27 11:23:10 +010019208and the default TCP format is defined this way :
19209
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019210 log-format "%ci:%cp [%t] %ft %b/%s %Tw/%Tc/%Tt %B %ts \
19211 %ac/%fc/%bc/%sc/%rc %sq/%bq"
William Lallemandbddd4fd2012-02-27 11:23:10 +010019212
William Lallemand48940402012-01-30 16:47:22 +010019213Please refer to the table below for currently defined variables :
19214
William Lallemandbddd4fd2012-02-27 11:23:10 +010019215 +---+------+-----------------------------------------------+-------------+
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020019216 | R | var | field name (8.2.2 and 8.2.3 for description) | type |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019217 +---+------+-----------------------------------------------+-------------+
19218 | | %o | special variable, apply flags on all next var | |
19219 +---+------+-----------------------------------------------+-------------+
Willy Tarreau2beef582012-12-20 17:22:52 +010019220 | | %B | bytes_read (from server to client) | numeric |
19221 | H | %CC | captured_request_cookie | string |
19222 | H | %CS | captured_response_cookie | string |
William Lallemand5f232402012-04-05 18:02:55 +020019223 | | %H | hostname | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000019224 | H | %HM | HTTP method (ex: POST) | string |
19225 | H | %HP | HTTP request URI without query string (path) | string |
Andrew Hayworthe63ac872015-07-31 16:14:16 +000019226 | H | %HQ | HTTP request URI query string (ex: ?bar=baz) | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000019227 | H | %HU | HTTP request URI (ex: /foo?bar=baz) | string |
19228 | H | %HV | HTTP version (ex: HTTP/1.0) | string |
William Lallemanda73203e2012-03-12 12:48:57 +010019229 | | %ID | unique-id | string |
Willy Tarreau4bf99632014-06-13 12:21:40 +020019230 | | %ST | status_code | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020019231 | | %T | gmt_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019232 | | %Ta | Active time of the request (from TR to end) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019233 | | %Tc | Tc | numeric |
Willy Tarreau27b639d2016-05-17 17:55:27 +020019234 | | %Td | Td = Tt - (Tq + Tw + Tc + Tr) | numeric |
Yuxans Yao4e25b012012-10-19 10:36:09 +080019235 | | %Tl | local_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019236 | | %Th | connection handshake time (SSL, PROXY proto) | numeric |
19237 | H | %Ti | idle time before the HTTP request | numeric |
19238 | H | %Tq | Th + Ti + TR | numeric |
19239 | H | %TR | time to receive the full request from 1st byte| numeric |
19240 | H | %Tr | Tr (response time) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020019241 | | %Ts | timestamp | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019242 | | %Tt | Tt | numeric |
Damien Claisse57c8eb92020-04-28 12:09:19 +000019243 | | %Tu | Tu | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019244 | | %Tw | Tw | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010019245 | | %U | bytes_uploaded (from client to server) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019246 | | %ac | actconn | numeric |
19247 | | %b | backend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010019248 | | %bc | beconn (backend concurrent connections) | numeric |
19249 | | %bi | backend_source_ip (connecting address) | IP |
19250 | | %bp | backend_source_port (connecting address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019251 | | %bq | backend_queue | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010019252 | | %ci | client_ip (accepted address) | IP |
19253 | | %cp | client_port (accepted address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019254 | | %f | frontend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010019255 | | %fc | feconn (frontend concurrent connections) | numeric |
19256 | | %fi | frontend_ip (accepting address) | IP |
19257 | | %fp | frontend_port (accepting address) | numeric |
Willy Tarreau773d65f2012-10-12 14:56:11 +020019258 | | %ft | frontend_name_transport ('~' suffix for SSL) | string |
Willy Tarreau7346acb2014-08-28 15:03:15 +020019259 | | %lc | frontend_log_counter | numeric |
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020019260 | | %hr | captured_request_headers default style | string |
19261 | | %hrl | captured_request_headers CLF style | string list |
19262 | | %hs | captured_response_headers default style | string |
19263 | | %hsl | captured_response_headers CLF style | string list |
Willy Tarreau812c88e2015-08-09 10:56:35 +020019264 | | %ms | accept date milliseconds (left-padded with 0) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020019265 | | %pid | PID | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020019266 | H | %r | http_request | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019267 | | %rc | retries | numeric |
Willy Tarreau1f0da242014-01-25 11:01:50 +010019268 | | %rt | request_counter (HTTP req or TCP session) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019269 | | %s | server_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010019270 | | %sc | srv_conn (server concurrent connections) | numeric |
19271 | | %si | server_IP (target address) | IP |
19272 | | %sp | server_port (target address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019273 | | %sq | srv_queue | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020019274 | S | %sslc| ssl_ciphers (ex: AES-SHA) | string |
19275 | S | %sslv| ssl_version (ex: TLSv1) | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010019276 | | %t | date_time (with millisecond resolution) | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019277 | H | %tr | date_time of HTTP request | date |
19278 | H | %trg | gmt_date_time of start of HTTP request | date |
Jens Bissinger15c64ff2018-08-23 14:11:27 +020019279 | H | %trl | local_date_time of start of HTTP request | date |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019280 | | %ts | termination_state | string |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020019281 | H | %tsc | termination_state with cookie status | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019282 +---+------+-----------------------------------------------+-------------+
William Lallemand48940402012-01-30 16:47:22 +010019283
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020019284 R = Restrictions : H = mode http only ; S = SSL only
William Lallemand48940402012-01-30 16:47:22 +010019285
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010019286
192878.2.5. Error log format
19288-----------------------
19289
19290When an incoming connection fails due to an SSL handshake or an invalid PROXY
19291protocol header, haproxy will log the event using a shorter, fixed line format.
19292By default, logs are emitted at the LOG_INFO level, unless the option
19293"log-separate-errors" is set in the backend, in which case the LOG_ERR level
Davor Ocelice9ed2812017-12-25 17:49:28 +010019294will be used. Connections on which no data are exchanged (e.g. probes) are not
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010019295logged if the "dontlognull" option is set.
19296
19297The format looks like this :
19298
19299 >>> Dec 3 18:27:14 localhost \
19300 haproxy[6103]: 127.0.0.1:56059 [03/Dec/2012:17:35:10.380] frt/f1: \
19301 Connection error during SSL handshake
19302
19303 Field Format Extract from the example above
19304 1 process_name '[' pid ']:' haproxy[6103]:
19305 2 client_ip ':' client_port 127.0.0.1:56059
19306 3 '[' accept_date ']' [03/Dec/2012:17:35:10.380]
19307 4 frontend_name "/" bind_name ":" frt/f1:
19308 5 message Connection error during SSL handshake
19309
19310These fields just provide minimal information to help debugging connection
19311failures.
19312
19313
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200193148.3. Advanced logging options
19315-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019316
19317Some advanced logging options are often looked for but are not easy to find out
19318just by looking at the various options. Here is an entry point for the few
19319options which can enable better logging. Please refer to the keywords reference
19320for more information about their usage.
19321
19322
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200193238.3.1. Disabling logging of external tests
19324------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019325
19326It is quite common to have some monitoring tools perform health checks on
19327haproxy. Sometimes it will be a layer 3 load-balancer such as LVS or any
19328commercial load-balancer, and sometimes it will simply be a more complete
19329monitoring system such as Nagios. When the tests are very frequent, users often
19330ask how to disable logging for those checks. There are three possibilities :
19331
19332 - if connections come from everywhere and are just TCP probes, it is often
19333 desired to simply disable logging of connections without data exchange, by
19334 setting "option dontlognull" in the frontend. It also disables logging of
19335 port scans, which may or may not be desired.
19336
19337 - if the connection come from a known source network, use "monitor-net" to
19338 declare this network as monitoring only. Any host in this network will then
19339 only be able to perform health checks, and their requests will not be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030019340 logged. This is generally appropriate to designate a list of equipment
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019341 such as other load-balancers.
19342
19343 - if the tests are performed on a known URI, use "monitor-uri" to declare
19344 this URI as dedicated to monitoring. Any host sending this request will
19345 only get the result of a health-check, and the request will not be logged.
19346
19347
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200193488.3.2. Logging before waiting for the session to terminate
19349----------------------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019350
19351The problem with logging at end of connection is that you have no clue about
19352what is happening during very long sessions, such as remote terminal sessions
19353or large file downloads. This problem can be worked around by specifying
Davor Ocelice9ed2812017-12-25 17:49:28 +010019354"option logasap" in the frontend. HAProxy will then log as soon as possible,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019355just before data transfer begins. This means that in case of TCP, it will still
19356log the connection status to the server, and in case of HTTP, it will log just
19357after processing the server headers. In this case, the number of bytes reported
19358is the number of header bytes sent to the client. In order to avoid confusion
19359with normal logs, the total time field and the number of bytes are prefixed
19360with a '+' sign which means that real numbers are certainly larger.
