blob: cc5fbf2d06eb95db143f2c8407361c2889d012e6 [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 Tarreau2e077f82019-11-25 20:36:16 +01005 version 2.2
Willy Tarreau6a06a402007-07-15 20:15:28 +02006 willy tarreau
Willy Tarreau5a753bd2020-03-09 14:57:20 +01007 2020/03/09
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
Willy Tarreauc57f0e22009-05-10 13:12:33 +020056
574. Proxies
584.1. Proxy keywords matrix
594.2. Alphabetically sorted keywords reference
60
Davor Ocelice9ed2812017-12-25 17:49:28 +0100615. Bind and server options
Willy Tarreau086fbf52012-09-24 20:34:51 +0200625.1. Bind options
635.2. Server and default-server options
Baptiste Assmann1fa66662015-04-14 00:28:47 +0200645.3. Server DNS resolution
655.3.1. Global overview
665.3.2. The resolvers section
Willy Tarreauc57f0e22009-05-10 13:12:33 +020067
Julien Pivotto6ccee412019-11-27 15:49:54 +0100686. Cache
696.1. Limitation
706.2. Setup
716.2.1. Cache section
726.2.2. Proxy section
73
Willy Tarreau74ca5042013-06-11 23:12:07 +0200747. Using ACLs and fetching samples
757.1. ACL basics
767.1.1. Matching booleans
777.1.2. Matching integers
787.1.3. Matching strings
797.1.4. Matching regular expressions (regexes)
807.1.5. Matching arbitrary data blocks
817.1.6. Matching IPv4 and IPv6 addresses
827.2. Using ACLs to form conditions
837.3. Fetching samples
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200847.3.1. Converters
857.3.2. Fetching samples from internal states
867.3.3. Fetching samples at Layer 4
877.3.4. Fetching samples at Layer 5
887.3.5. Fetching samples from buffer contents (Layer 6)
897.3.6. Fetching HTTP samples (Layer 7)
Christopher Fauletd47941d2020-01-08 14:40:19 +0100907.3.7. Fetching samples for developers
Willy Tarreau74ca5042013-06-11 23:12:07 +0200917.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020092
938. Logging
948.1. Log levels
958.2. Log formats
968.2.1. Default log format
978.2.2. TCP log format
988.2.3. HTTP log format
William Lallemand48940402012-01-30 16:47:22 +0100998.2.4. Custom log format
Willy Tarreau5f51e1a2012-12-03 18:40:10 +01001008.2.5. Error log format
Willy Tarreauc57f0e22009-05-10 13:12:33 +02001018.3. Advanced logging options
1028.3.1. Disabling logging of external tests
1038.3.2. Logging before waiting for the session to terminate
1048.3.3. Raising log level upon errors
1058.3.4. Disabling logging of successful connections
1068.4. Timing events
1078.5. Session state at disconnection
1088.6. Non-printable characters
1098.7. Capturing HTTP cookies
1108.8. Capturing HTTP headers
1118.9. Examples of logs
112
Christopher Fauletc3fe5332016-04-07 15:30:10 +02001139. Supported filters
1149.1. Trace
1159.2. HTTP compression
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +02001169.3. Stream Processing Offload Engine (SPOE)
Christopher Faulet99a17a22018-12-11 09:18:27 +01001179.4. Cache
Christopher Fauletb30b3102019-09-12 23:03:09 +02001189.5. fcgi-app
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200119
Christopher Fauletb30b3102019-09-12 23:03:09 +020012010. FastCGI applications
12110.1. Setup
12210.1.1. Fcgi-app section
12310.1.2. Proxy section
12410.1.3. Example
12510.2. Default parameters
12610.3. Limitations
127
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200128
1291. Quick reminder about HTTP
130----------------------------
131
Davor Ocelice9ed2812017-12-25 17:49:28 +0100132When HAProxy is running in HTTP mode, both the request and the response are
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200133fully analyzed and indexed, thus it becomes possible to build matching criteria
134on almost anything found in the contents.
135
136However, it is important to understand how HTTP requests and responses are
137formed, and how HAProxy decomposes them. It will then become easier to write
138correct rules and to debug existing configurations.
139
140
1411.1. The HTTP transaction model
142-------------------------------
143
144The HTTP protocol is transaction-driven. This means that each request will lead
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100145to one and only one response. Traditionally, a TCP connection is established
Davor Ocelice9ed2812017-12-25 17:49:28 +0100146from the client to the server, a request is sent by the client through the
147connection, the server responds, and the connection is closed. A new request
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200148will involve a new connection :
149
150 [CON1] [REQ1] ... [RESP1] [CLO1] [CON2] [REQ2] ... [RESP2] [CLO2] ...
151
152In this mode, called the "HTTP close" mode, there are as many connection
153establishments as there are HTTP transactions. Since the connection is closed
154by the server after the response, the client does not need to know the content
155length.
156
157Due to the transactional nature of the protocol, it was possible to improve it
158to avoid closing a connection between two subsequent transactions. In this mode
159however, it is mandatory that the server indicates the content length for each
160response so that the client does not wait indefinitely. For this, a special
161header is used: "Content-length". This mode is called the "keep-alive" mode :
162
163 [CON] [REQ1] ... [RESP1] [REQ2] ... [RESP2] [CLO] ...
164
165Its advantages are a reduced latency between transactions, and less processing
166power required on the server side. It is generally better than the close mode,
167but not always because the clients often limit their concurrent connections to
Patrick Mezard9ec2ec42010-06-12 17:02:45 +0200168a smaller value.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200169
Willy Tarreau95c4e142017-11-26 12:18:55 +0100170Another improvement in the communications is the pipelining mode. It still uses
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200171keep-alive, but the client does not wait for the first response to send the
172second request. This is useful for fetching large number of images composing a
173page :
174
175 [CON] [REQ1] [REQ2] ... [RESP1] [RESP2] [CLO] ...
176
177This can obviously have a tremendous benefit on performance because the network
178latency is eliminated between subsequent requests. Many HTTP agents do not
179correctly support pipelining since there is no way to associate a response with
180the corresponding request in HTTP. For this reason, it is mandatory for the
Cyril Bonté78caf842010-03-10 22:41:43 +0100181server to reply in the exact same order as the requests were received.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200182
Willy Tarreau95c4e142017-11-26 12:18:55 +0100183The next improvement is the multiplexed mode, as implemented in HTTP/2. This
184time, each transaction is assigned a single stream identifier, and all streams
185are multiplexed over an existing connection. Many requests can be sent in
186parallel by the client, and responses can arrive in any order since they also
187carry the stream identifier.
188
Willy Tarreau70dffda2014-01-30 03:07:23 +0100189By default HAProxy operates in keep-alive mode with regards to persistent
190connections: for each connection it processes each request and response, and
191leaves the connection idle on both sides between the end of a response and the
Willy Tarreau95c4e142017-11-26 12:18:55 +0100192start of a new request. When it receives HTTP/2 connections from a client, it
193processes all the requests in parallel and leaves the connection idling,
194waiting for new requests, just as if it was a keep-alive HTTP connection.
Patrick Mezard9ec2ec42010-06-12 17:02:45 +0200195
Christopher Faulet315b39c2018-09-21 16:26:19 +0200196HAProxy supports 4 connection modes :
Willy Tarreau70dffda2014-01-30 03:07:23 +0100197 - keep alive : all requests and responses are processed (default)
198 - tunnel : only the first request and response are processed,
Christopher Faulet6c9bbb22019-03-26 21:37:23 +0100199 everything else is forwarded with no analysis (deprecated).
Willy Tarreau70dffda2014-01-30 03:07:23 +0100200 - server close : the server-facing connection is closed after the response.
Christopher Faulet315b39c2018-09-21 16:26:19 +0200201 - close : the connection is actively closed after end of response.
Willy Tarreau70dffda2014-01-30 03:07:23 +0100202
Davor Ocelice9ed2812017-12-25 17:49:28 +0100203For HTTP/2, the connection mode resembles more the "server close" mode : given
204the independence of all streams, there is currently no place to hook the idle
Willy Tarreau95c4e142017-11-26 12:18:55 +0100205server connection after a response, so it is closed after the response. HTTP/2
206is only supported for incoming connections, not on connections going to
207servers.
208
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200209
2101.2. HTTP request
211-----------------
212
213First, let's consider this HTTP request :
214
215 Line Contents
Willy Tarreaud72758d2010-01-12 10:42:19 +0100216 number
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200217 1 GET /serv/login.php?lang=en&profile=2 HTTP/1.1
218 2 Host: www.mydomain.com
219 3 User-agent: my small browser
220 4 Accept: image/jpeg, image/gif
221 5 Accept: image/png
222
223
2241.2.1. The Request line
225-----------------------
226
227Line 1 is the "request line". It is always composed of 3 fields :
228
229 - a METHOD : GET
230 - a URI : /serv/login.php?lang=en&profile=2
231 - a version tag : HTTP/1.1
232
233All of them are delimited by what the standard calls LWS (linear white spaces),
234which are commonly spaces, but can also be tabs or line feeds/carriage returns
235followed by spaces/tabs. The method itself cannot contain any colon (':') and
236is limited to alphabetic letters. All those various combinations make it
237desirable that HAProxy performs the splitting itself rather than leaving it to
238the user to write a complex or inaccurate regular expression.
239
240The URI itself can have several forms :
241
242 - A "relative URI" :
243
244 /serv/login.php?lang=en&profile=2
245
246 It is a complete URL without the host part. This is generally what is
247 received by servers, reverse proxies and transparent proxies.
248
249 - An "absolute URI", also called a "URL" :
250
251 http://192.168.0.12:8080/serv/login.php?lang=en&profile=2
252
253 It is composed of a "scheme" (the protocol name followed by '://'), a host
254 name or address, optionally a colon (':') followed by a port number, then
255 a relative URI beginning at the first slash ('/') after the address part.
256 This is generally what proxies receive, but a server supporting HTTP/1.1
257 must accept this form too.
258
259 - a star ('*') : this form is only accepted in association with the OPTIONS
260 method and is not relayable. It is used to inquiry a next hop's
261 capabilities.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100262
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200263 - an address:port combination : 192.168.0.12:80
264 This is used with the CONNECT method, which is used to establish TCP
265 tunnels through HTTP proxies, generally for HTTPS, but sometimes for
266 other protocols too.
267
268In a relative URI, two sub-parts are identified. The part before the question
269mark is called the "path". It is typically the relative path to static objects
270on the server. The part after the question mark is called the "query string".
271It is mostly used with GET requests sent to dynamic scripts and is very
272specific to the language, framework or application in use.
273
Willy Tarreau95c4e142017-11-26 12:18:55 +0100274HTTP/2 doesn't convey a version information with the request, so the version is
Davor Ocelice9ed2812017-12-25 17:49:28 +0100275assumed to be the same as the one of the underlying protocol (i.e. "HTTP/2").
Willy Tarreau95c4e142017-11-26 12:18:55 +0100276However, haproxy natively processes HTTP/1.x requests and headers, so requests
277received over an HTTP/2 connection are transcoded to HTTP/1.1 before being
278processed. This explains why they still appear as "HTTP/1.1" in haproxy's logs
279as well as in server logs.
280
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200281
2821.2.2. The request headers
283--------------------------
284
285The headers start at the second line. They are composed of a name at the
286beginning of the line, immediately followed by a colon (':'). Traditionally,
287an LWS is added after the colon but that's not required. Then come the values.
288Multiple identical headers may be folded into one single line, delimiting the
289values with commas, provided that their order is respected. This is commonly
290encountered in the "Cookie:" field. A header may span over multiple lines if
291the subsequent lines begin with an LWS. In the example in 1.2, lines 4 and 5
292define a total of 3 values for the "Accept:" header.
293
Davor Ocelice9ed2812017-12-25 17:49:28 +0100294Contrary to a common misconception, header names are not case-sensitive, and
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200295their values are not either if they refer to other header names (such as the
Willy Tarreau95c4e142017-11-26 12:18:55 +0100296"Connection:" header). In HTTP/2, header names are always sent in lower case,
297as can be seen when running in debug mode.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200298
299The end of the headers is indicated by the first empty line. People often say
300that it's a double line feed, which is not exact, even if a double line feed
301is one valid form of empty line.
302
303Fortunately, HAProxy takes care of all these complex combinations when indexing
304headers, checking values and counting them, so there is no reason to worry
305about the way they could be written, but it is important not to accuse an
306application of being buggy if it does unusual, valid things.
307
308Important note:
Lukas Tribus23953682017-04-28 13:24:30 +0000309 As suggested by RFC7231, HAProxy normalizes headers by replacing line breaks
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200310 in the middle of headers by LWS in order to join multi-line headers. This
311 is necessary for proper analysis and helps less capable HTTP parsers to work
312 correctly and not to be fooled by such complex constructs.
313
314
3151.3. HTTP response
316------------------
317
318An HTTP response looks very much like an HTTP request. Both are called HTTP
319messages. Let's consider this HTTP response :
320
321 Line Contents
Willy Tarreaud72758d2010-01-12 10:42:19 +0100322 number
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200323 1 HTTP/1.1 200 OK
324 2 Content-length: 350
325 3 Content-Type: text/html
326
Willy Tarreau816b9792009-09-15 21:25:21 +0200327As a special case, HTTP supports so called "Informational responses" as status
328codes 1xx. These messages are special in that they don't convey any part of the
329response, they're just used as sort of a signaling message to ask a client to
Willy Tarreau5843d1a2010-02-01 15:13:32 +0100330continue to post its request for instance. In the case of a status 100 response
331the requested information will be carried by the next non-100 response message
332following the informational one. This implies that multiple responses may be
333sent to a single request, and that this only works when keep-alive is enabled
334(1xx messages are HTTP/1.1 only). HAProxy handles these messages and is able to
335correctly forward and skip them, and only process the next non-100 response. As
336such, these messages are neither logged nor transformed, unless explicitly
337state otherwise. Status 101 messages indicate that the protocol is changing
338over the same connection and that haproxy must switch to tunnel mode, just as
339if a CONNECT had occurred. Then the Upgrade header would contain additional
340information about the type of protocol the connection is switching to.
Willy Tarreau816b9792009-09-15 21:25:21 +0200341
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200342
Davor Ocelice9ed2812017-12-25 17:49:28 +01003431.3.1. The response line
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200344------------------------
345
346Line 1 is the "response line". It is always composed of 3 fields :
347
348 - a version tag : HTTP/1.1
349 - a status code : 200
350 - a reason : OK
351
352The status code is always 3-digit. The first digit indicates a general status :
Davor Ocelice9ed2812017-12-25 17:49:28 +0100353 - 1xx = informational message to be skipped (e.g. 100, 101)
354 - 2xx = OK, content is following (e.g. 200, 206)
355 - 3xx = OK, no content following (e.g. 302, 304)
356 - 4xx = error caused by the client (e.g. 401, 403, 404)
357 - 5xx = error caused by the server (e.g. 500, 502, 503)
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200358
Lukas Tribus23953682017-04-28 13:24:30 +0000359Please refer to RFC7231 for the detailed meaning of all such codes. The
Willy Tarreaud72758d2010-01-12 10:42:19 +0100360"reason" field is just a hint, but is not parsed by clients. Anything can be
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200361found there, but it's a common practice to respect the well-established
362messages. It can be composed of one or multiple words, such as "OK", "Found",
363or "Authentication Required".
364
Davor Ocelice9ed2812017-12-25 17:49:28 +0100365HAProxy may emit the following status codes by itself :
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200366
367 Code When / reason
368 200 access to stats page, and when replying to monitoring requests
369 301 when performing a redirection, depending on the configured code
370 302 when performing a redirection, depending on the configured code
371 303 when performing a redirection, depending on the configured code
Willy Tarreaub67fdc42013-03-29 19:28:11 +0100372 307 when performing a redirection, depending on the configured code
373 308 when performing a redirection, depending on the configured code
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200374 400 for an invalid or too large request
375 401 when an authentication is required to perform the action (when
376 accessing the stats page)
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200377 403 when a request is forbidden by a "http-request deny" rule
Florian Tham9205fea2020-01-08 13:35:30 +0100378 404 when the requested resource could not be found
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200379 408 when the request timeout strikes before the request is complete
Florian Tham272e29b2020-01-08 10:19:05 +0100380 410 when the requested resource is no longer available and will not
381 be available again
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200382 500 when haproxy encounters an unrecoverable internal error, such as a
383 memory allocation failure, which should never happen
384 502 when the server returns an empty, invalid or incomplete response, or
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200385 when an "http-response deny" rule blocks the response.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200386 503 when no server was available to handle the request, or in response to
387 monitoring requests which match the "monitor fail" condition
388 504 when the response timeout strikes before the server responds
389
390The error 4xx and 5xx codes above may be customized (see "errorloc" in section
3914.2).
392
393
3941.3.2. The response headers
395---------------------------
396
397Response headers work exactly like request headers, and as such, HAProxy uses
398the same parsing function for both. Please refer to paragraph 1.2.2 for more
399details.
400
401
4022. Configuring HAProxy
403----------------------
404
4052.1. Configuration file format
406------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +0200407
408HAProxy's configuration process involves 3 major sources of parameters :
409
410 - the arguments from the command-line, which always take precedence
411 - the "global" section, which sets process-wide parameters
412 - the proxies sections which can take form of "defaults", "listen",
413 "frontend" and "backend".
414
Willy Tarreau0ba27502007-12-24 16:55:16 +0100415The configuration file syntax consists in lines beginning with a keyword
416referenced in this manual, optionally followed by one or several parameters
William Lallemandf9873ba2015-05-05 17:37:14 +0200417delimited by spaces.
Willy Tarreau0ba27502007-12-24 16:55:16 +0100418
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200419
William Lallemandf9873ba2015-05-05 17:37:14 +02004202.2. Quoting and escaping
421-------------------------
422
423HAProxy's configuration introduces a quoting and escaping system similar to
424many programming languages. The configuration file supports 3 types: escaping
425with a backslash, weak quoting with double quotes, and strong quoting with
426single quotes.
427
428If spaces have to be entered in strings, then they must be escaped by preceding
429them by a backslash ('\') or by quoting them. Backslashes also have to be
430escaped by doubling or strong quoting them.
431
432Escaping is achieved by preceding a special character by a backslash ('\'):
433
434 \ to mark a space and differentiate it from a delimiter
435 \# to mark a hash and differentiate it from a comment
436 \\ to use a backslash
437 \' to use a single quote and differentiate it from strong quoting
438 \" to use a double quote and differentiate it from weak quoting
439
440Weak quoting is achieved by using double quotes (""). Weak quoting prevents
441the interpretation of:
442
443 space as a parameter separator
444 ' single quote as a strong quoting delimiter
445 # hash as a comment start
446
William Lallemandb2f07452015-05-12 14:27:13 +0200447Weak quoting permits the interpretation of variables, if you want to use a non
448-interpreted dollar within a double quoted string, you should escape it with a
449backslash ("\$"), it does not work outside weak quoting.
450
451Interpretation of escaping and special characters are not prevented by weak
William Lallemandf9873ba2015-05-05 17:37:14 +0200452quoting.
453
454Strong quoting is achieved by using single quotes (''). Inside single quotes,
455nothing is interpreted, it's the efficient way to quote regexes.
456
457Quoted and escaped strings are replaced in memory by their interpreted
458equivalent, it allows you to perform concatenation.
459
460 Example:
461 # those are equivalents:
462 log-format %{+Q}o\ %t\ %s\ %{-Q}r
463 log-format "%{+Q}o %t %s %{-Q}r"
464 log-format '%{+Q}o %t %s %{-Q}r'
465 log-format "%{+Q}o %t"' %s %{-Q}r'
466 log-format "%{+Q}o %t"' %s'\ %{-Q}r
467
468 # those are equivalents:
469 reqrep "^([^\ :]*)\ /static/(.*)" \1\ /\2
470 reqrep "^([^ :]*)\ /static/(.*)" '\1 /\2'
471 reqrep "^([^ :]*)\ /static/(.*)" "\1 /\2"
472 reqrep "^([^ :]*)\ /static/(.*)" "\1\ /\2"
473
474
William Lallemandb2f07452015-05-12 14:27:13 +02004752.3. Environment variables
476--------------------------
477
478HAProxy's configuration supports environment variables. Those variables are
479interpreted only within double quotes. Variables are expanded during the
480configuration parsing. Variable names must be preceded by a dollar ("$") and
481optionally enclosed with braces ("{}") similarly to what is done in Bourne
482shell. Variable names can contain alphanumerical characters or the character
483underscore ("_") but should not start with a digit.
484
485 Example:
486
487 bind "fd@${FD_APP1}"
488
489 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
490
491 user "$HAPROXY_USER"
492
William Lallemand4d03e432019-06-14 15:35:37 +0200493Some variables are defined by HAProxy, they can be used in the configuration
494file, or could be inherited by a program (See 3.7. Programs):
William Lallemanddaf4cd22018-04-17 16:46:13 +0200495
William Lallemand4d03e432019-06-14 15:35:37 +0200496* HAPROXY_LOCALPEER: defined at the startup of the process which contains the
497 name of the local peer. (See "-L" in the management guide.)
498
499* HAPROXY_CFGFILES: list of the configuration files loaded by HAProxy,
500 separated by semicolons. Can be useful in the case you specified a
501 directory.
502
503* HAPROXY_MWORKER: In master-worker mode, this variable is set to 1.
504
John Roeslerfb2fce12019-07-10 15:45:51 -0500505* HAPROXY_CLI: configured listeners addresses of the stats socket for every
William Lallemand4d03e432019-06-14 15:35:37 +0200506 processes, separated by semicolons.
507
John Roeslerfb2fce12019-07-10 15:45:51 -0500508* HAPROXY_MASTER_CLI: In master-worker mode, listeners addresses of the master
William Lallemand4d03e432019-06-14 15:35:37 +0200509 CLI, separated by semicolons.
510
511See also "external-check command" for other variables.
William Lallemandb2f07452015-05-12 14:27:13 +0200512
5132.4. Time format
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200514----------------
515
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100516Some parameters involve values representing time, such as timeouts. These
Willy Tarreau0ba27502007-12-24 16:55:16 +0100517values are generally expressed in milliseconds (unless explicitly stated
518otherwise) but may be expressed in any other unit by suffixing the unit to the
519numeric value. It is important to consider this because it will not be repeated
520for every keyword. Supported units are :
521
522 - us : microseconds. 1 microsecond = 1/1000000 second
523 - ms : milliseconds. 1 millisecond = 1/1000 second. This is the default.
524 - s : seconds. 1s = 1000ms
525 - m : minutes. 1m = 60s = 60000ms
526 - h : hours. 1h = 60m = 3600s = 3600000ms
527 - d : days. 1d = 24h = 1440m = 86400s = 86400000ms
528
529
Lukas Tribusaa83a312017-03-21 09:25:09 +00005302.5. Examples
Patrick Mezard35da19c2010-06-12 17:02:47 +0200531-------------
532
533 # Simple configuration for an HTTP proxy listening on port 80 on all
534 # interfaces and forwarding requests to a single backend "servers" with a
535 # single server "server1" listening on 127.0.0.1:8000
536 global
537 daemon
538 maxconn 256
539
540 defaults
541 mode http
542 timeout connect 5000ms
543 timeout client 50000ms
544 timeout server 50000ms
545
546 frontend http-in
547 bind *:80
548 default_backend servers
549
550 backend servers
551 server server1 127.0.0.1:8000 maxconn 32
552
553
554 # The same configuration defined with a single listen block. Shorter but
555 # less expressive, especially in HTTP mode.
556 global
557 daemon
558 maxconn 256
559
560 defaults
561 mode http
562 timeout connect 5000ms
563 timeout client 50000ms
564 timeout server 50000ms
565
566 listen http-in
567 bind *:80
568 server server1 127.0.0.1:8000 maxconn 32
569
570
571Assuming haproxy is in $PATH, test these configurations in a shell with:
572
Willy Tarreauccb289d2010-12-11 20:19:38 +0100573 $ sudo haproxy -f configuration.conf -c
Patrick Mezard35da19c2010-06-12 17:02:47 +0200574
575
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005763. Global parameters
Willy Tarreau6a06a402007-07-15 20:15:28 +0200577--------------------
578
579Parameters in the "global" section are process-wide and often OS-specific. They
580are generally set once for all and do not need being changed once correct. Some
581of them have command-line equivalents.
582
583The following keywords are supported in the "global" section :
584
585 * Process management and security
Emeric Brunc8e8d122012-10-02 18:42:10 +0200586 - ca-base
Willy Tarreau6a06a402007-07-15 20:15:28 +0200587 - chroot
Emeric Brunc8e8d122012-10-02 18:42:10 +0200588 - crt-base
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200589 - cpu-map
Willy Tarreau6a06a402007-07-15 20:15:28 +0200590 - daemon
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200591 - description
592 - deviceatlas-json-file
593 - deviceatlas-log-level
594 - deviceatlas-separator
595 - deviceatlas-properties-cookie
Simon Horman98637e52014-06-20 12:30:16 +0900596 - external-check
Willy Tarreau6a06a402007-07-15 20:15:28 +0200597 - gid
598 - group
Cyril Bonté203ec5a2017-03-23 22:44:13 +0100599 - hard-stop-after
Christopher Faulet98fbe952019-07-22 16:18:24 +0200600 - h1-case-adjust
601 - h1-case-adjust-file
Willy Tarreaud96f1122019-12-03 07:07:36 +0100602 - insecure-fork-wanted
Willy Tarreaua45a8b52019-12-06 16:31:45 +0100603 - insecure-setuid-wanted
Emmanuel Hocdet70df7bf2019-01-04 11:08:20 +0100604 - issuers-chain-path
Willy Tarreau6a06a402007-07-15 20:15:28 +0200605 - log
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200606 - log-tag
Joe Williamsdf5b38f2010-12-29 17:05:48 +0100607 - log-send-hostname
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200608 - lua-load
Tim Duesterhusdd74b5f2020-01-12 13:55:40 +0100609 - lua-prepend-path
William Lallemand27edc4b2019-05-07 17:49:33 +0200610 - mworker-max-reloads
Willy Tarreau6a06a402007-07-15 20:15:28 +0200611 - nbproc
Christopher Fauletbe0faa22017-08-29 15:37:10 +0200612 - nbthread
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200613 - node
Willy Tarreau6a06a402007-07-15 20:15:28 +0200614 - pidfile
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
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200625 - ssl-default-bind-options
626 - ssl-default-server-ciphers
Dirkjan Bussink415150f2018-09-14 11:14:21 +0200627 - ssl-default-server-ciphersuites
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200628 - ssl-default-server-options
629 - ssl-dh-param-file
Emeric Brun850efd52014-01-29 12:24:34 +0100630 - ssl-server-verify
Willy Tarreauceb24bc2010-11-09 12:46:41 +0100631 - unix-bind
Willy Tarreau1d549722016-02-16 12:41:57 +0100632 - unsetenv
Thomas Holmesdb04f192015-05-18 13:21:39 +0100633 - 51degrees-data-file
634 - 51degrees-property-name-list
Dragan Dosen93b38d92015-06-29 16:43:25 +0200635 - 51degrees-property-separator
Dragan Dosenae6d39a2015-06-29 16:43:27 +0200636 - 51degrees-cache-size
Willy Tarreaub3cc9f22019-04-19 16:03:32 +0200637 - wurfl-data-file
638 - wurfl-information-list
639 - wurfl-information-list-separator
Willy Tarreaub3cc9f22019-04-19 16:03:32 +0200640 - wurfl-cache-size
William Dauchy0fec3ab2019-10-27 20:08:11 +0100641 - strict-limits
Willy Tarreaud72758d2010-01-12 10:42:19 +0100642
Willy Tarreau6a06a402007-07-15 20:15:28 +0200643 * Performance tuning
William Dauchy0a8824f2019-10-27 20:08:09 +0100644 - busy-polling
Willy Tarreau1746eec2014-04-25 10:46:47 +0200645 - max-spread-checks
Willy Tarreau6a06a402007-07-15 20:15:28 +0200646 - maxconn
Willy Tarreau81c25d02011-09-07 15:17:21 +0200647 - maxconnrate
William Lallemandd85f9172012-11-09 17:05:39 +0100648 - maxcomprate
William Lallemand072a2bf2012-11-20 17:01:01 +0100649 - maxcompcpuusage
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100650 - maxpipes
Willy Tarreau93e7c002013-10-07 18:51:07 +0200651 - maxsessrate
Willy Tarreau403edff2012-09-06 11:58:37 +0200652 - maxsslconn
Willy Tarreaue43d5322013-10-07 20:01:52 +0200653 - maxsslrate
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200654 - maxzlibmem
Willy Tarreau6a06a402007-07-15 20:15:28 +0200655 - noepoll
656 - nokqueue
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +0000657 - noevports
Willy Tarreau6a06a402007-07-15 20:15:28 +0200658 - nopoll
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100659 - nosplice
Jarno Huuskonen0e82b922014-04-12 18:22:19 +0300660 - nogetaddrinfo
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +0000661 - noreuseport
Willy Tarreau75c62c22018-11-22 11:02:09 +0100662 - profiling.tasks
Willy Tarreaufe255b72007-10-14 23:09:26 +0200663 - spread-checks
Baptiste Assmann5626f482015-08-23 10:00:10 +0200664 - server-state-base
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +0200665 - server-state-file
Grant Zhang872f9c22017-01-21 01:10:18 +0000666 - ssl-engine
Grant Zhangfa6c7ee2017-01-14 01:42:15 +0000667 - ssl-mode-async
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200668 - tune.buffers.limit
669 - tune.buffers.reserve
Willy Tarreau27a674e2009-08-17 07:23:33 +0200670 - tune.bufsize
Willy Tarreau43961d52010-10-04 20:39:20 +0200671 - tune.chksize
William Lallemandf3747832012-11-09 12:33:10 +0100672 - tune.comp.maxlevel
Willy Tarreaufe20e5b2017-07-27 11:42:14 +0200673 - tune.h2.header-table-size
Willy Tarreaue6baec02017-07-27 11:45:11 +0200674 - tune.h2.initial-window-size
Willy Tarreau5242ef82017-07-27 11:47:28 +0200675 - tune.h2.max-concurrent-streams
Willy Tarreau193b8c62012-11-22 00:17:38 +0100676 - tune.http.cookielen
Stéphane Cottin23e9e932017-05-18 08:58:41 +0200677 - tune.http.logurilen
Willy Tarreauac1932d2011-10-24 19:14:41 +0200678 - tune.http.maxhdr
Willy Tarreau7e312732014-02-12 16:35:14 +0100679 - tune.idletimer
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100680 - tune.lua.forced-yield
Willy Tarreau32f61e22015-03-18 17:54:59 +0100681 - tune.lua.maxmem
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100682 - tune.lua.session-timeout
683 - tune.lua.task-timeout
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +0200684 - tune.lua.service-timeout
Willy Tarreaua0250ba2008-01-06 11:22:57 +0100685 - tune.maxaccept
686 - tune.maxpollevents
Willy Tarreau27a674e2009-08-17 07:23:33 +0200687 - tune.maxrewrite
Willy Tarreauf3045d22015-04-29 16:24:50 +0200688 - tune.pattern.cache-size
Willy Tarreaubd9a0a72011-10-23 21:14:29 +0200689 - tune.pipesize
Willy Tarreaue803de22010-01-21 17:43:04 +0100690 - tune.rcvbuf.client
691 - tune.rcvbuf.server
Willy Tarreaub22fc302015-12-14 12:04:35 +0100692 - tune.recv_enough
Olivier Houchard1599b802018-05-24 18:59:04 +0200693 - tune.runqueue-depth
Willy Tarreaue803de22010-01-21 17:43:04 +0100694 - tune.sndbuf.client
695 - tune.sndbuf.server
Willy Tarreau6ec58db2012-11-16 16:32:15 +0100696 - tune.ssl.cachesize
Willy Tarreaubfd59462013-02-21 07:46:09 +0100697 - tune.ssl.lifetime
Emeric Brun8dc60392014-05-09 13:52:00 +0200698 - tune.ssl.force-private-cache
Willy Tarreaubfd59462013-02-21 07:46:09 +0100699 - tune.ssl.maxrecord
Remi Gacognef46cd6e2014-06-12 14:58:40 +0200700 - tune.ssl.default-dh-param
Christopher Faulet31af49d2015-06-09 17:29:50 +0200701 - tune.ssl.ssl-ctx-cache-size
Thierry FOURNIER5bf77322017-02-25 12:45:22 +0100702 - tune.ssl.capture-cipherlist-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200703 - tune.vars.global-max-size
Christopher Fauletff2613e2016-11-09 11:36:17 +0100704 - tune.vars.proc-max-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200705 - tune.vars.reqres-max-size
706 - tune.vars.sess-max-size
707 - tune.vars.txn-max-size
William Lallemanda509e4c2012-11-07 16:54:34 +0100708 - tune.zlib.memlevel
709 - tune.zlib.windowsize
Willy Tarreaud72758d2010-01-12 10:42:19 +0100710
Willy Tarreau6a06a402007-07-15 20:15:28 +0200711 * Debugging
712 - debug
713 - quiet
Willy Tarreau6a06a402007-07-15 20:15:28 +0200714
715
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007163.1. Process management and security
Willy Tarreau6a06a402007-07-15 20:15:28 +0200717------------------------------------
718
Emeric Brunc8e8d122012-10-02 18:42:10 +0200719ca-base <dir>
720 Assigns a default directory to fetch SSL CA certificates and CRLs from when a
Emmanuel Hocdet842e94e2019-12-16 16:39:17 +0100721 relative path is used with "ca-file", "ca-verify-file" or "crl-file"
722 directives. Absolute locations specified in "ca-file", "ca-verify-file" and
723 "crl-file" prevail and ignore "ca-base".
Emeric Brunc8e8d122012-10-02 18:42:10 +0200724
Willy Tarreau6a06a402007-07-15 20:15:28 +0200725chroot <jail dir>
726 Changes current directory to <jail dir> and performs a chroot() there before
727 dropping privileges. This increases the security level in case an unknown
728 vulnerability would be exploited, since it would make it very hard for the
729 attacker to exploit the system. This only works when the process is started
730 with superuser privileges. It is important to ensure that <jail_dir> is both
Davor Ocelice9ed2812017-12-25 17:49:28 +0100731 empty and non-writable to anyone.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100732
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100733cpu-map [auto:]<process-set>[/<thread-set>] <cpu-set>...
734 On Linux 2.6 and above, it is possible to bind a process or a thread to a
735 specific CPU set. This means that the process or the thread will never run on
736 other CPUs. The "cpu-map" directive specifies CPU sets for process or thread
737 sets. The first argument is a process set, eventually followed by a thread
738 set. These sets have the format
739
740 all | odd | even | number[-[number]]
741
742 <number>> must be a number between 1 and 32 or 64, depending on the machine's
Davor Ocelice9ed2812017-12-25 17:49:28 +0100743 word size. Any process IDs above nbproc and any thread IDs above nbthread are
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100744 ignored. It is possible to specify a range with two such number delimited by
745 a dash ('-'). It also is possible to specify all processes at once using
Christopher Faulet1dcb9cb2017-11-22 10:24:40 +0100746 "all", only odd numbers using "odd" or even numbers using "even", just like
747 with the "bind-process" directive. The second and forthcoming arguments are
Davor Ocelice9ed2812017-12-25 17:49:28 +0100748 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 +0100749 range with two such numbers delimited by a dash ('-'). Multiple CPU numbers
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100750 or ranges may be specified, and the processes or threads will be allowed to
Davor Ocelice9ed2812017-12-25 17:49:28 +0100751 bind to all of them. Obviously, multiple "cpu-map" directives may be
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100752 specified. Each "cpu-map" directive will replace the previous ones when they
753 overlap. A thread will be bound on the intersection of its mapping and the
754 one of the process on which it is attached. If the intersection is null, no
755 specific binding will be set for the thread.
Willy Tarreaufc6c0322012-11-16 16:12:27 +0100756
Christopher Fauletff4121f2017-11-22 16:38:49 +0100757 Ranges can be partially defined. The higher bound can be omitted. In such
758 case, it is replaced by the corresponding maximum value, 32 or 64 depending
759 on the machine's word size.