19361
19362
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200193638.3.3. Raising log level upon errors
19364------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020019365
19366Sometimes it is more convenient to separate normal traffic from errors logs,
19367for instance in order to ease error monitoring from log files. When the option
19368"log-separate-errors" is used, connections which experience errors, timeouts,
19369retries, redispatches or HTTP status codes 5xx will see their syslog level
19370raised from "info" to "err". This will help a syslog daemon store the log in
19371a separate file. It is very important to keep the errors in the normal traffic
19372file too, so that log ordering is not altered. You should also be careful if
19373you already have configured your syslog daemon to store all logs higher than
19374"notice" in an "admin" file, because the "err" level is higher than "notice".
19375
19376
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200193778.3.4. Disabling logging of successful connections
19378--------------------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020019379
19380Although this may sound strange at first, some large sites have to deal with
19381multiple thousands of logs per second and are experiencing difficulties keeping
19382them intact for a long time or detecting errors within them. If the option
19383"dontlog-normal" is set on the frontend, all normal connections will not be
19384logged. In this regard, a normal connection is defined as one without any
19385error, timeout, retry nor redispatch. In HTTP, the status code is checked too,
19386and a response with a status 5xx is not considered normal and will be logged
19387too. Of course, doing is is really discouraged as it will remove most of the
19388useful information from the logs. Do this only if you have no other
19389alternative.
19390
19391
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200193928.4. Timing events
19393------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019394
19395Timers provide a great help in troubleshooting network problems. All values are
19396reported in milliseconds (ms). These timers should be used in conjunction with
19397the session termination flags. In TCP mode with "option tcplog" set on the
19398frontend, 3 control points are reported under the form "Tw/Tc/Tt", and in HTTP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019399mode, 5 control points are reported under the form "TR/Tw/Tc/Tr/Ta". In
19400addition, three other measures are provided, "Th", "Ti", and "Tq".
19401
Guillaume de Lafondf27cddc2016-12-23 17:32:43 +010019402Timings events in HTTP mode:
19403
19404 first request 2nd request
19405 |<-------------------------------->|<-------------- ...
19406 t tr t tr ...
19407 ---|----|----|----|----|----|----|----|----|--
19408 : Th Ti TR Tw Tc Tr Td : Ti ...
19409 :<---- Tq ---->: :
19410 :<-------------- Tt -------------->:
Damien Claisse57c8eb92020-04-28 12:09:19 +000019411 :<-- -----Tu--------------->:
Guillaume de Lafondf27cddc2016-12-23 17:32:43 +010019412 :<--------- Ta --------->:
19413
19414Timings events in TCP mode:
19415
19416 TCP session
19417 |<----------------->|
19418 t t
19419 ---|----|----|----|----|---
19420 | Th Tw Tc Td |
19421 |<------ Tt ------->|
19422
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019423 - Th: total time to accept tcp connection and execute handshakes for low level
Davor Ocelice9ed2812017-12-25 17:49:28 +010019424 protocols. Currently, these protocols are proxy-protocol and SSL. This may
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019425 only happen once during the whole connection's lifetime. A large time here
19426 may indicate that the client only pre-established the connection without
19427 speaking, that it is experiencing network issues preventing it from
Davor Ocelice9ed2812017-12-25 17:49:28 +010019428 completing a handshake in a reasonable time (e.g. MTU issues), or that an
Willy Tarreau590a0512018-09-05 11:56:48 +020019429 SSL handshake was very expensive to compute. Please note that this time is
19430 reported only before the first request, so it is safe to average it over
19431 all request to calculate the amortized value. The second and subsequent
19432 request will always report zero here.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019433
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019434 - Ti: is the idle time before the HTTP request (HTTP mode only). This timer
19435 counts between the end of the handshakes and the first byte of the HTTP
19436 request. When dealing with a second request in keep-alive mode, it starts
Willy Tarreau590a0512018-09-05 11:56:48 +020019437 to count after the end of the transmission the previous response. When a
19438 multiplexed protocol such as HTTP/2 is used, it starts to count immediately
19439 after the previous request. Some browsers pre-establish connections to a
19440 server in order to reduce the latency of a future request, and keep them
19441 pending until they need it. This delay will be reported as the idle time. A
19442 value of -1 indicates that nothing was received on the connection.
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019443
19444 - TR: total time to get the client request (HTTP mode only). It's the time
19445 elapsed between the first bytes received and the moment the proxy received
19446 the empty line marking the end of the HTTP headers. The value "-1"
19447 indicates that the end of headers has never been seen. This happens when
19448 the client closes prematurely or times out. This time is usually very short
19449 since most requests fit in a single packet. A large time may indicate a
19450 request typed by hand during a test.
19451
19452 - Tq: total time to get the client request from the accept date or since the
19453 emission of the last byte of the previous response (HTTP mode only). It's
Davor Ocelice9ed2812017-12-25 17:49:28 +010019454 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 +020019455 returns -1 as well. This timer used to be very useful before the arrival of
19456 HTTP keep-alive and browsers' pre-connect feature. It's recommended to drop
19457 it in favor of TR nowadays, as the idle time adds a lot of noise to the
19458 reports.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019459
19460 - Tw: total time spent in the queues waiting for a connection slot. It
19461 accounts for backend queue as well as the server queues, and depends on the
19462 queue size, and the time needed for the server to complete previous
19463 requests. The value "-1" means that the request was killed before reaching
19464 the queue, which is generally what happens with invalid or denied requests.
19465
19466 - Tc: total time to establish the TCP connection to the server. It's the time
19467 elapsed between the moment the proxy sent the connection request, and the
19468 moment it was acknowledged by the server, or between the TCP SYN packet and
19469 the matching SYN/ACK packet in return. The value "-1" means that the
19470 connection never established.
19471
19472 - Tr: server response time (HTTP mode only). It's the time elapsed between
19473 the moment the TCP connection was established to the server and the moment
19474 the server sent its complete response headers. It purely shows its request
19475 processing time, without the network overhead due to the data transmission.
19476 It is worth noting that when the client has data to send to the server, for
19477 instance during a POST request, the time already runs, and this can distort
19478 apparent response time. For this reason, it's generally wise not to trust
19479 too much this field for POST requests initiated from clients behind an
19480 untrusted network. A value of "-1" here means that the last the response
19481 header (empty line) was never seen, most likely because the server timeout
19482 stroke before the server managed to process the request.
19483
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019484 - Ta: total active time for the HTTP request, between the moment the proxy
19485 received the first byte of the request header and the emission of the last
19486 byte of the response body. The exception is when the "logasap" option is
19487 specified. In this case, it only equals (TR+Tw+Tc+Tr), and is prefixed with
19488 a '+' sign. From this field, we can deduce "Td", the data transmission time,
19489 by subtracting other timers when valid :
19490
19491 Td = Ta - (TR + Tw + Tc + Tr)
19492
19493 Timers with "-1" values have to be excluded from this equation. Note that
19494 "Ta" can never be negative.
19495
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019496 - Tt: total session duration time, between the moment the proxy accepted it
19497 and the moment both ends were closed. The exception is when the "logasap"
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019498 option is specified. In this case, it only equals (Th+Ti+TR+Tw+Tc+Tr), and
19499 is prefixed with a '+' sign. From this field, we can deduce "Td", the data
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030019500 transmission time, by subtracting other timers when valid :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019501
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019502 Td = Tt - (Th + Ti + TR + Tw + Tc + Tr)
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019503
19504 Timers with "-1" values have to be excluded from this equation. In TCP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019505 mode, "Ti", "Tq" and "Tr" have to be excluded too. Note that "Tt" can never
19506 be negative and that for HTTP, Tt is simply equal to (Th+Ti+Ta).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019507
Damien Claisse57c8eb92020-04-28 12:09:19 +000019508 - Tu: total estimated time as seen from client, between the moment the proxy
19509 accepted it and the moment both ends were closed, without idle time.
19510 This is useful to roughly measure end-to-end time as a user would see it,
19511 without idle time pollution from keep-alive time between requests. This
19512 timer in only an estimation of time seen by user as it assumes network
19513 latency is the same in both directions. The exception is when the "logasap"
19514 option is specified. In this case, it only equals (Th+TR+Tw+Tc+Tr), and is
19515 prefixed with a '+' sign.
19516
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019517These timers provide precious indications on trouble causes. Since the TCP
19518protocol defines retransmit delays of 3, 6, 12... seconds, we know for sure
19519that timers close to multiples of 3s are nearly always related to lost packets
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019520due to network problems (wires, negotiation, congestion). Moreover, if "Ta" or
19521"Tt" is close to a timeout value specified in the configuration, it often means
19522that a session has been aborted on timeout.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019523
19524Most common cases :
19525
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019526 - If "Th" or "Ti" are close to 3000, a packet has probably been lost between
19527 the client and the proxy. This is very rare on local networks but might
19528 happen when clients are on far remote networks and send large requests. It
19529 may happen that values larger than usual appear here without any network
19530 cause. Sometimes, during an attack or just after a resource starvation has
19531 ended, haproxy may accept thousands of connections in a few milliseconds.
19532 The time spent accepting these connections will inevitably slightly delay
19533 processing of other connections, and it can happen that request times in the
19534 order of a few tens of milliseconds are measured after a few thousands of
19535 new connections have been accepted at once. Using one of the keep-alive
19536 modes may display larger idle times since "Ti" measures the time spent
Patrick Mezard105faca2010-06-12 17:02:46 +020019537 waiting for additional requests.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019538
19539 - If "Tc" is close to 3000, a packet has probably been lost between the
19540 server and the proxy during the server connection phase. This value should
19541 always be very low, such as 1 ms on local networks and less than a few tens
19542 of ms on remote networks.
19543
Willy Tarreau55165fe2009-05-10 12:02:55 +020019544 - If "Tr" is nearly always lower than 3000 except some rare values which seem
19545 to be the average majored by 3000, there are probably some packets lost
19546 between the proxy and the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019547
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019548 - If "Ta" is large even for small byte counts, it generally is because
19549 neither the client nor the server decides to close the connection while
19550 haproxy is running in tunnel mode and both have agreed on a keep-alive
19551 connection mode. In order to solve this issue, it will be needed to specify
19552 one of the HTTP options to manipulate keep-alive or close options on either
19553 the frontend or the backend. Having the smallest possible 'Ta' or 'Tt' is
19554 important when connection regulation is used with the "maxconn" option on
19555 the servers, since no new connection will be sent to the server until
19556 another one is released.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019557
19558Other noticeable HTTP log cases ('xx' means any value to be ignored) :
19559
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019560 TR/Tw/Tc/Tr/+Ta The "option logasap" is present on the frontend and the log
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019561 was emitted before the data phase. All the timers are valid
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019562 except "Ta" which is shorter than reality.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019563
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019564 -1/xx/xx/xx/Ta The client was not able to send a complete request in time
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019565 or it aborted too early. Check the session termination flags
19566 then "timeout http-request" and "timeout client" settings.
19567
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019568 TR/-1/xx/xx/Ta It was not possible to process the request, maybe because
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019569 servers were out of order, because the request was invalid
19570 or forbidden by ACL rules. Check the session termination
19571 flags.
19572
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019573 TR/Tw/-1/xx/Ta The connection could not establish on the server. Either it
19574 actively refused it or it timed out after Ta-(TR+Tw) ms.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019575 Check the session termination flags, then check the
19576 "timeout connect" setting. Note that the tarpit action might
19577 return similar-looking patterns, with "Tw" equal to the time
19578 the client connection was maintained open.
19579
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019580 TR/Tw/Tc/-1/Ta The server has accepted the connection but did not return
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030019581 a complete response in time, or it closed its connection
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019582 unexpectedly after Ta-(TR+Tw+Tc) ms. Check the session
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019583 termination flags, then check the "timeout server" setting.
19584
19585
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200195868.5. Session state at disconnection
19587-----------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019588
19589TCP and HTTP logs provide a session termination indicator in the
19590"termination_state" field, just before the number of active connections. It is
195912-characters long in TCP mode, and is extended to 4 characters in HTTP mode,
19592each of which has a special meaning :
19593
19594 - On the first character, a code reporting the first event which caused the
19595 session to terminate :
19596
19597 C : the TCP session was unexpectedly aborted by the client.
19598
19599 S : the TCP session was unexpectedly aborted by the server, or the
19600 server explicitly refused it.
19601
19602 P : the session was prematurely aborted by the proxy, because of a
19603 connection limit enforcement, because a DENY filter was matched,
19604 because of a security check which detected and blocked a dangerous
19605 error in server response which might have caused information leak
Davor Ocelice9ed2812017-12-25 17:49:28 +010019606 (e.g. cacheable cookie).
Willy Tarreau570f2212013-06-10 16:42:09 +020019607
19608 L : the session was locally processed by haproxy and was not passed to
19609 a server. This is what happens for stats and redirects.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019610
19611 R : a resource on the proxy has been exhausted (memory, sockets, source
19612 ports, ...). Usually, this appears during the connection phase, and
19613 system logs should contain a copy of the precise error. If this
19614 happens, it must be considered as a very serious anomaly which
19615 should be fixed as soon as possible by any means.
19616
19617 I : an internal error was identified by the proxy during a self-check.
19618 This should NEVER happen, and you are encouraged to report any log
19619 containing this, because this would almost certainly be a bug. It
19620 would be wise to preventively restart the process after such an
19621 event too, in case it would be caused by memory corruption.
19622
Simon Horman752dc4a2011-06-21 14:34:59 +090019623 D : the session was killed by haproxy because the server was detected
19624 as down and was configured to kill all connections when going down.
19625
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070019626 U : the session was killed by haproxy on this backup server because an
19627 active server was detected as up and was configured to kill all
19628 backup connections when going up.
19629
Willy Tarreaua2a64e92011-09-07 23:01:56 +020019630 K : the session was actively killed by an admin operating on haproxy.
19631
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019632 c : the client-side timeout expired while waiting for the client to
19633 send or receive data.
19634
19635 s : the server-side timeout expired while waiting for the server to
19636 send or receive data.
19637
19638 - : normal session completion, both the client and the server closed
19639 with nothing left in the buffers.
19640
19641 - on the second character, the TCP or HTTP session state when it was closed :
19642
Willy Tarreauf7b30a92010-12-06 22:59:17 +010019643 R : the proxy was waiting for a complete, valid REQUEST from the client
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019644 (HTTP mode only). Nothing was sent to any server.
19645
19646 Q : the proxy was waiting in the QUEUE for a connection slot. This can
19647 only happen when servers have a 'maxconn' parameter set. It can
19648 also happen in the global queue after a redispatch consecutive to
19649 a failed attempt to connect to a dying server. If no redispatch is
19650 reported, then no connection attempt was made to any server.
19651
19652 C : the proxy was waiting for the CONNECTION to establish on the
19653 server. The server might at most have noticed a connection attempt.
19654
19655 H : the proxy was waiting for complete, valid response HEADERS from the
19656 server (HTTP only).
19657
19658 D : the session was in the DATA phase.
19659
19660 L : the proxy was still transmitting LAST data to the client while the
19661 server had already finished. This one is very rare as it can only
19662 happen when the client dies while receiving the last packets.
19663
19664 T : the request was tarpitted. It has been held open with the client
19665 during the whole "timeout tarpit" duration or until the client
19666 closed, both of which will be reported in the "Tw" timer.
19667
19668 - : normal session completion after end of data transfer.
19669
19670 - the third character tells whether the persistence cookie was provided by
19671 the client (only in HTTP mode) :
19672
19673 N : the client provided NO cookie. This is usually the case for new
19674 visitors, so counting the number of occurrences of this flag in the
19675 logs generally indicate a valid trend for the site frequentation.
19676
19677 I : the client provided an INVALID cookie matching no known server.
19678 This might be caused by a recent configuration change, mixed
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +020019679 cookies between HTTP/HTTPS sites, persistence conditionally
19680 ignored, or an attack.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019681
19682 D : the client provided a cookie designating a server which was DOWN,
19683 so either "option persist" was used and the client was sent to
19684 this server, or it was not set and the client was redispatched to
19685 another server.
19686
Willy Tarreau996a92c2010-10-13 19:30:47 +020019687 V : the client provided a VALID cookie, and was sent to the associated
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019688 server.
19689
Willy Tarreau996a92c2010-10-13 19:30:47 +020019690 E : the client provided a valid cookie, but with a last date which was
19691 older than what is allowed by the "maxidle" cookie parameter, so
19692 the cookie is consider EXPIRED and is ignored. The request will be
19693 redispatched just as if there was no cookie.
19694
19695 O : the client provided a valid cookie, but with a first date which was
19696 older than what is allowed by the "maxlife" cookie parameter, so
19697 the cookie is consider too OLD and is ignored. The request will be
19698 redispatched just as if there was no cookie.