760
Christopher Faulet26028f62017-11-22 15:01:51 +0100761 The prefix "auto:" can be added before the process set to let HAProxy
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100762 automatically bind a process or a thread to a CPU by incrementing
763 process/thread and CPU sets. To be valid, both sets must have the same
764 size. No matter the declaration order of the CPU sets, it will be bound from
765 the lowest to the highest bound. Having a process and a thread range with the
766 "auto:" prefix is not supported. Only one range is supported, the other one
767 must be a fixed number.
Christopher Faulet26028f62017-11-22 15:01:51 +0100768
769 Examples:
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100770 cpu-map 1-4 0-3 # bind processes 1 to 4 on the first 4 CPUs
771
772 cpu-map 1/all 0-3 # bind all threads of the first process on the
773 # first 4 CPUs
774
775 cpu-map 1- 0- # will be replaced by "cpu-map 1-64 0-63"
776 # or "cpu-map 1-32 0-31" depending on the machine's
777 # word size.
778
Christopher Faulet26028f62017-11-22 15:01:51 +0100779 # all these lines bind the process 1 to the cpu 0, the process 2 to cpu 1
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100780 # and so on.
Christopher Faulet26028f62017-11-22 15:01:51 +0100781 cpu-map auto:1-4 0-3
782 cpu-map auto:1-4 0-1 2-3
783 cpu-map auto:1-4 3 2 1 0
784
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100785 # all these lines bind the thread 1 to the cpu 0, the thread 2 to cpu 1
786 # and so on.
787 cpu-map auto:1/1-4 0-3
788 cpu-map auto:1/1-4 0-1 2-3
789 cpu-map auto:1/1-4 3 2 1 0
790
Davor Ocelice9ed2812017-12-25 17:49:28 +0100791 # bind each process to exactly one CPU using all/odd/even keyword
Christopher Faulet26028f62017-11-22 15:01:51 +0100792 cpu-map auto:all 0-63
793 cpu-map auto:even 0-31
794 cpu-map auto:odd 32-63
795
796 # invalid cpu-map because process and CPU sets have different sizes.
797 cpu-map auto:1-4 0 # invalid
798 cpu-map auto:1 0-3 # invalid
799
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100800 # invalid cpu-map because automatic binding is used with a process range
801 # and a thread range.
802 cpu-map auto:all/all 0 # invalid
803 cpu-map auto:all/1-4 0 # invalid
804 cpu-map auto:1-4/all 0 # invalid
805
Emeric Brunc8e8d122012-10-02 18:42:10 +0200806crt-base <dir>
807 Assigns a default directory to fetch SSL certificates from when a relative
William Dauchy238ea3b2020-01-11 13:09:12 +0100808 path is used with "crtfile" or "crt" directives. Absolute locations specified
809 prevail and ignore "crt-base".
Emeric Brunc8e8d122012-10-02 18:42:10 +0200810
Willy Tarreau6a06a402007-07-15 20:15:28 +0200811daemon
812 Makes the process fork into background. This is the recommended mode of
813 operation. It is equivalent to the command line "-D" argument. It can be
Lukas Tribusf46bf952017-11-21 12:39:34 +0100814 disabled by the command line "-db" argument. This option is ignored in
815 systemd mode.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200816
David Carlier8167f302015-06-01 13:50:06 +0200817deviceatlas-json-file <path>
818 Sets the path of the DeviceAtlas JSON data file to be loaded by the API.
Davor Ocelice9ed2812017-12-25 17:49:28 +0100819 The path must be a valid JSON data file and accessible by HAProxy process.
David Carlier8167f302015-06-01 13:50:06 +0200820
821deviceatlas-log-level <value>
Davor Ocelice9ed2812017-12-25 17:49:28 +0100822 Sets the level of information returned by the API. This directive is
David Carlier8167f302015-06-01 13:50:06 +0200823 optional and set to 0 by default if not set.
824
825deviceatlas-separator <char>
826 Sets the character separator for the API properties results. This directive
827 is optional and set to | by default if not set.
828
Cyril Bonté0306c4a2015-10-26 22:37:38 +0100829deviceatlas-properties-cookie <name>
Cyril Bonté307ee1e2015-09-28 23:16:06 +0200830 Sets the client cookie's name used for the detection if the DeviceAtlas
831 Client-side component was used during the request. This directive is optional
832 and set to DAPROPS by default if not set.
David Carlier29b3ca32015-09-25 14:09:21 +0100833
Simon Horman98637e52014-06-20 12:30:16 +0900834external-check
Willy Tarreaud96f1122019-12-03 07:07:36 +0100835 Allows the use of an external agent to perform health checks. This is
836 disabled by default as a security precaution, and even when enabled, checks
Willy Tarreaua45a8b52019-12-06 16:31:45 +0100837 may still fail unless "insecure-fork-wanted" is enabled as well. If the
838 program launched makes use of a setuid executable (it should really not),
839 you may also need to set "insecure-setuid-wanted" in the global section.
840 See "option external-check", and "insecure-fork-wanted", and
841 "insecure-setuid-wanted".
Simon Horman98637e52014-06-20 12:30:16 +0900842
Willy Tarreau6a06a402007-07-15 20:15:28 +0200843gid <number>
844 Changes the process' group ID to <number>. It is recommended that the group
845 ID is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
846 be started with a user belonging to this group, or with superuser privileges.
Michael Schererab012dd2013-01-12 18:35:19 +0100847 Note that if haproxy is started from a user having supplementary groups, it
848 will only be able to drop these groups if started with superuser privileges.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200849 See also "group" and "uid".
Willy Tarreaud72758d2010-01-12 10:42:19 +0100850
Willy Tarreau11770ce2019-12-03 08:29:22 +0100851group <group name>
852 Similar to "gid" but uses the GID of group name <group name> from /etc/group.
853 See also "gid" and "user".
854
Cyril Bonté203ec5a2017-03-23 22:44:13 +0100855hard-stop-after <time>
856 Defines the maximum time allowed to perform a clean soft-stop.
857
858 Arguments :
859 <time> is the maximum time (by default in milliseconds) for which the
860 instance will remain alive when a soft-stop is received via the
861 SIGUSR1 signal.
862
863 This may be used to ensure that the instance will quit even if connections
864 remain opened during a soft-stop (for example with long timeouts for a proxy
865 in tcp mode). It applies both in TCP and HTTP mode.
866
867 Example:
868 global
869 hard-stop-after 30s
870
Christopher Faulet98fbe952019-07-22 16:18:24 +0200871h1-case-adjust <from> <to>
872 Defines the case adjustment to apply, when enabled, to the header name
873 <from>, to change it to <to> before sending it to HTTP/1 clients or
874 servers. <from> must be in lower case, and <from> and <to> must not differ
875 except for their case. It may be repeated if several header names need to be
Ilya Shipitsin8525fd92020-02-29 12:34:59 +0500876 adjusted. Duplicate entries are not allowed. If a lot of header names have to
Christopher Faulet98fbe952019-07-22 16:18:24 +0200877 be adjusted, it might be more convenient to use "h1-case-adjust-file".
878 Please note that no transformation will be applied unless "option
879 h1-case-adjust-bogus-client" or "option h1-case-adjust-bogus-server" is
880 specified in a proxy.
881
882 There is no standard case for header names because, as stated in RFC7230,
883 they are case-insensitive. So applications must handle them in a case-
884 insensitive manner. But some bogus applications violate the standards and
885 erroneously rely on the cases most commonly used by browsers. This problem
886 becomes critical with HTTP/2 because all header names must be exchanged in
887 lower case, and HAProxy follows the same convention. All header names are
888 sent in lower case to clients and servers, regardless of the HTTP version.
889
890 Applications which fail to properly process requests or responses may require
891 to temporarily use such workarounds to adjust header names sent to them for
892 the time it takes the application to be fixed. Please note that an
893 application which requires such workarounds might be vulnerable to content
894 smuggling attacks and must absolutely be fixed.
895
896 Example:
897 global
898 h1-case-adjust content-length Content-Length
899
900 See "h1-case-adjust-file", "option h1-case-adjust-bogus-client" and
901 "option h1-case-adjust-bogus-server".
902
903h1-case-adjust-file <hdrs-file>
904 Defines a file containing a list of key/value pairs used to adjust the case
905 of some header names before sending them to HTTP/1 clients or servers. The
906 file <hdrs-file> must contain 2 header names per line. The first one must be
907 in lower case and both must not differ except for their case. Lines which
908 start with '#' are ignored, just like empty lines. Leading and trailing tabs
909 and spaces are stripped. Duplicate entries are not allowed. Please note that
910 no transformation will be applied unless "option h1-case-adjust-bogus-client"
911 or "option h1-case-adjust-bogus-server" is specified in a proxy.
912
913 If this directive is repeated, only the last one will be processed. It is an
914 alternative to the directive "h1-case-adjust" if a lot of header names need
915 to be adjusted. Please read the risks associated with using this.
916
917 See "h1-case-adjust", "option h1-case-adjust-bogus-client" and
918 "option h1-case-adjust-bogus-server".
919
Willy Tarreaud96f1122019-12-03 07:07:36 +0100920insecure-fork-wanted
921 By default haproxy tries hard to prevent any thread and process creation
922 after it starts. Doing so is particularly important when using Lua files of
923 uncertain origin, and when experimenting with development versions which may
924 still contain bugs whose exploitability is uncertain. And generally speaking
925 it's good hygiene to make sure that no unexpected background activity can be
926 triggered by traffic. But this prevents external checks from working, and may
927 break some very specific Lua scripts which actively rely on the ability to
928 fork. This option is there to disable this protection. Note that it is a bad
929 idea to disable it, as a vulnerability in a library or within haproxy itself
930 will be easier to exploit once disabled. In addition, forking from Lua or
931 anywhere else is not reliable as the forked process may randomly embed a lock
932 set by another thread and never manage to finish an operation. As such it is
933 highly recommended that this option is never used and that any workload
934 requiring such a fork be reconsidered and moved to a safer solution (such as
935 agents instead of external checks). This option supports the "no" prefix to
936 disable it.
937
Willy Tarreaua45a8b52019-12-06 16:31:45 +0100938insecure-setuid-wanted
939 HAProxy doesn't need to call executables at run time (except when using
940 external checks which are strongly recommended against), and is even expected
941 to isolate itself into an empty chroot. As such, there basically is no valid
942 reason to allow a setuid executable to be called without the user being fully
943 aware of the risks. In a situation where haproxy would need to call external
944 checks and/or disable chroot, exploiting a vulnerability in a library or in
945 haproxy itself could lead to the execution of an external program. On Linux
946 it is possible to lock the process so that any setuid bit present on such an
947 executable is ignored. This significantly reduces the risk of privilege
948 escalation in such a situation. This is what haproxy does by default. In case
949 this causes a problem to an external check (for example one which would need
950 the "ping" command), then it is possible to disable this protection by
951 explicitly adding this directive in the global section. If enabled, it is
952 possible to turn it back off by prefixing it with the "no" keyword.
953
Emmanuel Hocdet70df7bf2019-01-04 11:08:20 +0100954issuers-chain-path <dir>
955 Assigns a directory to load certificate chain for issuer completion. All
956 files must be in PEM format. For certificates loaded with "crt" or "crt-list",
957 if certificate chain is not included in PEM (also commonly known as
958 intermediate certificate), haproxy will complete chain if the issuer of the
959 certificate corresponds to the first certificate of the chain loaded with
960 "issuers-chain-path".
961 A "crt" file with PrivateKey+Certificate+IntermediateCA2+IntermediateCA1
962 could be replaced with PrivateKey+Certificate. HAProxy will complete the
963 chain if a file with IntermediateCA2+IntermediateCA1 is present in
964 "issuers-chain-path" directory. All other certificates with the same issuer
965 will share the chain in memory.
966
Frédéric Lécailled690dfa2019-04-25 10:52:17 +0200967log <address> [len <length>] [format <format>] [sample <ranges>:<smp_size>]
968 <facility> [max level [min level]]
Cyril Bonté3e954872018-03-20 23:30:27 +0100969 Adds a global syslog server. Several global servers can be defined. They
Davor Ocelice9ed2812017-12-25 17:49:28 +0100970 will receive logs for starts and exits, as well as all logs from proxies
Robert Tsai81ae1952007-12-05 10:47:29 +0100971 configured with "log global".
972
973 <address> can be one of:
974
Willy Tarreau2769aa02007-12-27 18:26:09 +0100975 - An IPv4 address optionally followed by a colon and a UDP port. If
Robert Tsai81ae1952007-12-05 10:47:29 +0100976 no port is specified, 514 is used by default (the standard syslog
977 port).
978
David du Colombier24bb5f52011-03-17 10:40:23 +0100979 - An IPv6 address followed by a colon and optionally a UDP port. If
980 no port is specified, 514 is used by default (the standard syslog
981 port).
982
Willy Tarreau5a32ecc2018-11-12 07:34:59 +0100983 - A filesystem path to a datagram UNIX domain socket, keeping in mind
Robert Tsai81ae1952007-12-05 10:47:29 +0100984 considerations for chroot (be sure the path is accessible inside
985 the chroot) and uid/gid (be sure the path is appropriately
Davor Ocelice9ed2812017-12-25 17:49:28 +0100986 writable).
Robert Tsai81ae1952007-12-05 10:47:29 +0100987
Willy Tarreau5a32ecc2018-11-12 07:34:59 +0100988 - A file descriptor number in the form "fd@<number>", which may point
989 to a pipe, terminal, or socket. In this case unbuffered logs are used
990 and one writev() call per log is performed. This is a bit expensive
991 but acceptable for most workloads. Messages sent this way will not be
992 truncated but may be dropped, in which case the DroppedLogs counter
993 will be incremented. The writev() call is atomic even on pipes for
994 messages up to PIPE_BUF size, which POSIX recommends to be at least
995 512 and which is 4096 bytes on most modern operating systems. Any
996 larger message may be interleaved with messages from other processes.
997 Exceptionally for debugging purposes the file descriptor may also be
998 directed to a file, but doing so will significantly slow haproxy down
999 as non-blocking calls will be ignored. Also there will be no way to
1000 purge nor rotate this file without restarting the process. Note that
1001 the configured syslog format is preserved, so the output is suitable
Willy Tarreauc1b06452018-11-12 11:57:56 +01001002 for use with a TCP syslog server. See also the "short" and "raw"
1003 format below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01001004
1005 - "stdout" / "stderr", which are respectively aliases for "fd@1" and
1006 "fd@2", see above.
1007
Willy Tarreauc046d162019-08-30 15:24:59 +02001008 - A ring buffer in the form "ring@<name>", which will correspond to an
1009 in-memory ring buffer accessible over the CLI using the "show events"
1010 command, which will also list existing rings and their sizes. Such
1011 buffers are lost on reload or restart but when used as a complement
1012 this can help troubleshooting by having the logs instantly available.
1013
William Lallemandb2f07452015-05-12 14:27:13 +02001014 You may want to reference some environment variables in the address
1015 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01001016
Willy Tarreau18324f52014-06-27 18:10:07 +02001017 <length> is an optional maximum line length. Log lines larger than this value
1018 will be truncated before being sent. The reason is that syslog
1019 servers act differently on log line length. All servers support the
1020 default value of 1024, but some servers simply drop larger lines
1021 while others do log them. If a server supports long lines, it may
1022 make sense to set this value here in order to avoid truncating long
1023 lines. Similarly, if a server drops long lines, it is preferable to
1024 truncate them before sending them. Accepted values are 80 to 65535
1025 inclusive. The default value of 1024 is generally fine for all
1026 standard usages. Some specific cases of long captures or
Davor Ocelice9ed2812017-12-25 17:49:28 +01001027 JSON-formatted logs may require larger values. You may also need to
1028 increase "tune.http.logurilen" if your request URIs are truncated.
Willy Tarreau18324f52014-06-27 18:10:07 +02001029
Dragan Dosen7ad31542015-09-28 17:16:47 +02001030 <format> is the log format used when generating syslog messages. It may be
1031 one of the following :
1032
1033 rfc3164 The RFC3164 syslog message format. This is the default.
1034 (https://tools.ietf.org/html/rfc3164)
1035
1036 rfc5424 The RFC5424 syslog message format.
1037 (https://tools.ietf.org/html/rfc5424)
1038
Willy Tarreaue8746a02018-11-12 08:45:00 +01001039 short A message containing only a level between angle brackets such as
1040 '<3>', followed by the text. The PID, date, time, process name
1041 and system name are omitted. This is designed to be used with a
1042 local log server. This format is compatible with what the systemd
1043 logger consumes.
1044
Willy Tarreauc1b06452018-11-12 11:57:56 +01001045 raw A message containing only the text. The level, PID, date, time,
1046 process name and system name are omitted. This is designed to be
1047 used in containers or during development, where the severity only
1048 depends on the file descriptor used (stdout/stderr).
1049
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02001050 <ranges> A list of comma-separated ranges to identify the logs to sample.
1051 This is used to balance the load of the logs to send to the log
1052 server. The limits of the ranges cannot be null. They are numbered
1053 from 1. The size or period (in number of logs) of the sample must be
1054 set with <sample_size> parameter.
1055
1056 <sample_size>
1057 The size of the sample in number of logs to consider when balancing
1058 their logging loads. It is used to balance the load of the logs to
1059 send to the syslog server. This size must be greater or equal to the
1060 maximum of the high limits of the ranges.
1061 (see also <ranges> parameter).
1062
Robert Tsai81ae1952007-12-05 10:47:29 +01001063 <facility> must be one of the 24 standard syslog facilities :
Willy Tarreau6a06a402007-07-15 20:15:28 +02001064
Willy Tarreaue8746a02018-11-12 08:45:00 +01001065 kern user mail daemon auth syslog lpr news
1066 uucp cron auth2 ftp ntp audit alert cron2
1067 local0 local1 local2 local3 local4 local5 local6 local7
1068
Willy Tarreauc1b06452018-11-12 11:57:56 +01001069 Note that the facility is ignored for the "short" and "raw"
1070 formats, but still required as a positional field. It is
1071 recommended to use "daemon" in this case to make it clear that
1072 it's only supposed to be used locally.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001073
1074 An optional level can be specified to filter outgoing messages. By default,
Willy Tarreauf7edefa2009-05-10 17:20:05 +02001075 all messages are sent. If a maximum level is specified, only messages with a
1076 severity at least as important as this level will be sent. An optional minimum
1077 level can be specified. If it is set, logs emitted with a more severe level
1078 than this one will be capped to this level. This is used to avoid sending
1079 "emerg" messages on all terminals on some default syslog configurations.
1080 Eight levels are known :
Willy Tarreau6a06a402007-07-15 20:15:28 +02001081
Cyril Bontédc4d9032012-04-08 21:57:39 +02001082 emerg alert crit err warning notice info debug
Willy Tarreau6a06a402007-07-15 20:15:28 +02001083
Joe Williamsdf5b38f2010-12-29 17:05:48 +01001084log-send-hostname [<string>]
1085 Sets the hostname field in the syslog header. If optional "string" parameter
1086 is set the header is set to the string contents, otherwise uses the hostname
1087 of the system. Generally used if one is not relaying logs through an
1088 intermediate syslog server or for simply customizing the hostname printed in
1089 the logs.
1090
Kevinm48936af2010-12-22 16:08:21 +00001091log-tag <string>
1092 Sets the tag field in the syslog header to this string. It defaults to the
1093 program name as launched from the command line, which usually is "haproxy".
1094 Sometimes it can be useful to differentiate between multiple processes
Willy Tarreau094af4e2015-01-07 15:03:42 +01001095 running on the same host. See also the per-proxy "log-tag" directive.
Kevinm48936af2010-12-22 16:08:21 +00001096
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001097lua-load <file>
1098 This global directive loads and executes a Lua file. This directive can be
1099 used multiple times.
1100
Tim Duesterhusdd74b5f2020-01-12 13:55:40 +01001101lua-prepend-path <string> [<type>]
1102 Prepends the given string followed by a semicolon to Lua's package.<type>
1103 variable.
1104 <type> must either be "path" or "cpath". If <type> is not given it defaults
1105 to "path".
1106
1107 Lua's paths are semicolon delimited lists of patterns that specify how the
1108 `require` function attempts to find the source file of a library. Question
1109 marks (?) within a pattern will be replaced by module name. The path is
1110 evaluated left to right. This implies that paths that are prepended later
1111 will be checked earlier.
1112
1113 As an example by specifying the following path:
1114
1115 lua-prepend-path /usr/share/haproxy-lua/?/init.lua
1116 lua-prepend-path /usr/share/haproxy-lua/?.lua
1117
1118 When `require "example"` is being called Lua will first attempt to load the
1119 /usr/share/haproxy-lua/example.lua script, if that does not exist the
1120 /usr/share/haproxy-lua/example/init.lua will be attempted and the default
1121 paths if that does not exist either.
1122
1123 See https://www.lua.org/pil/8.1.html for the details within the Lua
1124 documentation.
1125
William Lallemand4cfede82017-11-24 22:02:34 +01001126master-worker [no-exit-on-failure]
William Lallemande202b1e2017-06-01 17:38:56 +02001127 Master-worker mode. It is equivalent to the command line "-W" argument.
1128 This mode will launch a "master" which will monitor the "workers". Using
1129 this mode, you can reload HAProxy directly by sending a SIGUSR2 signal to
Davor Ocelice9ed2812017-12-25 17:49:28 +01001130 the master. The master-worker mode is compatible either with the foreground
William Lallemande202b1e2017-06-01 17:38:56 +02001131 or daemon mode. It is recommended to use this mode with multiprocess and
1132 systemd.
William Lallemand4cfede82017-11-24 22:02:34 +01001133 By default, if a worker exits with a bad return code, in the case of a
1134 segfault for example, all workers will be killed, and the master will leave.
1135 It is convenient to combine this behavior with Restart=on-failure in a
1136 systemd unit file in order to relaunch the whole process. If you don't want
1137 this behavior, you must use the keyword "no-exit-on-failure".
William Lallemande202b1e2017-06-01 17:38:56 +02001138
William Lallemand4cfede82017-11-24 22:02:34 +01001139 See also "-W" in the management guide.
William Lallemande202b1e2017-06-01 17:38:56 +02001140
William Lallemand27edc4b2019-05-07 17:49:33 +02001141mworker-max-reloads <number>
1142 In master-worker mode, this option limits the number of time a worker can
John Roeslerfb2fce12019-07-10 15:45:51 -05001143 survive to a reload. If the worker did not leave after a reload, once its
William Lallemand27edc4b2019-05-07 17:49:33 +02001144 number of reloads is greater than this number, the worker will receive a
1145 SIGTERM. This option helps to keep under control the number of workers.
1146 See also "show proc" in the Management Guide.
1147
Willy Tarreau6a06a402007-07-15 20:15:28 +02001148nbproc <number>
1149 Creates <number> processes when going daemon. This requires the "daemon"
1150 mode. By default, only one process is created, which is the recommended mode
1151 of operation. For systems limited to small sets of file descriptors per
Willy Tarreau149ab772019-01-26 14:27:06 +01001152 process, it may be needed to fork multiple daemons. When set to a value
1153 larger than 1, threads are automatically disabled. USING MULTIPLE PROCESSES
Willy Tarreau1f672a82019-01-26 14:20:55 +01001154 IS HARDER TO DEBUG AND IS REALLY DISCOURAGED. See also "daemon" and
1155 "nbthread".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001156
Christopher Fauletbe0faa22017-08-29 15:37:10 +02001157nbthread <number>
1158 This setting is only available when support for threads was built in. It
Willy Tarreau26f6ae12019-02-02 12:56:15 +01001159 makes haproxy run on <number> threads. This is exclusive with "nbproc". While
1160 "nbproc" historically used to be the only way to use multiple processors, it
1161 also involved a number of shortcomings related to the lack of synchronization
1162 between processes (health-checks, peers, stick-tables, stats, ...) which do
1163 not affect threads. As such, any modern configuration is strongly encouraged
Willy Tarreau149ab772019-01-26 14:27:06 +01001164 to migrate away from "nbproc" to "nbthread". "nbthread" also works when
1165 HAProxy is started in foreground. On some platforms supporting CPU affinity,
1166 when nbproc is not used, the default "nbthread" value is automatically set to
1167 the number of CPUs the process is bound to upon startup. This means that the
1168 thread count can easily be adjusted from the calling process using commands
1169 like "taskset" or "cpuset". Otherwise, this value defaults to 1. The default
1170 value is reported in the output of "haproxy -vv". See also "nbproc".
Christopher Fauletbe0faa22017-08-29 15:37:10 +02001171
Willy Tarreau6a06a402007-07-15 20:15:28 +02001172pidfile <pidfile>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001173 Writes PIDs of all daemons into file <pidfile>. This option is equivalent to
Willy Tarreau6a06a402007-07-15 20:15:28 +02001174 the "-p" command line argument. The file must be accessible to the user
1175 starting the process. See also "daemon".
1176
Willy Tarreau1d549722016-02-16 12:41:57 +01001177presetenv <name> <value>
1178 Sets environment variable <name> to value <value>. If the variable exists, it
1179 is NOT overwritten. The changes immediately take effect so that the next line
1180 in the configuration file sees the new value. See also "setenv", "resetenv",
1181 and "unsetenv".
1182
1183resetenv [<name> ...]
1184 Removes all environment variables except the ones specified in argument. It
1185 allows to use a clean controlled environment before setting new values with
1186 setenv or unsetenv. Please note that some internal functions may make use of
1187 some environment variables, such as time manipulation functions, but also
1188 OpenSSL or even external checks. This must be used with extreme care and only
1189 after complete validation. The changes immediately take effect so that the
1190 next line in the configuration file sees the new environment. See also
1191 "setenv", "presetenv", and "unsetenv".
1192
Christopher Fauletff4121f2017-11-22 16:38:49 +01001193stats bind-process [ all | odd | even | <process_num>[-[process_num>]] ] ...
Willy Tarreau35b7b162012-10-22 23:17:18 +02001194 Limits the stats socket to a certain set of processes numbers. By default the
1195 stats socket is bound to all processes, causing a warning to be emitted when
1196 nbproc is greater than 1 because there is no way to select the target process
1197 when connecting. However, by using this setting, it becomes possible to pin
1198 the stats socket to a specific set of processes, typically the first one. The
1199 warning will automatically be disabled when this setting is used, whatever
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001200 the number of processes used. The maximum process ID depends on the machine's
Christopher Fauletff4121f2017-11-22 16:38:49 +01001201 word size (32 or 64). Ranges can be partially defined. The higher bound can
1202 be omitted. In such case, it is replaced by the corresponding maximum
1203 value. A better option consists in using the "process" setting of the "stats
1204 socket" line to force the process on each line.
Willy Tarreau35b7b162012-10-22 23:17:18 +02001205
Baptiste Assmann5626f482015-08-23 10:00:10 +02001206server-state-base <directory>
1207 Specifies the directory prefix to be prepended in front of all servers state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001208 file names which do not start with a '/'. See also "server-state-file",
1209 "load-server-state-from-file" and "server-state-file-name".
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +02001210
1211server-state-file <file>
1212 Specifies the path to the file containing state of servers. If the path starts
1213 with a slash ('/'), it is considered absolute, otherwise it is considered
1214 relative to the directory specified using "server-state-base" (if set) or to
1215 the current directory. Before reloading HAProxy, it is possible to save the
1216 servers' current state using the stats command "show servers state". The
1217 output of this command must be written in the file pointed by <file>. When
1218 starting up, before handling traffic, HAProxy will read, load and apply state
1219 for each server found in the file and available in its current running
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001220 configuration. See also "server-state-base" and "show servers state",
1221 "load-server-state-from-file" and "server-state-file-name"
Baptiste Assmann5626f482015-08-23 10:00:10 +02001222
Willy Tarreau1d549722016-02-16 12:41:57 +01001223setenv <name> <value>
1224 Sets environment variable <name> to value <value>. If the variable exists, it
1225 is overwritten. The changes immediately take effect so that the next line in
1226 the configuration file sees the new value. See also "presetenv", "resetenv",
1227 and "unsetenv".
1228
Willy Tarreau636848a2019-04-15 19:38:50 +02001229set-dumpable
1230 This option is better left disabled by default and enabled only upon a
William Dauchyec730982019-10-27 20:08:10 +01001231 developer's request. If it has been enabled, it may still be forcibly
1232 disabled by prefixing it with the "no" keyword. It has no impact on
1233 performance nor stability but will try hard to re-enable core dumps that were
1234 possibly disabled by file size limitations (ulimit -f), core size limitations
1235 (ulimit -c), or "dumpability" of a process after changing its UID/GID (such
1236 as /proc/sys/fs/suid_dumpable on Linux). Core dumps might still be limited by
1237 the current directory's permissions (check what directory the file is started
1238 from), the chroot directory's permission (it may be needed to temporarily
1239 disable the chroot directive or to move it to a dedicated writable location),
1240 or any other system-specific constraint. For example, some Linux flavours are
1241 notorious for replacing the default core file with a path to an executable
1242 not even installed on the system (check /proc/sys/kernel/core_pattern). Often,
1243 simply writing "core", "core.%p" or "/var/log/core/core.%p" addresses the
1244 issue. When trying to enable this option waiting for a rare issue to
1245 re-appear, it's often a good idea to first try to obtain such a dump by
1246 issuing, for example, "kill -11" to the haproxy process and verify that it
1247 leaves a core where expected when dying.
Willy Tarreau636848a2019-04-15 19:38:50 +02001248
Willy Tarreau610f04b2014-02-13 11:36:41 +01001249ssl-default-bind-ciphers <ciphers>
1250 This setting is only available when support for OpenSSL was built in. It sets
1251 the default string describing the list of cipher algorithms ("cipher suite")
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +00001252 that are negotiated during the SSL/TLS handshake up to TLSv1.2 for all
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001253 "bind" lines which do not explicitly define theirs. The format of the string
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001254 is defined in "man 1 ciphers" from OpenSSL man pages. For background
1255 information and recommendations see e.g.
1256 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
1257 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
1258 cipher configuration, please check the "ssl-default-bind-ciphersuites" keyword.
1259 Please check the "bind" keyword for more information.
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001260
1261ssl-default-bind-ciphersuites <ciphersuites>
1262 This setting is only available when support for OpenSSL was built in and
1263 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default string
1264 describing the list of cipher algorithms ("cipher suite") that are negotiated
1265 during the TLSv1.3 handshake for all "bind" lines which do not explicitly define
1266 theirs. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001267 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites". For
1268 cipher configuration for TLSv1.2 and earlier, please check the
1269 "ssl-default-bind-ciphers" keyword. Please check the "bind" keyword for more
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001270 information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001271
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001272ssl-default-bind-options [<option>]...
1273 This setting is only available when support for OpenSSL was built in. It sets
1274 default ssl-options to force on all "bind" lines. Please check the "bind"
1275 keyword to see available options.
1276
1277 Example:
1278 global
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +02001279 ssl-default-bind-options ssl-min-ver TLSv1.0 no-tls-tickets
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001280
Willy Tarreau610f04b2014-02-13 11:36:41 +01001281ssl-default-server-ciphers <ciphers>
1282 This setting is only available when support for OpenSSL was built in. It
1283 sets the default string describing the list of cipher algorithms that are
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +00001284 negotiated during the SSL/TLS handshake up to TLSv1.2 with the server,
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001285 for all "server" lines which do not explicitly define theirs. The format of
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001286 the string is defined in "man 1 ciphers" from OpenSSL man pages. For background
1287 information and recommendations see e.g.
1288 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
1289 (https://mozilla.github.io/server-side-tls/ssl-config-generator/).
1290 For TLSv1.3 cipher configuration, please check the
1291 "ssl-default-server-ciphersuites" keyword. Please check the "server" keyword
1292 for more information.
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001293
1294ssl-default-server-ciphersuites <ciphersuites>
1295 This setting is only available when support for OpenSSL was built in and
1296 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default
1297 string describing the list of cipher algorithms that are negotiated during
1298 the TLSv1.3 handshake with the server, for all "server" lines which do not
1299 explicitly define theirs. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001300 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites". For
1301 cipher configuration for TLSv1.2 and earlier, please check the
1302 "ssl-default-server-ciphers" keyword. Please check the "server" keyword for
1303 more information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001304
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001305ssl-default-server-options [<option>]...
1306 This setting is only available when support for OpenSSL was built in. It sets
1307 default ssl-options to force on all "server" lines. Please check the "server"
1308 keyword to see available options.
1309
Remi Gacogne47783ef2015-05-29 15:53:22 +02001310ssl-dh-param-file <file>
1311 This setting is only available when support for OpenSSL was built in. It sets
1312 the default DH parameters that are used during the SSL/TLS handshake when
1313 ephemeral Diffie-Hellman (DHE) key exchange is used, for all "bind" lines
Davor Ocelice9ed2812017-12-25 17:49:28 +01001314 which do not explicitly define theirs. It will be overridden by custom DH
Remi Gacogne47783ef2015-05-29 15:53:22 +02001315 parameters found in a bind certificate file if any. If custom DH parameters
Cyril Bonté307ee1e2015-09-28 23:16:06 +02001316 are not specified either by using ssl-dh-param-file or by setting them
1317 directly in the certificate file, pre-generated DH parameters of the size
1318 specified by tune.ssl.default-dh-param will be used. Custom parameters are
1319 known to be more secure and therefore their use is recommended.
Remi Gacogne47783ef2015-05-29 15:53:22 +02001320 Custom DH parameters may be generated by using the OpenSSL command
1321 "openssl dhparam <size>", where size should be at least 2048, as 1024-bit DH
1322 parameters should not be considered secure anymore.
1323
William Lallemand4c5adbf2020-02-24 14:23:22 +01001324ssl-load-extra-files <none|all|bundle|sctl|ocsp|issuer|key>*
William Lallemand3af48e72020-02-03 17:15:52 +01001325 This setting alters the way HAProxy will look for unspecified files during
1326 the loading of the SSL certificates.
1327
1328 By default, HAProxy discovers automatically a lot of files not specified in
1329 the configuration, and you may want to disable this behavior if you want to
1330 optimize the startup time.
1331
1332 "none": Only load the files specified in the configuration. Don't try to load
1333 a certificate bundle if the file does not exist. In the case of a directory,
1334 it won't try to bundle the certificates if they have the same basename.
1335
1336 "all": This is the default behavior, it will try to load everything,
William Lallemand4c5adbf2020-02-24 14:23:22 +01001337 bundles, sctl, ocsp, issuer, key.
William Lallemand3af48e72020-02-03 17:15:52 +01001338
1339 "bundle": When a file specified in the configuration does not exist, HAProxy
1340 will try to load a certificate bundle. This is done by looking for
1341 <basename>.rsa, .ecdsa and .dsa. In the case of directories, HAProxy will
1342 try to gather the files with the same basename in a multi-certificate bundle.
1343 The bundles were introduced with OpenSSL 1.0.2 and were the only way back
1344 then to load an ECDSA certificate and a RSA one, with the same SNI. Since
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05001345 OpenSSL 1.1.1 it is not recommended anymore, you can specify both the ECDSA
William Lallemand3af48e72020-02-03 17:15:52 +01001346 and the RSA file on the bind line.
1347
1348 "sctl": Try to load "<basename>.sctl" for each crt keyword.
1349
1350 "ocsp": Try to load "<basename>.ocsp" for each crt keyword.
1351
1352 "issuer": Try to load "<basename>.issuer" if the issuer of the OCSP file is
1353 not provided in the PEM file.
1354
William Lallemand4c5adbf2020-02-24 14:23:22 +01001355 "key": If the private key was not provided by the PEM file, try to load a
1356 file "<basename>.key" containing a private key.
1357
William Lallemand3af48e72020-02-03 17:15:52 +01001358 The default behavior is "all".
1359
1360 Example:
1361 ssl-load-extra-files bundle sctl
1362 ssl-load-extra-files sctl ocsp issuer
1363 ssl-load-extra-files none
1364
1365 See also: "crt", section 5.1 about bind options.
1366
Emeric Brun850efd52014-01-29 12:24:34 +01001367ssl-server-verify [none|required]
1368 The default behavior for SSL verify on servers side. If specified to 'none',
1369 servers certificates are not verified. The default is 'required' except if
1370 forced using cmdline option '-dV'.
1371
Willy Tarreauabb175f2012-09-24 12:43:26 +02001372stats socket [<address:port>|<path>] [param*]
1373 Binds a UNIX socket to <path> or a TCPv4/v6 address to <address:port>.
1374 Connections to this socket will return various statistics outputs and even
1375 allow some commands to be issued to change some runtime settings. Please
Willy Tarreau1af20c72017-06-23 16:01:14 +02001376 consult section 9.3 "Unix Socket commands" of Management Guide for more
Kevin Decherf949c7202015-10-13 23:26:44 +02001377 details.