19699
Willy Tarreauc89ccb62012-04-05 21:18:22 +020019700 U : a cookie was present but was not used to select the server because
19701 some other server selection mechanism was used instead (typically a
19702 "use-server" rule).
19703
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019704 - : does not apply (no cookie set in configuration).
19705
19706 - the last character reports what operations were performed on the persistence
19707 cookie returned by the server (only in HTTP mode) :
19708
19709 N : NO cookie was provided by the server, and none was inserted either.
19710
19711 I : no cookie was provided by the server, and the proxy INSERTED one.
19712 Note that in "cookie insert" mode, if the server provides a cookie,
19713 it will still be overwritten and reported as "I" here.
19714
Willy Tarreau996a92c2010-10-13 19:30:47 +020019715 U : the proxy UPDATED the last date in the cookie that was presented by
19716 the client. This can only happen in insert mode with "maxidle". It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030019717 happens every time there is activity at a different date than the
Willy Tarreau996a92c2010-10-13 19:30:47 +020019718 date indicated in the cookie. If any other change happens, such as
19719 a redispatch, then the cookie will be marked as inserted instead.
19720
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019721 P : a cookie was PROVIDED by the server and transmitted as-is.
19722
19723 R : the cookie provided by the server was REWRITTEN by the proxy, which
19724 happens in "cookie rewrite" or "cookie prefix" modes.
19725
19726 D : the cookie provided by the server was DELETED by the proxy.
19727
19728 - : does not apply (no cookie set in configuration).
19729
Willy Tarreau996a92c2010-10-13 19:30:47 +020019730The combination of the two first flags gives a lot of information about what
19731was happening when the session terminated, and why it did terminate. It can be
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019732helpful to detect server saturation, network troubles, local system resource
19733starvation, attacks, etc...
19734
19735The most common termination flags combinations are indicated below. They are
19736alphabetically sorted, with the lowercase set just after the upper case for
19737easier finding and understanding.
19738
19739 Flags Reason
19740
19741 -- Normal termination.
19742
19743 CC The client aborted before the connection could be established to the
19744 server. This can happen when haproxy tries to connect to a recently
19745 dead (or unchecked) server, and the client aborts while haproxy is
19746 waiting for the server to respond or for "timeout connect" to expire.
19747
19748 CD The client unexpectedly aborted during data transfer. This can be
19749 caused by a browser crash, by an intermediate equipment between the
19750 client and haproxy which decided to actively break the connection,
19751 by network routing issues between the client and haproxy, or by a
19752 keep-alive session between the server and the client terminated first
19753 by the client.
Willy Tarreaud72758d2010-01-12 10:42:19 +010019754
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019755 cD The client did not send nor acknowledge any data for as long as the
19756 "timeout client" delay. This is often caused by network failures on
Cyril Bontédc4d9032012-04-08 21:57:39 +020019757 the client side, or the client simply leaving the net uncleanly.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019758
19759 CH The client aborted while waiting for the server to start responding.
19760 It might be the server taking too long to respond or the client
19761 clicking the 'Stop' button too fast.
19762
19763 cH The "timeout client" stroke while waiting for client data during a
19764 POST request. This is sometimes caused by too large TCP MSS values
19765 for PPPoE networks which cannot transport full-sized packets. It can
19766 also happen when client timeout is smaller than server timeout and
19767 the server takes too long to respond.
19768
19769 CQ The client aborted while its session was queued, waiting for a server
19770 with enough empty slots to accept it. It might be that either all the
19771 servers were saturated or that the assigned server was taking too
19772 long a time to respond.
19773
19774 CR The client aborted before sending a full HTTP request. Most likely
19775 the request was typed by hand using a telnet client, and aborted
19776 too early. The HTTP status code is likely a 400 here. Sometimes this
19777 might also be caused by an IDS killing the connection between haproxy
Willy Tarreau0f228a02015-05-01 15:37:53 +020019778 and the client. "option http-ignore-probes" can be used to ignore
19779 connections without any data transfer.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019780
19781 cR The "timeout http-request" stroke before the client sent a full HTTP
19782 request. This is sometimes caused by too large TCP MSS values on the
19783 client side for PPPoE networks which cannot transport full-sized
19784 packets, or by clients sending requests by hand and not typing fast
19785 enough, or forgetting to enter the empty line at the end of the
Willy Tarreau2705a612014-05-23 17:38:34 +020019786 request. The HTTP status code is likely a 408 here. Note: recently,
Willy Tarreau0f228a02015-05-01 15:37:53 +020019787 some browsers started to implement a "pre-connect" feature consisting
19788 in speculatively connecting to some recently visited web sites just
19789 in case the user would like to visit them. This results in many
19790 connections being established to web sites, which end up in 408
19791 Request Timeout if the timeout strikes first, or 400 Bad Request when
19792 the browser decides to close them first. These ones pollute the log
19793 and feed the error counters. Some versions of some browsers have even
19794 been reported to display the error code. It is possible to work
Davor Ocelice9ed2812017-12-25 17:49:28 +010019795 around the undesirable effects of this behavior by adding "option
Willy Tarreau0f228a02015-05-01 15:37:53 +020019796 http-ignore-probes" in the frontend, resulting in connections with
19797 zero data transfer to be totally ignored. This will definitely hide
19798 the errors of people experiencing connectivity issues though.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019799
19800 CT The client aborted while its session was tarpitted. It is important to
19801 check if this happens on valid requests, in order to be sure that no
Willy Tarreau55165fe2009-05-10 12:02:55 +020019802 wrong tarpit rules have been written. If a lot of them happen, it
19803 might make sense to lower the "timeout tarpit" value to something
19804 closer to the average reported "Tw" timer, in order not to consume
19805 resources for just a few attackers.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019806
Willy Tarreau570f2212013-06-10 16:42:09 +020019807 LR The request was intercepted and locally handled by haproxy. Generally
19808 it means that this was a redirect or a stats request.
19809
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010019810 SC The server or an equipment between it and haproxy explicitly refused
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019811 the TCP connection (the proxy received a TCP RST or an ICMP message
19812 in return). Under some circumstances, it can also be the network
Davor Ocelice9ed2812017-12-25 17:49:28 +010019813 stack telling the proxy that the server is unreachable (e.g. no route,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019814 or no ARP response on local network). When this happens in HTTP mode,
19815 the status code is likely a 502 or 503 here.
19816
19817 sC The "timeout connect" stroke before a connection to the server could
19818 complete. When this happens in HTTP mode, the status code is likely a
19819 503 or 504 here.
19820
19821 SD The connection to the server died with an error during the data
19822 transfer. This usually means that haproxy has received an RST from
19823 the server or an ICMP message from an intermediate equipment while
19824 exchanging data with the server. This can be caused by a server crash
19825 or by a network issue on an intermediate equipment.
19826
19827 sD The server did not send nor acknowledge any data for as long as the
19828 "timeout server" setting during the data phase. This is often caused
Davor Ocelice9ed2812017-12-25 17:49:28 +010019829 by too short timeouts on L4 equipment before the server (firewalls,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019830 load-balancers, ...), as well as keep-alive sessions maintained
19831 between the client and the server expiring first on haproxy.
19832
19833 SH The server aborted before sending its full HTTP response headers, or
19834 it crashed while processing the request. Since a server aborting at
19835 this moment is very rare, it would be wise to inspect its logs to
19836 control whether it crashed and why. The logged request may indicate a
19837 small set of faulty requests, demonstrating bugs in the application.
19838 Sometimes this might also be caused by an IDS killing the connection
19839 between haproxy and the server.
19840
19841 sH The "timeout server" stroke before the server could return its
19842 response headers. This is the most common anomaly, indicating too
19843 long transactions, probably caused by server or database saturation.
19844 The immediate workaround consists in increasing the "timeout server"
19845 setting, but it is important to keep in mind that the user experience
19846 will suffer from these long response times. The only long term
19847 solution is to fix the application.
19848
19849 sQ The session spent too much time in queue and has been expired. See
19850 the "timeout queue" and "timeout connect" settings to find out how to
19851 fix this if it happens too often. If it often happens massively in
19852 short periods, it may indicate general problems on the affected
19853 servers due to I/O or database congestion, or saturation caused by
19854 external attacks.
19855
19856 PC The proxy refused to establish a connection to the server because the
19857 process' socket limit has been reached while attempting to connect.
Cyril Bontédc4d9032012-04-08 21:57:39 +020019858 The global "maxconn" parameter may be increased in the configuration
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019859 so that it does not happen anymore. This status is very rare and
19860 might happen when the global "ulimit-n" parameter is forced by hand.