Willy Tarreau6162db22009-10-10 17:13:00 +02001378
Willy Tarreauabb175f2012-09-24 12:43:26 +02001379 All parameters supported by "bind" lines are supported, for instance to
1380 restrict access to some users or their access rights. Please consult
1381 section 5.1 for more information.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001382
1383stats timeout <timeout, in milliseconds>
1384 The default timeout on the stats socket is set to 10 seconds. It is possible
1385 to change this value with "stats timeout". The value must be passed in
Willy Tarreaubefdff12007-12-02 22:27:38 +01001386 milliseconds, or be suffixed by a time unit among { us, ms, s, m, h, d }.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001387
1388stats maxconn <connections>
1389 By default, the stats socket is limited to 10 concurrent connections. It is
1390 possible to change this value with "stats maxconn".
1391
Willy Tarreau6a06a402007-07-15 20:15:28 +02001392uid <number>
1393 Changes the process' user ID to <number>. It is recommended that the user ID
1394 is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
1395 be started with superuser privileges in order to be able to switch to another
1396 one. See also "gid" and "user".
1397
1398ulimit-n <number>
1399 Sets the maximum number of per-process file-descriptors to <number>. By
1400 default, it is automatically computed, so it is recommended not to use this
1401 option.
1402
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001403unix-bind [ prefix <prefix> ] [ mode <mode> ] [ user <user> ] [ uid <uid> ]
1404 [ group <group> ] [ gid <gid> ]
1405
1406 Fixes common settings to UNIX listening sockets declared in "bind" statements.
1407 This is mainly used to simplify declaration of those UNIX sockets and reduce
1408 the risk of errors, since those settings are most commonly required but are
1409 also process-specific. The <prefix> setting can be used to force all socket
1410 path to be relative to that directory. This might be needed to access another
1411 component's chroot. Note that those paths are resolved before haproxy chroots
1412 itself, so they are absolute. The <mode>, <user>, <uid>, <group> and <gid>
1413 all have the same meaning as their homonyms used by the "bind" statement. If
1414 both are specified, the "bind" statement has priority, meaning that the
1415 "unix-bind" settings may be seen as process-wide default settings.
1416
Willy Tarreau1d549722016-02-16 12:41:57 +01001417unsetenv [<name> ...]
1418 Removes environment variables specified in arguments. This can be useful to
1419 hide some sensitive information that are occasionally inherited from the
1420 user's environment during some operations. Variables which did not exist are
1421 silently ignored so that after the operation, it is certain that none of
1422 these variables remain. The changes immediately take effect so that the next
1423 line in the configuration file will not see these variables. See also
1424 "setenv", "presetenv", and "resetenv".
1425
Willy Tarreau6a06a402007-07-15 20:15:28 +02001426user <user name>
1427 Similar to "uid" but uses the UID of user name <user name> from /etc/passwd.
1428 See also "uid" and "group".
1429
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02001430node <name>
1431 Only letters, digits, hyphen and underscore are allowed, like in DNS names.
1432
1433 This statement is useful in HA configurations where two or more processes or
1434 servers share the same IP address. By setting a different node-name on all
1435 nodes, it becomes easy to immediately spot what server is handling the
1436 traffic.
1437
1438description <text>
1439 Add a text that describes the instance.
1440
1441 Please note that it is required to escape certain characters (# for example)
1442 and this text is inserted into a html page so you should avoid using
1443 "<" and ">" characters.
1444
Thomas Holmesdb04f192015-05-18 13:21:39 +0100144551degrees-data-file <file path>
1446 The path of the 51Degrees data file to provide device detection services. The
Davor Ocelice9ed2812017-12-25 17:49:28 +01001447 file should be unzipped and accessible by HAProxy with relevant permissions.
Thomas Holmesdb04f192015-05-18 13:21:39 +01001448
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001449 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001450 compiled with USE_51DEGREES.
1451
Ben Shillitof25e8e52016-12-02 14:25:37 +0000145251degrees-property-name-list [<string> ...]
Thomas Holmesdb04f192015-05-18 13:21:39 +01001453 A list of 51Degrees property names to be load from the dataset. A full list
1454 of names is available on the 51Degrees website:
1455 https://51degrees.com/resources/property-dictionary
1456
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001457 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001458 compiled with USE_51DEGREES.
1459
Dragan Dosen93b38d92015-06-29 16:43:25 +0200146051degrees-property-separator <char>
Thomas Holmesdb04f192015-05-18 13:21:39 +01001461 A char that will be appended to every property value in a response header
1462 containing 51Degrees results. If not set that will be set as ','.
1463
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001464 Please note that this option is only available when haproxy has been
1465 compiled with USE_51DEGREES.
1466
146751degrees-cache-size <number>
1468 Sets the size of the 51Degrees converter cache to <number> entries. This
1469 is an LRU cache which reminds previous device detections and their results.
1470 By default, this cache is disabled.
1471
1472 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001473 compiled with USE_51DEGREES.
1474
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001475wurfl-data-file <file path>
1476 The path of the WURFL data file to provide device detection services. The
1477 file should be accessible by HAProxy with relevant permissions.
1478
1479 Please note that this option is only available when haproxy has been compiled
1480 with USE_WURFL=1.
1481
1482wurfl-information-list [<capability>]*
1483 A space-delimited list of WURFL capabilities, virtual capabilities, property
1484 names we plan to use in injected headers. A full list of capability and
1485 virtual capability names is available on the Scientiamobile website :
1486
1487 https://www.scientiamobile.com/wurflCapability
1488
1489 Valid WURFL properties are:
1490 - wurfl_id Contains the device ID of the matched device.
1491
1492 - wurfl_root_id Contains the device root ID of the matched
1493 device.
1494
1495 - wurfl_isdevroot Tells if the matched device is a root device.
1496 Possible values are "TRUE" or "FALSE".
1497
1498 - wurfl_useragent The original useragent coming with this
1499 particular web request.
1500
1501 - wurfl_api_version Contains a string representing the currently
1502 used Libwurfl API version.
1503
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001504 - wurfl_info A string containing information on the parsed
1505 wurfl.xml and its full path.
1506
1507 - wurfl_last_load_time Contains the UNIX timestamp of the last time
1508 WURFL has been loaded successfully.
1509
1510 - wurfl_normalized_useragent The normalized useragent.
1511
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001512 Please note that this option is only available when haproxy has been compiled
1513 with USE_WURFL=1.
1514
1515wurfl-information-list-separator <char>
1516 A char that will be used to separate values in a response header containing
1517 WURFL results. If not set that a comma (',') will be used by default.
1518
1519 Please note that this option is only available when haproxy has been compiled
1520 with USE_WURFL=1.
1521
1522wurfl-patch-file [<file path>]
1523 A list of WURFL patch file paths. Note that patches are loaded during startup
1524 thus before the chroot.
1525
1526 Please note that this option is only available when haproxy has been compiled
1527 with USE_WURFL=1.
1528
paulborilebad132c2019-04-18 11:57:04 +02001529wurfl-cache-size <size>
1530 Sets the WURFL Useragent cache size. For faster lookups, already processed user
1531 agents are kept in a LRU cache :
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001532 - "0" : no cache is used.
paulborilebad132c2019-04-18 11:57:04 +02001533 - <size> : size of lru cache in elements.
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001534
1535 Please note that this option is only available when haproxy has been compiled
1536 with USE_WURFL=1.
1537
William Dauchy0fec3ab2019-10-27 20:08:11 +01001538strict-limits
1539 Makes process fail at startup when a setrlimit fails. Haproxy is tries to set
1540 the best setrlimit according to what has been calculated. If it fails, it
1541 will emit a warning. Use this option if you want an explicit failure of
1542 haproxy when those limits fail. This option is disabled by default. If it has
1543 been enabled, it may still be forcibly disabled by prefixing it with the "no"
1544 keyword.
1545
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015463.2. Performance tuning
Willy Tarreau6a06a402007-07-15 20:15:28 +02001547-----------------------
1548
Willy Tarreaubeb859a2018-11-22 18:07:59 +01001549busy-polling
1550 In some situations, especially when dealing with low latency on processors
1551 supporting a variable frequency or when running inside virtual machines, each
1552 time the process waits for an I/O using the poller, the processor goes back
1553 to sleep or is offered to another VM for a long time, and it causes
1554 excessively high latencies. This option provides a solution preventing the
1555 processor from sleeping by always using a null timeout on the pollers. This
1556 results in a significant latency reduction (30 to 100 microseconds observed)
1557 at the expense of a risk to overheat the processor. It may even be used with
1558 threads, in which case improperly bound threads may heavily conflict,
1559 resulting in a worse performance and high values for the CPU stolen fields
1560 in "show info" output, indicating which threads are misconfigured. It is
1561 important not to let the process run on the same processor as the network
1562 interrupts when this option is used. It is also better to avoid using it on
1563 multiple CPU threads sharing the same core. This option is disabled by
1564 default. If it has been enabled, it may still be forcibly disabled by
1565 prefixing it with the "no" keyword. It is ignored by the "select" and
1566 "poll" pollers.
1567
William Dauchy3894d972019-12-28 15:36:02 +01001568 This option is automatically disabled on old processes in the context of
1569 seamless reload; it avoids too much cpu conflicts when multiple processes
1570 stay around for some time waiting for the end of their current connections.
1571
Willy Tarreau1746eec2014-04-25 10:46:47 +02001572max-spread-checks <delay in milliseconds>
1573 By default, haproxy tries to spread the start of health checks across the
1574 smallest health check interval of all the servers in a farm. The principle is
1575 to avoid hammering services running on the same server. But when using large
1576 check intervals (10 seconds or more), the last servers in the farm take some
1577 time before starting to be tested, which can be a problem. This parameter is
1578 used to enforce an upper bound on delay between the first and the last check,
1579 even if the servers' check intervals are larger. When servers run with
1580 shorter intervals, their intervals will be respected though.
1581
Willy Tarreau6a06a402007-07-15 20:15:28 +02001582maxconn <number>
1583 Sets the maximum per-process number of concurrent connections to <number>. It
1584 is equivalent to the command-line argument "-n". Proxies will stop accepting
1585 connections when this limit is reached. The "ulimit-n" parameter is
Willy Tarreau8274e102014-06-19 15:31:25 +02001586 automatically adjusted according to this value. See also "ulimit-n". Note:
1587 the "select" poller cannot reliably use more than 1024 file descriptors on
1588 some platforms. If your platform only supports select and reports "select
1589 FAILED" on startup, you need to reduce maxconn until it works (slightly
Willy Tarreaub28f3442019-03-04 08:13:43 +01001590 below 500 in general). If this value is not set, it will automatically be
1591 calculated based on the current file descriptors limit reported by the
1592 "ulimit -n" command, possibly reduced to a lower value if a memory limit
1593 is enforced, based on the buffer size, memory allocated to compression, SSL
1594 cache size, and use or not of SSL and the associated maxsslconn (which can
1595 also be automatic).
Willy Tarreau6a06a402007-07-15 20:15:28 +02001596
Willy Tarreau81c25d02011-09-07 15:17:21 +02001597maxconnrate <number>
1598 Sets the maximum per-process number of connections per second to <number>.
1599 Proxies will stop accepting connections when this limit is reached. It can be
1600 used to limit the global capacity regardless of each frontend capacity. It is
1601 important to note that this can only be used as a service protection measure,
1602 as there will not necessarily be a fair share between frontends when the
1603 limit is reached, so it's a good idea to also limit each frontend to some
1604 value close to its expected share. Also, lowering tune.maxaccept can improve
1605 fairness.
1606
William Lallemandd85f9172012-11-09 17:05:39 +01001607maxcomprate <number>
1608 Sets the maximum per-process input compression rate to <number> kilobytes
Davor Ocelice9ed2812017-12-25 17:49:28 +01001609 per second. For each session, if the maximum is reached, the compression
William Lallemandd85f9172012-11-09 17:05:39 +01001610 level will be decreased during the session. If the maximum is reached at the
1611 beginning of a session, the session will not compress at all. If the maximum
1612 is not reached, the compression level will be increased up to
Davor Ocelice9ed2812017-12-25 17:49:28 +01001613 tune.comp.maxlevel. A value of zero means there is no limit, this is the
William Lallemandd85f9172012-11-09 17:05:39 +01001614 default value.
1615
William Lallemand072a2bf2012-11-20 17:01:01 +01001616maxcompcpuusage <number>
1617 Sets the maximum CPU usage HAProxy can reach before stopping the compression
1618 for new requests or decreasing the compression level of current requests.
1619 It works like 'maxcomprate' but measures CPU usage instead of incoming data
1620 bandwidth. The value is expressed in percent of the CPU used by haproxy. In
1621 case of multiple processes (nbproc > 1), each process manages its individual
1622 usage. A value of 100 disable the limit. The default value is 100. Setting
1623 a lower value will prevent the compression work from slowing the whole
1624 process down and from introducing high latencies.
1625
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001626maxpipes <number>
1627 Sets the maximum per-process number of pipes to <number>. Currently, pipes
1628 are only used by kernel-based tcp splicing. Since a pipe contains two file
1629 descriptors, the "ulimit-n" value will be increased accordingly. The default
1630 value is maxconn/4, which seems to be more than enough for most heavy usages.
1631 The splice code dynamically allocates and releases pipes, and can fall back
1632 to standard copy, so setting this value too low may only impact performance.
1633
Willy Tarreau93e7c002013-10-07 18:51:07 +02001634maxsessrate <number>
1635 Sets the maximum per-process number of sessions per second to <number>.
1636 Proxies will stop accepting connections when this limit is reached. It can be
1637 used to limit the global capacity regardless of each frontend capacity. It is
1638 important to note that this can only be used as a service protection measure,
1639 as there will not necessarily be a fair share between frontends when the
1640 limit is reached, so it's a good idea to also limit each frontend to some
1641 value close to its expected share. Also, lowering tune.maxaccept can improve
1642 fairness.
1643
Willy Tarreau403edff2012-09-06 11:58:37 +02001644maxsslconn <number>
1645 Sets the maximum per-process number of concurrent SSL connections to
1646 <number>. By default there is no SSL-specific limit, which means that the
1647 global maxconn setting will apply to all connections. Setting this limit
1648 avoids having openssl use too much memory and crash when malloc returns NULL
1649 (since it unfortunately does not reliably check for such conditions). Note
1650 that the limit applies both to incoming and outgoing connections, so one
1651 connection which is deciphered then ciphered accounts for 2 SSL connections.
Willy Tarreaud0256482015-01-15 21:45:22 +01001652 If this value is not set, but a memory limit is enforced, this value will be
1653 automatically computed based on the memory limit, maxconn, the buffer size,
1654 memory allocated to compression, SSL cache size, and use of SSL in either
1655 frontends, backends or both. If neither maxconn nor maxsslconn are specified
1656 when there is a memory limit, haproxy will automatically adjust these values
1657 so that 100% of the connections can be made over SSL with no risk, and will
1658 consider the sides where it is enabled (frontend, backend, both).
Willy Tarreau403edff2012-09-06 11:58:37 +02001659
Willy Tarreaue43d5322013-10-07 20:01:52 +02001660maxsslrate <number>
1661 Sets the maximum per-process number of SSL sessions per second to <number>.
1662 SSL listeners will stop accepting connections when this limit is reached. It
1663 can be used to limit the global SSL CPU usage regardless of each frontend
1664 capacity. It is important to note that this can only be used as a service
1665 protection measure, as there will not necessarily be a fair share between
1666 frontends when the limit is reached, so it's a good idea to also limit each
1667 frontend to some value close to its expected share. It is also important to
1668 note that the sessions are accounted before they enter the SSL stack and not
1669 after, which also protects the stack against bad handshakes. Also, lowering
1670 tune.maxaccept can improve fairness.
1671
William Lallemand9d5f5482012-11-07 16:12:57 +01001672maxzlibmem <number>
1673 Sets the maximum amount of RAM in megabytes per process usable by the zlib.
1674 When the maximum amount is reached, future sessions will not compress as long
1675 as RAM is unavailable. When sets to 0, there is no limit.
William Lallemande3a7d992012-11-20 11:25:20 +01001676 The default value is 0. The value is available in bytes on the UNIX socket
1677 with "show info" on the line "MaxZlibMemUsage", the memory used by zlib is
1678 "ZlibMemUsage" in bytes.
1679
Willy Tarreau6a06a402007-07-15 20:15:28 +02001680noepoll
1681 Disables the use of the "epoll" event polling system on Linux. It is
1682 equivalent to the command-line argument "-de". The next polling system
Willy Tarreaue9f49e72012-11-11 17:42:00 +01001683 used will generally be "poll". See also "nopoll".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001684
1685nokqueue
1686 Disables the use of the "kqueue" event polling system on BSD. It is
1687 equivalent to the command-line argument "-dk". The next polling system
1688 used will generally be "poll". See also "nopoll".
1689
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +00001690noevports
1691 Disables the use of the event ports event polling system on SunOS systems
1692 derived from Solaris 10 and later. It is equivalent to the command-line
1693 argument "-dv". The next polling system used will generally be "poll". See
1694 also "nopoll".
1695
Willy Tarreau6a06a402007-07-15 20:15:28 +02001696nopoll
1697 Disables the use of the "poll" event polling system. It is equivalent to the
1698 command-line argument "-dp". The next polling system used will be "select".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001699 It should never be needed to disable "poll" since it's available on all
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +00001700 platforms supported by HAProxy. See also "nokqueue", "noepoll" and
1701 "noevports".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001702
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001703nosplice
1704 Disables the use of kernel tcp splicing between sockets on Linux. It is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001705 equivalent to the command line argument "-dS". Data will then be copied
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001706 using conventional and more portable recv/send calls. Kernel tcp splicing is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001707 limited to some very recent instances of kernel 2.6. Most versions between
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001708 2.6.25 and 2.6.28 are buggy and will forward corrupted data, so they must not
1709 be used. This option makes it easier to globally disable kernel splicing in
1710 case of doubt. See also "option splice-auto", "option splice-request" and
1711 "option splice-response".
1712
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001713nogetaddrinfo
1714 Disables the use of getaddrinfo(3) for name resolving. It is equivalent to
1715 the command line argument "-dG". Deprecated gethostbyname(3) will be used.
1716
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +00001717noreuseport
1718 Disables the use of SO_REUSEPORT - see socket(7). It is equivalent to the
1719 command line argument "-dR".
1720
Willy Tarreaud2d33482019-04-25 17:09:07 +02001721profiling.tasks { auto | on | off }
1722 Enables ('on') or disables ('off') per-task CPU profiling. When set to 'auto'
1723 the profiling automatically turns on a thread when it starts to suffer from
1724 an average latency of 1000 microseconds or higher as reported in the
1725 "avg_loop_us" activity field, and automatically turns off when the latency
John Roeslerfb2fce12019-07-10 15:45:51 -05001726 returns below 990 microseconds (this value is an average over the last 1024
Willy Tarreaud2d33482019-04-25 17:09:07 +02001727 loops so it does not vary quickly and tends to significantly smooth short
1728 spikes). It may also spontaneously trigger from time to time on overloaded
1729 systems, containers, or virtual machines, or when the system swaps (which
1730 must absolutely never happen on a load balancer).
1731
1732 CPU profiling per task can be very convenient to report where the time is
1733 spent and which requests have what effect on which other request. Enabling
1734 it will typically affect the overall's performance by less than 1%, thus it
1735 is recommended to leave it to the default 'auto' value so that it only
1736 operates when a problem is identified. This feature requires a system
Willy Tarreau75c62c22018-11-22 11:02:09 +01001737 supporting the clock_gettime(2) syscall with clock identifiers
1738 CLOCK_MONOTONIC and CLOCK_THREAD_CPUTIME_ID, otherwise the reported time will
1739 be zero. This option may be changed at run time using "set profiling" on the
1740 CLI.
1741
Willy Tarreaufe255b72007-10-14 23:09:26 +02001742spread-checks <0..50, in percent>
Simon Hormand60d6912013-11-25 10:46:36 +09001743 Sometimes it is desirable to avoid sending agent and health checks to
1744 servers at exact intervals, for instance when many logical servers are
1745 located on the same physical server. With the help of this parameter, it
1746 becomes possible to add some randomness in the check interval between 0
1747 and +/- 50%. A value between 2 and 5 seems to show good results. The
1748 default value remains at 0.
Willy Tarreaufe255b72007-10-14 23:09:26 +02001749
Davor Ocelice9ed2812017-12-25 17:49:28 +01001750ssl-engine <name> [algo <comma-separated list of algorithms>]
Grant Zhang872f9c22017-01-21 01:10:18 +00001751 Sets the OpenSSL engine to <name>. List of valid values for <name> may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01001752 obtained using the command "openssl engine". This statement may be used
Grant Zhang872f9c22017-01-21 01:10:18 +00001753 multiple times, it will simply enable multiple crypto engines. Referencing an
1754 unsupported engine will prevent haproxy from starting. Note that many engines
1755 will lead to lower HTTPS performance than pure software with recent
1756 processors. The optional command "algo" sets the default algorithms an ENGINE
1757 will supply using the OPENSSL function ENGINE_set_default_string(). A value
Davor Ocelice9ed2812017-12-25 17:49:28 +01001758 of "ALL" uses the engine for all cryptographic operations. If no list of
1759 algo is specified then the value of "ALL" is used. A comma-separated list
Grant Zhang872f9c22017-01-21 01:10:18 +00001760 of different algorithms may be specified, including: RSA, DSA, DH, EC, RAND,
1761 CIPHERS, DIGESTS, PKEY, PKEY_CRYPTO, PKEY_ASN1. This is the same format that
1762 openssl configuration file uses:
1763 https://www.openssl.org/docs/man1.0.2/apps/config.html
1764
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00001765ssl-mode-async
1766 Adds SSL_MODE_ASYNC mode to the SSL context. This enables asynchronous TLS
Emeric Brun3854e012017-05-17 20:42:48 +02001767 I/O operations if asynchronous capable SSL engines are used. The current
Emeric Brunb5e42a82017-06-06 12:35:14 +00001768 implementation supports a maximum of 32 engines. The Openssl ASYNC API
1769 doesn't support moving read/write buffers and is not compliant with
1770 haproxy's buffer management. So the asynchronous mode is disabled on
John Roeslerfb2fce12019-07-10 15:45:51 -05001771 read/write operations (it is only enabled during initial and renegotiation
Emeric Brunb5e42a82017-06-06 12:35:14 +00001772 handshakes).
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00001773
Willy Tarreau33cb0652014-12-23 22:52:37 +01001774tune.buffers.limit <number>
1775 Sets a hard limit on the number of buffers which may be allocated per process.
1776 The default value is zero which means unlimited. The minimum non-zero value
1777 will always be greater than "tune.buffers.reserve" and should ideally always
1778 be about twice as large. Forcing this value can be particularly useful to
1779 limit the amount of memory a process may take, while retaining a sane
Davor Ocelice9ed2812017-12-25 17:49:28 +01001780 behavior. When this limit is reached, sessions which need a buffer wait for
Willy Tarreau33cb0652014-12-23 22:52:37 +01001781 another one to be released by another session. Since buffers are dynamically
1782 allocated and released, the waiting time is very short and not perceptible
1783 provided that limits remain reasonable. In fact sometimes reducing the limit
1784 may even increase performance by increasing the CPU cache's efficiency. Tests
1785 have shown good results on average HTTP traffic with a limit to 1/10 of the
1786 expected global maxconn setting, which also significantly reduces memory
1787 usage. The memory savings come from the fact that a number of connections
1788 will not allocate 2*tune.bufsize. It is best not to touch this value unless
1789 advised to do so by an haproxy core developer.
1790
Willy Tarreau1058ae72014-12-23 22:40:40 +01001791tune.buffers.reserve <number>
1792 Sets the number of buffers which are pre-allocated and reserved for use only
1793 during memory shortage conditions resulting in failed memory allocations. The
1794 minimum value is 2 and is also the default. There is no reason a user would
1795 want to change this value, it's mostly aimed at haproxy core developers.
1796
Willy Tarreau27a674e2009-08-17 07:23:33 +02001797tune.bufsize <number>
1798 Sets the buffer size to this size (in bytes). Lower values allow more
1799 sessions to coexist in the same amount of RAM, and higher values allow some
1800 applications with very large cookies to work. The default value is 16384 and
1801 can be changed at build time. It is strongly recommended not to change this
1802 from the default value, as very low values will break some services such as
1803 statistics, and values larger than default size will increase memory usage,
1804 possibly causing the system to run out of memory. At least the global maxconn
Willy Tarreau45a66cc2017-11-24 11:28:00 +01001805 parameter should be decreased by the same factor as this one is increased. In
1806 addition, use of HTTP/2 mandates that this value must be 16384 or more. If an
1807 HTTP request is larger than (tune.bufsize - tune.maxrewrite), haproxy will
Dmitry Sivachenkof6f4f7b2012-10-21 18:10:25 +04001808 return HTTP 400 (Bad Request) error. Similarly if an HTTP response is larger
Willy Tarreauc77d3642018-12-12 06:19:42 +01001809 than this size, haproxy will return HTTP 502 (Bad Gateway). Note that the
1810 value set using this parameter will automatically be rounded up to the next
1811 multiple of 8 on 32-bit machines and 16 on 64-bit machines.
Willy Tarreau27a674e2009-08-17 07:23:33 +02001812
Willy Tarreau43961d52010-10-04 20:39:20 +02001813tune.chksize <number>
1814 Sets the check buffer size to this size (in bytes). Higher values may help
1815 find string or regex patterns in very large pages, though doing so may imply
1816 more memory and CPU usage. The default value is 16384 and can be changed at
1817 build time. It is not recommended to change this value, but to use better
1818 checks whenever possible.
1819
William Lallemandf3747832012-11-09 12:33:10 +01001820tune.comp.maxlevel <number>
1821 Sets the maximum compression level. The compression level affects CPU
1822 usage during compression. This value affects CPU usage during compression.
1823 Each session using compression initializes the compression algorithm with
1824 this value. The default value is 1.
1825
Willy Tarreauc299e1e2019-02-27 11:35:12 +01001826tune.fail-alloc
1827 If compiled with DEBUG_FAIL_ALLOC, gives the percentage of chances an
1828 allocation attempt fails. Must be between 0 (no failure) and 100 (no
1829 success). This is useful to debug and make sure memory failures are handled
1830 gracefully.
1831
Willy Tarreaufe20e5b2017-07-27 11:42:14 +02001832tune.h2.header-table-size <number>
1833 Sets the HTTP/2 dynamic header table size. It defaults to 4096 bytes and
1834 cannot be larger than 65536 bytes. A larger value may help certain clients
1835 send more compact requests, depending on their capabilities. This amount of
1836 memory is consumed for each HTTP/2 connection. It is recommended not to
1837 change it.
1838
Willy Tarreaue6baec02017-07-27 11:45:11 +02001839tune.h2.initial-window-size <number>
1840 Sets the HTTP/2 initial window size, which is the number of bytes the client
Davor Ocelice9ed2812017-12-25 17:49:28 +01001841 can upload before waiting for an acknowledgment from haproxy. This setting
1842 only affects payload contents (i.e. the body of POST requests), not headers.
Willy Tarreaue6baec02017-07-27 11:45:11 +02001843 The default value is 65535, which roughly allows up to 5 Mbps of upload
1844 bandwidth per client over a network showing a 100 ms ping time, or 500 Mbps
1845 over a 1-ms local network. It can make sense to increase this value to allow
1846 faster uploads, or to reduce it to increase fairness when dealing with many
1847 clients. It doesn't affect resource usage.
1848
Willy Tarreau5242ef82017-07-27 11:47:28 +02001849tune.h2.max-concurrent-streams <number>
1850 Sets the HTTP/2 maximum number of concurrent streams per connection (ie the
1851 number of outstanding requests on a single connection). The default value is
1852 100. A larger one may slightly improve page load time for complex sites when
1853 visited over high latency networks, but increases the amount of resources a
1854 single client may allocate. A value of zero disables the limit so a single
1855 client may create as many streams as allocatable by haproxy. It is highly
1856 recommended not to change this value.
1857
Willy Tarreaua24b35c2019-02-21 13:24:36 +01001858tune.h2.max-frame-size <number>
1859 Sets the HTTP/2 maximum frame size that haproxy announces it is willing to
1860 receive to its peers. The default value is the largest between 16384 and the
1861 buffer size (tune.bufsize). In any case, haproxy will not announce support
1862 for frame sizes larger than buffers. The main purpose of this setting is to
1863 allow to limit the maximum frame size setting when using large buffers. Too
1864 large frame sizes might have performance impact or cause some peers to
1865 misbehave. It is highly recommended not to change this value.
1866
Willy Tarreau193b8c62012-11-22 00:17:38 +01001867tune.http.cookielen <number>
1868 Sets the maximum length of captured cookies. This is the maximum value that
1869 the "capture cookie xxx len yyy" will be allowed to take, and any upper value
1870 will automatically be truncated to this one. It is important not to set too
1871 high a value because all cookie captures still allocate this size whatever
1872 their configured value (they share a same pool). This value is per request
1873 per response, so the memory allocated is twice this value per connection.
1874 When not specified, the limit is set to 63 characters. It is recommended not
1875 to change this value.
1876
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001877tune.http.logurilen <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001878 Sets the maximum length of request URI in logs. This prevents truncating long
1879 request URIs with valuable query strings in log lines. This is not related
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001880 to syslog limits. If you increase this limit, you may also increase the
Davor Ocelice9ed2812017-12-25 17:49:28 +01001881 'log ... len yyy' parameter. Your syslog daemon may also need specific
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001882 configuration directives too.
1883 The default value is 1024.
1884
Willy Tarreauac1932d2011-10-24 19:14:41 +02001885tune.http.maxhdr <number>
1886 Sets the maximum number of headers in a request. When a request comes with a
1887 number of headers greater than this value (including the first line), it is
1888 rejected with a "400 Bad Request" status code. Similarly, too large responses
1889 are blocked with "502 Bad Gateway". The default value is 101, which is enough
1890 for all usages, considering that the widely deployed Apache server uses the
1891 same limit. It can be useful to push this limit further to temporarily allow
Christopher Faulet50174f32017-06-21 16:31:35 +02001892 a buggy application to work by the time it gets fixed. The accepted range is
1893 1..32767. Keep in mind that each new header consumes 32bits of memory for
1894 each session, so don't push this limit too high.
Willy Tarreauac1932d2011-10-24 19:14:41 +02001895
Willy Tarreau7e312732014-02-12 16:35:14 +01001896tune.idletimer <timeout>
1897 Sets the duration after which haproxy will consider that an empty buffer is
1898 probably associated with an idle stream. This is used to optimally adjust
1899 some packet sizes while forwarding large and small data alternatively. The
1900 decision to use splice() or to send large buffers in SSL is modulated by this
1901 parameter. The value is in milliseconds between 0 and 65535. A value of zero
1902 means that haproxy will not try to detect idle streams. The default is 1000,
Davor Ocelice9ed2812017-12-25 17:49:28 +01001903 which seems to correctly detect end user pauses (e.g. read a page before
John Roeslerfb2fce12019-07-10 15:45:51 -05001904 clicking). There should be no reason for changing this value. Please check
Willy Tarreau7e312732014-02-12 16:35:14 +01001905 tune.ssl.maxrecord below.
1906
Willy Tarreau7ac908b2019-02-27 12:02:18 +01001907tune.listener.multi-queue { on | off }
1908 Enables ('on') or disables ('off') the listener's multi-queue accept which
1909 spreads the incoming traffic to all threads a "bind" line is allowed to run
1910 on instead of taking them for itself. This provides a smoother traffic
1911 distribution and scales much better, especially in environments where threads
1912 may be unevenly loaded due to external activity (network interrupts colliding
1913 with one thread for example). This option is enabled by default, but it may
1914 be forcefully disabled for troubleshooting or for situations where it is
1915 estimated that the operating system already provides a good enough
1916 distribution and connections are extremely short-lived.
1917
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001918tune.lua.forced-yield <number>
1919 This directive forces the Lua engine to execute a yield each <number> of
Tim Düsterhus4896c442016-11-29 02:15:19 +01001920 instructions executed. This permits interrupting a long script and allows the
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001921 HAProxy scheduler to process other tasks like accepting connections or
1922 forwarding traffic. The default value is 10000 instructions. If HAProxy often
Davor Ocelice9ed2812017-12-25 17:49:28 +01001923 executes some Lua code but more responsiveness is required, this value can be
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001924 lowered. If the Lua code is quite long and its result is absolutely required
1925 to process the data, the <number> can be increased.
1926
Willy Tarreau32f61e22015-03-18 17:54:59 +01001927tune.lua.maxmem
1928 Sets the maximum amount of RAM in megabytes per process usable by Lua. By
1929 default it is zero which means unlimited. It is important to set a limit to
1930 ensure that a bug in a script will not result in the system running out of
1931 memory.
1932
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001933tune.lua.session-timeout <timeout>
1934 This is the execution timeout for the Lua sessions. This is useful for
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001935 preventing infinite loops or spending too much time in Lua. This timeout
1936 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001937 not taken in account. The default timeout is 4s.
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001938
1939tune.lua.task-timeout <timeout>
1940 Purpose is the same as "tune.lua.session-timeout", but this timeout is
1941 dedicated to the tasks. By default, this timeout isn't set because a task may
1942 remain alive during of the lifetime of HAProxy. For example, a task used to
1943 check servers.
1944
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001945tune.lua.service-timeout <timeout>
1946 This is the execution timeout for the Lua services. This is useful for
1947 preventing infinite loops or spending too much time in Lua. This timeout
1948 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001949 not taken in account. The default timeout is 4s.
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001950
Willy Tarreaua0250ba2008-01-06 11:22:57 +01001951tune.maxaccept <number>
Willy Tarreau16a21472012-11-19 12:39:59 +01001952 Sets the maximum number of consecutive connections a process may accept in a
1953 row before switching to other work. In single process mode, higher numbers
1954 give better performance at high connection rates. However in multi-process
1955 modes, keeping a bit of fairness between processes generally is better to
1956 increase performance. This value applies individually to each listener, so
1957 that the number of processes a listener is bound to is taken into account.
1958 This value defaults to 64. In multi-process mode, it is divided by twice
1959 the number of processes the listener is bound to. Setting this value to -1
1960 completely disables the limitation. It should normally not be needed to tweak
1961 this value.
Willy Tarreaua0250ba2008-01-06 11:22:57 +01001962
1963tune.maxpollevents <number>
1964 Sets the maximum amount of events that can be processed at once in a call to
1965 the polling system. The default value is adapted to the operating system. It
1966 has been noticed that reducing it below 200 tends to slightly decrease
1967 latency at the expense of network bandwidth, and increasing it above 200
1968 tends to trade latency for slightly increased bandwidth.
1969
Willy Tarreau27a674e2009-08-17 07:23:33 +02001970tune.maxrewrite <number>
1971 Sets the reserved buffer space to this size in bytes. The reserved space is
1972 used for header rewriting or appending. The first reads on sockets will never
1973 fill more than bufsize-maxrewrite. Historically it has defaulted to half of
1974 bufsize, though that does not make much sense since there are rarely large
1975 numbers of headers to add. Setting it too high prevents processing of large
1976 requests or responses. Setting it too low prevents addition of new headers
1977 to already large requests or to POST requests. It is generally wise to set it
1978 to about 1024. It is automatically readjusted to half of bufsize if it is
1979 larger than that. This means you don't have to worry about it when changing
1980 bufsize.
1981
Willy Tarreauf3045d22015-04-29 16:24:50 +02001982tune.pattern.cache-size <number>
1983 Sets the size of the pattern lookup cache to <number> entries. This is an LRU
1984 cache which reminds previous lookups and their results. It is used by ACLs
1985 and maps on slow pattern lookups, namely the ones using the "sub", "reg",
1986 "dir", "dom", "end", "bin" match methods as well as the case-insensitive
1987 strings. It applies to pattern expressions which means that it will be able
1988 to memorize the result of a lookup among all the patterns specified on a
1989 configuration line (including all those loaded from files). It automatically
1990 invalidates entries which are updated using HTTP actions or on the CLI. The
1991 default cache size is set to 10000 entries, which limits its footprint to
Willy Tarreau403bfbb2019-10-23 06:59:31 +02001992 about 5 MB per process/thread on 32-bit systems and 8 MB per process/thread
1993 on 64-bit systems, as caches are thread/process local. There is a very low
Willy Tarreauf3045d22015-04-29 16:24:50 +02001994 risk of collision in this cache, which is in the order of the size of the
1995 cache divided by 2^64. Typically, at 10000 requests per second with the
1996 default cache size of 10000 entries, there's 1% chance that a brute force
1997 attack could cause a single collision after 60 years, or 0.1% after 6 years.
1998 This is considered much lower than the risk of a memory corruption caused by
1999 aging components. If this is not acceptable, the cache can be disabled by
2000 setting this parameter to 0.
2001
Willy Tarreaubd9a0a72011-10-23 21:14:29 +02002002tune.pipesize <number>
2003 Sets the kernel pipe buffer size to this size (in bytes). By default, pipes
2004 are the default size for the system. But sometimes when using TCP splicing,
2005 it can improve performance to increase pipe sizes, especially if it is
2006 suspected that pipes are not filled and that many calls to splice() are
2007 performed. This has an impact on the kernel's memory footprint, so this must
2008 not be changed if impacts are not understood.
2009
Olivier Houchard88698d92019-04-16 19:07:22 +02002010tune.pool-low-fd-ratio <number>
2011 This setting sets the max number of file descriptors (in percentage) used by
2012 haproxy globally against the maximum number of file descriptors haproxy can
2013 use before we stop putting connection into the idle pool for reuse. The
2014 default is 20.
2015
2016tune.pool-high-fd-ratio <number>
2017 This setting sets the max number of file descriptors (in percentage) used by
2018 haproxy globally against the maximum number of file descriptors haproxy can
2019 use before we start killing idle connections when we can't reuse a connection
2020 and we have to create a new one. The default is 25 (one quarter of the file
2021 descriptor will mean that roughly half of the maximum front connections can
2022 keep an idle connection behind, anything beyond this probably doesn't make
John Roeslerfb2fce12019-07-10 15:45:51 -05002023 much sense in the general case when targeting connection reuse).
Olivier Houchard88698d92019-04-16 19:07:22 +02002024
Willy Tarreaue803de22010-01-21 17:43:04 +01002025tune.rcvbuf.client <number>
2026tune.rcvbuf.server <number>
2027 Forces the kernel socket receive buffer size on the client or the server side
2028 to the specified value in bytes. This value applies to all TCP/HTTP frontends
2029 and backends. It should normally never be set, and the default size (0) lets
John Roeslerfb2fce12019-07-10 15:45:51 -05002030 the kernel auto-tune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01002031 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01002032 order to save kernel memory by preventing it from buffering too large amounts
2033 of received data. Lower values will significantly increase CPU usage though.
2034
Willy Tarreaub22fc302015-12-14 12:04:35 +01002035tune.recv_enough <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01002036 HAProxy uses some hints to detect that a short read indicates the end of the
Willy Tarreaub22fc302015-12-14 12:04:35 +01002037 socket buffers. One of them is that a read returns more than <recv_enough>
2038 bytes, which defaults to 10136 (7 segments of 1448 each). This default value
2039 may be changed by this setting to better deal with workloads involving lots
2040 of short messages such as telnet or SSH sessions.
2041
Olivier Houchard1599b802018-05-24 18:59:04 +02002042tune.runqueue-depth <number>
John Roeslerfb2fce12019-07-10 15:45:51 -05002043 Sets the maximum amount of task that can be processed at once when running
Olivier Houchard1599b802018-05-24 18:59:04 +02002044 tasks. The default value is 200. Increasing it may incur latency when
2045 dealing with I/Os, making it too small can incur extra overhead.
2046
Willy Tarreaue803de22010-01-21 17:43:04 +01002047tune.sndbuf.client <number>
2048tune.sndbuf.server <number>
2049 Forces the kernel socket send buffer size on the client or the server side to
2050 the specified value in bytes. This value applies to all TCP/HTTP frontends
2051 and backends. It should normally never be set, and the default size (0) lets
John Roeslerfb2fce12019-07-10 15:45:51 -05002052 the kernel auto-tune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01002053 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01002054 order to save kernel memory by preventing it from buffering too large amounts
2055 of received data. Lower values will significantly increase CPU usage though.
2056 Another use case is to prevent write timeouts with extremely slow clients due
2057 to the kernel waiting for a large part of the buffer to be read before
2058 notifying haproxy again.
2059
Willy Tarreau6ec58db2012-11-16 16:32:15 +01002060tune.ssl.cachesize <number>
Emeric Brunaf9619d2012-11-28 18:47:52 +01002061 Sets the size of the global SSL session cache, in a number of blocks. A block
2062 is large enough to contain an encoded session without peer certificate.
2063 An encoded session with peer certificate is stored in multiple blocks
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002064 depending on the size of the peer certificate. A block uses approximately
Emeric Brunaf9619d2012-11-28 18:47:52 +01002065 200 bytes of memory. The default value may be forced at build time, otherwise
Davor Ocelice9ed2812017-12-25 17:49:28 +01002066 defaults to 20000. When the cache is full, the most idle entries are purged
Emeric Brunaf9619d2012-11-28 18:47:52 +01002067 and reassigned. Higher values reduce the occurrence of such a purge, hence
2068 the number of CPU-intensive SSL handshakes by ensuring that all users keep
2069 their session as long as possible. All entries are pre-allocated upon startup
Emeric Brun22890a12012-12-28 14:41:32 +01002070 and are shared between all processes if "nbproc" is greater than 1. Setting
2071 this value to 0 disables the SSL session cache.
Willy Tarreau6ec58db2012-11-16 16:32:15 +01002072
Emeric Brun8dc60392014-05-09 13:52:00 +02002073tune.ssl.force-private-cache
Lukas Tribus27935782018-10-01 02:00:16 +02002074 This option disables SSL session cache sharing between all processes. It
Emeric Brun8dc60392014-05-09 13:52:00 +02002075 should normally not be used since it will force many renegotiations due to
2076 clients hitting a random process. But it may be required on some operating
2077 systems where none of the SSL cache synchronization method may be used. In
2078 this case, adding a first layer of hash-based load balancing before the SSL
2079 layer might limit the impact of the lack of session sharing.
2080
Emeric Brun4f65bff2012-11-16 15:11:00 +01002081tune.ssl.lifetime <timeout>
2082 Sets how long a cached SSL session may remain valid. This time is expressed
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002083 in seconds and defaults to 300 (5 min). It is important to understand that it
Emeric Brun4f65bff2012-11-16 15:11:00 +01002084 does not guarantee that sessions will last that long, because if the cache is
2085 full, the longest idle sessions will be purged despite their configured
2086 lifetime. The real usefulness of this setting is to prevent sessions from
2087 being used for too long.
2088
Willy Tarreaubfd59462013-02-21 07:46:09 +01002089tune.ssl.maxrecord <number>
2090 Sets the maximum amount of bytes passed to SSL_write() at a time. Default
2091 value 0 means there is no limit. Over SSL/TLS, the client can decipher the
2092 data only once it has received a full record. With large records, it means
2093 that clients might have to download up to 16kB of data before starting to
2094 process them. Limiting the value can improve page load times on browsers
2095 located over high latency or low bandwidth networks. It is suggested to find
2096 optimal values which fit into 1 or 2 TCP segments (generally 1448 bytes over
2097 Ethernet with TCP timestamps enabled, or 1460 when timestamps are disabled),
2098 keeping in mind that SSL/TLS add some overhead. Typical values of 1419 and
2099 2859 gave good results during tests. Use "strace -e trace=write" to find the
Davor Ocelice9ed2812017-12-25 17:49:28 +01002100 best value. HAProxy will automatically switch to this setting after an idle
Willy Tarreau7e312732014-02-12 16:35:14 +01002101 stream has been detected (see tune.idletimer above).
Willy Tarreaubfd59462013-02-21 07:46:09 +01002102
Remi Gacognef46cd6e2014-06-12 14:58:40 +02002103tune.ssl.default-dh-param <number>
2104 Sets the maximum size of the Diffie-Hellman parameters used for generating
2105 the ephemeral/temporary Diffie-Hellman key in case of DHE key exchange. The
2106 final size will try to match the size of the server's RSA (or DSA) key (e.g,
2107 a 2048 bits temporary DH key for a 2048 bits RSA key), but will not exceed
2108 this maximum value. Default value if 1024. Only 1024 or higher values are
2109 allowed. Higher values will increase the CPU load, and values greater than
2110 1024 bits are not supported by Java 7 and earlier clients. This value is not
Remi Gacogne47783ef2015-05-29 15:53:22 +02002111 used if static Diffie-Hellman parameters are supplied either directly
2112 in the certificate file or by using the ssl-dh-param-file parameter.
Remi Gacognef46cd6e2014-06-12 14:58:40 +02002113
Christopher Faulet31af49d2015-06-09 17:29:50 +02002114tune.ssl.ssl-ctx-cache-size <number>
2115 Sets the size of the cache used to store generated certificates to <number>
2116 entries. This is a LRU cache. Because generating a SSL certificate
2117 dynamically is expensive, they are cached. The default cache size is set to
2118 1000 entries.
2119
Thierry FOURNIER5bf77322017-02-25 12:45:22 +01002120tune.ssl.capture-cipherlist-size <number>
2121 Sets the maximum size of the buffer used for capturing client-hello cipher
2122 list. If the value is 0 (default value) the capture is disabled, otherwise
2123 a buffer is allocated for each SSL/TLS connection.
2124
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002125tune.vars.global-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01002126tune.vars.proc-max-size <size>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002127tune.vars.reqres-max-size <size>
2128tune.vars.sess-max-size <size>
2129tune.vars.txn-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01002130 These five tunes help to manage the maximum amount of memory used by the
2131 variables system. "global" limits the overall amount of memory available for
2132 all scopes. "proc" limits the memory for the process scope, "sess" limits the
2133 memory for the session scope, "txn" for the transaction scope, and "reqres"
2134 limits the memory for each request or response processing.
2135 Memory accounting is hierarchical, meaning more coarse grained limits include
2136 the finer grained ones: "proc" includes "sess", "sess" includes "txn", and
2137 "txn" includes "reqres".
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002138
Daniel Schneller0b547052016-03-21 20:46:57 +01002139 For example, when "tune.vars.sess-max-size" is limited to 100,
2140 "tune.vars.txn-max-size" and "tune.vars.reqres-max-size" cannot exceed
2141 100 either. If we create a variable "txn.var" that contains 100 bytes,
2142 all available space is consumed.
2143 Notice that exceeding the limits at runtime will not result in an error
2144 message, but values might be cut off or corrupted. So make sure to accurately
2145 plan for the amount of space needed to store all your variables.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002146
William Lallemanda509e4c2012-11-07 16:54:34 +01002147tune.zlib.memlevel <number>
2148 Sets the memLevel parameter in zlib initialization for each session. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002149 defines how much memory should be allocated for the internal compression
William Lallemanda509e4c2012-11-07 16:54:34 +01002150 state. A value of 1 uses minimum memory but is slow and reduces compression
Davor Ocelice9ed2812017-12-25 17:49:28 +01002151 ratio, a value of 9 uses maximum memory for optimal speed. Can be a value
William Lallemanda509e4c2012-11-07 16:54:34 +01002152 between 1 and 9. The default value is 8.
2153
2154tune.zlib.windowsize <number>
2155 Sets the window size (the size of the history buffer) as a parameter of the
2156 zlib initialization for each session. Larger values of this parameter result
Davor Ocelice9ed2812017-12-25 17:49:28 +01002157 in better compression at the expense of memory usage. Can be a value between
2158 8 and 15. The default value is 15.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002159
Willy Tarreauc57f0e22009-05-10 13:12:33 +020021603.3. Debugging
2161--------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02002162
Willy Tarreau1b857852020-02-25 11:27:22 +01002163debug (deprecated)
Willy Tarreau6a06a402007-07-15 20:15:28 +02002164 Enables debug mode which dumps to stdout all exchanges, and disables forking
2165 into background. It is the equivalent of the command-line argument "-d". It
2166 should never be used in a production configuration since it may prevent full
2167 system startup.
2168
2169quiet
2170 Do not display any message during startup. It is equivalent to the command-
2171 line argument "-q".
2172
Emeric Brunf099e792010-09-27 12:05:28 +02002173
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010021743.4. Userlists
2175--------------
2176It is possible to control access to frontend/backend/listen sections or to
2177http stats by allowing only authenticated and authorized users. To do this,
2178it is required to create at least one userlist and to define users.
2179
2180userlist <listname>
Cyril Bonté78caf842010-03-10 22:41:43 +01002181 Creates new userlist with name <listname>. Many independent userlists can be
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002182 used to store authentication & authorization data for independent customers.
2183
2184group <groupname> [users <user>,<user>,(...)]
Cyril Bonté78caf842010-03-10 22:41:43 +01002185 Adds group <groupname> to the current userlist. It is also possible to
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002186 attach users to this group by using a comma separated list of names
2187 proceeded by "users" keyword.
2188
Cyril Bontéf0c60612010-02-06 14:44:47 +01002189user <username> [password|insecure-password <password>]
2190 [groups <group>,<group>,(...)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002191 Adds user <username> to the current userlist. Both secure (encrypted) and
2192 insecure (unencrypted) passwords can be used. Encrypted passwords are
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01002193 evaluated using the crypt(3) function, so depending on the system's
2194 capabilities, different algorithms are supported. For example, modern Glibc
2195 based Linux systems support MD5, SHA-256, SHA-512, and, of course, the
2196 classic DES-based method of encrypting passwords.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002197
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01002198 Attention: Be aware that using encrypted passwords might cause significantly
2199 increased CPU usage, depending on the number of requests, and the algorithm
2200 used. For any of the hashed variants, the password for each request must
2201 be processed through the chosen algorithm, before it can be compared to the
2202 value specified in the config file. Most current algorithms are deliberately
2203 designed to be expensive to compute to achieve resistance against brute
2204 force attacks. They do not simply salt/hash the clear text password once,
2205 but thousands of times. This can quickly become a major factor in haproxy's
2206 overall CPU consumption!
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002207
2208 Example:
Cyril Bontéf0c60612010-02-06 14:44:47 +01002209 userlist L1
2210 group G1 users tiger,scott
2211 group G2 users xdb,scott
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002212
Cyril Bontéf0c60612010-02-06 14:44:47 +01002213 user tiger password $6$k6y3o.eP$JlKBx9za9667qe4(...)xHSwRv6J.C0/D7cV91
2214 user scott insecure-password elgato
2215 user xdb insecure-password hello
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002216
Cyril Bontéf0c60612010-02-06 14:44:47 +01002217 userlist L2
2218 group G1
2219 group G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002220
Cyril Bontéf0c60612010-02-06 14:44:47 +01002221 user tiger password $6$k6y3o.eP$JlKBx(...)xHSwRv6J.C0/D7cV91 groups G1
2222 user scott insecure-password elgato groups G1,G2
2223 user xdb insecure-password hello groups G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002224
2225 Please note that both lists are functionally identical.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002226
Emeric Brunf099e792010-09-27 12:05:28 +02002227
22283.5. Peers
Cyril Bontédc4d9032012-04-08 21:57:39 +02002229----------
Emeric Brun94900952015-06-11 18:25:54 +02002230It is possible to propagate entries of any data-types in stick-tables between
2231several haproxy instances over TCP connections in a multi-master fashion. Each
2232instance pushes its local updates and insertions to remote peers. The pushed
2233values overwrite remote ones without aggregation. Interrupted exchanges are
2234automatically detected and recovered from the last known point.
2235In addition, during a soft restart, the old process connects to the new one
2236using such a TCP connection to push all its entries before the new process
2237tries to connect to other peers. That ensures very fast replication during a
2238reload, it typically takes a fraction of a second even for large tables.
2239Note that Server IDs are used to identify servers remotely, so it is important
2240that configurations look similar or at least that the same IDs are forced on
2241each server on all participants.
Emeric Brunf099e792010-09-27 12:05:28 +02002242
2243peers <peersect>
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002244 Creates a new peer list with name <peersect>. It is an independent section,
Emeric Brunf099e792010-09-27 12:05:28 +02002245 which is referenced by one or more stick-tables.
2246
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002247bind [<address>]:<port_range> [, ...] [param*]
2248 Defines the binding parameters of the local peer of this "peers" section.
2249 Such lines are not supported with "peer" line in the same "peers" section.
2250
Willy Tarreau77e4bd12015-05-01 20:02:17 +02002251disabled
2252 Disables a peers section. It disables both listening and any synchronization
2253 related to this section. This is provided to disable synchronization of stick
2254 tables without having to comment out all "peers" references.
2255
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002256default-bind [param*]
2257 Defines the binding parameters for the local peer, excepted its address.
2258
2259default-server [param*]
2260 Change default options for a server in a "peers" section.
2261
2262 Arguments:
2263 <param*> is a list of parameters for this server. The "default-server"
2264 keyword accepts an important number of options and has a complete
2265 section dedicated to it. Please refer to section 5 for more
2266 details.
2267
2268
2269 See also: "server" and section 5 about server options
2270
Willy Tarreau77e4bd12015-05-01 20:02:17 +02002271enable
2272 This re-enables a disabled peers section which was previously disabled.
2273
Frédéric Lécailleb6f759b2019-11-05 09:57:45 +01002274log <address> [len <length>] [format <format>] [sample <ranges>:<smp_size>]
2275 <facility> [<level> [<minlevel>]]
2276 "peers" sections support the same "log" keyword as for the proxies to
2277 log information about the "peers" listener. See "log" option for proxies for
2278 more details.
2279
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002280peer <peername> <ip>:<port> [param*]
Emeric Brunf099e792010-09-27 12:05:28 +02002281 Defines a peer inside a peers section.
2282 If <peername> is set to the local peer name (by default hostname, or forced
2283 using "-L" command line option), haproxy will listen for incoming remote peer
2284 connection on <ip>:<port>. Otherwise, <ip>:<port> defines where to connect to
2285 to join the remote peer, and <peername> is used at the protocol level to
2286 identify and validate the remote peer on the server side.
2287
2288 During a soft restart, local peer <ip>:<port> is used by the old instance to
2289 connect the new one and initiate a complete replication (teaching process).
2290
2291 It is strongly recommended to have the exact same peers declaration on all
2292 peers and to only rely on the "-L" command line argument to change the local
2293 peer name. This makes it easier to maintain coherent configuration files
2294 across all peers.
2295
William Lallemandb2f07452015-05-12 14:27:13 +02002296 You may want to reference some environment variables in the address
2297 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01002298
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002299 Note: "peer" keyword may transparently be replaced by "server" keyword (see
2300 "server" keyword explanation below).
2301
2302server <peername> [<ip>:<port>] [param*]
Michael Prokop4438c602019-05-24 10:25:45 +02002303 As previously mentioned, "peer" keyword may be replaced by "server" keyword
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002304 with a support for all "server" parameters found in 5.2 paragraph.
2305 If the underlying peer is local, <ip>:<port> parameters must not be present.
2306 These parameters must be provided on a "bind" line (see "bind" keyword
2307 of this "peers" section).
2308 Some of these parameters are irrelevant for "peers" sections.
2309
2310
Cyril Bontédc4d9032012-04-08 21:57:39 +02002311 Example:
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002312 # The old way.
Emeric Brunf099e792010-09-27 12:05:28 +02002313 peers mypeers
Willy Tarreauf7b30a92010-12-06 22:59:17 +01002314 peer haproxy1 192.168.0.1:1024
2315 peer haproxy2 192.168.0.2:1024
2316 peer haproxy3 10.2.0.1:1024
Emeric Brunf099e792010-09-27 12:05:28 +02002317
2318 backend mybackend
2319 mode tcp
2320 balance roundrobin
2321 stick-table type ip size 20k peers mypeers
2322 stick on src
2323
Willy Tarreauf7b30a92010-12-06 22:59:17 +01002324 server srv1 192.168.0.30:80
2325 server srv2 192.168.0.31:80
Emeric Brunf099e792010-09-27 12:05:28 +02002326
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002327 Example:
2328 peers mypeers
2329 bind 127.0.0.11:10001 ssl crt mycerts/pem
2330 default-server ssl verify none
2331 server hostA 127.0.0.10:10000
2332 server hostB #local peer
Emeric Brunf099e792010-09-27 12:05:28 +02002333
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01002334
2335table <tablename> type {ip | integer | string [len <length>] | binary [len <length>]}
2336 size <size> [expire <expire>] [nopurge] [store <data_type>]*
2337
2338 Configure a stickiness table for the current section. This line is parsed
2339 exactly the same way as the "stick-table" keyword in others section, except
John Roeslerfb2fce12019-07-10 15:45:51 -05002340 for the "peers" argument which is not required here and with an additional
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01002341 mandatory first parameter to designate the stick-table. Contrary to others
2342 sections, there may be several "table" lines in "peers" sections (see also
2343 "stick-table" keyword).
2344
2345 Also be aware of the fact that "peers" sections have their own stick-table
2346 namespaces to avoid collisions between stick-table names identical in
2347 different "peers" section. This is internally handled prepending the "peers"
2348 sections names to the name of the stick-tables followed by a '/' character.
2349 If somewhere else in the configuration file you have to refer to such
2350 stick-tables declared in "peers" sections you must use the prefixed version
2351 of the stick-table name as follows:
2352
2353 peers mypeers
2354 peer A ...
2355 peer B ...
2356 table t1 ...
2357
2358 frontend fe1
2359 tcp-request content track-sc0 src table mypeers/t1
2360
2361 This is also this prefixed version of the stick-table names which must be
2362 used to refer to stick-tables through the CLI.
2363
2364 About "peers" protocol, as only "peers" belonging to the same section may
2365 communicate with each others, there is no need to do such a distinction.
2366 Several "peers" sections may declare stick-tables with the same name.
2367 This is shorter version of the stick-table name which is sent over the network.
2368 There is only a '/' character as prefix to avoid stick-table name collisions between
2369 stick-tables declared as backends and stick-table declared in "peers" sections
2370 as follows in this weird but supported configuration:
2371
2372 peers mypeers
2373 peer A ...
2374 peer B ...
2375 table t1 type string size 10m store gpc0
2376
2377 backend t1
2378 stick-table type string size 10m store gpc0 peers mypeers
2379
2380 Here "t1" table declared in "mypeeers" section has "mypeers/t1" as global name.
2381 "t1" table declared as a backend as "t1" as global name. But at peer protocol
2382 level the former table is named "/t1", the latter is again named "t1".
2383
Simon Horman51a1cf62015-02-03 13:00:44 +090023843.6. Mailers
2385------------
2386It is possible to send email alerts when the state of servers changes.
2387If configured email alerts are sent to each mailer that is configured
2388in a mailers section. Email is sent to mailers using SMTP.
2389
Pieter Baauw386a1272015-08-16 15:26:24 +02002390mailers <mailersect>
Simon Horman51a1cf62015-02-03 13:00:44 +09002391 Creates a new mailer list with the name <mailersect>. It is an
2392 independent section which is referenced by one or more proxies.
2393
2394mailer <mailername> <ip>:<port>
2395 Defines a mailer inside a mailers section.
2396
2397 Example:
2398 mailers mymailers
2399 mailer smtp1 192.168.0.1:587
2400 mailer smtp2 192.168.0.2:587
2401
2402 backend mybackend
2403 mode tcp
2404 balance roundrobin
2405
2406 email-alert mailers mymailers
2407 email-alert from test1@horms.org
2408 email-alert to test2@horms.org
2409
2410 server srv1 192.168.0.30:80
2411 server srv2 192.168.0.31:80
2412
Pieter Baauw235fcfc2016-02-13 15:33:40 +01002413timeout mail <time>
2414 Defines the time available for a mail/connection to be made and send to
2415 the mail-server. If not defined the default value is 10 seconds. To allow
2416 for at least two SYN-ACK packets to be send during initial TCP handshake it
2417 is advised to keep this value above 4 seconds.
2418
2419 Example:
2420 mailers mymailers
2421 timeout mail 20s
2422 mailer smtp1 192.168.0.1:587
Simon Horman51a1cf62015-02-03 13:00:44 +09002423
William Lallemandc9515522019-06-12 16:32:11 +020024243.7. Programs
2425-------------
2426In master-worker mode, it is possible to launch external binaries with the
2427master, these processes are called programs. These programs are launched and
2428managed the same way as the workers.
2429
2430During a reload of HAProxy, those processes are dealing with the same
2431sequence as a worker:
2432
2433 - the master is re-executed
2434 - the master sends a SIGUSR1 signal to the program
2435 - if "option start-on-reload" is not disabled, the master launches a new
2436 instance of the program
2437
2438During a stop, or restart, a SIGTERM is sent to the programs.
2439
2440program <name>
2441 This is a new program section, this section will create an instance <name>
2442 which is visible in "show proc" on the master CLI. (See "9.4. Master CLI" in
2443 the management guide).
2444
2445command <command> [arguments*]
2446 Define the command to start with optional arguments. The command is looked
2447 up in the current PATH if it does not include an absolute path. This is a
2448 mandatory option of the program section. Arguments containing spaces must
2449 be enclosed in quotes or double quotes or be prefixed by a backslash.
2450
Andrew Heberle97236962019-07-12 11:50:26 +08002451user <user name>
2452 Changes the executed command user ID to the <user name> from /etc/passwd.
2453 See also "group".
2454
2455group <group name>
2456 Changes the executed command group ID to the <group name> from /etc/group.
2457 See also "user".
2458
William Lallemandc9515522019-06-12 16:32:11 +02002459option start-on-reload
2460no option start-on-reload
2461 Start (or not) a new instance of the program upon a reload of the master.
2462 The default is to start a new instance. This option may only be used in a
2463 program section.
2464
2465
Christopher Faulet76edc0f2020-01-13 15:52:01 +010024663.8. HTTP-errors
2467----------------
2468
2469It is possible to globally declare several groups of HTTP errors, to be
2470imported afterwards in any proxy section. Same group may be referenced at
2471several places and can be fully or partially imported.
2472
2473http-errors <name>
2474 Create a new http-errors group with the name <name>. It is an independent
2475 section that may be referenced by one or more proxies using its name.
2476
2477errorfile <code> <file>
2478 Associate a file contents to an HTTP error code
2479
2480 Arguments :
2481 <code> is the HTTP status code. Currently, HAProxy is capable of
2482 generating codes 200, 400, 403, 404, 405, 408, 410, 425, 429,
2483 500, 502, 503, and 504.
2484
2485 <file> designates a file containing the full HTTP response. It is
2486 recommended to follow the common practice of appending ".http" to
2487 the filename so that people do not confuse the response with HTML
2488 error pages, and to use absolute paths, since files are read
2489 before any chroot is performed.
2490
2491 Please referrers to "errorfile" keyword in section 4 for details.
2492
2493 Example:
2494 http-errors website-1
2495 errorfile 400 /etc/haproxy/errorfiles/site1/400.http
2496 errorfile 404 /etc/haproxy/errorfiles/site1/404.http
2497 errorfile 408 /dev/null # work around Chrome pre-connect bug
2498
2499 http-errors website-2
2500 errorfile 400 /etc/haproxy/errorfiles/site2/400.http
2501 errorfile 404 /etc/haproxy/errorfiles/site2/404.http
2502 errorfile 408 /dev/null # work around Chrome pre-connect bug
2503
2504
Willy Tarreauc57f0e22009-05-10 13:12:33 +020025054. Proxies
Willy Tarreau6a06a402007-07-15 20:15:28 +02002506----------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002507
Willy Tarreau6a06a402007-07-15 20:15:28 +02002508Proxy configuration can be located in a set of sections :
William Lallemand6e62fb62015-04-28 16:55:23 +02002509 - defaults [<name>]
Willy Tarreau6a06a402007-07-15 20:15:28 +02002510 - frontend <name>
2511 - backend <name>
2512 - listen <name>
2513
2514A "defaults" section sets default parameters for all other sections following
2515its declaration. Those default parameters are reset by the next "defaults"
2516section. See below for the list of parameters which can be set in a "defaults"
Willy Tarreau0ba27502007-12-24 16:55:16 +01002517section. The name is optional but its use is encouraged for better readability.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002518
2519A "frontend" section describes a set of listening sockets accepting client
2520connections.
2521
2522A "backend" section describes a set of servers to which the proxy will connect
2523to forward incoming connections.
2524
2525A "listen" section defines a complete proxy with its frontend and backend
2526parts combined in one section. It is generally useful for TCP-only traffic.
2527
Willy Tarreau0ba27502007-12-24 16:55:16 +01002528All proxy names must be formed from upper and lower case letters, digits,
2529'-' (dash), '_' (underscore) , '.' (dot) and ':' (colon). ACL names are
2530case-sensitive, which means that "www" and "WWW" are two different proxies.
2531
2532Historically, all proxy names could overlap, it just caused troubles in the
2533logs. Since the introduction of content switching, it is mandatory that two
2534proxies with overlapping capabilities (frontend/backend) have different names.
2535However, it is still permitted that a frontend and a backend share the same
2536name, as this configuration seems to be commonly encountered.
2537
2538Right now, two major proxy modes are supported : "tcp", also known as layer 4,
2539and "http", also known as layer 7. In layer 4 mode, HAProxy simply forwards
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002540bidirectional traffic between two sides. In layer 7 mode, HAProxy analyzes the
Willy Tarreau0ba27502007-12-24 16:55:16 +01002541protocol, and can interact with it by allowing, blocking, switching, adding,
2542modifying, or removing arbitrary contents in requests or responses, based on
2543arbitrary criteria.
2544
Willy Tarreau70dffda2014-01-30 03:07:23 +01002545In HTTP mode, the processing applied to requests and responses flowing over
2546a connection depends in the combination of the frontend's HTTP options and
Julien Pivotto21ad3152019-12-10 13:11:17 +01002547the backend's. HAProxy supports 3 connection modes :
Willy Tarreau70dffda2014-01-30 03:07:23 +01002548
2549 - KAL : keep alive ("option http-keep-alive") which is the default mode : all
2550 requests and responses are processed, and connections remain open but idle
2551 between responses and new requests.
2552
Willy Tarreau70dffda2014-01-30 03:07:23 +01002553 - SCL: server close ("option http-server-close") : the server-facing
2554 connection is closed after the end of the response is received, but the
2555 client-facing connection remains open.
2556
Christopher Faulet315b39c2018-09-21 16:26:19 +02002557 - CLO: close ("option httpclose"): the connection is closed after the end of
2558 the response and "Connection: close" appended in both directions.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002559
2560The effective mode that will be applied to a connection passing through a
2561frontend and a backend can be determined by both proxy modes according to the
2562following matrix, but in short, the modes are symmetric, keep-alive is the
Christopher Faulet315b39c2018-09-21 16:26:19 +02002563weakest option and close is the strongest.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002564
Christopher Faulet315b39c2018-09-21 16:26:19 +02002565 Backend mode
Willy Tarreau70dffda2014-01-30 03:07:23 +01002566
Christopher Faulet315b39c2018-09-21 16:26:19 +02002567 | KAL | SCL | CLO
2568 ----+-----+-----+----
2569 KAL | KAL | SCL | CLO
2570 ----+-----+-----+----
Christopher Faulet315b39c2018-09-21 16:26:19 +02002571 mode SCL | SCL | SCL | CLO
2572 ----+-----+-----+----
2573 CLO | CLO | CLO | CLO
Willy Tarreau70dffda2014-01-30 03:07:23 +01002574
Willy Tarreau0ba27502007-12-24 16:55:16 +01002575
Willy Tarreau70dffda2014-01-30 03:07:23 +01002576
Willy Tarreauc57f0e22009-05-10 13:12:33 +020025774.1. Proxy keywords matrix
2578--------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002579
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002580The following list of keywords is supported. Most of them may only be used in a
2581limited set of section types. Some of them are marked as "deprecated" because
2582they are inherited from an old syntax which may be confusing or functionally
2583limited, and there are new recommended keywords to replace them. Keywords
Davor Ocelice9ed2812017-12-25 17:49:28 +01002584marked with "(*)" can be optionally inverted using the "no" prefix, e.g. "no
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002585option contstats". This makes sense when the option has been enabled by default
Willy Tarreau3842f002009-06-14 11:39:52 +02002586and must be disabled for a specific instance. Such options may also be prefixed
2587with "default" in order to restore default settings regardless of what has been
2588specified in a previous "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002589
Willy Tarreau6a06a402007-07-15 20:15:28 +02002590
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002591 keyword defaults frontend listen backend
2592------------------------------------+----------+----------+---------+---------
2593acl - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002594backlog X X X -
2595balance X - X X
2596bind - X X -
2597bind-process X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002598capture cookie - X X -
2599capture request header - X X -
2600capture response header - X X -
William Lallemand82fe75c2012-10-23 10:25:10 +02002601compression X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002602cookie X - X X
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02002603declare capture - X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002604default-server X - X X
2605default_backend X X X -
2606description - X X X
2607disabled X X X X
2608dispatch - - X X
Simon Horman51a1cf62015-02-03 13:00:44 +09002609email-alert from X X X X
Simon Horman64e34162015-02-06 11:11:57 +09002610email-alert level X X X X
Simon Horman51a1cf62015-02-03 13:00:44 +09002611email-alert mailers X X X X
2612email-alert myhostname X X X X
2613email-alert to X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002614enabled X X X X
2615errorfile X X X X
Christopher Faulet76edc0f2020-01-13 15:52:01 +01002616errorfiles X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002617errorloc X X X X
2618errorloc302 X X X X
2619-- keyword -------------------------- defaults - frontend - listen -- backend -
2620errorloc303 X X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01002621force-persist - - X X
Christopher Fauletc3fe5332016-04-07 15:30:10 +02002622filter - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002623fullconn X - X X
2624grace X X X X
2625hash-type X - X X
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01002626http-after-response - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002627http-check disable-on-404 X - X X
Willy Tarreaubd741542010-03-16 18:46:54 +01002628http-check expect - - X X
Willy Tarreau7ab6aff2010-10-12 06:30:16 +02002629http-check send-state X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002630http-request - X X X
Willy Tarreaue365c0b2013-06-11 16:06:12 +02002631http-response - X X X
Willy Tarreau30631952015-08-06 15:05:24 +02002632http-reuse X - X X
Baptiste Assmann2c42ef52013-10-09 21:57:02 +02002633http-send-name-header - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002634id - X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01002635ignore-persist - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02002636load-server-state-from-file X - X X
William Lallemand0f99e342011-10-12 17:50:54 +02002637log (*) X X X X
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01002638log-format X X X -
Dragan Dosen7ad31542015-09-28 17:16:47 +02002639log-format-sd X X X -
Willy Tarreau094af4e2015-01-07 15:03:42 +01002640log-tag X X X X
Willy Tarreauc35362a2014-04-25 13:58:37 +02002641max-keep-alive-queue X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002642maxconn X X X -
2643mode X X X X
2644monitor fail - X X -
2645monitor-net X X X -
2646monitor-uri X X X -
2647option abortonclose (*) X - X X
2648option accept-invalid-http-request (*) X X X -
2649option accept-invalid-http-response (*) X - X X
2650option allbackups (*) X - X X
2651option checkcache (*) X - X X
2652option clitcpka (*) X X X -
2653option contstats (*) X X X -
2654option dontlog-normal (*) X X X -
2655option dontlognull (*) X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002656-- keyword -------------------------- defaults - frontend - listen -- backend -
2657option forwardfor X X X X
Christopher Faulet98fbe952019-07-22 16:18:24 +02002658option h1-case-adjust-bogus-client (*) X X X -
2659option h1-case-adjust-bogus-server (*) X - X X
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02002660option http-buffer-request (*) X X X X
Willy Tarreau82649f92015-05-01 22:40:51 +02002661option http-ignore-probes (*) X X X -
Willy Tarreau16bfb022010-01-16 19:48:41 +01002662option http-keep-alive (*) X X X X
Willy Tarreau96e31212011-05-30 18:10:30 +02002663option http-no-delay (*) X X X X
Christopher Faulet98db9762018-09-21 10:25:19 +02002664option http-pretend-keepalive (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002665option http-server-close (*) X X X X
2666option http-use-proxy-header (*) X X X -
2667option httpchk X - X X
2668option httpclose (*) X X X X
Freddy Spierenburge88b7732019-03-25 14:35:17 +01002669option httplog X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002670option http_proxy (*) X X X X
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002671option independent-streams (*) X X X X
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02002672option ldap-check X - X X
Simon Horman98637e52014-06-20 12:30:16 +09002673option external-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002674option log-health-checks (*) X - X X
2675option log-separate-errors (*) X X X -
2676option logasap (*) X X X -
2677option mysql-check X - X X
2678option nolinger (*) X X X X
2679option originalto X X X X
2680option persist (*) X - X X
Baptiste Assmann809e22a2015-10-12 20:22:55 +02002681option pgsql-check X - X X
2682option prefer-last-server (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002683option redispatch (*) X - X X
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02002684option redis-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002685option smtpchk X - X X
2686option socket-stats (*) X X X -
2687option splice-auto (*) X X X X
2688option splice-request (*) X X X X
2689option splice-response (*) X X X X
Christopher Fauletba7bc162016-11-07 21:07:38 +01002690option spop-check - - - X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002691option srvtcpka (*) X - X X
2692option ssl-hello-chk X - X X
2693-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreaued179852013-12-16 01:07:00 +01002694option tcp-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002695option tcp-smart-accept (*) X X X -
2696option tcp-smart-connect (*) X - X X
2697option tcpka X X X X
2698option tcplog X X X X
2699option transparent (*) X - X X
Simon Horman98637e52014-06-20 12:30:16 +09002700external-check command X - X X
2701external-check path X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002702persist rdp-cookie X - X X
2703rate-limit sessions X X X -
2704redirect - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002705-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002706retries X - X X
Olivier Houcharda254a372019-04-05 15:30:12 +02002707retry-on X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002708server - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02002709server-state-file-name X - X X
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02002710server-template - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002711source X - X X
Baptiste Assmann5a549212015-10-12 20:30:24 +02002712stats admin - X X X
2713stats auth X X X X
2714stats enable X X X X
2715stats hide-version X X X X
2716stats http-request - X X X
2717stats realm X X X X
2718stats refresh X X X X
2719stats scope X X X X
2720stats show-desc X X X X
2721stats show-legends X X X X
2722stats show-node X X X X
2723stats uri X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002724-- keyword -------------------------- defaults - frontend - listen -- backend -
2725stick match - - X X
2726stick on - - X X
2727stick store-request - - X X
Willy Tarreaud8dc99f2011-07-01 11:33:25 +02002728stick store-response - - X X
Adam Spiers68af3c12017-04-06 16:31:39 +01002729stick-table - X X X
Willy Tarreau938c7fe2014-04-25 14:21:39 +02002730tcp-check connect - - X X
2731tcp-check expect - - X X
2732tcp-check send - - X X
2733tcp-check send-binary - - X X
Willy Tarreaue9656522010-08-17 15:40:09 +02002734tcp-request connection - X X -
2735tcp-request content - X X X
Willy Tarreaua56235c2010-09-14 11:31:36 +02002736tcp-request inspect-delay - X X X
Willy Tarreau4f614292016-10-21 17:49:36 +02002737tcp-request session - X X -
Emeric Brun0a3b67f2010-09-24 15:34:53 +02002738tcp-response content - - X X
2739tcp-response inspect-delay - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002740timeout check X - X X
2741timeout client X X X -
Willy Tarreau05cdd962014-05-10 14:30:07 +02002742timeout client-fin X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002743timeout connect X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002744timeout http-keep-alive X X X X
2745timeout http-request X X X X
2746timeout queue X - X X
2747timeout server X - X X
Willy Tarreau05cdd962014-05-10 14:30:07 +02002748timeout server-fin X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002749timeout tarpit X X X X
Willy Tarreauce887fd2012-05-12 12:50:00 +02002750timeout tunnel X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002751transparent (deprecated) X - X X
William Lallemanda73203e2012-03-12 12:48:57 +01002752unique-id-format X X X -
2753unique-id-header X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002754use_backend - X X -
Christopher Fauletb30b3102019-09-12 23:03:09 +02002755use-fcgi-app - - X X
Willy Tarreau4a5cade2012-04-05 21:09:48 +02002756use-server - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002757------------------------------------+----------+----------+---------+---------
2758 keyword defaults frontend listen backend
Willy Tarreau6a06a402007-07-15 20:15:28 +02002759
Willy Tarreau0ba27502007-12-24 16:55:16 +01002760
Willy Tarreauc57f0e22009-05-10 13:12:33 +020027614.2. Alphabetically sorted keywords reference
2762---------------------------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002763
2764This section provides a description of each keyword and its usage.