19861
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010019862 PD The proxy blocked an incorrectly formatted chunked encoded message in
19863 a request or a response, after the server has emitted its headers. In
19864 most cases, this will indicate an invalid message from the server to
Davor Ocelice9ed2812017-12-25 17:49:28 +010019865 the client. HAProxy supports chunk sizes of up to 2GB - 1 (2147483647
Willy Tarreauf3a3e132013-08-31 08:16:26 +020019866 bytes). Any larger size will be considered as an error.
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010019867
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019868 PH The proxy blocked the server's response, because it was invalid,
19869 incomplete, dangerous (cache control), or matched a security filter.
19870 In any case, an HTTP 502 error is sent to the client. One possible
19871 cause for this error is an invalid syntax in an HTTP header name
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010019872 containing unauthorized characters. It is also possible but quite
19873 rare, that the proxy blocked a chunked-encoding request from the
19874 client due to an invalid syntax, before the server responded. In this
19875 case, an HTTP 400 error is sent to the client and reported in the
19876 logs.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019877
19878 PR The proxy blocked the client's HTTP request, either because of an
19879 invalid HTTP syntax, in which case it returned an HTTP 400 error to
19880 the client, or because a deny filter matched, in which case it
19881 returned an HTTP 403 error.
19882
19883 PT The proxy blocked the client's request and has tarpitted its
19884 connection before returning it a 500 server error. Nothing was sent
19885 to the server. The connection was maintained open for as long as
19886 reported by the "Tw" timer field.
19887
19888 RC A local resource has been exhausted (memory, sockets, source ports)
19889 preventing the connection to the server from establishing. The error
19890 logs will tell precisely what was missing. This is very rare and can
19891 only be solved by proper system tuning.
19892
Willy Tarreau996a92c2010-10-13 19:30:47 +020019893The combination of the two last flags gives a lot of information about how
19894persistence was handled by the client, the server and by haproxy. This is very
19895important to troubleshoot disconnections, when users complain they have to
19896re-authenticate. The commonly encountered flags are :
19897
19898 -- Persistence cookie is not enabled.
19899
19900 NN No cookie was provided by the client, none was inserted in the
19901 response. For instance, this can be in insert mode with "postonly"
19902 set on a GET request.
19903
19904 II A cookie designating an invalid server was provided by the client,
19905 a valid one was inserted in the response. This typically happens when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040019906 a "server" entry is removed from the configuration, since its cookie
Willy Tarreau996a92c2010-10-13 19:30:47 +020019907 value can be presented by a client when no other server knows it.
19908
19909 NI No cookie was provided by the client, one was inserted in the
19910 response. This typically happens for first requests from every user
19911 in "insert" mode, which makes it an easy way to count real users.
19912
19913 VN A cookie was provided by the client, none was inserted in the
19914 response. This happens for most responses for which the client has
19915 already got a cookie.
19916
19917 VU A cookie was provided by the client, with a last visit date which is
19918 not completely up-to-date, so an updated cookie was provided in
19919 response. This can also happen if there was no date at all, or if
19920 there was a date but the "maxidle" parameter was not set, so that the
19921 cookie can be switched to unlimited time.
19922
19923 EI A cookie was provided by the client, with a last visit date which is
19924 too old for the "maxidle" parameter, so the cookie was ignored and a
19925 new cookie was inserted in the response.
19926
19927 OI A cookie was provided by the client, with a first visit date which is
19928 too old for the "maxlife" parameter, so the cookie was ignored and a
19929 new cookie was inserted in the response.
19930
19931 DI The server designated by the cookie was down, a new server was
19932 selected and a new cookie was emitted in the response.
19933
19934 VI The server designated by the cookie was not marked dead but could not
19935 be reached. A redispatch happened and selected another one, which was
19936 then advertised in the response.
19937
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019938
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200199398.6. Non-printable characters
19940-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019941
19942In order not to cause trouble to log analysis tools or terminals during log
19943consulting, non-printable characters are not sent as-is into log files, but are
19944converted to the two-digits hexadecimal representation of their ASCII code,
19945prefixed by the character '#'. The only characters that can be logged without
19946being escaped are comprised between 32 and 126 (inclusive). Obviously, the
19947escape character '#' itself is also encoded to avoid any ambiguity ("#23"). It
19948is the same for the character '"' which becomes "#22", as well as '{', '|' and
19949'}' when logging headers.
19950
19951Note that the space character (' ') is not encoded in headers, which can cause
19952issues for tools relying on space count to locate fields. A typical header
19953containing spaces is "User-Agent".
19954
19955Last, it has been observed that some syslog daemons such as syslog-ng escape
19956the quote ('"') with a backslash ('\'). The reverse operation can safely be
19957performed since no quote may appear anywhere else in the logs.
19958
19959
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200199608.7. Capturing HTTP cookies
19961---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019962
19963Cookie capture simplifies the tracking a complete user session. This can be
19964achieved using the "capture cookie" statement in the frontend. Please refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019965section 4.2 for more details. Only one cookie can be captured, and the same
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019966cookie will simultaneously be checked in the request ("Cookie:" header) and in
19967the response ("Set-Cookie:" header). The respective values will be reported in
19968the HTTP logs at the "captured_request_cookie" and "captured_response_cookie"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019969locations (see section 8.2.3 about HTTP log format). When either cookie is
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019970not seen, a dash ('-') replaces the value. This way, it's easy to detect when a
19971user switches to a new session for example, because the server will reassign it
19972a new cookie. It is also possible to detect if a server unexpectedly sets a
19973wrong cookie to a client, leading to session crossing.
19974
19975 Examples :
19976 # capture the first cookie whose name starts with "ASPSESSION"
19977 capture cookie ASPSESSION len 32
19978
19979 # capture the first cookie whose name is exactly "vgnvisitor"
19980 capture cookie vgnvisitor= len 32
19981
19982
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200199838.8. Capturing HTTP headers
19984---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019985
19986Header captures are useful to track unique request identifiers set by an upper
19987proxy, virtual host names, user-agents, POST content-length, referrers, etc. In
19988the response, one can search for information about the response length, how the
19989server asked the cache to behave, or an object location during a redirection.
19990
19991Header captures are performed using the "capture request header" and "capture
19992response header" statements in the frontend. Please consult their definition in
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019993section 4.2 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019994
19995It is possible to include both request headers and response headers at the same
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010019996time. Non-existent headers are logged as empty strings, and if one header
19997appears more than once, only its last occurrence will be logged. Request headers
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019998are grouped within braces '{' and '}' in the same order as they were declared,
19999and delimited with a vertical bar '|' without any space. Response headers
20000follow the same representation, but are displayed after a space following the
20001request headers block. These blocks are displayed just before the HTTP request
20002in the logs.
20003
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020020004As a special case, it is possible to specify an HTTP header capture in a TCP
20005frontend. The purpose is to enable logging of headers which will be parsed in
20006an HTTP backend if the request is then switched to this HTTP backend.
20007
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020008 Example :
20009 # This instance chains to the outgoing proxy
20010 listen proxy-out
20011 mode http
20012 option httplog
20013 option logasap
20014 log global
20015 server cache1 192.168.1.1:3128
20016
20017 # log the name of the virtual server
20018 capture request header Host len 20
20019
20020 # log the amount of data uploaded during a POST
20021 capture request header Content-Length len 10
20022
20023 # log the beginning of the referrer
20024 capture request header Referer len 20
20025
20026 # server name (useful for outgoing proxies only)
20027 capture response header Server len 20
20028
20029 # logging the content-length is useful with "option logasap"
20030 capture response header Content-Length len 10
20031
Davor Ocelice9ed2812017-12-25 17:49:28 +010020032 # log the expected cache behavior on the response
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020033 capture response header Cache-Control len 8
20034
20035 # the Via header will report the next proxy's name
20036 capture response header Via len 20
20037
20038 # log the URL location during a redirection
20039 capture response header Location len 20
20040
20041 >>> Aug 9 20:26:09 localhost \
20042 haproxy[2022]: 127.0.0.1:34014 [09/Aug/2004:20:26:09] proxy-out \
20043 proxy-out/cache1 0/0/0/162/+162 200 +350 - - ---- 0/0/0/0/0 0/0 \
20044 {fr.adserver.yahoo.co||http://fr.f416.mail.} {|864|private||} \
20045 "GET http://fr.adserver.yahoo.com/"
20046
20047 >>> Aug 9 20:30:46 localhost \
20048 haproxy[2022]: 127.0.0.1:34020 [09/Aug/2004:20:30:46] proxy-out \
20049 proxy-out/cache1 0/0/0/182/+182 200 +279 - - ---- 0/0/0/0/0 0/0 \
20050 {w.ods.org||} {Formilux/0.1.8|3495|||} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010020051 "GET http://trafic.1wt.eu/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020052
20053 >>> Aug 9 20:30:46 localhost \
20054 haproxy[2022]: 127.0.0.1:34028 [09/Aug/2004:20:30:46] proxy-out \
20055 proxy-out/cache1 0/0/2/126/+128 301 +223 - - ---- 0/0/0/0/0 0/0 \
20056 {www.sytadin.equipement.gouv.fr||http://trafic.1wt.eu/} \
20057 {Apache|230|||http://www.sytadin.} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010020058 "GET http://www.sytadin.equipement.gouv.fr/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020059
20060
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200200618.9. Examples of logs
20062---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020063
20064These are real-world examples of logs accompanied with an explanation. Some of
20065them have been made up by hand. The syslog part has been removed for better
20066reading. Their sole purpose is to explain how to decipher them.