2765
2766
2767acl <aclname> <criterion> [flags] [operator] <value> ...
2768 Declare or complete an access list.
2769 May be used in sections : defaults | frontend | listen | backend
2770 no | yes | yes | yes
2771 Example:
2772 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
2773 acl invalid_src src_port 0:1023
2774 acl local_dst hdr(host) -i localhost
2775
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002776 See section 7 about ACL usage.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002777
2778
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01002779backlog <conns>
2780 Give hints to the system about the approximate listen backlog desired size
2781 May be used in sections : defaults | frontend | listen | backend
2782 yes | yes | yes | no
2783 Arguments :
2784 <conns> is the number of pending connections. Depending on the operating
2785 system, it may represent the number of already acknowledged
Cyril Bontédc4d9032012-04-08 21:57:39 +02002786 connections, of non-acknowledged ones, or both.
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01002787
2788 In order to protect against SYN flood attacks, one solution is to increase
2789 the system's SYN backlog size. Depending on the system, sometimes it is just
2790 tunable via a system parameter, sometimes it is not adjustable at all, and
2791 sometimes the system relies on hints given by the application at the time of
2792 the listen() syscall. By default, HAProxy passes the frontend's maxconn value
2793 to the listen() syscall. On systems which can make use of this value, it can
2794 sometimes be useful to be able to specify a different value, hence this
2795 backlog parameter.
2796
2797 On Linux 2.4, the parameter is ignored by the system. On Linux 2.6, it is
2798 used as a hint and the system accepts up to the smallest greater power of
2799 two, and never more than some limits (usually 32768).
2800
2801 See also : "maxconn" and the target operating system's tuning guide.
2802
2803
Willy Tarreau0ba27502007-12-24 16:55:16 +01002804balance <algorithm> [ <arguments> ]
Willy Tarreau226071e2014-04-10 11:55:45 +02002805balance url_param <param> [check_post]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002806 Define the load balancing algorithm to be used in a backend.
2807 May be used in sections : defaults | frontend | listen | backend
2808 yes | no | yes | yes
2809 Arguments :
2810 <algorithm> is the algorithm used to select a server when doing load
2811 balancing. This only applies when no persistence information
2812 is available, or when a connection is redispatched to another
2813 server. <algorithm> may be one of the following :
2814
2815 roundrobin Each server is used in turns, according to their weights.
2816 This is the smoothest and fairest algorithm when the server's
2817 processing time remains equally distributed. This algorithm
2818 is dynamic, which means that server weights may be adjusted
Willy Tarreau9757a382009-10-03 12:56:50 +02002819 on the fly for slow starts for instance. It is limited by
Godbacha34bdc02013-07-22 07:44:53 +08002820 design to 4095 active servers per backend. Note that in some
Willy Tarreau9757a382009-10-03 12:56:50 +02002821 large farms, when a server becomes up after having been down
2822 for a very short time, it may sometimes take a few hundreds
2823 requests for it to be re-integrated into the farm and start
2824 receiving traffic. This is normal, though very rare. It is
2825 indicated here in case you would have the chance to observe
2826 it, so that you don't worry.
2827
2828 static-rr Each server is used in turns, according to their weights.
2829 This algorithm is as similar to roundrobin except that it is
2830 static, which means that changing a server's weight on the
2831 fly will have no effect. On the other hand, it has no design
2832 limitation on the number of servers, and when a server goes
2833 up, it is always immediately reintroduced into the farm, once
2834 the full map is recomputed. It also uses slightly less CPU to
2835 run (around -1%).
Willy Tarreau0ba27502007-12-24 16:55:16 +01002836
Willy Tarreau2d2a7f82008-03-17 12:07:56 +01002837 leastconn The server with the lowest number of connections receives the
2838 connection. Round-robin is performed within groups of servers
2839 of the same load to ensure that all servers will be used. Use
2840 of this algorithm is recommended where very long sessions are
2841 expected, such as LDAP, SQL, TSE, etc... but is not very well
2842 suited for protocols using short sessions such as HTTP. This
2843 algorithm is dynamic, which means that server weights may be
2844 adjusted on the fly for slow starts for instance.
2845
Willy Tarreauf09c6602012-02-13 17:12:08 +01002846 first The first server with available connection slots receives the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002847 connection. The servers are chosen from the lowest numeric
Willy Tarreauf09c6602012-02-13 17:12:08 +01002848 identifier to the highest (see server parameter "id"), which
2849 defaults to the server's position in the farm. Once a server
Willy Tarreau64559c52012-04-07 09:08:45 +02002850 reaches its maxconn value, the next server is used. It does
Willy Tarreauf09c6602012-02-13 17:12:08 +01002851 not make sense to use this algorithm without setting maxconn.
2852 The purpose of this algorithm is to always use the smallest
2853 number of servers so that extra servers can be powered off
2854 during non-intensive hours. This algorithm ignores the server
2855 weight, and brings more benefit to long session such as RDP
Willy Tarreau64559c52012-04-07 09:08:45 +02002856 or IMAP than HTTP, though it can be useful there too. In
2857 order to use this algorithm efficiently, it is recommended
2858 that a cloud controller regularly checks server usage to turn
2859 them off when unused, and regularly checks backend queue to
2860 turn new servers on when the queue inflates. Alternatively,
2861 using "http-check send-state" may inform servers on the load.
Willy Tarreauf09c6602012-02-13 17:12:08 +01002862
Willy Tarreau0ba27502007-12-24 16:55:16 +01002863 source The source IP address is hashed and divided by the total
2864 weight of the running servers to designate which server will
2865 receive the request. This ensures that the same client IP
2866 address will always reach the same server as long as no
2867 server goes down or up. If the hash result changes due to the
2868 number of running servers changing, many clients will be
2869 directed to a different server. This algorithm is generally
2870 used in TCP mode where no cookie may be inserted. It may also
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002871 be used on the Internet to provide a best-effort stickiness
Willy Tarreau0ba27502007-12-24 16:55:16 +01002872 to clients which refuse session cookies. This algorithm is
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002873 static by default, which means that changing a server's
2874 weight on the fly will have no effect, but this can be
2875 changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002876
Oskar Stolc8dc41842012-05-19 10:19:54 +01002877 uri This algorithm hashes either the left part of the URI (before
2878 the question mark) or the whole URI (if the "whole" parameter
2879 is present) and divides the hash value by the total weight of
2880 the running servers. The result designates which server will
2881 receive the request. This ensures that the same URI will
2882 always be directed to the same server as long as no server
2883 goes up or down. This is used with proxy caches and
2884 anti-virus proxies in order to maximize the cache hit rate.
2885 Note that this algorithm may only be used in an HTTP backend.
2886 This algorithm is static by default, which means that
2887 changing a server's weight on the fly will have no effect,
2888 but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002889
Oskar Stolc8dc41842012-05-19 10:19:54 +01002890 This algorithm supports two optional parameters "len" and
Marek Majkowski9c30fc12008-04-27 23:25:55 +02002891 "depth", both followed by a positive integer number. These
2892 options may be helpful when it is needed to balance servers
2893 based on the beginning of the URI only. The "len" parameter
2894 indicates that the algorithm should only consider that many
2895 characters at the beginning of the URI to compute the hash.
2896 Note that having "len" set to 1 rarely makes sense since most
2897 URIs start with a leading "/".
2898
2899 The "depth" parameter indicates the maximum directory depth
2900 to be used to compute the hash. One level is counted for each
2901 slash in the request. If both parameters are specified, the
2902 evaluation stops when either is reached.
2903
Willy Tarreau0ba27502007-12-24 16:55:16 +01002904 url_param The URL parameter specified in argument will be looked up in
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002905 the query string of each HTTP GET request.
2906
2907 If the modifier "check_post" is used, then an HTTP POST
Cyril Bontédc4d9032012-04-08 21:57:39 +02002908 request entity will be searched for the parameter argument,
2909 when it is not found in a query string after a question mark
Willy Tarreau226071e2014-04-10 11:55:45 +02002910 ('?') in the URL. The message body will only start to be
2911 analyzed once either the advertised amount of data has been
2912 received or the request buffer is full. In the unlikely event
2913 that chunked encoding is used, only the first chunk is
Cyril Bontédc4d9032012-04-08 21:57:39 +02002914 scanned. Parameter values separated by a chunk boundary, may
Willy Tarreau226071e2014-04-10 11:55:45 +02002915 be randomly balanced if at all. This keyword used to support
2916 an optional <max_wait> parameter which is now ignored.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002917
2918 If the parameter is found followed by an equal sign ('=') and
2919 a value, then the value is hashed and divided by the total
2920 weight of the running servers. The result designates which
2921 server will receive the request.
2922
2923 This is used to track user identifiers in requests and ensure
2924 that a same user ID will always be sent to the same server as
2925 long as no server goes up or down. If no value is found or if
2926 the parameter is not found, then a round robin algorithm is
2927 applied. Note that this algorithm may only be used in an HTTP
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002928 backend. This algorithm is static by default, which means
2929 that changing a server's weight on the fly will have no
2930 effect, but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002931
Cyril Bontédc4d9032012-04-08 21:57:39 +02002932 hdr(<name>) The HTTP header <name> will be looked up in each HTTP
2933 request. Just as with the equivalent ACL 'hdr()' function,
2934 the header name in parenthesis is not case sensitive. If the
2935 header is absent or if it does not contain any value, the
2936 roundrobin algorithm is applied instead.
Benoitaffb4812009-03-25 13:02:10 +01002937
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002938 An optional 'use_domain_only' parameter is available, for
Benoitaffb4812009-03-25 13:02:10 +01002939 reducing the hash algorithm to the main domain part with some
2940 specific headers such as 'Host'. For instance, in the Host
2941 value "haproxy.1wt.eu", only "1wt" will be considered.
2942
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002943 This algorithm is static by default, which means that
2944 changing a server's weight on the fly will have no effect,
2945 but this can be changed using "hash-type".
2946
Willy Tarreau21c741a2019-01-14 18:14:27 +01002947 random
2948 random(<draws>)
2949 A random number will be used as the key for the consistent
Willy Tarreau760e81d2018-05-03 07:20:40 +02002950 hashing function. This means that the servers' weights are
2951 respected, dynamic weight changes immediately take effect, as
2952 well as new server additions. Random load balancing can be
2953 useful with large farms or when servers are frequently added
Willy Tarreau21c741a2019-01-14 18:14:27 +01002954 or removed as it may avoid the hammering effect that could
2955 result from roundrobin or leastconn in this situation. The
2956 hash-balance-factor directive can be used to further improve
2957 fairness of the load balancing, especially in situations
2958 where servers show highly variable response times. When an
2959 argument <draws> is present, it must be an integer value one
2960 or greater, indicating the number of draws before selecting
2961 the least loaded of these servers. It was indeed demonstrated
2962 that picking the least loaded of two servers is enough to
2963 significantly improve the fairness of the algorithm, by
2964 always avoiding to pick the most loaded server within a farm
2965 and getting rid of any bias that could be induced by the
2966 unfair distribution of the consistent list. Higher values N
2967 will take away N-1 of the highest loaded servers at the
2968 expense of performance. With very high values, the algorithm
2969 will converge towards the leastconn's result but much slower.
2970 The default value is 2, which generally shows very good
2971 distribution and performance. This algorithm is also known as
2972 the Power of Two Random Choices and is described here :
2973 http://www.eecs.harvard.edu/~michaelm/postscripts/handbook2001.pdf
Willy Tarreau760e81d2018-05-03 07:20:40 +02002974
Emeric Brun736aa232009-06-30 17:56:00 +02002975 rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02002976 rdp-cookie(<name>)
Emeric Brun736aa232009-06-30 17:56:00 +02002977 The RDP cookie <name> (or "mstshash" if omitted) will be
2978 looked up and hashed for each incoming TCP request. Just as
2979 with the equivalent ACL 'req_rdp_cookie()' function, the name
2980 is not case-sensitive. This mechanism is useful as a degraded
2981 persistence mode, as it makes it possible to always send the
2982 same user (or the same session ID) to the same server. If the
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002983 cookie is not found, the normal roundrobin algorithm is
Emeric Brun736aa232009-06-30 17:56:00 +02002984 used instead.
2985
2986 Note that for this to work, the frontend must ensure that an
2987 RDP cookie is already present in the request buffer. For this
2988 you must use 'tcp-request content accept' rule combined with
2989 a 'req_rdp_cookie_cnt' ACL.
2990
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002991 This algorithm is static by default, which means that
2992 changing a server's weight on the fly will have no effect,
2993 but this can be changed using "hash-type".
2994
Cyril Bontédc4d9032012-04-08 21:57:39 +02002995 See also the rdp_cookie pattern fetch function.
Simon Hormanab814e02011-06-24 14:50:20 +09002996
Willy Tarreau0ba27502007-12-24 16:55:16 +01002997 <arguments> is an optional list of arguments which may be needed by some
Marek Majkowski9c30fc12008-04-27 23:25:55 +02002998 algorithms. Right now, only "url_param" and "uri" support an
2999 optional argument.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003000
Willy Tarreau3cd9af22009-03-15 14:06:41 +01003001 The load balancing algorithm of a backend is set to roundrobin when no other
3002 algorithm, mode nor option have been set. The algorithm may only be set once
3003 for each backend.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003004
Lukas Tribus80512b12018-10-27 20:07:40 +02003005 With authentication schemes that require the same connection like NTLM, URI
John Roeslerfb2fce12019-07-10 15:45:51 -05003006 based algorithms must not be used, as they would cause subsequent requests
Lukas Tribus80512b12018-10-27 20:07:40 +02003007 to be routed to different backend servers, breaking the invalid assumptions
3008 NTLM relies on.
3009
Willy Tarreau0ba27502007-12-24 16:55:16 +01003010 Examples :
3011 balance roundrobin
3012 balance url_param userid
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003013 balance url_param session_id check_post 64
Benoitaffb4812009-03-25 13:02:10 +01003014 balance hdr(User-Agent)
3015 balance hdr(host)
3016 balance hdr(Host) use_domain_only
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003017
3018 Note: the following caveats and limitations on using the "check_post"
3019 extension with "url_param" must be considered :
3020
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003021 - all POST requests are eligible for consideration, because there is no way
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003022 to determine if the parameters will be found in the body or entity which
3023 may contain binary data. Therefore another method may be required to
3024 restrict consideration of POST requests that have no URL parameters in
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02003025 the body. (see acl http_end)
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003026
3027 - using a <max_wait> value larger than the request buffer size does not
3028 make sense and is useless. The buffer size is set at build time, and
3029 defaults to 16 kB.
3030
3031 - Content-Encoding is not supported, the parameter search will probably
3032 fail; and load balancing will fall back to Round Robin.
3033
3034 - Expect: 100-continue is not supported, load balancing will fall back to
3035 Round Robin.
3036
Lukas Tribus23953682017-04-28 13:24:30 +00003037 - Transfer-Encoding (RFC7230 3.3.1) is only supported in the first chunk.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003038 If the entire parameter value is not present in the first chunk, the
3039 selection of server is undefined (actually, defined by how little
3040 actually appeared in the first chunk).
3041
3042 - This feature does not support generation of a 100, 411 or 501 response.
3043
3044 - In some cases, requesting "check_post" MAY attempt to scan the entire
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003045 contents of a message body. Scanning normally terminates when linear
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003046 white space or control characters are found, indicating the end of what
3047 might be a URL parameter list. This is probably not a concern with SGML
3048 type message bodies.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003049
Willy Tarreau294d0f02015-08-10 19:40:12 +02003050 See also : "dispatch", "cookie", "transparent", "hash-type" and "http_proxy".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003051
3052
Willy Tarreaub6205fd2012-09-24 12:27:33 +02003053bind [<address>]:<port_range> [, ...] [param*]
3054bind /<path> [, ...] [param*]
Willy Tarreau0ba27502007-12-24 16:55:16 +01003055 Define one or several listening addresses and/or ports in a frontend.
3056 May be used in sections : defaults | frontend | listen | backend
3057 no | yes | yes | no
3058 Arguments :
Willy Tarreaub1e52e82008-01-13 14:49:51 +01003059 <address> is optional and can be a host name, an IPv4 address, an IPv6
3060 address, or '*'. It designates the address the frontend will
3061 listen on. If unset, all IPv4 addresses of the system will be
3062 listened on. The same will apply for '*' or the system's
David du Colombier9c938da2011-03-17 10:40:27 +01003063 special address "0.0.0.0". The IPv6 equivalent is '::'.
Willy Tarreau24709282013-03-10 21:32:12 +01003064 Optionally, an address family prefix may be used before the
3065 address to force the family regardless of the address format,
3066 which can be useful to specify a path to a unix socket with
3067 no slash ('/'). Currently supported prefixes are :
3068 - 'ipv4@' -> address is always IPv4
3069 - 'ipv6@' -> address is always IPv6
3070 - 'unix@' -> address is a path to a local unix socket
Willy Tarreau70f72e02014-07-08 00:37:50 +02003071 - 'abns@' -> address is in abstract namespace (Linux only).
3072 Note: since abstract sockets are not "rebindable", they
3073 do not cope well with multi-process mode during
3074 soft-restart, so it is better to avoid them if
3075 nbproc is greater than 1. The effect is that if the
3076 new process fails to start, only one of the old ones
3077 will be able to rebind to the socket.
Willy Tarreau40aa0702013-03-10 23:51:38 +01003078 - 'fd@<n>' -> use file descriptor <n> inherited from the
3079 parent. The fd must be bound and may or may not already
3080 be listening.
William Lallemand2fe7dd02018-09-11 16:51:29 +02003081 - 'sockpair@<n>'-> like fd@ but you must use the fd of a
3082 connected unix socket or of a socketpair. The bind waits
3083 to receive a FD over the unix socket and uses it as if it
3084 was the FD of an accept(). Should be used carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02003085 You may want to reference some environment variables in the
3086 address parameter, see section 2.3 about environment
3087 variables.
Willy Tarreaub1e52e82008-01-13 14:49:51 +01003088
Willy Tarreauc5011ca2010-03-22 11:53:56 +01003089 <port_range> is either a unique TCP port, or a port range for which the
3090 proxy will accept connections for the IP address specified
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003091 above. The port is mandatory for TCP listeners. Note that in
3092 the case of an IPv6 address, the port is always the number
3093 after the last colon (':'). A range can either be :
Willy Tarreauc5011ca2010-03-22 11:53:56 +01003094 - a numerical port (ex: '80')
3095 - a dash-delimited ports range explicitly stating the lower
3096 and upper bounds (ex: '2000-2100') which are included in
3097 the range.
3098
3099 Particular care must be taken against port ranges, because
3100 every <address:port> couple consumes one socket (= a file
3101 descriptor), so it's easy to consume lots of descriptors
3102 with a simple range, and to run out of sockets. Also, each
3103 <address:port> couple must be used only once among all
3104 instances running on a same system. Please note that binding
3105 to ports lower than 1024 generally require particular
Jamie Gloudon801a0a32012-08-25 00:18:33 -04003106 privileges to start the program, which are independent of
Willy Tarreauc5011ca2010-03-22 11:53:56 +01003107 the 'uid' parameter.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003108
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003109 <path> is a UNIX socket path beginning with a slash ('/'). This is
Davor Ocelice9ed2812017-12-25 17:49:28 +01003110 alternative to the TCP listening port. HAProxy will then
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003111 receive UNIX connections on the socket located at this place.
3112 The path must begin with a slash and by default is absolute.
3113 It can be relative to the prefix defined by "unix-bind" in
3114 the global section. Note that the total length of the prefix
3115 followed by the socket path cannot exceed some system limits
3116 for UNIX sockets, which commonly are set to 107 characters.
3117
Willy Tarreaub6205fd2012-09-24 12:27:33 +02003118 <param*> is a list of parameters common to all sockets declared on the
3119 same line. These numerous parameters depend on OS and build
3120 options and have a complete section dedicated to them. Please
3121 refer to section 5 to for more details.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02003122
Willy Tarreau0ba27502007-12-24 16:55:16 +01003123 It is possible to specify a list of address:port combinations delimited by
3124 commas. The frontend will then listen on all of these addresses. There is no
3125 fixed limit to the number of addresses and ports which can be listened on in
3126 a frontend, as well as there is no limit to the number of "bind" statements
3127 in a frontend.
3128
3129 Example :
3130 listen http_proxy
3131 bind :80,:443
3132 bind 10.0.0.1:10080,10.0.0.1:10443
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003133 bind /var/run/ssl-frontend.sock user root mode 600 accept-proxy
Willy Tarreau0ba27502007-12-24 16:55:16 +01003134
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02003135 listen http_https_proxy
3136 bind :80
Cyril Bonté0d44fc62012-10-09 22:45:33 +02003137 bind :443 ssl crt /etc/haproxy/site.pem
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02003138
Willy Tarreau24709282013-03-10 21:32:12 +01003139 listen http_https_proxy_explicit
3140 bind ipv6@:80
3141 bind ipv4@public_ssl:443 ssl crt /etc/haproxy/site.pem
3142 bind unix@ssl-frontend.sock user root mode 600 accept-proxy
3143
Willy Tarreaudad36a32013-03-11 01:20:04 +01003144 listen external_bind_app1
William Lallemandb2f07452015-05-12 14:27:13 +02003145 bind "fd@${FD_APP1}"
Willy Tarreaudad36a32013-03-11 01:20:04 +01003146
Willy Tarreau55dcaf62015-09-27 15:03:15 +02003147 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
3148 sun_path length is used for the address length. Some other programs
3149 such as socat use the string length only by default. Pass the option
3150 ",unix-tightsocklen=0" to any abstract socket definition in socat to
3151 make it compatible with HAProxy's.
3152
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003153 See also : "source", "option forwardfor", "unix-bind" and the PROXY protocol
Willy Tarreaub6205fd2012-09-24 12:27:33 +02003154 documentation, and section 5 about bind options.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003155
3156
Christopher Fauletff4121f2017-11-22 16:38:49 +01003157bind-process [ all | odd | even | <process_num>[-[<process_num>]] ] ...
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003158 Limit visibility of an instance to a certain set of processes numbers.
3159 May be used in sections : defaults | frontend | listen | backend
3160 yes | yes | yes | yes
3161 Arguments :
3162 all All process will see this instance. This is the default. It
3163 may be used to override a default value.
3164
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003165 odd This instance will be enabled on processes 1,3,5,...63. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003166 option may be combined with other numbers.
3167
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003168 even This instance will be enabled on processes 2,4,6,...64. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003169 option may be combined with other numbers. Do not use it
3170 with less than 2 processes otherwise some instances might be
3171 missing from all processes.
3172
Christopher Fauletff4121f2017-11-22 16:38:49 +01003173 process_num The instance will be enabled on this process number or range,
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003174 whose values must all be between 1 and 32 or 64 depending on
Christopher Fauletff4121f2017-11-22 16:38:49 +01003175 the machine's word size. Ranges can be partially defined. The
3176 higher bound can be omitted. In such case, it is replaced by
3177 the corresponding maximum value. If a proxy is bound to
3178 process numbers greater than the configured global.nbproc, it
3179 will either be forced to process #1 if a single process was
Willy Tarreau102df612014-05-07 23:56:38 +02003180 specified, or to all processes otherwise.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003181
3182 This keyword limits binding of certain instances to certain processes. This
3183 is useful in order not to have too many processes listening to the same
3184 ports. For instance, on a dual-core machine, it might make sense to set
3185 'nbproc 2' in the global section, then distributes the listeners among 'odd'
3186 and 'even' instances.
3187
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003188 At the moment, it is not possible to reference more than 32 or 64 processes
3189 using this keyword, but this should be more than enough for most setups.
3190 Please note that 'all' really means all processes regardless of the machine's
3191 word size, and is not limited to the first 32 or 64.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003192
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02003193 Each "bind" line may further be limited to a subset of the proxy's processes,
3194 please consult the "process" bind keyword in section 5.1.
3195
Willy Tarreaub369a042014-09-16 13:21:03 +02003196 When a frontend has no explicit "bind-process" line, it tries to bind to all
3197 the processes referenced by its "bind" lines. That means that frontends can
3198 easily adapt to their listeners' processes.
3199
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003200 If some backends are referenced by frontends bound to other processes, the
3201 backend automatically inherits the frontend's processes.
3202
3203 Example :
3204 listen app_ip1
3205 bind 10.0.0.1:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02003206 bind-process odd
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003207
3208 listen app_ip2
3209 bind 10.0.0.2:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02003210 bind-process even
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003211
3212 listen management
3213 bind 10.0.0.3:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02003214 bind-process 1 2 3 4
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003215
Willy Tarreau110ecc12012-11-15 17:50:01 +01003216 listen management
3217 bind 10.0.0.4:80
3218 bind-process 1-4
3219
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02003220 See also : "nbproc" in global section, and "process" in section 5.1.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003221
3222
Willy Tarreau0ba27502007-12-24 16:55:16 +01003223capture cookie <name> len <length>
3224 Capture and log a cookie in the request and in the response.
3225 May be used in sections : defaults | frontend | listen | backend
3226 no | yes | yes | no
3227 Arguments :
3228 <name> is the beginning of the name of the cookie to capture. In order
3229 to match the exact name, simply suffix the name with an equal
3230 sign ('='). The full name will appear in the logs, which is
3231 useful with application servers which adjust both the cookie name
Davor Ocelice9ed2812017-12-25 17:49:28 +01003232 and value (e.g. ASPSESSIONXXX).
Willy Tarreau0ba27502007-12-24 16:55:16 +01003233
3234 <length> is the maximum number of characters to report in the logs, which
3235 include the cookie name, the equal sign and the value, all in the
3236 standard "name=value" form. The string will be truncated on the
3237 right if it exceeds <length>.
3238
3239 Only the first cookie is captured. Both the "cookie" request headers and the
3240 "set-cookie" response headers are monitored. This is particularly useful to
3241 check for application bugs causing session crossing or stealing between
3242 users, because generally the user's cookies can only change on a login page.
3243
3244 When the cookie was not presented by the client, the associated log column
3245 will report "-". When a request does not cause a cookie to be assigned by the
3246 server, a "-" is reported in the response column.
3247
3248 The capture is performed in the frontend only because it is necessary that
3249 the log format does not change for a given frontend depending on the
3250 backends. This may change in the future. Note that there can be only one
Willy Tarreau193b8c62012-11-22 00:17:38 +01003251 "capture cookie" statement in a frontend. The maximum capture length is set
3252 by the global "tune.http.cookielen" setting and defaults to 63 characters. It
3253 is not possible to specify a capture in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003254
3255 Example:
3256 capture cookie ASPSESSION len 32
3257
3258 See also : "capture request header", "capture response header" as well as
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003259 section 8 about logging.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003260
3261
3262capture request header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01003263 Capture and log the last occurrence of the specified request header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003264 May be used in sections : defaults | frontend | listen | backend
3265 no | yes | yes | no
3266 Arguments :
3267 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003268 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01003269 appear in the requests, with the first letter of each word in
3270 upper case. The header name will not appear in the logs, only the
3271 value is reported, but the position in the logs is respected.
3272
3273 <length> is the maximum number of characters to extract from the value and
3274 report in the logs. The string will be truncated on the right if
3275 it exceeds <length>.
3276
Willy Tarreau4460d032012-11-21 23:37:37 +01003277 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01003278 value will be added to the logs between braces ('{}'). If multiple headers
3279 are captured, they will be delimited by a vertical bar ('|') and will appear
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003280 in the same order they were declared in the configuration. Non-existent
3281 headers will be logged just as an empty string. Common uses for request
3282 header captures include the "Host" field in virtual hosting environments, the
3283 "Content-length" when uploads are supported, "User-agent" to quickly
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003284 differentiate between real users and robots, and "X-Forwarded-For" in proxied
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003285 environments to find where the request came from.
3286
3287 Note that when capturing headers such as "User-agent", some spaces may be
3288 logged, making the log analysis more difficult. Thus be careful about what
3289 you log if you know your log parser is not smart enough to rely on the
3290 braces.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003291
Willy Tarreau0900abb2012-11-22 00:21:46 +01003292 There is no limit to the number of captured request headers nor to their
3293 length, though it is wise to keep them low to limit memory usage per session.
3294 In order to keep log format consistent for a same frontend, header captures
3295 can only be declared in a frontend. It is not possible to specify a capture
3296 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003297
3298 Example:
3299 capture request header Host len 15
3300 capture request header X-Forwarded-For len 15
Cyril Bontéd1b0f7c2015-10-26 22:37:39 +01003301 capture request header Referer len 15
Willy Tarreau0ba27502007-12-24 16:55:16 +01003302
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003303 See also : "capture cookie", "capture response header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01003304 about logging.
3305
3306
3307capture response header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01003308 Capture and log the last occurrence of the specified response header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003309 May be used in sections : defaults | frontend | listen | backend
3310 no | yes | yes | no
3311 Arguments :
3312 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003313 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01003314 appear in the response, with the first letter of each word in
3315 upper case. The header name will not appear in the logs, only the
3316 value is reported, but the position in the logs is respected.
3317
3318 <length> is the maximum number of characters to extract from the value and
3319 report in the logs. The string will be truncated on the right if
3320 it exceeds <length>.
3321
Willy Tarreau4460d032012-11-21 23:37:37 +01003322 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01003323 result will be added to the logs between braces ('{}') after the captured
3324 request headers. If multiple headers are captured, they will be delimited by
3325 a vertical bar ('|') and will appear in the same order they were declared in
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003326 the configuration. Non-existent headers will be logged just as an empty
3327 string. Common uses for response header captures include the "Content-length"
3328 header which indicates how many bytes are expected to be returned, the
3329 "Location" header to track redirections.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003330
Willy Tarreau0900abb2012-11-22 00:21:46 +01003331 There is no limit to the number of captured response headers nor to their
3332 length, though it is wise to keep them low to limit memory usage per session.
3333 In order to keep log format consistent for a same frontend, header captures
3334 can only be declared in a frontend. It is not possible to specify a capture
3335 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003336
3337 Example:
3338 capture response header Content-length len 9
3339 capture response header Location len 15
3340
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003341 See also : "capture cookie", "capture request header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01003342 about logging.
3343
3344
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003345compression algo <algorithm> ...
3346compression type <mime type> ...
Willy Tarreau70737d12012-10-27 00:34:28 +02003347compression offload
William Lallemand82fe75c2012-10-23 10:25:10 +02003348 Enable HTTP compression.
3349 May be used in sections : defaults | frontend | listen | backend
3350 yes | yes | yes | yes
3351 Arguments :
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003352 algo is followed by the list of supported compression algorithms.
3353 type is followed by the list of MIME types that will be compressed.
3354 offload makes haproxy work as a compression offloader only (see notes).
3355
3356 The currently supported algorithms are :
Willy Tarreauc91840a2015-03-28 17:00:39 +01003357 identity this is mostly for debugging, and it was useful for developing
3358 the compression feature. Identity does not apply any change on
3359 data.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003360
Willy Tarreauc91840a2015-03-28 17:00:39 +01003361 gzip applies gzip compression. This setting is only available when
Baptiste Assmannf085d632015-12-21 17:57:32 +01003362 support for zlib or libslz was built in.
Willy Tarreauc91840a2015-03-28 17:00:39 +01003363
3364 deflate same as "gzip", but with deflate algorithm and zlib format.
3365 Note that this algorithm has ambiguous support on many
3366 browsers and no support at all from recent ones. It is
3367 strongly recommended not to use it for anything else than
3368 experimentation. This setting is only available when support
Baptiste Assmannf085d632015-12-21 17:57:32 +01003369 for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003370
Willy Tarreauc91840a2015-03-28 17:00:39 +01003371 raw-deflate same as "deflate" without the zlib wrapper, and used as an
3372 alternative when the browser wants "deflate". All major
3373 browsers understand it and despite violating the standards,
3374 it is known to work better than "deflate", at least on MSIE
3375 and some versions of Safari. Do not use it in conjunction
3376 with "deflate", use either one or the other since both react
3377 to the same Accept-Encoding token. This setting is only
Baptiste Assmannf085d632015-12-21 17:57:32 +01003378 available when support for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003379
Dmitry Sivachenko87c208b2012-11-22 20:03:26 +04003380 Compression will be activated depending on the Accept-Encoding request
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003381 header. With identity, it does not take care of that header.