20067
20068 >>> haproxy[674]: 127.0.0.1:33318 [15/Oct/2003:08:31:57.130] px-http \
20069 px-http/srv1 6559/0/7/147/6723 200 243 - - ---- 5/3/3/1/0 0/0 \
20070 "HEAD / HTTP/1.0"
20071
20072 => long request (6.5s) entered by hand through 'telnet'. The server replied
20073 in 147 ms, and the session ended normally ('----')
20074
20075 >>> haproxy[674]: 127.0.0.1:33319 [15/Oct/2003:08:31:57.149] px-http \
20076 px-http/srv1 6559/1230/7/147/6870 200 243 - - ---- 324/239/239/99/0 \
20077 0/9 "HEAD / HTTP/1.0"
20078
20079 => Idem, but the request was queued in the global queue behind 9 other
20080 requests, and waited there for 1230 ms.
20081
20082 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.654] px-http \
20083 px-http/srv1 9/0/7/14/+30 200 +243 - - ---- 3/3/3/1/0 0/0 \
20084 "GET /image.iso HTTP/1.0"
20085
20086 => request for a long data transfer. The "logasap" option was specified, so
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010020087 the log was produced just before transferring data. The server replied in
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020088 14 ms, 243 bytes of headers were sent to the client, and total time from
20089 accept to first data byte is 30 ms.
20090
20091 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.925] px-http \
20092 px-http/srv1 9/0/7/14/30 502 243 - - PH-- 3/2/2/0/0 0/0 \
20093 "GET /cgi-bin/bug.cgi? HTTP/1.0"
20094
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020020095 => the proxy blocked a server response either because of an "http-response
20096 deny" rule, or because the response was improperly formatted and not
20097 HTTP-compliant, or because it blocked sensitive information which risked
20098 being cached. In this case, the response is replaced with a "502 bad
20099 gateway". The flags ("PH--") tell us that it was haproxy who decided to
20100 return the 502 and not the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020101
20102 >>> haproxy[18113]: 127.0.0.1:34548 [15/Oct/2003:15:18:55.798] px-http \
Willy Tarreaud72758d2010-01-12 10:42:19 +010020103 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 +010020104
20105 => the client never completed its request and aborted itself ("C---") after
20106 8.5s, while the proxy was waiting for the request headers ("-R--").
20107 Nothing was sent to any server.
20108
20109 >>> haproxy[18113]: 127.0.0.1:34549 [15/Oct/2003:15:19:06.103] px-http \
20110 px-http/<NOSRV> -1/-1/-1/-1/50001 408 0 - - cR-- 2/2/2/0/0 0/0 ""
20111
20112 => The client never completed its request, which was aborted by the
20113 time-out ("c---") after 50s, while the proxy was waiting for the request
Davor Ocelice9ed2812017-12-25 17:49:28 +010020114 headers ("-R--"). Nothing was sent to any server, but the proxy could
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020115 send a 408 return code to the client.
20116
20117 >>> haproxy[18989]: 127.0.0.1:34550 [15/Oct/2003:15:24:28.312] px-tcp \
20118 px-tcp/srv1 0/0/5007 0 cD 0/0/0/0/0 0/0
20119
20120 => This log was produced with "option tcplog". The client timed out after
20121 5 seconds ("c----").
20122
20123 >>> haproxy[18989]: 10.0.0.1:34552 [15/Oct/2003:15:26:31.462] px-http \
20124 px-http/srv1 3183/-1/-1/-1/11215 503 0 - - SC-- 205/202/202/115/3 \
Willy Tarreaud72758d2010-01-12 10:42:19 +010020125 0/0 "HEAD / HTTP/1.0"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020126
20127 => The request took 3s to complete (probably a network problem), and the
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020128 connection to the server failed ('SC--') after 4 attempts of 2 seconds
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020129 (config says 'retries 3'), and no redispatch (otherwise we would have
20130 seen "/+3"). Status code 503 was returned to the client. There were 115
20131 connections on this server, 202 connections on this proxy, and 205 on
20132 the global process. It is possible that the server refused the
20133 connection because of too many already established.
Willy Tarreau844e3c52008-01-11 16:28:18 +010020134
Willy Tarreau52b2d222011-09-07 23:48:48 +020020135
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200201369. Supported filters
20137--------------------
20138
20139Here are listed officially supported filters with the list of parameters they
20140accept. Depending on compile options, some of these filters might be
20141unavailable. The list of available filters is reported in haproxy -vv.
20142
20143See also : "filter"
20144
201459.1. Trace
20146----------
20147
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010020148filter trace [name <name>] [random-parsing] [random-forwarding] [hexdump]
Christopher Fauletc3fe5332016-04-07 15:30:10 +020020149
20150 Arguments:
20151 <name> is an arbitrary name that will be reported in
20152 messages. If no name is provided, "TRACE" is used.
20153
20154 <random-parsing> enables the random parsing of data exchanged between
20155 the client and the server. By default, this filter
20156 parses all available data. With this parameter, it
20157 only parses a random amount of the available data.
20158
Davor Ocelice9ed2812017-12-25 17:49:28 +010020159 <random-forwarding> enables the random forwarding of parsed data. By
Christopher Fauletc3fe5332016-04-07 15:30:10 +020020160 default, this filter forwards all previously parsed
20161 data. With this parameter, it only forwards a random
20162 amount of the parsed data.
20163
Davor Ocelice9ed2812017-12-25 17:49:28 +010020164 <hexdump> dumps all forwarded data to the server and the client.
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010020165
Christopher Fauletc3fe5332016-04-07 15:30:10 +020020166This filter can be used as a base to develop new filters. It defines all
20167callbacks and print a message on the standard error stream (stderr) with useful
20168information for all of them. It may be useful to debug the activity of other
20169filters or, quite simply, HAProxy's activity.
20170
20171Using <random-parsing> and/or <random-forwarding> parameters is a good way to
20172tests the behavior of a filter that parses data exchanged between a client and
20173a server by adding some latencies in the processing.
20174
20175
201769.2. HTTP compression
20177---------------------
20178
20179filter compression
20180
20181The HTTP compression has been moved in a filter in HAProxy 1.7. "compression"
20182keyword must still be used to enable and configure the HTTP compression. And
Christopher Fauletb30b3102019-09-12 23:03:09 +020020183when no other filter is used, it is enough. When used with the cache or the
20184fcgi-app enabled, it is also enough. In this case, the compression is always
20185done after the response is stored in the cache. But it is mandatory to
20186explicitly use a filter line to enable the HTTP compression when at least one
20187filter other than the cache or the fcgi-app is used for the same
20188listener/frontend/backend. This is important to know the filters evaluation
20189order.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020020190
Christopher Fauletb30b3102019-09-12 23:03:09 +020020191See also : "compression", section 9.4 about the cache filter and section 9.5
20192 about the fcgi-app filter.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020020193
20194
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +0200201959.3. Stream Processing Offload Engine (SPOE)
20196--------------------------------------------
20197
20198filter spoe [engine <name>] config <file>
20199
20200 Arguments :
20201
20202 <name> is the engine name that will be used to find the right scope in
20203 the configuration file. If not provided, all the file will be
20204 parsed.
20205
20206 <file> is the path of the engine configuration file. This file can
20207 contain configuration of several engines. In this case, each
20208 part must be placed in its own scope.
20209
20210The Stream Processing Offload Engine (SPOE) is a filter communicating with
20211external components. It allows the offload of some specifics processing on the
Davor Ocelice9ed2812017-12-25 17:49:28 +010020212streams in tiered applications. These external components and information
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020020213exchanged with them are configured in dedicated files, for the main part. It
20214also requires dedicated backends, defined in HAProxy configuration.
20215
20216SPOE communicates with external components using an in-house binary protocol,
20217the Stream Processing Offload Protocol (SPOP).
20218
Tim Düsterhus4896c442016-11-29 02:15:19 +010020219For all information about the SPOE configuration and the SPOP specification, see
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020020220"doc/SPOE.txt".