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04003382 If backend servers support HTTP compression, these directives
3383 will be no-op: haproxy will see the compressed response and will not
3384 compress again. If backend servers do not support HTTP compression and
3385 there is Accept-Encoding header in request, haproxy will compress the
3386 matching response.
Willy Tarreau70737d12012-10-27 00:34:28 +02003387
3388 The "offload" setting makes haproxy remove the Accept-Encoding header to
3389 prevent backend servers from compressing responses. It is strongly
3390 recommended not to do this because this means that all the compression work
3391 will be done on the single point where haproxy is located. However in some
3392 deployment scenarios, haproxy may be installed in front of a buggy gateway
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04003393 with broken HTTP compression implementation which can't be turned off.
3394 In that case haproxy can be used to prevent that gateway from emitting
3395 invalid payloads. In this case, simply removing the header in the
3396 configuration does not work because it applies before the header is parsed,
3397 so that prevents haproxy from compressing. The "offload" setting should
Willy Tarreauffea9fd2014-07-12 16:37:02 +02003398 then be used for such scenarios. Note: for now, the "offload" setting is
3399 ignored when set in a defaults section.
William Lallemand82fe75c2012-10-23 10:25:10 +02003400
William Lallemand05097442012-11-20 12:14:28 +01003401 Compression is disabled when:
Baptiste Assmann650d53d2013-01-05 15:44:44 +01003402 * the request does not advertise a supported compression algorithm in the
3403 "Accept-Encoding" header
3404 * the response message is not HTTP/1.1
Tim Duesterhusbb48c9a2019-01-30 23:46:04 +01003405 * HTTP status code is not one of 200, 201, 202, or 203
Baptiste Assmann650d53d2013-01-05 15:44:44 +01003406 * response contain neither a "Content-Length" header nor a
3407 "Transfer-Encoding" whose last value is "chunked"
3408 * response contains a "Content-Type" header whose first value starts with
3409 "multipart"
3410 * the response contains the "no-transform" value in the "Cache-control"
3411 header
3412 * User-Agent matches "Mozilla/4" unless it is MSIE 6 with XP SP2, or MSIE 7
3413 and later
3414 * The response contains a "Content-Encoding" header, indicating that the
3415 response is already compressed (see compression offload)
Tim Duesterhusbb48c9a2019-01-30 23:46:04 +01003416 * The response contains an invalid "ETag" header or multiple ETag headers
William Lallemand05097442012-11-20 12:14:28 +01003417
Tim Duesterhusb229f012019-01-29 16:38:56 +01003418 Note: The compression does not emit the Warning header.
William Lallemand05097442012-11-20 12:14:28 +01003419
William Lallemand82fe75c2012-10-23 10:25:10 +02003420 Examples :
3421 compression algo gzip
3422 compression type text/html text/plain
Willy Tarreau0ba27502007-12-24 16:55:16 +01003423
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003424
Willy Tarreau55165fe2009-05-10 12:02:55 +02003425cookie <name> [ rewrite | insert | prefix ] [ indirect ] [ nocache ]
Willy Tarreau4992dd22012-05-31 21:02:17 +02003426 [ postonly ] [ preserve ] [ httponly ] [ secure ]
3427 [ domain <domain> ]* [ maxidle <idle> ] [ maxlife <life> ]
Christopher Faulet2f533902020-01-21 11:06:48 +01003428 [ dynamic ] [ attr <value> ]*
Willy Tarreau0ba27502007-12-24 16:55:16 +01003429 Enable cookie-based persistence in a backend.
3430 May be used in sections : defaults | frontend | listen | backend
3431 yes | no | yes | yes
3432 Arguments :
3433 <name> is the name of the cookie which will be monitored, modified or
3434 inserted in order to bring persistence. This cookie is sent to
3435 the client via a "Set-Cookie" header in the response, and is
3436 brought back by the client in a "Cookie" header in all requests.
3437 Special care should be taken to choose a name which does not
3438 conflict with any likely application cookie. Also, if the same
Davor Ocelice9ed2812017-12-25 17:49:28 +01003439 backends are subject to be used by the same clients (e.g.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003440 HTTP/HTTPS), care should be taken to use different cookie names
3441 between all backends if persistence between them is not desired.
3442
3443 rewrite This keyword indicates that the cookie will be provided by the
3444 server and that haproxy will have to modify its value to set the
3445 server's identifier in it. This mode is handy when the management
3446 of complex combinations of "Set-cookie" and "Cache-control"
3447 headers is left to the application. The application can then
3448 decide whether or not it is appropriate to emit a persistence
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003449 cookie. Since all responses should be monitored, this mode
3450 doesn't work in HTTP tunnel mode. Unless the application
Davor Ocelice9ed2812017-12-25 17:49:28 +01003451 behavior is very complex and/or broken, it is advised not to
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003452 start with this mode for new deployments. This keyword is
3453 incompatible with "insert" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003454
3455 insert This keyword indicates that the persistence cookie will have to
Willy Tarreaua79094d2010-08-31 22:54:15 +02003456 be inserted by haproxy in server responses if the client did not
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003457
Willy Tarreaua79094d2010-08-31 22:54:15 +02003458 already have a cookie that would have permitted it to access this
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003459 server. When used without the "preserve" option, if the server
Michael Prokop4438c602019-05-24 10:25:45 +02003460 emits a cookie with the same name, it will be removed before
Davor Ocelice9ed2812017-12-25 17:49:28 +01003461 processing. For this reason, this mode can be used to upgrade
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003462 existing configurations running in the "rewrite" mode. The cookie
3463 will only be a session cookie and will not be stored on the
3464 client's disk. By default, unless the "indirect" option is added,
3465 the server will see the cookies emitted by the client. Due to
3466 caching effects, it is generally wise to add the "nocache" or
3467 "postonly" keywords (see below). The "insert" keyword is not
3468 compatible with "rewrite" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003469
3470 prefix This keyword indicates that instead of relying on a dedicated
3471 cookie for the persistence, an existing one will be completed.
3472 This may be needed in some specific environments where the client
3473 does not support more than one single cookie and the application
3474 already needs it. In this case, whenever the server sets a cookie
3475 named <name>, it will be prefixed with the server's identifier
3476 and a delimiter. The prefix will be removed from all client
3477 requests so that the server still finds the cookie it emitted.
3478 Since all requests and responses are subject to being modified,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003479 this mode doesn't work with tunnel mode. The "prefix" keyword is
Willy Tarreau37229df2011-10-17 12:24:55 +02003480 not compatible with "rewrite" and "insert". Note: it is highly
3481 recommended not to use "indirect" with "prefix", otherwise server
3482 cookie updates would not be sent to clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003483
Willy Tarreaua79094d2010-08-31 22:54:15 +02003484 indirect When this option is specified, no cookie will be emitted to a
3485 client which already has a valid one for the server which has
3486 processed the request. If the server sets such a cookie itself,
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003487 it will be removed, unless the "preserve" option is also set. In
3488 "insert" mode, this will additionally remove cookies from the
3489 requests transmitted to the server, making the persistence
3490 mechanism totally transparent from an application point of view.
Willy Tarreau37229df2011-10-17 12:24:55 +02003491 Note: it is highly recommended not to use "indirect" with
3492 "prefix", otherwise server cookie updates would not be sent to
3493 clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003494
3495 nocache This option is recommended in conjunction with the insert mode
3496 when there is a cache between the client and HAProxy, as it
3497 ensures that a cacheable response will be tagged non-cacheable if
3498 a cookie needs to be inserted. This is important because if all
3499 persistence cookies are added on a cacheable home page for
3500 instance, then all customers will then fetch the page from an
3501 outer cache and will all share the same persistence cookie,
3502 leading to one server receiving much more traffic than others.
3503 See also the "insert" and "postonly" options.
3504
3505 postonly This option ensures that cookie insertion will only be performed
3506 on responses to POST requests. It is an alternative to the
3507 "nocache" option, because POST responses are not cacheable, so
3508 this ensures that the persistence cookie will never get cached.
3509 Since most sites do not need any sort of persistence before the
3510 first POST which generally is a login request, this is a very
3511 efficient method to optimize caching without risking to find a
3512 persistence cookie in the cache.
3513 See also the "insert" and "nocache" options.
3514
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003515 preserve This option may only be used with "insert" and/or "indirect". It
3516 allows the server to emit the persistence cookie itself. In this
3517 case, if a cookie is found in the response, haproxy will leave it
3518 untouched. This is useful in order to end persistence after a
3519 logout request for instance. For this, the server just has to
Davor Ocelice9ed2812017-12-25 17:49:28 +01003520 emit a cookie with an invalid value (e.g. empty) or with a date in
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003521 the past. By combining this mechanism with the "disable-on-404"
3522 check option, it is possible to perform a completely graceful
3523 shutdown because users will definitely leave the server after
3524 they logout.
3525
Willy Tarreau4992dd22012-05-31 21:02:17 +02003526 httponly This option tells haproxy to add an "HttpOnly" cookie attribute
3527 when a cookie is inserted. This attribute is used so that a
3528 user agent doesn't share the cookie with non-HTTP components.
3529 Please check RFC6265 for more information on this attribute.
3530
3531 secure This option tells haproxy to add a "Secure" cookie attribute when
3532 a cookie is inserted. This attribute is used so that a user agent
3533 never emits this cookie over non-secure channels, which means
3534 that a cookie learned with this flag will be presented only over
3535 SSL/TLS connections. Please check RFC6265 for more information on
3536 this attribute.
3537
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02003538 domain This option allows to specify the domain at which a cookie is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003539 inserted. It requires exactly one parameter: a valid domain
Willy Tarreau68a897b2009-12-03 23:28:34 +01003540 name. If the domain begins with a dot, the browser is allowed to
3541 use it for any host ending with that name. It is also possible to
3542 specify several domain names by invoking this option multiple
3543 times. Some browsers might have small limits on the number of
3544 domains, so be careful when doing that. For the record, sending
3545 10 domains to MSIE 6 or Firefox 2 works as expected.
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02003546
Willy Tarreau996a92c2010-10-13 19:30:47 +02003547 maxidle This option allows inserted cookies to be ignored after some idle
3548 time. It only works with insert-mode cookies. When a cookie is
3549 sent to the client, the date this cookie was emitted is sent too.
3550 Upon further presentations of this cookie, if the date is older
3551 than the delay indicated by the parameter (in seconds), it will
3552 be ignored. Otherwise, it will be refreshed if needed when the
3553 response is sent to the client. This is particularly useful to
3554 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01003555 too long on the same server (e.g. after a farm size change). When
Willy Tarreau996a92c2010-10-13 19:30:47 +02003556 this option is set and a cookie has no date, it is always
3557 accepted, but gets refreshed in the response. This maintains the
3558 ability for admins to access their sites. Cookies that have a
3559 date in the future further than 24 hours are ignored. Doing so
3560 lets admins fix timezone issues without risking kicking users off
3561 the site.
3562
3563 maxlife This option allows inserted cookies to be ignored after some life
3564 time, whether they're in use or not. It only works with insert
3565 mode cookies. When a cookie is first sent to the client, the date
3566 this cookie was emitted is sent too. Upon further presentations
3567 of this cookie, if the date is older than the delay indicated by
3568 the parameter (in seconds), it will be ignored. If the cookie in
3569 the request has no date, it is accepted and a date will be set.
3570 Cookies that have a date in the future further than 24 hours are
3571 ignored. Doing so lets admins fix timezone issues without risking
3572 kicking users off the site. Contrary to maxidle, this value is
3573 not refreshed, only the first visit date counts. Both maxidle and
3574 maxlife may be used at the time. This is particularly useful to
3575 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01003576 too long on the same server (e.g. after a farm size change). This
Willy Tarreau996a92c2010-10-13 19:30:47 +02003577 is stronger than the maxidle method in that it forces a
3578 redispatch after some absolute delay.
3579
Olivier Houchard4e694042017-03-14 20:01:29 +01003580 dynamic Activate dynamic cookies. When used, a session cookie is
3581 dynamically created for each server, based on the IP and port
3582 of the server, and a secret key, specified in the
3583 "dynamic-cookie-key" backend directive.
3584 The cookie will be regenerated each time the IP address change,
3585 and is only generated for IPv4/IPv6.
3586
Christopher Faulet2f533902020-01-21 11:06:48 +01003587 attr This option tells haproxy to add an extra attribute when a
3588 cookie is inserted. The attribute value can contain any
3589 characters except control ones or ";". This option may be
3590 repeated.
3591
Willy Tarreau0ba27502007-12-24 16:55:16 +01003592 There can be only one persistence cookie per HTTP backend, and it can be
3593 declared in a defaults section. The value of the cookie will be the value
3594 indicated after the "cookie" keyword in a "server" statement. If no cookie
3595 is declared for a given server, the cookie is not set.
Willy Tarreau6a06a402007-07-15 20:15:28 +02003596
Willy Tarreau0ba27502007-12-24 16:55:16 +01003597 Examples :
3598 cookie JSESSIONID prefix
3599 cookie SRV insert indirect nocache
3600 cookie SRV insert postonly indirect
Willy Tarreau996a92c2010-10-13 19:30:47 +02003601 cookie SRV insert indirect nocache maxidle 30m maxlife 8h
Willy Tarreau0ba27502007-12-24 16:55:16 +01003602
Willy Tarreau294d0f02015-08-10 19:40:12 +02003603 See also : "balance source", "capture cookie", "server" and "ignore-persist".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003604
Willy Tarreau983e01e2010-01-11 18:42:06 +01003605
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02003606declare capture [ request | response ] len <length>
3607 Declares a capture slot.
3608 May be used in sections : defaults | frontend | listen | backend
3609 no | yes | yes | no
3610 Arguments:
3611 <length> is the length allowed for the capture.
3612
3613 This declaration is only available in the frontend or listen section, but the
3614 reserved slot can be used in the backends. The "request" keyword allocates a
3615 capture slot for use in the request, and "response" allocates a capture slot
3616 for use in the response.
3617
3618 See also: "capture-req", "capture-res" (sample converters),
Baptiste Assmann5ac425c2015-10-21 23:13:46 +02003619 "capture.req.hdr", "capture.res.hdr" (sample fetches),
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02003620 "http-request capture" and "http-response capture".
3621
3622
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003623default-server [param*]
3624 Change default options for a server in a backend
3625 May be used in sections : defaults | frontend | listen | backend
3626 yes | no | yes | yes
3627 Arguments:
Willy Tarreau983e01e2010-01-11 18:42:06 +01003628 <param*> is a list of parameters for this server. The "default-server"
3629 keyword accepts an important number of options and has a complete
3630 section dedicated to it. Please refer to section 5 for more
3631 details.
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003632
Willy Tarreau983e01e2010-01-11 18:42:06 +01003633 Example :
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003634 default-server inter 1000 weight 13
3635
3636 See also: "server" and section 5 about server options
Willy Tarreau0ba27502007-12-24 16:55:16 +01003637
Willy Tarreau983e01e2010-01-11 18:42:06 +01003638
Willy Tarreau0ba27502007-12-24 16:55:16 +01003639default_backend <backend>
3640 Specify the backend to use when no "use_backend" rule has been matched.
3641 May be used in sections : defaults | frontend | listen | backend
3642 yes | yes | yes | no
3643 Arguments :
3644 <backend> is the name of the backend to use.
3645
3646 When doing content-switching between frontend and backends using the
3647 "use_backend" keyword, it is often useful to indicate which backend will be
3648 used when no rule has matched. It generally is the dynamic backend which
3649 will catch all undetermined requests.
3650
Willy Tarreau0ba27502007-12-24 16:55:16 +01003651 Example :
3652
3653 use_backend dynamic if url_dyn
3654 use_backend static if url_css url_img extension_img
3655 default_backend dynamic
3656
Willy Tarreau98d04852015-05-26 12:18:29 +02003657 See also : "use_backend"
Willy Tarreau2769aa02007-12-27 18:26:09 +01003658
Willy Tarreau0ba27502007-12-24 16:55:16 +01003659
Baptiste Assmann27f51342013-10-09 06:51:49 +02003660description <string>
3661 Describe a listen, frontend or backend.
3662 May be used in sections : defaults | frontend | listen | backend
3663 no | yes | yes | yes
3664 Arguments : string
3665
3666 Allows to add a sentence to describe the related object in the HAProxy HTML
3667 stats page. The description will be printed on the right of the object name
3668 it describes.
3669 No need to backslash spaces in the <string> arguments.
3670
3671
Willy Tarreau0ba27502007-12-24 16:55:16 +01003672disabled
3673 Disable a proxy, frontend or backend.
3674 May be used in sections : defaults | frontend | listen | backend
3675 yes | yes | yes | yes
3676 Arguments : none
3677
3678 The "disabled" keyword is used to disable an instance, mainly in order to
3679 liberate a listening port or to temporarily disable a service. The instance
3680 will still be created and its configuration will be checked, but it will be
3681 created in the "stopped" state and will appear as such in the statistics. It
3682 will not receive any traffic nor will it send any health-checks or logs. It
3683 is possible to disable many instances at once by adding the "disabled"
3684 keyword in a "defaults" section.
3685
3686 See also : "enabled"
3687
3688
Willy Tarreau5ce94572010-06-07 14:35:41 +02003689dispatch <address>:<port>
3690 Set a default server address
3691 May be used in sections : defaults | frontend | listen | backend
3692 no | no | yes | yes
Cyril Bonté108cf6e2012-04-21 23:30:29 +02003693 Arguments :
Willy Tarreau5ce94572010-06-07 14:35:41 +02003694
3695 <address> is the IPv4 address of the default server. Alternatively, a
3696 resolvable hostname is supported, but this name will be resolved
3697 during start-up.
3698
3699 <ports> is a mandatory port specification. All connections will be sent
3700 to this port, and it is not permitted to use port offsets as is
3701 possible with normal servers.
3702
Willy Tarreau787aed52011-04-15 06:45:37 +02003703 The "dispatch" keyword designates a default server for use when no other
Willy Tarreau5ce94572010-06-07 14:35:41 +02003704 server can take the connection. In the past it was used to forward non
3705 persistent connections to an auxiliary load balancer. Due to its simple
3706 syntax, it has also been used for simple TCP relays. It is recommended not to
3707 use it for more clarity, and to use the "server" directive instead.
3708
3709 See also : "server"
3710
Olivier Houchard4e694042017-03-14 20:01:29 +01003711
3712dynamic-cookie-key <string>
3713 Set the dynamic cookie secret key for a backend.
3714 May be used in sections : defaults | frontend | listen | backend
3715 yes | no | yes | yes
3716 Arguments : The secret key to be used.
3717
3718 When dynamic cookies are enabled (see the "dynamic" directive for cookie),
Davor Ocelice9ed2812017-12-25 17:49:28 +01003719 a dynamic cookie is created for each server (unless one is explicitly
Olivier Houchard4e694042017-03-14 20:01:29 +01003720 specified on the "server" line), using a hash of the IP address of the
3721 server, the TCP port, and the secret key.
Davor Ocelice9ed2812017-12-25 17:49:28 +01003722 That way, we can ensure session persistence across multiple load-balancers,
Olivier Houchard4e694042017-03-14 20:01:29 +01003723 even if servers are dynamically added or removed.
Willy Tarreau5ce94572010-06-07 14:35:41 +02003724
Willy Tarreau0ba27502007-12-24 16:55:16 +01003725enabled
3726 Enable a proxy, frontend or backend.
3727 May be used in sections : defaults | frontend | listen | backend
3728 yes | yes | yes | yes
3729 Arguments : none
3730
3731 The "enabled" keyword is used to explicitly enable an instance, when the
3732 defaults has been set to "disabled". This is very rarely used.
3733
3734 See also : "disabled"
3735
3736
3737errorfile <code> <file>
3738 Return a file contents instead of errors generated by HAProxy
3739 May be used in sections : defaults | frontend | listen | backend
3740 yes | yes | yes | yes
3741 Arguments :
3742 <code> is the HTTP status code. Currently, HAProxy is capable of
Florian Tham9205fea2020-01-08 13:35:30 +01003743 generating codes 200, 400, 403, 404, 405, 408, 410, 425, 429, 500,
3744 502, 503, and 504.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003745
3746 <file> designates a file containing the full HTTP response. It is
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003747 recommended to follow the common practice of appending ".http" to
Willy Tarreau0ba27502007-12-24 16:55:16 +01003748 the filename so that people do not confuse the response with HTML
Willy Tarreau59140a22009-02-22 12:02:30 +01003749 error pages, and to use absolute paths, since files are read
3750 before any chroot is performed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003751
3752 It is important to understand that this keyword is not meant to rewrite
3753 errors returned by the server, but errors detected and returned by HAProxy.
3754 This is why the list of supported errors is limited to a small set.
3755
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003756 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3757
Willy Tarreau0ba27502007-12-24 16:55:16 +01003758 The files are returned verbatim on the TCP socket. This allows any trick such
3759 as redirections to another URL or site, as well as tricks to clean cookies,
3760 force enable or disable caching, etc... The package provides default error
3761 files returning the same contents as default errors.
3762
Willy Tarreau59140a22009-02-22 12:02:30 +01003763 The files should not exceed the configured buffer size (BUFSIZE), which
3764 generally is 8 or 16 kB, otherwise they will be truncated. It is also wise
Davor Ocelice9ed2812017-12-25 17:49:28 +01003765 not to put any reference to local contents (e.g. images) in order to avoid
Willy Tarreau59140a22009-02-22 12:02:30 +01003766 loops between the client and HAProxy when all servers are down, causing an
3767 error to be returned instead of an image. For better HTTP compliance, it is
3768 recommended that all header lines end with CR-LF and not LF alone.
3769
Willy Tarreau0ba27502007-12-24 16:55:16 +01003770 The files are read at the same time as the configuration and kept in memory.
3771 For this reason, the errors continue to be returned even when the process is
3772 chrooted, and no file change is considered while the process is running. A
Willy Tarreauc27debf2008-01-06 08:57:02 +01003773 simple method for developing those files consists in associating them to the
Willy Tarreau0ba27502007-12-24 16:55:16 +01003774 403 status code and interrogating a blocked URL.
3775
3776 See also : "errorloc", "errorloc302", "errorloc303"
3777
Willy Tarreau59140a22009-02-22 12:02:30 +01003778 Example :
3779 errorfile 400 /etc/haproxy/errorfiles/400badreq.http
Willy Tarreau989222a2016-01-15 10:26:26 +01003780 errorfile 408 /dev/null # work around Chrome pre-connect bug
Willy Tarreau59140a22009-02-22 12:02:30 +01003781 errorfile 403 /etc/haproxy/errorfiles/403forbid.http
3782 errorfile 503 /etc/haproxy/errorfiles/503sorry.http
3783
Willy Tarreau2769aa02007-12-27 18:26:09 +01003784
Christopher Faulet76edc0f2020-01-13 15:52:01 +01003785errorfiles <name> [<code> ...]
3786 Import, fully or partially, the error files defined in the <name> http-errors
3787 section.
3788 May be used in sections : defaults | frontend | listen | backend
3789 yes | yes | yes | yes
3790 Arguments :
3791 <name> is the name of an existing http-errors section.
3792
3793 <code> is a HTTP status code. Several status code may be listed.
3794 Currently, HAProxy is capable of generating codes 200, 400, 403,
3795 404, 405, 408, 410, 425, 429, 500, 502, 503, and 504.
3796
3797 Errors defined in the http-errors section with the name <name> are imported
3798 in the current proxy. If no status code is specified, all error files of the
3799 http-errors section are imported. Otherwise, only error files associated to
3800 the listed status code are imported. Those error files override the already
3801 defined custom errors for the proxy. And they may be overridden by following
3802 ones. Fonctionnly, it is exactly the same than declaring all error files by
3803 hand using "errorfile" directives.
3804
3805 See also : "errorfile", "errorloc", "errorloc302" , "errorloc303" and section
3806 3.8 about http-errors.
3807
3808 Example :
3809 errorfiles generic
3810 errorfiles site-1 403 404
3811
3812
Willy Tarreau2769aa02007-12-27 18:26:09 +01003813errorloc <code> <url>
3814errorloc302 <code> <url>
3815 Return an HTTP redirection to a URL instead of errors generated by HAProxy
3816 May be used in sections : defaults | frontend | listen | backend
3817 yes | yes | yes | yes
3818 Arguments :
3819 <code> is the HTTP status code. Currently, HAProxy is capable of
Florian Tham9205fea2020-01-08 13:35:30 +01003820 generating codes 200, 400, 403, 404, 405, 408, 410, 425, 429, 500,
3821 502, 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003822
3823 <url> it is the exact contents of the "Location" header. It may contain
3824 either a relative URI to an error page hosted on the same site,
3825 or an absolute URI designating an error page on another site.
3826 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01003827 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01003828
3829 It is important to understand that this keyword is not meant to rewrite
3830 errors returned by the server, but errors detected and returned by HAProxy.
3831 This is why the list of supported errors is limited to a small set.
3832
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003833 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3834
Willy Tarreau2769aa02007-12-27 18:26:09 +01003835 Note that both keyword return the HTTP 302 status code, which tells the
3836 client to fetch the designated URL using the same HTTP method. This can be
3837 quite problematic in case of non-GET methods such as POST, because the URL
3838 sent to the client might not be allowed for something other than GET. To
Willy Tarreau989222a2016-01-15 10:26:26 +01003839 work around this problem, please use "errorloc303" which send the HTTP 303
Willy Tarreau2769aa02007-12-27 18:26:09 +01003840 status code, indicating to the client that the URL must be fetched with a GET
3841 request.
3842
3843 See also : "errorfile", "errorloc303"
3844
3845
3846errorloc303 <code> <url>
3847 Return an HTTP redirection to a URL instead of errors generated by HAProxy
3848 May be used in sections : defaults | frontend | listen | backend
3849 yes | yes | yes | yes
3850 Arguments :
3851 <code> is the HTTP status code. Currently, HAProxy is capable of
Florian Tham9205fea2020-01-08 13:35:30 +01003852 generating codes 200, 400, 403, 404, 405, 408, 410, 425, 429, 500,
3853 502, 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003854
3855 <url> it is the exact contents of the "Location" header. It may contain
3856 either a relative URI to an error page hosted on the same site,
3857 or an absolute URI designating an error page on another site.
3858 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01003859 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01003860
3861 It is important to understand that this keyword is not meant to rewrite
3862 errors returned by the server, but errors detected and returned by HAProxy.
3863 This is why the list of supported errors is limited to a small set.
3864
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003865 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3866
Willy Tarreau2769aa02007-12-27 18:26:09 +01003867 Note that both keyword return the HTTP 303 status code, which tells the
3868 client to fetch the designated URL using the same HTTP GET method. This
3869 solves the usual problems associated with "errorloc" and the 302 code. It is
3870 possible that some very old browsers designed before HTTP/1.1 do not support
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003871 it, but no such problem has been reported till now.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003872
3873 See also : "errorfile", "errorloc", "errorloc302"
3874
3875
Simon Horman51a1cf62015-02-03 13:00:44 +09003876email-alert from <emailaddr>
3877 Declare the from email address to be used in both the envelope and header
Davor Ocelice9ed2812017-12-25 17:49:28 +01003878 of email alerts. This is the address that email alerts are sent from.
Simon Horman51a1cf62015-02-03 13:00:44 +09003879 May be used in sections: defaults | frontend | listen | backend
3880 yes | yes | yes | yes
3881
3882 Arguments :
3883
3884 <emailaddr> is the from email address to use when sending email alerts
3885
3886 Also requires "email-alert mailers" and "email-alert to" to be set
3887 and if so sending email alerts is enabled for the proxy.
3888
Simon Horman64e34162015-02-06 11:11:57 +09003889 See also : "email-alert level", "email-alert mailers",
Cyril Bonté307ee1e2015-09-28 23:16:06 +02003890 "email-alert myhostname", "email-alert to", section 3.6 about
3891 mailers.
Simon Horman64e34162015-02-06 11:11:57 +09003892
3893
3894email-alert level <level>
3895 Declare the maximum log level of messages for which email alerts will be
3896 sent. This acts as a filter on the sending of email alerts.
3897 May be used in sections: defaults | frontend | listen | backend
3898 yes | yes | yes | yes
3899
3900 Arguments :
3901
3902 <level> One of the 8 syslog levels:
3903 emerg alert crit err warning notice info debug
3904 The above syslog levels are ordered from lowest to highest.
3905
3906 By default level is alert
3907
3908 Also requires "email-alert from", "email-alert mailers" and
3909 "email-alert to" to be set and if so sending email alerts is enabled
3910 for the proxy.
3911
Simon Horman1421e212015-04-30 13:10:35 +09003912 Alerts are sent when :
3913
3914 * An un-paused server is marked as down and <level> is alert or lower
3915 * A paused server is marked as down and <level> is notice or lower
3916 * A server is marked as up or enters the drain state and <level>
3917 is notice or lower
3918 * "option log-health-checks" is enabled, <level> is info or lower,
3919 and a health check status update occurs
3920
Simon Horman64e34162015-02-06 11:11:57 +09003921 See also : "email-alert from", "email-alert mailers",
3922 "email-alert myhostname", "email-alert to",
Simon Horman51a1cf62015-02-03 13:00:44 +09003923 section 3.6 about mailers.
3924
3925
3926email-alert mailers <mailersect>
3927 Declare the mailers to be used when sending email alerts
3928 May be used in sections: defaults | frontend | listen | backend
3929 yes | yes | yes | yes
3930
3931 Arguments :
3932
3933 <mailersect> is the name of the mailers section to send email alerts.
3934
3935 Also requires "email-alert from" and "email-alert to" to be set
3936 and if so sending email alerts is enabled for the proxy.
3937
Simon Horman64e34162015-02-06 11:11:57 +09003938 See also : "email-alert from", "email-alert level", "email-alert myhostname",
3939 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09003940
3941
3942email-alert myhostname <hostname>
3943 Declare the to hostname address to be used when communicating with
3944 mailers.
3945 May be used in sections: defaults | frontend | listen | backend
3946 yes | yes | yes | yes
3947
3948 Arguments :
3949
Baptiste Assmann738bad92015-12-21 15:27:53 +01003950 <hostname> is the hostname to use when communicating with mailers
Simon Horman51a1cf62015-02-03 13:00:44 +09003951
3952 By default the systems hostname is used.
3953
3954 Also requires "email-alert from", "email-alert mailers" and
3955 "email-alert to" to be set and if so sending email alerts is enabled
3956 for the proxy.
3957
Simon Horman64e34162015-02-06 11:11:57 +09003958 See also : "email-alert from", "email-alert level", "email-alert mailers",
3959 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09003960
3961
3962email-alert to <emailaddr>
Davor Ocelice9ed2812017-12-25 17:49:28 +01003963 Declare both the recipient address in the envelope and to address in the
Simon Horman51a1cf62015-02-03 13:00:44 +09003964 header of email alerts. This is the address that email alerts are sent to.
3965 May be used in sections: defaults | frontend | listen | backend
3966 yes | yes | yes | yes
3967
3968 Arguments :
3969
3970 <emailaddr> is the to email address to use when sending email alerts
3971
3972 Also requires "email-alert mailers" and "email-alert to" to be set
3973 and if so sending email alerts is enabled for the proxy.
3974
Simon Horman64e34162015-02-06 11:11:57 +09003975 See also : "email-alert from", "email-alert level", "email-alert mailers",
Simon Horman51a1cf62015-02-03 13:00:44 +09003976 "email-alert myhostname", section 3.6 about mailers.
3977
3978
Willy Tarreau4de91492010-01-22 19:10:05 +01003979force-persist { if | unless } <condition>
3980 Declare a condition to force persistence on down servers
3981 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01003982 no | no | yes | yes
Willy Tarreau4de91492010-01-22 19:10:05 +01003983
3984 By default, requests are not dispatched to down servers. It is possible to
3985 force this using "option persist", but it is unconditional and redispatches
3986 to a valid server if "option redispatch" is set. That leaves with very little
3987 possibilities to force some requests to reach a server which is artificially
3988 marked down for maintenance operations.
3989
3990 The "force-persist" statement allows one to declare various ACL-based
3991 conditions which, when met, will cause a request to ignore the down status of
3992 a server and still try to connect to it. That makes it possible to start a
3993 server, still replying an error to the health checks, and run a specially
3994 configured browser to test the service. Among the handy methods, one could
3995 use a specific source IP address, or a specific cookie. The cookie also has
3996 the advantage that it can easily be added/removed on the browser from a test
3997 page. Once the service is validated, it is then possible to open the service
3998 to the world by returning a valid response to health checks.
3999
4000 The forced persistence is enabled when an "if" condition is met, or unless an
4001 "unless" condition is met. The final redispatch is always disabled when this
4002 is used.
4003
Cyril Bonté0d4bf012010-04-25 23:21:46 +02004004 See also : "option redispatch", "ignore-persist", "persist",
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +02004005 and section 7 about ACL usage.
Willy Tarreau4de91492010-01-22 19:10:05 +01004006
Christopher Fauletc3fe5332016-04-07 15:30:10 +02004007
4008filter <name> [param*]
4009 Add the filter <name> in the filter list attached to the proxy.
4010 May be used in sections : defaults | frontend | listen | backend
4011 no | yes | yes | yes
4012 Arguments :
4013 <name> is the name of the filter. Officially supported filters are
4014 referenced in section 9.
4015
Tim Düsterhus4896c442016-11-29 02:15:19 +01004016 <param*> is a list of parameters accepted by the filter <name>. The
Christopher Fauletc3fe5332016-04-07 15:30:10 +02004017 parsing of these parameters are the responsibility of the
Tim Düsterhus4896c442016-11-29 02:15:19 +01004018 filter. Please refer to the documentation of the corresponding
4019 filter (section 9) for all details on the supported parameters.
Christopher Fauletc3fe5332016-04-07 15:30:10 +02004020
4021 Multiple occurrences of the filter line can be used for the same proxy. The
4022 same filter can be referenced many times if needed.
4023
4024 Example:
4025 listen
4026 bind *:80
4027
4028 filter trace name BEFORE-HTTP-COMP
4029 filter compression
4030 filter trace name AFTER-HTTP-COMP
4031
4032 compression algo gzip
4033 compression offload
4034
4035 server srv1 192.168.0.1:80
4036
4037 See also : section 9.
4038
Willy Tarreau4de91492010-01-22 19:10:05 +01004039
Willy Tarreau2769aa02007-12-27 18:26:09 +01004040fullconn <conns>
4041 Specify at what backend load the servers will reach their maxconn
4042 May be used in sections : defaults | frontend | listen | backend
4043 yes | no | yes | yes
4044 Arguments :
4045 <conns> is the number of connections on the backend which will make the
4046 servers use the maximal number of connections.
4047
Willy Tarreau198a7442008-01-17 12:05:32 +01004048 When a server has a "maxconn" parameter specified, it means that its number
Willy Tarreau2769aa02007-12-27 18:26:09 +01004049 of concurrent connections will never go higher. Additionally, if it has a
Willy Tarreau198a7442008-01-17 12:05:32 +01004050 "minconn" parameter, it indicates a dynamic limit following the backend's
Willy Tarreau2769aa02007-12-27 18:26:09 +01004051 load. The server will then always accept at least <minconn> connections,
4052 never more than <maxconn>, and the limit will be on the ramp between both
4053 values when the backend has less than <conns> concurrent connections. This
4054 makes it possible to limit the load on the servers during normal loads, but
4055 push it further for important loads without overloading the servers during
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004056 exceptional loads.
Willy Tarreau2769aa02007-12-27 18:26:09 +01004057
Willy Tarreaufbb78422011-06-05 15:38:35 +02004058 Since it's hard to get this value right, haproxy automatically sets it to
4059 10% of the sum of the maxconns of all frontends that may branch to this
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01004060 backend (based on "use_backend" and "default_backend" rules). That way it's
4061 safe to leave it unset. However, "use_backend" involving dynamic names are
4062 not counted since there is no way to know if they could match or not.
Willy Tarreaufbb78422011-06-05 15:38:35 +02004063
Willy Tarreau2769aa02007-12-27 18:26:09 +01004064 Example :
4065 # The servers will accept between 100 and 1000 concurrent connections each
4066 # and the maximum of 1000 will be reached when the backend reaches 10000
4067 # connections.
4068 backend dynamic
4069 fullconn 10000
4070 server srv1 dyn1:80 minconn 100 maxconn 1000
4071 server srv2 dyn2:80 minconn 100 maxconn 1000
4072
4073 See also : "maxconn", "server"
4074
4075
4076grace <time>
4077 Maintain a proxy operational for some time after a soft stop
4078 May be used in sections : defaults | frontend | listen | backend
Cyril Bonté99ed3272010-01-24 23:29:44 +01004079 yes | yes | yes | yes
Willy Tarreau2769aa02007-12-27 18:26:09 +01004080 Arguments :
4081 <time> is the time (by default in milliseconds) for which the instance
4082 will remain operational with the frontend sockets still listening
4083 when a soft-stop is received via the SIGUSR1 signal.
4084
4085 This may be used to ensure that the services disappear in a certain order.
4086 This was designed so that frontends which are dedicated to monitoring by an
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004087 external equipment fail immediately while other ones remain up for the time
Willy Tarreau2769aa02007-12-27 18:26:09 +01004088 needed by the equipment to detect the failure.
4089
4090 Note that currently, there is very little benefit in using this parameter,
4091 and it may in fact complicate the soft-reconfiguration process more than
4092 simplify it.
4093
Willy Tarreau0ba27502007-12-24 16:55:16 +01004094
Andrew Rodland17be45e2016-10-25 17:04:12 -04004095hash-balance-factor <factor>
4096 Specify the balancing factor for bounded-load consistent hashing
4097 May be used in sections : defaults | frontend | listen | backend
4098 yes | no | no | yes
4099 Arguments :
4100 <factor> is the control for the maximum number of concurrent requests to
4101 send to a server, expressed as a percentage of the average number
Frédéric Lécaille93d33162019-03-06 09:35:59 +01004102 of concurrent requests across all of the active servers.
Andrew Rodland17be45e2016-10-25 17:04:12 -04004103
4104 Specifying a "hash-balance-factor" for a server with "hash-type consistent"
4105 enables an algorithm that prevents any one server from getting too many
4106 requests at once, even if some hash buckets receive many more requests than
4107 others. Setting <factor> to 0 (the default) disables the feature. Otherwise,
4108 <factor> is a percentage greater than 100. For example, if <factor> is 150,
4109 then no server will be allowed to have a load more than 1.5 times the average.
4110 If server weights are used, they will be respected.
4111
4112 If the first-choice server is disqualified, the algorithm will choose another
4113 server based on the request hash, until a server with additional capacity is
4114 found. A higher <factor> allows more imbalance between the servers, while a
4115 lower <factor> means that more servers will be checked on average, affecting
4116 performance. Reasonable values are from 125 to 200.
4117
Willy Tarreau760e81d2018-05-03 07:20:40 +02004118 This setting is also used by "balance random" which internally relies on the
4119 consistent hashing mechanism.
4120
Andrew Rodland17be45e2016-10-25 17:04:12 -04004121 See also : "balance" and "hash-type".
4122
4123
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004124hash-type <method> <function> <modifier>
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004125 Specify a method to use for mapping hashes to servers
4126 May be used in sections : defaults | frontend | listen | backend
4127 yes | no | yes | yes
4128 Arguments :
Bhaskar98634f02013-10-29 23:30:51 -04004129 <method> is the method used to select a server from the hash computed by
4130 the <function> :
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004131
Bhaskar98634f02013-10-29 23:30:51 -04004132 map-based the hash table is a static array containing all alive servers.
4133 The hashes will be very smooth, will consider weights, but
4134 will be static in that weight changes while a server is up
4135 will be ignored. This means that there will be no slow start.
4136 Also, since a server is selected by its position in the array,
4137 most mappings are changed when the server count changes. This
4138 means that when a server goes up or down, or when a server is
4139 added to a farm, most connections will be redistributed to
4140 different servers. This can be inconvenient with caches for
4141 instance.
Willy Tarreau798a39c2010-11-24 15:04:29 +01004142
Bhaskar98634f02013-10-29 23:30:51 -04004143 consistent the hash table is a tree filled with many occurrences of each
4144 server. The hash key is looked up in the tree and the closest
4145 server is chosen. This hash is dynamic, it supports changing
4146 weights while the servers are up, so it is compatible with the
4147 slow start feature. It has the advantage that when a server
4148 goes up or down, only its associations are moved. When a
4149 server is added to the farm, only a few part of the mappings
4150 are redistributed, making it an ideal method for caches.
4151 However, due to its principle, the distribution will never be
4152 very smooth and it may sometimes be necessary to adjust a
4153 server's weight or its ID to get a more balanced distribution.
4154 In order to get the same distribution on multiple load
4155 balancers, it is important that all servers have the exact
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004156 same IDs. Note: consistent hash uses sdbm and avalanche if no
4157 hash function is specified.
Bhaskar98634f02013-10-29 23:30:51 -04004158
4159 <function> is the hash function to be used :
4160
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03004161 sdbm this function was created initially for sdbm (a public-domain
Bhaskar98634f02013-10-29 23:30:51 -04004162 reimplementation of ndbm) database library. It was found to do
4163 well in scrambling bits, causing better distribution of the keys
4164 and fewer splits. It also happens to be a good general hashing
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004165 function with good distribution, unless the total server weight
4166 is a multiple of 64, in which case applying the avalanche
4167 modifier may help.
Bhaskar98634f02013-10-29 23:30:51 -04004168
4169 djb2 this function was first proposed by Dan Bernstein many years ago
4170 on comp.lang.c. Studies have shown that for certain workload this
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004171 function provides a better distribution than sdbm. It generally
4172 works well with text-based inputs though it can perform extremely
4173 poorly with numeric-only input or when the total server weight is
4174 a multiple of 33, unless the avalanche modifier is also used.
4175
Willy Tarreaua0f42712013-11-14 14:30:35 +01004176 wt6 this function was designed for haproxy while testing other
4177 functions in the past. It is not as smooth as the other ones, but
4178 is much less sensible to the input data set or to the number of
4179 servers. It can make sense as an alternative to sdbm+avalanche or
4180 djb2+avalanche for consistent hashing or when hashing on numeric
4181 data such as a source IP address or a visitor identifier in a URL
4182 parameter.
4183
Willy Tarreau324f07f2015-01-20 19:44:50 +01004184 crc32 this is the most common CRC32 implementation as used in Ethernet,
4185 gzip, PNG, etc. It is slower than the other ones but may provide
4186 a better distribution or less predictable results especially when
4187 used on strings.
4188
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004189 <modifier> indicates an optional method applied after hashing the key :
4190
4191 avalanche This directive indicates that the result from the hash
4192 function above should not be used in its raw form but that
4193 a 4-byte full avalanche hash must be applied first. The
4194 purpose of this step is to mix the resulting bits from the
4195 previous hash in order to avoid any undesired effect when
4196 the input contains some limited values or when the number of
4197 servers is a multiple of one of the hash's components (64
4198 for SDBM, 33 for DJB2). Enabling avalanche tends to make the
4199 result less predictable, but it's also not as smooth as when
4200 using the original function. Some testing might be needed
4201 with some workloads. This hash is one of the many proposed
4202 by Bob Jenkins.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004203
Bhaskar98634f02013-10-29 23:30:51 -04004204 The default hash type is "map-based" and is recommended for most usages. The
4205 default function is "sdbm", the selection of a function should be based on
4206 the range of the values being hashed.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004207
Andrew Rodland17be45e2016-10-25 17:04:12 -04004208 See also : "balance", "hash-balance-factor", "server"
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004209
4210
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01004211http-after-response <action> <options...> [ { if | unless } <condition> ]
4212 Access control for all Layer 7 responses (server, applet/service and internal
4213 ones).
4214
4215 May be used in sections: defaults | frontend | listen | backend
4216 no | yes | yes | yes
4217
4218 The http-after-response statement defines a set of rules which apply to layer
4219 7 processing. The rules are evaluated in their declaration order when they
4220 are met in a frontend, listen or backend section. Any rule may optionally be
4221 followed by an ACL-based condition, in which case it will only be evaluated
4222 if the condition is true. Since these rules apply on responses, the backend
4223 rules are applied first, followed by the frontend's rules.
4224
4225 Unlike http-response rules, these ones are applied on all responses, the
4226 server ones but also to all responses generated by HAProxy. These rules are
4227 evaluated at the end of the responses analysis, before the data forwarding.
4228
4229 The first keyword is the rule's action. The supported actions are described
4230 below.
4231
4232 There is no limit to the number of http-after-response statements per
4233 instance.
4234
4235 Example:
4236 http-after-response set-header Strict-Transport-Security "max-age=31536000"
4237 http-after-response set-header Cache-Control "no-store,no-cache,private"
4238 http-after-response set-header Pragma "no-cache"
4239
4240http-after-response add-header <name> <fmt> [ { if | unless } <condition> ]
4241
4242 This appends an HTTP header field whose name is specified in <name> and whose
4243 value is defined by <fmt> which follows the log-format rules (see Custom Log
4244 Format in section 8.2.4). This may be used to send a cookie to a client for
4245 example, or to pass some internal information.
4246 This rule is not final, so it is possible to add other similar rules.
4247 Note that header addition is performed immediately, so one rule might reuse
4248 the resulting header from a previous rule.
4249
4250http-after-response allow [ { if | unless } <condition> ]
4251
4252 This stops the evaluation of the rules and lets the response pass the check.
4253 No further "http-after-response" rules are evaluated.
4254
4255http-after-response del-header <name> [ { if | unless } <condition> ]
4256
4257 This removes all HTTP header fields whose name is specified in <name>.
4258
4259http-after-response replace-header <name> <regex-match> <replace-fmt>
4260 [ { if | unless } <condition> ]
4261
4262 This works like "http-response replace-header".
4263
4264 Example:
4265 http-after-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2
4266
4267 # applied to:
4268 Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT
4269
4270 # outputs:
4271 Set-Cookie: C=1;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT
4272
4273 # assuming the backend IP is 192.168.1.20.
4274
4275http-after-response replace-value <name> <regex-match> <replace-fmt>
4276 [ { if | unless } <condition> ]
4277
4278 This works like "http-response replace-value".
4279
4280 Example:
4281 http-after-response replace-value Cache-control ^public$ private
4282
4283 # applied to:
4284 Cache-Control: max-age=3600, public
4285
4286 # outputs:
4287 Cache-Control: max-age=3600, private
4288
4289http-after-response set-header <name> <fmt> [ { if | unless } <condition> ]
4290
4291 This does the same as "add-header" except that the header name is first
4292 removed if it existed. This is useful when passing security information to
4293 the server, where the header must not be manipulated by external users.
4294
4295http-after-response set-status <status> [reason <str>]
4296 [ { if | unless } <condition> ]
4297
4298 This replaces the response status code with <status> which must be an integer
4299 between 100 and 999. Optionally, a custom reason text can be provided defined
4300 by <str>, or the default reason for the specified code will be used as a
4301 fallback.
4302
4303 Example:
4304 # return "431 Request Header Fields Too Large"
4305 http-response set-status 431
4306 # return "503 Slow Down", custom reason
4307 http-response set-status 503 reason "Slow Down"
4308
4309http-after-response set-var(<var-name>) <expr> [ { if | unless } <condition> ]
4310
4311 This is used to set the contents of a variable. The variable is declared
4312 inline.
4313
4314 Arguments:
4315 <var-name> The name of the variable starts with an indication about its
4316 scope. The scopes allowed are:
4317 "proc" : the variable is shared with the whole process
4318 "sess" : the variable is shared with the whole session
4319 "txn" : the variable is shared with the transaction
4320 (request and response)
4321 "req" : the variable is shared only during request
4322 processing
4323 "res" : the variable is shared only during response
4324 processing
4325 This prefix is followed by a name. The separator is a '.'.
4326 The name may only contain characters 'a-z', 'A-Z', '0-9', '.'
4327 and '_'.
4328
4329 <expr> Is a standard HAProxy expression formed by a sample-fetch
4330 followed by some converters.
4331
4332 Example:
4333 http-after-response set-var(sess.last_redir) res.hdr(location)
4334
4335http-after-response strict-mode { on | off }
4336
4337 This enables or disables the strict rewriting mode for following rules. It
4338 does not affect rules declared before it and it is only applicable on rules
4339 performing a rewrite on the responses. When the strict mode is enabled, any
4340 rewrite failure triggers an internal error. Otherwise, such errors are
4341 silently ignored. The purpose of the strict rewriting mode is to make some
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05004342 rewrites optional while others must be performed to continue the response
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01004343 processing.
4344
4345 By default, the strict rewriting mode is enabled. Its value is also reset
4346 when a ruleset evaluation ends. So, for instance, if you change the mode on
4347 the bacnkend, the default mode is restored when HAProxy starts the frontend
4348 rules evaluation.
4349
4350http-after-response unset-var(<var-name>) [ { if | unless } <condition> ]
4351
4352 This is used to unset a variable. See "http-after-response set-var" for
4353 details about <var-name>.
4354
4355 Example:
4356 http-after-response unset-var(sess.last_redir)
4357
4358
Willy Tarreau0ba27502007-12-24 16:55:16 +01004359http-check disable-on-404
4360 Enable a maintenance mode upon HTTP/404 response to health-checks
4361 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau2769aa02007-12-27 18:26:09 +01004362 yes | no | yes | yes
Willy Tarreau0ba27502007-12-24 16:55:16 +01004363 Arguments : none
4364
4365 When this option is set, a server which returns an HTTP code 404 will be
4366 excluded from further load-balancing, but will still receive persistent
4367 connections. This provides a very convenient method for Web administrators
4368 to perform a graceful shutdown of their servers. It is also important to note
4369 that a server which is detected as failed while it was in this mode will not
4370 generate an alert, just a notice. If the server responds 2xx or 3xx again, it
4371 will immediately be reinserted into the farm. The status on the stats page
4372 reports "NOLB" for a server in this mode. It is important to note that this
Willy Tarreaubd741542010-03-16 18:46:54 +01004373 option only works in conjunction with the "httpchk" option. If this option
4374 is used with "http-check expect", then it has precedence over it so that 404
4375 responses will still be considered as soft-stop.
4376
4377 See also : "option httpchk", "http-check expect"
4378
4379
4380http-check expect [!] <match> <pattern>
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004381 Make HTTP health checks consider response contents or specific status codes
Willy Tarreaubd741542010-03-16 18:46:54 +01004382 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau1ee51a62011-08-19 20:04:17 +02004383 yes | no | yes | yes
Willy Tarreaubd741542010-03-16 18:46:54 +01004384 Arguments :
4385 <match> is a keyword indicating how to look for a specific pattern in the
4386 response. The keyword may be one of "status", "rstatus",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004387 "string", or "rstring". The keyword may be preceded by an
Willy Tarreaubd741542010-03-16 18:46:54 +01004388 exclamation mark ("!") to negate the match. Spaces are allowed
4389 between the exclamation mark and the keyword. See below for more
4390 details on the supported keywords.
4391
4392 <pattern> is the pattern to look for. It may be a string or a regular
4393 expression. If the pattern contains spaces, they must be escaped
4394 with the usual backslash ('\').
4395
4396 By default, "option httpchk" considers that response statuses 2xx and 3xx
4397 are valid, and that others are invalid. When "http-check expect" is used,
4398 it defines what is considered valid or invalid. Only one "http-check"
4399 statement is supported in a backend. If a server fails to respond or times
4400 out, the check obviously fails. The available matches are :
4401
4402 status <string> : test the exact string match for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004403 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01004404 response's status code is exactly this string. If the
4405 "status" keyword is prefixed with "!", then the response
4406 will be considered invalid if the status code matches.
4407
4408 rstatus <regex> : test a regular expression for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004409 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01004410 response's status code matches the expression. If the
4411 "rstatus" keyword is prefixed with "!", then the response
4412 will be considered invalid if the status code matches.
4413 This is mostly used to check for multiple codes.
4414
4415 string <string> : test the exact string match in the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004416 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01004417 response's body contains this exact string. If the
4418 "string" keyword is prefixed with "!", then the response
4419 will be considered invalid if the body contains this
4420 string. This can be used to look for a mandatory word at
4421 the end of a dynamic page, or to detect a failure when a
Davor Ocelice9ed2812017-12-25 17:49:28 +01004422 specific error appears on the check page (e.g. a stack
Willy Tarreaubd741542010-03-16 18:46:54 +01004423 trace).
4424
4425 rstring <regex> : test a regular expression on the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004426 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01004427 response's body matches this expression. If the "rstring"
4428 keyword is prefixed with "!", then the response will be
4429 considered invalid if the body matches the expression.
4430 This can be used to look for a mandatory word at the end
4431 of a dynamic page, or to detect a failure when a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01004432 error appears on the check page (e.g. a stack trace).
Willy Tarreaubd741542010-03-16 18:46:54 +01004433
4434 It is important to note that the responses will be limited to a certain size
4435 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
4436 Thus, too large responses may not contain the mandatory pattern when using
4437 "string" or "rstring". If a large response is absolutely required, it is
4438 possible to change the default max size by setting the global variable.
4439 However, it is worth keeping in mind that parsing very large responses can
4440 waste some CPU cycles, especially when regular expressions are used, and that
4441 it is always better to focus the checks on smaller resources.
4442
Cyril Bonté32602d22015-01-30 00:07:07 +01004443 Also "http-check expect" doesn't support HTTP keep-alive. Keep in mind that it
4444 will automatically append a "Connection: close" header, meaning that this
4445 header should not be present in the request provided by "option httpchk".
4446
Willy Tarreaubd741542010-03-16 18:46:54 +01004447 Last, if "http-check expect" is combined with "http-check disable-on-404",
4448 then this last one has precedence when the server responds with 404.
4449
4450 Examples :
4451 # only accept status 200 as valid
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01004452 http-check expect status 200
Willy Tarreaubd741542010-03-16 18:46:54 +01004453
4454 # consider SQL errors as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01004455 http-check expect ! string SQL\ Error
Willy Tarreaubd741542010-03-16 18:46:54 +01004456
4457 # consider status 5xx only as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01004458 http-check expect ! rstatus ^5
Willy Tarreaubd741542010-03-16 18:46:54 +01004459
4460 # check that we have a correct hexadecimal tag before /html
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03004461 http-check expect rstring <!--tag:[0-9a-f]*--></html>
Willy Tarreau0ba27502007-12-24 16:55:16 +01004462
Willy Tarreaubd741542010-03-16 18:46:54 +01004463 See also : "option httpchk", "http-check disable-on-404"
Willy Tarreau2769aa02007-12-27 18:26:09 +01004464
4465
Willy Tarreauef781042010-01-27 11:53:01 +01004466http-check send-state
4467 Enable emission of a state header with HTTP health checks
4468 May be used in sections : defaults | frontend | listen | backend
4469 yes | no | yes | yes
4470 Arguments : none
4471
4472 When this option is set, haproxy will systematically send a special header
4473 "X-Haproxy-Server-State" with a list of parameters indicating to each server
4474 how they are seen by haproxy. This can be used for instance when a server is
4475 manipulated without access to haproxy and the operator needs to know whether
4476 haproxy still sees it up or not, or if the server is the last one in a farm.
4477
4478 The header is composed of fields delimited by semi-colons, the first of which
4479 is a word ("UP", "DOWN", "NOLB"), possibly followed by a number of valid
4480 checks on the total number before transition, just as appears in the stats
4481 interface. Next headers are in the form "<variable>=<value>", indicating in
4482 no specific order some values available in the stats interface :
Joseph Lynch514061c2015-01-15 17:52:59 -08004483 - a variable "address", containing the address of the backend server.
4484 This corresponds to the <address> field in the server declaration. For
4485 unix domain sockets, it will read "unix".
4486
4487 - a variable "port", containing the port of the backend server. This
4488 corresponds to the <port> field in the server declaration. For unix
4489 domain sockets, it will read "unix".
4490
Willy Tarreauef781042010-01-27 11:53:01 +01004491 - a variable "name", containing the name of the backend followed by a slash
4492 ("/") then the name of the server. This can be used when a server is
4493 checked in multiple backends.
4494
4495 - a variable "node" containing the name of the haproxy node, as set in the
4496 global "node" variable, otherwise the system's hostname if unspecified.
4497
4498 - a variable "weight" indicating the weight of the server, a slash ("/")
4499 and the total weight of the farm (just counting usable servers). This
4500 helps to know if other servers are available to handle the load when this
4501 one fails.
4502
4503 - a variable "scur" indicating the current number of concurrent connections
4504 on the server, followed by a slash ("/") then the total number of
4505 connections on all servers of the same backend.
4506
4507 - a variable "qcur" indicating the current number of requests in the
4508 server's queue.
4509
4510 Example of a header received by the application server :
4511 >>> X-Haproxy-Server-State: UP 2/3; name=bck/srv2; node=lb1; weight=1/2; \
4512 scur=13/22; qcur=0
4513
4514 See also : "option httpchk", "http-check disable-on-404"
4515
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004516
4517http-request <action> [options...] [ { if | unless } <condition> ]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004518 Access control for Layer 7 requests
4519
4520 May be used in sections: defaults | frontend | listen | backend
4521 no | yes | yes | yes
4522
Willy Tarreau20b0de52012-12-24 15:45:22 +01004523 The http-request statement defines a set of rules which apply to layer 7
4524 processing. The rules are evaluated in their declaration order when they are
4525 met in a frontend, listen or backend section. Any rule may optionally be
4526 followed by an ACL-based condition, in which case it will only be evaluated
4527 if the condition is true.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004528
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004529 The first keyword is the rule's action. The supported actions are described
4530 below.
Willy Tarreau20b0de52012-12-24 15:45:22 +01004531
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004532 There is no limit to the number of http-request statements per instance.
Willy Tarreau20b0de52012-12-24 15:45:22 +01004533
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004534 Example:
4535 acl nagios src 192.168.129.3
4536 acl local_net src 192.168.0.0/16
4537 acl auth_ok http_auth(L1)
Willy Tarreau20b0de52012-12-24 15:45:22 +01004538
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004539 http-request allow if nagios
4540 http-request allow if local_net auth_ok
4541 http-request auth realm Gimme if local_net auth_ok
4542 http-request deny
Willy Tarreau81499eb2012-12-27 12:19:02 +01004543
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004544 Example:
4545 acl key req.hdr(X-Add-Acl-Key) -m found
4546 acl add path /addacl
4547 acl del path /delacl
Willy Tarreau20b0de52012-12-24 15:45:22 +01004548
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004549 acl myhost hdr(Host) -f myhost.lst
Willy Tarreau20b0de52012-12-24 15:45:22 +01004550
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004551 http-request add-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key add
4552 http-request del-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key del
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02004553
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004554 Example:
4555 acl value req.hdr(X-Value) -m found
4556 acl setmap path /setmap
4557 acl delmap path /delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06004558
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004559 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06004560
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004561 http-request set-map(map.lst) %[src] %[req.hdr(X-Value)] if setmap value
4562 http-request del-map(map.lst) %[src] if delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06004563
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004564 See also : "stats http-request", section 3.4 about userlists and section 7
4565 about ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06004566
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004567http-request add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004568
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004569 This is used to add a new entry into an ACL. The ACL must be loaded from a
4570 file (even a dummy empty file). The file name of the ACL to be updated is
4571 passed between parentheses. It takes one argument: <key fmt>, which follows
4572 log-format rules, to collect content of the new entry. It performs a lookup
4573 in the ACL before insertion, to avoid duplicated (or more) values. This
4574 lookup is done by a linear search and can be expensive with large lists!
4575 It is the equivalent of the "add acl" command from the stats socket, but can
4576 be triggered by an HTTP request.
Sasha Pachev218f0642014-06-16 12:05:59 -06004577
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004578http-request add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004579
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004580 This appends an HTTP header field whose name is specified in <name> and
4581 whose value is defined by <fmt> which follows the log-format rules (see
4582 Custom Log Format in section 8.2.4). This is particularly useful to pass
4583 connection-specific information to the server (e.g. the client's SSL
4584 certificate), or to combine several headers into one. This rule is not
4585 final, so it is possible to add other similar rules. Note that header
4586 addition is performed immediately, so one rule might reuse the resulting
4587 header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06004588
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004589http-request allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004590
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004591 This stops the evaluation of the rules and lets the request pass the check.
4592 No further "http-request" rules are evaluated.
Sasha Pachev218f0642014-06-16 12:05:59 -06004593
Sasha Pachev218f0642014-06-16 12:05:59 -06004594
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004595http-request auth [realm <realm>] [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004596
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004597 This stops the evaluation of the rules and immediately responds with an
4598 HTTP 401 or 407 error code to invite the user to present a valid user name
4599 and password. No further "http-request" rules are evaluated. An optional
4600 "realm" parameter is supported, it sets the authentication realm that is
4601 returned with the response (typically the application's name).
Sasha Pachev218f0642014-06-16 12:05:59 -06004602
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004603 Example:
4604 acl auth_ok http_auth_group(L1) G1
4605 http-request auth unless auth_ok
Sasha Pachev218f0642014-06-16 12:05:59 -06004606
Jarno Huuskonen251a6b72019-01-04 14:05:02 +02004607http-request cache-use <name> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004608
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02004609 See section 6.2 about cache setup.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004610
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004611http-request capture <sample> [ len <length> | id <id> ]
4612 [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004613
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004614 This captures sample expression <sample> from the request buffer, and
4615 converts it to a string of at most <len> characters. The resulting string is
4616 stored into the next request "capture" slot, so it will possibly appear next
4617 to some captured HTTP headers. It will then automatically appear in the logs,
4618 and it will be possible to extract it using sample fetch rules to feed it
4619 into headers or anything. The length should be limited given that this size
4620 will be allocated for each capture during the whole session life.
4621 Please check section 7.3 (Fetching samples) and "capture request header" for
4622 more information.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004623
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004624 If the keyword "id" is used instead of "len", the action tries to store the
4625 captured string in a previously declared capture slot. This is useful to run
4626 captures in backends. The slot id can be declared by a previous directive
Baptiste Assmann19a69b32020-01-16 14:34:22 +01004627 "http-request capture" or with the "declare capture" keyword.
4628
4629 When using this action in a backend, double check that the relevant
4630 frontend(s) have the required capture slots otherwise, this rule will be
4631 ignored at run time. This can't be detected at configuration parsing time
4632 due to HAProxy's ability to dynamically resolve backend name at runtime.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004633
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004634http-request del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004635
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004636 This is used to delete an entry from an ACL. The ACL must be loaded from a
4637 file (even a dummy empty file). The file name of the ACL to be updated is
4638 passed between parentheses. It takes one argument: <key fmt>, which follows
4639 log-format rules, to collect content of the entry to delete.
4640 It is the equivalent of the "del acl" command from the stats socket, but can
4641 be triggered by an HTTP request.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004642
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004643http-request del-header <name> [ { if | unless } <condition> ]
Willy Tarreauf4c43c12013-06-11 17:01:13 +02004644
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004645 This removes all HTTP header fields whose name is specified in <name>.
Willy Tarreau9a355ec2013-06-11 17:45:46 +02004646
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004647http-request del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau42cf39e2013-06-11 18:51:32 +02004648
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004649 This is used to delete an entry from a MAP. The MAP must be loaded from a
4650 file (even a dummy empty file). The file name of the MAP to be updated is
4651 passed between parentheses. It takes one argument: <key fmt>, which follows
4652 log-format rules, to collect content of the entry to delete.
4653 It takes one argument: "file name" It is the equivalent of the "del map"
4654 command from the stats socket, but can be triggered by an HTTP request.
Willy Tarreau51347ed2013-06-11 19:34:13 +02004655
Christopher Faulet554c0eb2020-01-14 12:00:28 +01004656http-request deny [deny_status <status>] [ { errorfile | errorfiles } <err> ]
4657 [ { if | unless } <condition> ]
Patrick Hemmer268a7072018-05-11 12:52:31 -04004658
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004659 This stops the evaluation of the rules and immediately rejects the request
4660 and emits an HTTP 403 error, or optionally the status code specified as an
4661 argument to "deny_status". The list of permitted status codes is limited to
Christopher Faulet554c0eb2020-01-14 12:00:28 +01004662 those that can be overridden by the "errorfile" directive. A specific error
4663 message may be specified. It may be an error file, using the "errorfile"
4664 keyword followed by the file containing the full HTTP response. It may also
4665 be an error from an http-errors section, using the "errorfiles" keyword
4666 followed by the section name.
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004667 No further "http-request" rules are evaluated.
Patrick Hemmer268a7072018-05-11 12:52:31 -04004668
Olivier Houchard602bf7d2019-05-10 13:59:15 +02004669http-request disable-l7-retry [ { if | unless } <condition> ]
4670 This disables any attempt to retry the request if it fails for any other
4671 reason than a connection failure. This can be useful for example to make
4672 sure POST requests aren't retried on failure.
4673
Baptiste Assmann333939c2019-01-21 08:34:50 +01004674http-request do-resolve(<var>,<resolvers>,[ipv4,ipv6]) <expr> :
4675
4676 This action performs a DNS resolution of the output of <expr> and stores
4677 the result in the variable <var>. It uses the DNS resolvers section
4678 pointed by <resolvers>.
4679 It is possible to choose a resolution preference using the optional
4680 arguments 'ipv4' or 'ipv6'.
4681 When performing the DNS resolution, the client side connection is on
4682 pause waiting till the end of the resolution.
4683 If an IP address can be found, it is stored into <var>. If any kind of
4684 error occurs, then <var> is not set.
4685 One can use this action to discover a server IP address at run time and
4686 based on information found in the request (IE a Host header).
4687 If this action is used to find the server's IP address (using the
4688 "set-dst" action), then the server IP address in the backend must be set
4689 to 0.0.0.0.
4690
4691 Example:
4692 resolvers mydns
4693 nameserver local 127.0.0.53:53
4694 nameserver google 8.8.8.8:53
4695 timeout retry 1s
4696 hold valid 10s
4697 hold nx 3s
4698 hold other 3s
4699 hold obsolete 0s
4700 accepted_payload_size 8192
4701
4702 frontend fe
4703 bind 10.42.0.1:80
4704 http-request do-resolve(txn.myip,mydns,ipv4) hdr(Host),lower
4705 http-request capture var(txn.myip) len 40
4706
4707 # return 503 when the variable is not set,
4708 # which mean DNS resolution error
4709 use_backend b_503 unless { var(txn.myip) -m found }
4710
4711 default_backend be
4712
4713 backend b_503
4714 # dummy backend used to return 503.
4715 # one can use the errorfile directive to send a nice
4716 # 503 error page to end users
4717
4718 backend be
4719 # rule to prevent HAProxy from reconnecting to services
4720 # on the local network (forged DNS name used to scan the network)
4721 http-request deny if { var(txn.myip) -m ip 127.0.0.0/8 10.0.0.0/8 }
4722 http-request set-dst var(txn.myip)
4723 server clear 0.0.0.0:0
4724
4725 NOTE: Don't forget to set the "protection" rules to ensure HAProxy won't
4726 be used to scan the network or worst won't loop over itself...
4727
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01004728http-request early-hint <name> <fmt> [ { if | unless } <condition> ]
4729
4730 This is used to build an HTTP 103 Early Hints response prior to any other one.
4731 This appends an HTTP header field to this response whose name is specified in
4732 <name> and whose value is defined by <fmt> which follows the log-format rules
4733 (see Custom Log Format in section 8.2.4). This is particularly useful to pass
Frédéric Lécaille3aac1062018-11-13 09:42:13 +01004734 to the client some Link headers to preload resources required to render the
4735 HTML documents.
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01004736
4737 See RFC 8297 for more information.
4738
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004739http-request redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004740
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004741 This performs an HTTP redirection based on a redirect rule. This is exactly
4742 the same as the "redirect" statement except that it inserts a redirect rule
4743 which can be processed in the middle of other "http-request" rules and that
4744 these rules use the "log-format" strings. See the "redirect" keyword for the
4745 rule's syntax.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004746
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004747http-request reject [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004748
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004749 This stops the evaluation of the rules and immediately closes the connection
4750 without sending any response. It acts similarly to the
4751 "tcp-request content reject" rules. It can be useful to force an immediate
4752 connection closure on HTTP/2 connections.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004753
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004754http-request replace-header <name> <match-regex> <replace-fmt>
4755 [ { if | unless } <condition> ]
Willy Tarreaua9083d02015-05-08 15:27:59 +02004756
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05004757 This matches the value of all occurrences of header field <name> against
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004758 <match-regex>. Matching is performed case-sensitively. Matching values are
4759 completely replaced by <replace-fmt>. Format characters are allowed in
4760 <replace-fmt> and work like <fmt> arguments in "http-request add-header".
4761 Standard back-references using the backslash ('\') followed by a number are
4762 supported.
Thierry FOURNIER82bf70d2015-05-26 17:58:29 +02004763
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004764 This action acts on whole header lines, regardless of the number of values
4765 they may contain. Thus it is well-suited to process headers naturally
4766 containing commas in their value, such as If-Modified-Since. Headers that
4767 contain a comma-separated list of values, such as Accept, should be processed
4768 using "http-request replace-value".
William Lallemand86d0df02017-11-24 21:36:45 +01004769
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004770 Example:
4771 http-request replace-header Cookie foo=([^;]*);(.*) foo=\1;ip=%bi;\2
4772
4773 # applied to:
4774 Cookie: foo=foobar; expires=Tue, 14-Jun-2016 01:40:45 GMT;
4775
4776 # outputs:
4777 Cookie: foo=foobar;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT;
4778
4779 # assuming the backend IP is 192.168.1.20
Willy Tarreau09448f72014-06-25 18:12:15 +02004780
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004781 http-request replace-header User-Agent curl foo
4782
4783 # applied to:
4784 User-Agent: curl/7.47.0
Willy Tarreau09448f72014-06-25 18:12:15 +02004785
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004786 # outputs:
4787 User-Agent: foo
Willy Tarreau09448f72014-06-25 18:12:15 +02004788
Willy Tarreau262c3f12019-12-17 06:52:51 +01004789http-request replace-path <match-regex> <replace-fmt>
4790 [ { if | unless } <condition> ]
4791
4792 This works like "replace-header" except that it works on the request's path
4793 component instead of a header. The path component starts at the first '/'
4794 after an optional scheme+authority. It does contain the query string if any
4795 is present. The replacement does not modify the scheme nor authority.
4796
4797 It is worth noting that regular expressions may be more expensive to evaluate
4798 than certain ACLs, so rare replacements may benefit from a condition to avoid
4799 performing the evaluation at all if it does not match.
4800
4801 Example:
4802 # prefix /foo : turn /bar?q=1 into /foo/bar?q=1 :
4803 http-request replace-path (.*) /foo\1
4804
4805 # suffix /foo : turn /bar?q=1 into /bar/foo?q=1 :
4806 http-request replace-path ([^?]*)(\?(.*))? \1/foo\2
4807
4808 # strip /foo : turn /foo/bar?q=1 into /bar?q=1
4809 http-request replace-path /foo/(.*) /\1
4810 # or more efficient if only some requests match :
4811 http-request replace-path /foo/(.*) /\1 if { url_beg /foo/ }
4812
Willy Tarreau33810222019-06-12 17:44:02 +02004813http-request replace-uri <match-regex> <replace-fmt>
4814 [ { if | unless } <condition> ]
4815
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004816 This works like "replace-header" except that it works on the request's URI part
4817 instead of a header. The URI part may contain an optional scheme, authority or
4818 query string. These are considered to be part of the value that is matched
4819 against.
4820
4821 It is worth noting that regular expressions may be more expensive to evaluate
4822 than certain ACLs, so rare replacements may benefit from a condition to avoid
4823 performing the evaluation at all if it does not match.
Willy Tarreau33810222019-06-12 17:44:02 +02004824
Willy Tarreau62b59132019-12-17 06:51:20 +01004825 IMPORTANT NOTE: historically in HTTP/1.x, the vast majority of requests sent
4826 by browsers use the "origin form", which differs from the "absolute form" in
4827 that they do not contain a scheme nor authority in the URI portion. Mostly
4828 only requests sent to proxies, those forged by hand and some emitted by
4829 certain applications use the absolute form. As such, "replace-uri" usually
4830 works fine most of the time in HTTP/1.x with rules starting with a "/". But
4831 with HTTP/2, clients are encouraged to send absolute URIs only, which look
4832 like the ones HTTP/1 clients use to talk to proxies. Such partial replace-uri
4833 rules may then fail in HTTP/2 when they work in HTTP/1. Either the rules need
Willy Tarreau262c3f12019-12-17 06:52:51 +01004834 to be adapted to optionally match a scheme and authority, or replace-path
4835 should be used.