20221
Christopher Faulet99a17a22018-12-11 09:18:27 +0100202229.4. Cache
20223----------
20224
20225filter cache <name>
20226
20227 Arguments :
20228
20229 <name> is name of the cache section this filter will use.
20230
20231The cache uses a filter to store cacheable responses. The HTTP rules
20232"cache-store" and "cache-use" must be used to define how and when to use a
John Roeslerfb2fce12019-07-10 15:45:51 -050020233cache. By default the corresponding filter is implicitly defined. And when no
Christopher Fauletb30b3102019-09-12 23:03:09 +020020234other filters than fcgi-app or compression are used, it is enough. In such
20235case, the compression filter is always evaluated after the cache filter. But it
20236is mandatory to explicitly use a filter line to use a cache when at least one
20237filter other than the compression or the fcgi-app is used for the same
Christopher Faulet27d93c32018-12-15 22:32:02 +010020238listener/frontend/backend. This is important to know the filters evaluation
20239order.
Christopher Faulet99a17a22018-12-11 09:18:27 +010020240
Christopher Fauletb30b3102019-09-12 23:03:09 +020020241See also : section 9.2 about the compression filter, section 9.5 about the
20242 fcgi-app filter and section 6 about cache.
20243
20244
202459.5. Fcgi-app
20246-------------
20247
Daniel Corbett67a82712020-07-06 23:01:19 -040020248filter fcgi-app <name>
Christopher Fauletb30b3102019-09-12 23:03:09 +020020249
20250 Arguments :
20251
20252 <name> is name of the fcgi-app section this filter will use.
20253
20254The FastCGI application uses a filter to evaluate all custom parameters on the
20255request path, and to process the headers on the response path. the <name> must
20256reference an existing fcgi-app section. The directive "use-fcgi-app" should be
20257used to define the application to use. By default the corresponding filter is
20258implicitly defined. And when no other filters than cache or compression are
20259used, it is enough. But it is mandatory to explicitly use a filter line to a
20260fcgi-app when at least one filter other than the compression or the cache is
20261used for the same backend. This is important to know the filters evaluation
20262order.
20263
20264See also: "use-fcgi-app", section 9.2 about the compression filter, section 9.4
20265 about the cache filter and section 10 about FastCGI application.
20266
20267
2026810. FastCGI applications
20269-------------------------
20270
20271HAProxy is able to send HTTP requests to Responder FastCGI applications. This
20272feature was added in HAProxy 2.1. To do so, servers must be configured to use
20273the FastCGI protocol (using the keyword "proto fcgi" on the server line) and a
20274FastCGI application must be configured and used by the backend managing these
20275servers (using the keyword "use-fcgi-app" into the proxy section). Several
20276FastCGI applications may be defined, but only one can be used at a time by a
20277backend.
20278
20279HAProxy implements all features of the FastCGI specification for Responder
20280application. Especially it is able to multiplex several requests on a simple
20281connection.
20282
2028310.1. Setup
20284-----------
20285
2028610.1.1. Fcgi-app section
20287--------------------------
20288
20289fcgi-app <name>
20290 Declare a FastCGI application named <name>. To be valid, at least the
20291 document root must be defined.
20292
20293acl <aclname> <criterion> [flags] [operator] <value> ...
20294 Declare or complete an access list.
20295
20296 See "acl" keyword in section 4.2 and section 7 about ACL usage for
20297 details. ACLs defined for a FastCGI application are private. They cannot be
20298 used by any other application or by any proxy. In the same way, ACLs defined
20299 in any other section are not usable by a FastCGI application. However,
20300 Pre-defined ACLs are available.
20301
20302docroot <path>
20303 Define the document root on the remote host. <path> will be used to build
20304 the default value of FastCGI parameters SCRIPT_FILENAME and
20305 PATH_TRANSLATED. It is a mandatory setting.
20306
20307index <script-name>
20308 Define the script name that will be appended after an URI that ends with a
20309 slash ("/") to set the default value of the FastCGI parameter SCRIPT_NAME. It
20310 is an optional setting.
20311
20312 Example :
20313 index index.php
20314
20315log-stderr global
20316log-stderr <address> [len <length>] [format <format>]
20317 [sample <ranges>:<smp_size>] <facility> [<level> [<minlevel>]]
20318 Enable logging of STDERR messages reported by the FastCGI application.
20319
20320 See "log" keyword in section 4.2 for details. It is an optional setting. By
20321 default STDERR messages are ignored.
20322
20323pass-header <name> [ { if | unless } <condition> ]
20324 Specify the name of a request header which will be passed to the FastCGI
20325 application. It may optionally be followed by an ACL-based condition, in
20326 which case it will only be evaluated if the condition is true.
20327
20328 Most request headers are already available to the FastCGI application,
20329 prefixed with "HTTP_". Thus, this directive is only required to pass headers
20330 that are purposefully omitted. Currently, the headers "Authorization",
20331 "Proxy-Authorization" and hop-by-hop headers are omitted.
20332
20333 Note that the headers "Content-type" and "Content-length" are never passed to
20334 the FastCGI application because they are already converted into parameters.
20335
20336path-info <regex>
Christopher Faulet28cb3662020-02-14 14:47:37 +010020337 Define a regular expression to extract the script-name and the path-info from
Christopher Faulet6c57f2d2020-02-14 16:55:52 +010020338 the URL-decoded path. Thus, <regex> may have two captures: the first one to
20339 capture the script name and the second one to capture the path-info. The
20340 first one is mandatory, the second one is optional. This way, it is possible
20341 to extract the script-name from the path ignoring the path-info. It is an
20342 optional setting. If it is not defined, no matching is performed on the
20343 path. and the FastCGI parameters PATH_INFO and PATH_TRANSLATED are not
20344 filled.
Christopher Faulet28cb3662020-02-14 14:47:37 +010020345
20346 For security reason, when this regular expression is defined, the newline and
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050020347 the null characters are forbidden from the path, once URL-decoded. The reason
Christopher Faulet28cb3662020-02-14 14:47:37 +010020348 to such limitation is because otherwise the matching always fails (due to a
20349 limitation one the way regular expression are executed in HAProxy). So if one
20350 of these two characters is found in the URL-decoded path, an error is
20351 returned to the client. The principle of least astonishment is applied here.
Christopher Fauletb30b3102019-09-12 23:03:09 +020020352
20353 Example :
Christopher Faulet6c57f2d2020-02-14 16:55:52 +010020354 path-info ^(/.+\.php)(/.*)?$ # both script-name and path-info may be set
20355 path-info ^(/.+\.php) # the path-info is ignored
Christopher Fauletb30b3102019-09-12 23:03:09 +020020356
20357option get-values
20358no option get-values
20359 Enable or disable the retrieve of variables about connection management.
20360
Daniel Corbett67a82712020-07-06 23:01:19 -040020361 HAProxy is able to send the record FCGI_GET_VALUES on connection
Christopher Fauletb30b3102019-09-12 23:03:09 +020020362 establishment to retrieve the value for following variables:
20363
20364 * FCGI_MAX_REQS The maximum number of concurrent requests this
20365 application will accept.
20366
William Lallemand93e548e2019-09-30 13:54:02 +020020367 * FCGI_MPXS_CONNS "0" if this application does not multiplex connections,
20368 "1" otherwise.
Christopher Fauletb30b3102019-09-12 23:03:09 +020020369
20370 Some FastCGI applications does not support this feature. Some others close
Ilya Shipitsin11057a32020-06-21 21:18:27 +050020371 the connection immediately after sending their response. So, by default, this
Christopher Fauletb30b3102019-09-12 23:03:09 +020020372 option is disabled.
20373
20374 Note that the maximum number of concurrent requests accepted by a FastCGI
20375 application is a connection variable. It only limits the number of streams
20376 per connection. If the global load must be limited on the application, the
20377 server parameters "maxconn" and "pool-max-conn" must be set. In addition, if
20378 an application does not support connection multiplexing, the maximum number
20379 of concurrent requests is automatically set to 1.
20380
20381option keep-conn
20382no option keep-conn
20383 Instruct the FastCGI application to keep the connection open or not after
20384 sending a response.
20385
20386 If disabled, the FastCGI application closes the connection after responding
20387 to this request. By default, this option is enabled.
20388
20389option max-reqs <reqs>
20390 Define the maximum number of concurrent requests this application will
20391 accept.
20392
20393 This option may be overwritten if the variable FCGI_MAX_REQS is retrieved
20394 during connection establishment. Furthermore, if the application does not
20395 support connection multiplexing, this option will be ignored. By default set
20396 to 1.
20397
20398option mpxs-conns
20399no option mpxs-conns
20400 Enable or disable the support of connection multiplexing.
20401
20402 This option may be overwritten if the variable FCGI_MPXS_CONNS is retrieved
20403 during connection establishment. It is disabled by default.
20404
20405set-param <name> <fmt> [ { if | unless } <condition> ]
20406 Set a FastCGI parameter that should be passed to this application. Its
20407 value, defined by <fmt> must follows the log-format rules (see section 8.2.4
20408 "Custom Log format"). It may optionally be followed by an ACL-based
20409 condition, in which case it will only be evaluated if the condition is true.
20410
20411 With this directive, it is possible to overwrite the value of default FastCGI
20412 parameters. If the value is evaluated to an empty string, the rule is
20413 ignored. These directives are evaluated in their declaration order.
20414
20415 Example :
20416 # PHP only, required if PHP was built with --enable-force-cgi-redirect
20417 set-param REDIRECT_STATUS 200
20418
20419 set-param PHP_AUTH_DIGEST %[req.hdr(Authorization)]
20420
20421
2042210.1.2. Proxy section
20423---------------------
20424
20425use-fcgi-app <name>
20426 Define the FastCGI application to use for the backend.
20427
20428 Arguments :
20429 <name> is the name of the FastCGI application to use.
20430
20431 This keyword is only available for HTTP proxies with the backend capability
20432 and with at least one FastCGI server. However, FastCGI servers can be mixed
20433 with HTTP servers. But except there is a good reason to do so, it is not
20434 recommended (see section 10.3 about the limitations for details). Only one
20435 application may be defined at a time per backend.
20436
20437 Note that, once a FastCGI application is referenced for a backend, depending
20438 on the configuration some processing may be done even if the request is not
20439 sent to a FastCGI server. Rules to set parameters or pass headers to an
20440 application are evaluated.
20441
20442
2044310.1.3. Example
20444---------------
20445
20446 frontend front-http
20447 mode http
20448 bind *:80
20449 bind *:
20450
20451 use_backend back-dynamic if { path_reg ^/.+\.php(/.*)?$ }
20452 default_backend back-static
20453
20454 backend back-static
20455 mode http
20456 server www A.B.C.D:80
20457
20458 backend back-dynamic
20459 mode http
20460 use-fcgi-app php-fpm
20461 server php-fpm A.B.C.D:9000 proto fcgi
20462
20463 fcgi-app php-fpm
20464 log-stderr global
20465 option keep-conn
20466
20467 docroot /var/www/my-app
20468 index index.php
20469 path-info ^(/.+\.php)(/.*)?$
20470
20471
2047210.2. Default parameters
20473------------------------
20474
20475A Responder FastCGI application has the same purpose as a CGI/1.1 program. In
20476the CGI/1.1 specification (RFC3875), several variables must be passed to the
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050020477script. So HAProxy set them and some others commonly used by FastCGI
Christopher Fauletb30b3102019-09-12 23:03:09 +020020478applications. All these variables may be overwritten, with caution though.
20479
20480 +-------------------+-----------------------------------------------------+
20481 | AUTH_TYPE | Identifies the mechanism, if any, used by HAProxy |
20482 | | to authenticate the user. Concretely, only the |
20483 | | BASIC authentication mechanism is supported. |
20484 | | |
20485 +-------------------+-----------------------------------------------------+
20486 | CONTENT_LENGTH | Contains the size of the message-body attached to |
20487 | | the request. It means only requests with a known |
20488 | | size are considered as valid and sent to the |
20489 | | application. |
20490 | | |
20491 +-------------------+-----------------------------------------------------+
20492 | CONTENT_TYPE | Contains the type of the message-body attached to |
20493 | | the request. It may not be set. |
20494 | | |
20495 +-------------------+-----------------------------------------------------+
20496 | DOCUMENT_ROOT | Contains the document root on the remote host under |
20497 | | which the script should be executed, as defined in |
20498 | | the application's configuration. |
20499 | | |
20500 +-------------------+-----------------------------------------------------+
20501 | GATEWAY_INTERFACE | Contains the dialect of CGI being used by HAProxy |
20502 | | to communicate with the FastCGI application. |
20503 | | Concretely, it is set to "CGI/1.1". |
20504 | | |
20505 +-------------------+-----------------------------------------------------+
20506 | PATH_INFO | Contains the portion of the URI path hierarchy |
20507 | | following the part that identifies the script |
20508 | | itself. To be set, the directive "path-info" must |
20509 | | be defined. |
20510 | | |
20511 +-------------------+-----------------------------------------------------+
20512 | PATH_TRANSLATED | If PATH_INFO is set, it is its translated version. |
20513 | | It is the concatenation of DOCUMENT_ROOT and |
20514 | | PATH_INFO. If PATH_INFO is not set, this parameters |
20515 | | is not set too. |
20516 | | |
20517 +-------------------+-----------------------------------------------------+
20518 | QUERY_STRING | Contains the request's query string. It may not be |
20519 | | set. |
20520 | | |
20521 +-------------------+-----------------------------------------------------+
20522 | REMOTE_ADDR | Contains the network address of the client sending |
20523 | | the request. |
20524 | | |
20525 +-------------------+-----------------------------------------------------+
20526 | REMOTE_USER | Contains the user identification string supplied by |
20527 | | client as part of user authentication. |
20528 | | |
20529 +-------------------+-----------------------------------------------------+
20530 | REQUEST_METHOD | Contains the method which should be used by the |
20531 | | script to process the request. |
20532 | | |
20533 +-------------------+-----------------------------------------------------+
20534 | REQUEST_URI | Contains the request's URI. |
20535 | | |
20536 +-------------------+-----------------------------------------------------+
20537 | SCRIPT_FILENAME | Contains the absolute pathname of the script. it is |
20538 | | the concatenation of DOCUMENT_ROOT and SCRIPT_NAME. |
20539 | | |
20540 +-------------------+-----------------------------------------------------+
20541 | SCRIPT_NAME | Contains the name of the script. If the directive |
20542 | | "path-info" is defined, it is the first part of the |
20543 | | URI path hierarchy, ending with the script name. |
20544 | | Otherwise, it is the entire URI path. |
20545 | | |
20546 +-------------------+-----------------------------------------------------+
20547 | SERVER_NAME | Contains the name of the server host to which the |
20548 | | client request is directed. It is the value of the |
20549 | | header "Host", if defined. Otherwise, the |
20550 | | destination address of the connection on the client |
20551 | | side. |
20552 | | |
20553 +-------------------+-----------------------------------------------------+
20554 | SERVER_PORT | Contains the destination TCP port of the connection |
20555 | | on the client side, which is the port the client |
20556 | | connected to. |
20557 | | |
20558 +-------------------+-----------------------------------------------------+
20559 | SERVER_PROTOCOL | Contains the request's protocol. |
20560 | | |
20561 +-------------------+-----------------------------------------------------+
20562 | HTTPS | Set to a non-empty value ("on") if the script was |
20563 | | queried through the HTTPS protocol. |
20564 | | |
20565 +-------------------+-----------------------------------------------------+
20566
20567
2056810.3. Limitations
20569------------------
20570
20571The current implementation have some limitations. The first one is about the
20572way some request headers are hidden to the FastCGI applications. This happens
20573during the headers analysis, on the backend side, before the connection
20574establishment. At this stage, HAProxy know the backend is using a FastCGI
20575application but it don't know if the request will be routed to a FastCGI server
20576or not. But to hide request headers, it simply removes them from the HTX
20577message. So, if the request is finally routed to an HTTP server, it never see
20578these headers. For this reason, it is not recommended to mix FastCGI servers
20579and HTTP servers under the same backend.
20580
20581Similarly, the rules "set-param" and "pass-header" are evaluated during the
20582request headers analysis. So the evaluation is always performed, even if the
20583requests is finally forwarded to an HTTP server.
20584
20585About the rules "set-param", when a rule is applied, a pseudo header is added
20586into the HTX message. So, the same way than for HTTP header rewrites, it may
20587fail if the buffer is full. The rules "set-param" will compete with
20588"http-request" ones.
20589
20590Finally, all FastCGI params and HTTP headers are sent into a unique record
20591FCGI_PARAM. Encoding of this record must be done in one pass, otherwise a
20592processing error is returned. It means the record FCGI_PARAM, once encoded,
20593must not exceeds the size of a buffer. However, there is no reserve to respect
20594here.
William Lallemand86d0df02017-11-24 21:36:45 +010020595
Willy Tarreau0ba27502007-12-24 16:55:16 +010020596/*
20597 * Local variables:
20598 * fill-column: 79
20599 * End:
20600 */