Willy Tarreau33810222019-06-12 17:44:02 +02004836
Willy Tarreau62b59132019-12-17 06:51:20 +01004837 Example:
4838 # rewrite all "http" absolute requests to "https":
4839 http-request replace-uri ^http://(.*) https://\1
Willy Tarreau33810222019-06-12 17:44:02 +02004840
Willy Tarreau62b59132019-12-17 06:51:20 +01004841 # prefix /foo : turn /bar?q=1 into /foo/bar?q=1 :
4842 http-request replace-uri ([^/:]*://[^/]*)?(.*) \1/foo\2
Willy Tarreau33810222019-06-12 17:44:02 +02004843
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004844http-request replace-value <name> <match-regex> <replace-fmt>
4845 [ { if | unless } <condition> ]
Willy Tarreau09448f72014-06-25 18:12:15 +02004846
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004847 This works like "replace-header" except that it matches the regex against
4848 every comma-delimited value of the header field <name> instead of the
4849 entire header. This is suited for all headers which are allowed to carry
4850 more than one value. An example could be the Accept header.
Willy Tarreau09448f72014-06-25 18:12:15 +02004851
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004852 Example:
4853 http-request replace-value X-Forwarded-For ^192\.168\.(.*)$ 172.16.\1
Thierry FOURNIER236657b2015-08-19 08:25:14 +02004854
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004855 # applied to:
4856 X-Forwarded-For: 192.168.10.1, 192.168.13.24, 10.0.0.37
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02004857
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004858 # outputs:
4859 X-Forwarded-For: 172.16.10.1, 172.16.13.24, 10.0.0.37
Frédéric Lécaille6778b272018-01-29 15:22:53 +01004860
Christopher Faulet24231ab2020-01-24 17:44:23 +01004861http-request return [status <code>] [content-type <type>]
4862 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
4863 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
Christopher Faulet4a2c1422020-01-31 17:36:01 +01004864 [ hdr <name> <fmt> ]*
Christopher Faulet24231ab2020-01-24 17:44:23 +01004865 [ { if | unless } <condition> ]
4866
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05004867 This stops the evaluation of the rules and immediately returns a response. The
Christopher Faulet24231ab2020-01-24 17:44:23 +01004868 default status code used for the response is 200. It can be optionally
4869 specified as an arguments to "status". The response content-type may also be
4870 specified as an argument to "content-type". Finally the response itselft may
4871 be defined. If can be a full HTTP response specifying the errorfile to use,
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05004872 or the response payload specifying the file or the string to use. These rules
Christopher Faulet24231ab2020-01-24 17:44:23 +01004873 are followed to create the response :
4874
4875 * If neither the errorfile nor the payload to use is defined, a dummy
4876 response is returned. Only the "status" argument is considered. It can be
4877 any code in the range [200, 599]. The "content-type" argument, if any, is
4878 ignored.
4879
4880 * If "default-errorfiles" argument is set, the proxy's errorfiles are
4881 considered. If the "status" argument is defined, it must be one of the
4882 status code handled by hparoxy (200, 400, 403, 404, 405, 408, 410, 425,
4883 429, 500, 502, 503, and 504). The "content-type" argument, if any, is
4884 ignored.
4885
4886 * If a specific errorfile is defined, with an "errorfile" argument, the
4887 corresponding file, containing a full HTTP response, is returned. Only the
4888 "status" argument is considered. It must be one of the status code handled
4889 by hparoxy (200, 400, 403, 404, 405, 408, 410, 425, 429, 500, 502, 503, and
4890 504). The "content-type" argument, if any, is ignored.
4891
4892 * If an http-errors section is defined, with an "errorfiles" argument, the
4893 corresponding file in the specified http-errors section, containing a full
4894 HTTP response, is returned. Only the "status" argument is considered. It
4895 must be one of the status code handled by hparoxy (200, 400, 403, 404, 405,
4896 408, 410, 425, 429, 500, 502, 503, and 504). The "content-type" argument,
4897 if any, is ignored.
4898
4899 * If a "file" or a "lf-file" argument is specified, the file's content is
4900 used as the response payload. If the file is not empty, its content-type
4901 must be set as argument to "content-type". Otherwise, any "content-type"
4902 argument is ignored. With a "lf-file" argument, the file's content is
4903 evaluated as a log-format string. With a "file" argument, it is considered
4904 as a raw content.
4905
4906 * If a "string" or "lf-string" argument is specified, the defined string is
4907 used as the response payload. The content-type must always be set as
4908 argument to "content-type". With a "lf-string" argument, the string is
4909 evaluated as a log-format string. With a "string" argument, it is
4910 considered as a raw string.
4911
Christopher Faulet4a2c1422020-01-31 17:36:01 +01004912 When the response is not based an errorfile, it is possible to appends HTTP
4913 header fields to the response using "hdr" arguments. Otherwise, all "hdr"
4914 arguments are ignored. For each one, the header name is specified in <name>
4915 and its value is defined by <fmt> which follows the log-format rules.
4916
Christopher Faulet24231ab2020-01-24 17:44:23 +01004917 Note that the generated response must be smaller than a buffer. And to avoid
4918 any warning, when an errorfile or a raw file is loaded, the buffer space
4919 reserved to the headers rewritting should also be free.
4920
4921 No further "http-request" rules are evaluated.
4922
4923 Example:
4924 http-request return errorfile /etc/haproy/errorfiles/200.http \
4925 if { path /ping }
4926
4927 http-request return content-type image/x-icon file /var/www/favicon.ico \
4928 if { path /favicon.ico }
4929
4930 http-request return status 403 content-type text/plain \
4931 lf-string "Access denied. IP %[src] is blacklisted." \
4932 if { src -f /etc/haproxy/blacklist.lst }
4933
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004934http-request sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
4935http-request sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004936
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004937 This actions increments the GPC0 or GPC1 counter according with the sticky
4938 counter designated by <sc-id>. If an error occurs, this action silently fails
4939 and the actions evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004940
Cédric Dufour0d7712d2019-11-06 18:38:53 +01004941http-request sc-set-gpt0(<sc-id>) { <int> | <expr> }
4942 [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004943
Cédric Dufour0d7712d2019-11-06 18:38:53 +01004944 This action sets the 32-bit unsigned GPT0 tag according to the sticky counter
4945 designated by <sc-id> and the value of <int>/<expr>. The expected result is a
4946 boolean. If an error occurs, this action silently fails and the actions
4947 evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004948
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004949http-request set-dst <expr> [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004950
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004951 This is used to set the destination IP address to the value of specified
4952 expression. Useful when a proxy in front of HAProxy rewrites destination IP,
4953 but provides the correct IP in a HTTP header; or you want to mask the IP for
4954 privacy. If you want to connect to the new address/port, use '0.0.0.0:0' as a
4955 server address in the backend.
Christopher Faulet85d79c92016-11-09 16:54:56 +01004956
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004957 Arguments:
4958 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
4959 by some converters.
Christopher Faulet85d79c92016-11-09 16:54:56 +01004960
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004961 Example:
4962 http-request set-dst hdr(x-dst)
4963 http-request set-dst dst,ipmask(24)
Christopher Faulet85d79c92016-11-09 16:54:56 +01004964
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004965 When possible, set-dst preserves the original destination port as long as the
4966 address family allows it, otherwise the destination port is set to 0.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004967
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004968http-request set-dst-port <expr> [ { if | unless } <condition> ]
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004969
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004970 This is used to set the destination port address to the value of specified
4971 expression. If you want to connect to the new address/port, use '0.0.0.0:0'
4972 as a server address in the backend.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004973
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004974 Arguments:
4975 <expr> Is a standard HAProxy expression formed by a sample-fetch
4976 followed by some converters.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004977
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004978 Example:
4979 http-request set-dst-port hdr(x-port)
4980 http-request set-dst-port int(4000)
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004981
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004982 When possible, set-dst-port preserves the original destination address as
4983 long as the address family supports a port, otherwise it forces the
4984 destination address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +02004985
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004986http-request set-header <name> <fmt> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02004987
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004988 This does the same as "http-request add-header" except that the header name
4989 is first removed if it existed. This is useful when passing security
4990 information to the server, where the header must not be manipulated by
4991 external users. Note that the new value is computed before the removal so it
4992 is possible to concatenate a value to an existing header.
William Lallemand44be6402016-05-25 01:51:35 +02004993
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004994 Example:
4995 http-request set-header X-Haproxy-Current-Date %T
4996 http-request set-header X-SSL %[ssl_fc]
4997 http-request set-header X-SSL-Session_ID %[ssl_fc_session_id,hex]
4998 http-request set-header X-SSL-Client-Verify %[ssl_c_verify]
4999 http-request set-header X-SSL-Client-DN %{+Q}[ssl_c_s_dn]
5000 http-request set-header X-SSL-Client-CN %{+Q}[ssl_c_s_dn(cn)]
5001 http-request set-header X-SSL-Issuer %{+Q}[ssl_c_i_dn]
5002 http-request set-header X-SSL-Client-NotBefore %{+Q}[ssl_c_notbefore]
5003 http-request set-header X-SSL-Client-NotAfter %{+Q}[ssl_c_notafter]
William Lallemand44be6402016-05-25 01:51:35 +02005004
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005005http-request set-log-level <level> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02005006
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005007 This is used to change the log level of the current request when a certain
5008 condition is met. Valid levels are the 8 syslog levels (see the "log"
5009 keyword) plus the special level "silent" which disables logging for this
5010 request. This rule is not final so the last matching rule wins. This rule
5011 can be useful to disable health checks coming from another equipment.
William Lallemand13e9b0c2016-05-25 02:34:07 +02005012
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005013http-request set-map(<file-name>) <key fmt> <value fmt>
5014 [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02005015
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005016 This is used to add a new entry into a MAP. The MAP must be loaded from a
5017 file (even a dummy empty file). The file name of the MAP to be updated is
5018 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
5019 log-format rules, used to collect MAP key, and <value fmt>, which follows
5020 log-format rules, used to collect content for the new entry.
5021 It performs a lookup in the MAP before insertion, to avoid duplicated (or
5022 more) values. This lookup is done by a linear search and can be expensive
5023 with large lists! It is the equivalent of the "set map" command from the
5024 stats socket, but can be triggered by an HTTP request.
William Lallemand13e9b0c2016-05-25 02:34:07 +02005025
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005026http-request set-mark <mark> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02005027
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005028 This is used to set the Netfilter MARK on all packets sent to the client to
5029 the value passed in <mark> on platforms which support it. This value is an
5030 unsigned 32 bit value which can be matched by netfilter and by the routing
5031 table. It can be expressed both in decimal or hexadecimal format (prefixed by
5032 "0x"). This can be useful to force certain packets to take a different route
5033 (for example a cheaper network path for bulk downloads). This works on Linux
5034 kernels 2.6.32 and above and requires admin privileges.
Willy Tarreau00005ce2016-10-21 15:07:45 +02005035
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005036http-request set-method <fmt> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02005037
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005038 This rewrites the request method with the result of the evaluation of format
5039 string <fmt>. There should be very few valid reasons for having to do so as
5040 this is more likely to break something than to fix it.
William Lallemand13e9b0c2016-05-25 02:34:07 +02005041
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005042http-request set-nice <nice> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02005043
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005044 This sets the "nice" factor of the current request being processed. It only
5045 has effect against the other requests being processed at the same time.
5046 The default value is 0, unless altered by the "nice" setting on the "bind"
5047 line. The accepted range is -1024..1024. The higher the value, the nicest
5048 the request will be. Lower values will make the request more important than
5049 other ones. This can be useful to improve the speed of some requests, or
5050 lower the priority of non-important requests. Using this setting without
5051 prior experimentation can cause some major slowdown.
William Lallemand13e9b0c2016-05-25 02:34:07 +02005052
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005053http-request set-path <fmt> [ { if | unless } <condition> ]
Willy Tarreau00005ce2016-10-21 15:07:45 +02005054
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005055 This rewrites the request path with the result of the evaluation of format
5056 string <fmt>. The query string, if any, is left intact. If a scheme and
5057 authority is found before the path, they are left intact as well. If the
5058 request doesn't have a path ("*"), this one is replaced with the format.
5059 This can be used to prepend a directory component in front of a path for
5060 example. See also "http-request set-query" and "http-request set-uri".
Willy Tarreau2d392c22015-08-24 01:43:45 +02005061
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005062 Example :
5063 # prepend the host name before the path
5064 http-request set-path /%[hdr(host)]%[path]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02005065
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005066http-request set-priority-class <expr> [ { if | unless } <condition> ]
Olivier Houchardccaa7de2017-10-02 11:51:03 +02005067
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005068 This is used to set the queue priority class of the current request.
5069 The value must be a sample expression which converts to an integer in the
5070 range -2047..2047. Results outside this range will be truncated.
5071 The priority class determines the order in which queued requests are
5072 processed. Lower values have higher priority.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02005073
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005074http-request set-priority-offset <expr> [ { if | unless } <condition> ]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02005075
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005076 This is used to set the queue priority timestamp offset of the current
5077 request. The value must be a sample expression which converts to an integer
5078 in the range -524287..524287. Results outside this range will be truncated.
5079 When a request is queued, it is ordered first by the priority class, then by
5080 the current timestamp adjusted by the given offset in milliseconds. Lower
5081 values have higher priority.
5082 Note that the resulting timestamp is is only tracked with enough precision
5083 for 524,287ms (8m44s287ms). If the request is queued long enough to where the
5084 adjusted timestamp exceeds this value, it will be misidentified as highest
5085 priority. Thus it is important to set "timeout queue" to a value, where when
5086 combined with the offset, does not exceed this limit.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02005087
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005088http-request set-query <fmt> [ { if | unless } <condition> ]
Willy Tarreau20b0de52012-12-24 15:45:22 +01005089
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005090 This rewrites the request's query string which appears after the first
5091 question mark ("?") with the result of the evaluation of format string <fmt>.
5092 The part prior to the question mark is left intact. If the request doesn't
5093 contain a question mark and the new value is not empty, then one is added at
5094 the end of the URI, followed by the new value. If a question mark was
5095 present, it will never be removed even if the value is empty. This can be
5096 used to add or remove parameters from the query string.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08005097
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005098 See also "http-request set-query" and "http-request set-uri".
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01005099
5100 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005101 # replace "%3D" with "=" in the query string
5102 http-request set-query %[query,regsub(%3D,=,g)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01005103
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005104http-request set-src <expr> [ { if | unless } <condition> ]
5105 This is used to set the source IP address to the value of specified
5106 expression. Useful when a proxy in front of HAProxy rewrites source IP, but
5107 provides the correct IP in a HTTP header; or you want to mask source IP for
5108 privacy.
5109
5110 Arguments :
5111 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
5112 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01005113
Cyril Bonté78caf842010-03-10 22:41:43 +01005114 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005115 http-request set-src hdr(x-forwarded-for)
5116 http-request set-src src,ipmask(24)
5117
5118 When possible, set-src preserves the original source port as long as the
5119 address family allows it, otherwise the source port is set to 0.
5120
5121http-request set-src-port <expr> [ { if | unless } <condition> ]
5122
5123 This is used to set the source port address to the value of specified
5124 expression.
5125
5126 Arguments:
5127 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
5128 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01005129
Willy Tarreau20b0de52012-12-24 15:45:22 +01005130 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005131 http-request set-src-port hdr(x-port)
5132 http-request set-src-port int(4000)
5133
5134 When possible, set-src-port preserves the original source address as long as
5135 the address family supports a port, otherwise it forces the source address to
5136 IPv4 "0.0.0.0" before rewriting the port.
5137
5138http-request set-tos <tos> [ { if | unless } <condition> ]
5139
5140 This is used to set the TOS or DSCP field value of packets sent to the client
5141 to the value passed in <tos> on platforms which support this. This value
5142 represents the whole 8 bits of the IP TOS field, and can be expressed both in
5143 decimal or hexadecimal format (prefixed by "0x"). Note that only the 6 higher
5144 bits are used in DSCP or TOS, and the two lower bits are always 0. This can
5145 be used to adjust some routing behavior on border routers based on some
5146 information from the request.
5147
5148 See RFC 2474, 2597, 3260 and 4594 for more information.
5149
5150http-request set-uri <fmt> [ { if | unless } <condition> ]
5151
5152 This rewrites the request URI with the result of the evaluation of format
5153 string <fmt>. The scheme, authority, path and query string are all replaced
5154 at once. This can be used to rewrite hosts in front of proxies, or to
5155 perform complex modifications to the URI such as moving parts between the
5156 path and the query string.
5157 See also "http-request set-path" and "http-request set-query".
5158
5159http-request set-var(<var-name>) <expr> [ { if | unless } <condition> ]
5160
5161 This is used to set the contents of a variable. The variable is declared
5162 inline.
5163
5164 Arguments:
5165 <var-name> The name of the variable starts with an indication about its
5166 scope. The scopes allowed are:
5167 "proc" : the variable is shared with the whole process
5168 "sess" : the variable is shared with the whole session
5169 "txn" : the variable is shared with the transaction
5170 (request and response)
5171 "req" : the variable is shared only during request
5172 processing
5173 "res" : the variable is shared only during response
5174 processing
5175 This prefix is followed by a name. The separator is a '.'.
5176 The name may only contain characters 'a-z', 'A-Z', '0-9'
5177 and '_'.
5178
5179 <expr> Is a standard HAProxy expression formed by a sample-fetch
5180 followed by some converters.
Willy Tarreau20b0de52012-12-24 15:45:22 +01005181
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005182 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005183 http-request set-var(req.my_var) req.fhdr(user-agent),lower
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005184
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005185http-request send-spoe-group <engine-name> <group-name>
5186 [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005187
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005188 This action is used to trigger sending of a group of SPOE messages. To do so,
5189 the SPOE engine used to send messages must be defined, as well as the SPOE
5190 group to send. Of course, the SPOE engine must refer to an existing SPOE
5191 filter. If not engine name is provided on the SPOE filter line, the SPOE
5192 agent name must be used.
5193
5194 Arguments:
5195 <engine-name> The SPOE engine name.
5196
5197 <group-name> The SPOE group name as specified in the engine
5198 configuration.
5199
5200http-request silent-drop [ { if | unless } <condition> ]
5201
5202 This stops the evaluation of the rules and makes the client-facing connection
5203 suddenly disappear using a system-dependent way that tries to prevent the
5204 client from being notified. The effect it then that the client still sees an
5205 established connection while there's none on HAProxy. The purpose is to
5206 achieve a comparable effect to "tarpit" except that it doesn't use any local
5207 resource at all on the machine running HAProxy. It can resist much higher
5208 loads than "tarpit", and slow down stronger attackers. It is important to
5209 understand the impact of using this mechanism. All stateful equipment placed
5210 between the client and HAProxy (firewalls, proxies, load balancers) will also
5211 keep the established connection for a long time and may suffer from this
5212 action.
5213 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
5214 option is used to block the emission of a TCP reset. On other systems, the
5215 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
5216 router, though it's still delivered to local networks. Do not use it unless
5217 you fully understand how it works.
5218
Christopher Faulet46f95542019-12-20 10:07:22 +01005219http-request strict-mode { on | off }
5220
5221 This enables or disables the strict rewriting mode for following rules. It
5222 does not affect rules declared before it and it is only applicable on rules
5223 performing a rewrite on the requests. When the strict mode is enabled, any
5224 rewrite failure triggers an internal error. Otherwise, such errors are
5225 silently ignored. The purpose of the strict rewriting mode is to make some
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05005226 rewrites optional while others must be performed to continue the request
Christopher Faulet46f95542019-12-20 10:07:22 +01005227 processing.
5228
Christopher Faulet1aea50e2020-01-17 16:03:53 +01005229 By default, the strict rewriting mode is enabled. Its value is also reset
Christopher Faulet46f95542019-12-20 10:07:22 +01005230 when a ruleset evaluation ends. So, for instance, if you change the mode on
5231 the frontend, the default mode is restored when HAProxy starts the backend
5232 rules evaluation.
5233
Christopher Faulet554c0eb2020-01-14 12:00:28 +01005234http-request tarpit [deny_status <status>] [ { errorfile | errorfiles } <err> ]
5235 [ { if | unless } <condition> ]
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005236
5237 This stops the evaluation of the rules and immediately blocks the request
5238 without responding for a delay specified by "timeout tarpit" or
5239 "timeout connect" if the former is not set. After that delay, if the client
5240 is still connected, an HTTP error 500 (or optionally the status code
5241 specified as an argument to "deny_status") is returned so that the client
5242 does not suspect it has been tarpitted. Logs will report the flags "PT".
5243 The goal of the tarpit rule is to slow down robots during an attack when
5244 they're limited on the number of concurrent requests. It can be very
5245 efficient against very dumb robots, and will significantly reduce the load
5246 on firewalls compared to a "deny" rule. But when facing "correctly"
5247 developed robots, it can make things worse by forcing haproxy and the front
Christopher Faulet554c0eb2020-01-14 12:00:28 +01005248 firewall to support insane number of concurrent connections. A specific error
5249 message may be specified. It may be an error file, using the "errorfile"
5250 keyword followed by the file containing the full HTTP response. It may also
5251 be an error from an http-errors section, using the "errorfiles" keyword
5252 followed by the section name.
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005253 See also the "silent-drop" action.
5254
5255http-request track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
5256http-request track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
5257http-request track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
5258
5259 This enables tracking of sticky counters from current request. These rules do
5260 not stop evaluation and do not change default action. The number of counters
5261 that may be simultaneously tracked by the same connection is set in
5262 MAX_SESS_STKCTR at build time (reported in haproxy -vv) which defaults to 3,
5263 so the track-sc number is between 0 and (MAX_SESS_STCKTR-1). The first
5264 "track-sc0" rule executed enables tracking of the counters of the specified
5265 table as the first set. The first "track-sc1" rule executed enables tracking
5266 of the counters of the specified table as the second set. The first
5267 "track-sc2" rule executed enables tracking of the counters of the specified
5268 table as the third set. It is a recommended practice to use the first set of
5269 counters for the per-frontend counters and the second set for the per-backend
5270 ones. But this is just a guideline, all may be used everywhere.
5271
5272 Arguments :
5273 <key> is mandatory, and is a sample expression rule as described in
5274 section 7.3. It describes what elements of the incoming request or
5275 connection will be analyzed, extracted, combined, and used to
5276 select which table entry to update the counters.
5277
5278 <table> is an optional table to be used instead of the default one, which
5279 is the stick-table declared in the current proxy. All the counters
5280 for the matches and updates for the key will then be performed in
5281 that table until the session ends.
5282
5283 Once a "track-sc*" rule is executed, the key is looked up in the table and if
5284 it is not found, an entry is allocated for it. Then a pointer to that entry
5285 is kept during all the session's life, and this entry's counters are updated
5286 as often as possible, every time the session's counters are updated, and also
5287 systematically when the session ends. Counters are only updated for events
5288 that happen after the tracking has been started. As an exception, connection
5289 counters and request counters are systematically updated so that they reflect
5290 useful information.
5291
5292 If the entry tracks concurrent connection counters, one connection is counted
5293 for as long as the entry is tracked, and the entry will not expire during
5294 that time. Tracking counters also provides a performance advantage over just
5295 checking the keys, because only one table lookup is performed for all ACL
5296 checks that make use of it.
5297
5298http-request unset-var(<var-name>) [ { if | unless } <condition> ]
5299
5300 This is used to unset a variable. See above for details about <var-name>.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005301
5302 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005303 http-request unset-var(req.my_var)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005304
Christopher Faulet579d83b2019-11-22 15:34:17 +01005305http-request use-service <service-name> [ { if | unless } <condition> ]
5306
5307 This directive executes the configured HTTP service to reply to the request
5308 and stops the evaluation of the rules. An HTTP service may choose to reply by
5309 sending any valid HTTP response or it may immediately close the connection
5310 without sending any response. Outside natives services, for instance the
5311 Prometheus exporter, it is possible to write your own services in Lua. No
5312 further "http-request" rules are evaluated.
5313
5314 Arguments :
5315 <service-name> is mandatory. It is the service to call
5316
5317 Example:
5318 http-request use-service prometheus-exporter if { path /metrics }
5319
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005320http-request wait-for-handshake [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005321
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005322 This will delay the processing of the request until the SSL handshake
5323 happened. This is mostly useful to delay processing early data until we're
5324 sure they are valid.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005325
Willy Tarreauef781042010-01-27 11:53:01 +01005326
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005327http-response <action> <options...> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02005328 Access control for Layer 7 responses
5329
5330 May be used in sections: defaults | frontend | listen | backend
5331 no | yes | yes | yes
5332
5333 The http-response statement defines a set of rules which apply to layer 7
5334 processing. The rules are evaluated in their declaration order when they are
5335 met in a frontend, listen or backend section. Any rule may optionally be
5336 followed by an ACL-based condition, in which case it will only be evaluated
5337 if the condition is true. Since these rules apply on responses, the backend
5338 rules are applied first, followed by the frontend's rules.
5339
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005340 The first keyword is the rule's action. The supported actions are described
5341 below.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02005342
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005343 There is no limit to the number of http-response statements per instance.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02005344
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005345 Example:
5346 acl key_acl res.hdr(X-Acl-Key) -m found
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02005347
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005348 acl myhost hdr(Host) -f myhost.lst
Sasha Pachev218f0642014-06-16 12:05:59 -06005349
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005350 http-response add-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
5351 http-response del-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
Sasha Pachev218f0642014-06-16 12:05:59 -06005352
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005353 Example:
5354 acl value res.hdr(X-Value) -m found
Sasha Pachev218f0642014-06-16 12:05:59 -06005355
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005356 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06005357
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005358 http-response set-map(map.lst) %[src] %[res.hdr(X-Value)] if value
5359 http-response del-map(map.lst) %[src] if ! value
Sasha Pachev218f0642014-06-16 12:05:59 -06005360
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005361 See also : "http-request", section 3.4 about userlists and section 7 about
5362 ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06005363
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005364http-response add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005365
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005366 This is used to add a new entry into an ACL. The ACL must be loaded from a
5367 file (even a dummy empty file). The file name of the ACL to be updated is
5368 passed between parentheses. It takes one argument: <key fmt>, which follows
5369 log-format rules, to collect content of the new entry. It performs a lookup
5370 in the ACL before insertion, to avoid duplicated (or more) values.
5371 This lookup is done by a linear search and can be expensive with large lists!
5372 It is the equivalent of the "add acl" command from the stats socket, but can
5373 be triggered by an HTTP response.
Sasha Pachev218f0642014-06-16 12:05:59 -06005374
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005375http-response add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005376
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005377 This appends an HTTP header field whose name is specified in <name> and whose
5378 value is defined by <fmt> which follows the log-format rules (see Custom Log
5379 Format in section 8.2.4). This may be used to send a cookie to a client for
5380 example, or to pass some internal information.
5381 This rule is not final, so it is possible to add other similar rules.
5382 Note that header addition is performed immediately, so one rule might reuse
5383 the resulting header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06005384
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005385http-response allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005386
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005387 This stops the evaluation of the rules and lets the response pass the check.
5388 No further "http-response" rules are evaluated for the current section.
Sasha Pachev218f0642014-06-16 12:05:59 -06005389
Jarno Huuskonen251a6b72019-01-04 14:05:02 +02005390http-response cache-store <name> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005391
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02005392 See section 6.2 about cache setup.
Sasha Pachev218f0642014-06-16 12:05:59 -06005393
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005394http-response capture <sample> id <id> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005395
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005396 This captures sample expression <sample> from the response buffer, and
5397 converts it to a string. The resulting string is stored into the next request
5398 "capture" slot, so it will possibly appear next to some captured HTTP
5399 headers. It will then automatically appear in the logs, and it will be
5400 possible to extract it using sample fetch rules to feed it into headers or
5401 anything. Please check section 7.3 (Fetching samples) and
5402 "capture response header" for more information.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02005403
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005404 The keyword "id" is the id of the capture slot which is used for storing the
5405 string. The capture slot must be defined in an associated frontend.
5406 This is useful to run captures in backends. The slot id can be declared by a
5407 previous directive "http-response capture" or with the "declare capture"
5408 keyword.
Baptiste Assmann19a69b32020-01-16 14:34:22 +01005409
5410 When using this action in a backend, double check that the relevant
5411 frontend(s) have the required capture slots otherwise, this rule will be
5412 ignored at run time. This can't be detected at configuration parsing time
5413 due to HAProxy's ability to dynamically resolve backend name at runtime.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02005414
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005415http-response del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02005416
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005417 This is used to delete an entry from an ACL. The ACL must be loaded from a
5418 file (even a dummy empty file). The file name of the ACL to be updated is
5419 passed between parentheses. It takes one argument: <key fmt>, which follows
5420 log-format rules, to collect content of the entry to delete.
5421 It is the equivalent of the "del acl" command from the stats socket, but can
5422 be triggered by an HTTP response.
Willy Tarreauf4c43c12013-06-11 17:01:13 +02005423
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005424http-response del-header <name> [ { if | unless } <condition> ]
Willy Tarreau9a355ec2013-06-11 17:45:46 +02005425
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005426 This removes all HTTP header fields whose name is specified in <name>.
Willy Tarreau42cf39e2013-06-11 18:51:32 +02005427
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005428http-response del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau51347ed2013-06-11 19:34:13 +02005429
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005430 This is used to delete an entry from a MAP. The MAP must be loaded from a
5431 file (even a dummy empty file). The file name of the MAP to be updated is
5432 passed between parentheses. It takes one argument: <key fmt>, which follows
5433 log-format rules, to collect content of the entry to delete.
5434 It takes one argument: "file name" It is the equivalent of the "del map"
5435 command from the stats socket, but can be triggered by an HTTP response.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005436
Christopher Faulet554c0eb2020-01-14 12:00:28 +01005437http-response deny [deny_status <status>] [ { errorfile | errorfiles } <err> ]
5438 [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005439
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005440 This stops the evaluation of the rules and immediately rejects the response
Christopher Faulet040c8cd2020-01-13 16:43:45 +01005441 and emits an HTTP 502 error, or optionally the status code specified as an
5442 argument to "deny_status". The list of permitted status codes is limited to
Christopher Faulet554c0eb2020-01-14 12:00:28 +01005443 those that can be overridden by the "errorfile" directive. A specific error
5444 message may be specified. It may be an error file, using the "errorfile"
5445 keyword followed by the file containing the full HTTP response. It may also
5446 be an error from an http-errors section, using the "errorfiles" keyword
5447 followed by the section name.
Christopher Faulet040c8cd2020-01-13 16:43:45 +01005448 No further "http-response" rules are evaluated.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005449
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005450http-response redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005451
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005452 This performs an HTTP redirection based on a redirect rule.
5453 This supports a format string similarly to "http-request redirect" rules,
5454 with the exception that only the "location" type of redirect is possible on
5455 the response. See the "redirect" keyword for the rule's syntax. When a
5456 redirect rule is applied during a response, connections to the server are
5457 closed so that no data can be forwarded from the server to the client.
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02005458
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005459http-response replace-header <name> <regex-match> <replace-fmt>
5460 [ { if | unless } <condition> ]
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02005461
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005462 This works like "http-request replace-header" except that it works on the
5463 server's response instead of the client's request.
William Lallemand86d0df02017-11-24 21:36:45 +01005464
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005465 Example:
5466 http-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2
Willy Tarreau51d861a2015-05-22 17:30:48 +02005467
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005468 # applied to:
5469 Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02005470
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005471 # outputs:
5472 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 +02005473
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005474 # assuming the backend IP is 192.168.1.20.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02005475
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005476http-response replace-value <name> <regex-match> <replace-fmt>
5477 [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02005478
Tim Duesterhus6bd909b2020-01-17 15:53:18 +01005479 This works like "http-request replace-value" except that it works on the
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005480 server's response instead of the client's request.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02005481
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005482 Example:
5483 http-response replace-value Cache-control ^public$ private
Christopher Faulet85d79c92016-11-09 16:54:56 +01005484
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005485 # applied to:
5486 Cache-Control: max-age=3600, public
Christopher Faulet85d79c92016-11-09 16:54:56 +01005487
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005488 # outputs:
5489 Cache-Control: max-age=3600, private
Christopher Faulet85d79c92016-11-09 16:54:56 +01005490
Christopher Faulet24231ab2020-01-24 17:44:23 +01005491http-response return [status <code>] [content-type <type>]
5492 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
5493 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
Christopher Faulet4a2c1422020-01-31 17:36:01 +01005494 [ hdr <name> <value> ]*
Christopher Faulet24231ab2020-01-24 17:44:23 +01005495 [ { if | unless } <condition> ]
5496
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05005497 This stops the evaluation of the rules and immediately returns a response. The
Christopher Faulet24231ab2020-01-24 17:44:23 +01005498 default status code used for the response is 200. It can be optionally
5499 specified as an arguments to "status". The response content-type may also be
5500 specified as an argument to "content-type". Finally the response itselft may
5501 be defined. If can be a full HTTP response specifying the errorfile to use,
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05005502 or the response payload specifying the file or the string to use. These rules
Christopher Faulet24231ab2020-01-24 17:44:23 +01005503 are followed to create the response :
5504
5505 * If neither the errorfile nor the payload to use is defined, a dummy
5506 response is returned. Only the "status" argument is considered. It can be
5507 any code in the range [200, 599]. The "content-type" argument, if any, is
5508 ignored.
5509
5510 * If "default-errorfiles" argument is set, the proxy's errorfiles are
5511 considered. If the "status" argument is defined, it must be one of the
5512 status code handled by hparoxy (200, 400, 403, 404, 405, 408, 410, 425,
5513 429, 500, 502, 503, and 504). The "content-type" argument, if any, is
5514 ignored.
5515
5516 * If a specific errorfile is defined, with an "errorfile" argument, the
5517 corresponding file, containing a full HTTP response, is returned. Only the
5518 "status" argument is considered. It must be one of the status code handled
5519 by hparoxy (200, 400, 403, 404, 405, 408, 410, 425, 429, 500, 502, 503, and
5520 504). The "content-type" argument, if any, is ignored.
5521
5522 * If an http-errors section is defined, with an "errorfiles" argument, the
5523 corresponding file in the specified http-errors section, containing a full
5524 HTTP response, is returned. Only the "status" argument is considered. It
5525 must be one of the status code handled by hparoxy (200, 400, 403, 404, 405,
5526 408, 410, 425, 429, 500, 502, 503, and 504). The "content-type" argument,
5527 if any, is ignored.
5528
5529 * If a "file" or a "lf-file" argument is specified, the file's content is
5530 used as the response payload. If the file is not empty, its content-type
5531 must be set as argument to "content-type". Otherwise, any "content-type"
5532 argument is ignored. With a "lf-file" argument, the file's content is
5533 evaluated as a log-format string. With a "file" argument, it is considered
5534 as a raw content.
5535
5536 * If a "string" or "lf-string" argument is specified, the defined string is
5537 used as the response payload. The content-type must always be set as
5538 argument to "content-type". With a "lf-string" argument, the string is
5539 evaluated as a log-format string. With a "string" argument, it is
5540 considered as a raw string.
5541
Christopher Faulet4a2c1422020-01-31 17:36:01 +01005542 When the response is not based an errorfile, it is possible to appends HTTP
5543 header fields to the response using "hdr" arguments. Otherwise, all "hdr"
5544 arguments are ignored. For each one, the header name is specified in <name>
5545 and its value is defined by <fmt> which follows the log-format rules.
5546
Christopher Faulet24231ab2020-01-24 17:44:23 +01005547 Note that the generated response must be smaller than a buffer. And to avoid
5548 any warning, when an errorfile or a raw file is loaded, the buffer space
5549 reserved to the headers rewritting should also be free.
5550
5551 No further "http-response" rules are evaluated.
5552
5553 Example:
5554 http-response return errorfile /etc/haproy/errorfiles/200.http \
5555 if { status eq 404 }
5556
5557 http-response return content-type text/plain \
5558 string "This is the end !" \
5559 if { status eq 500 }
5560
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005561http-response sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
5562http-response sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Ruoshan Huange4edc6b2016-07-14 15:07:45 +08005563
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005564 This action increments the GPC0 or GPC1 counter according with the sticky
5565 counter designated by <sc-id>. If an error occurs, this action silently fails
5566 and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +02005567
Cédric Dufour0d7712d2019-11-06 18:38:53 +01005568http-response sc-set-gpt0(<sc-id>) { <int> | <expr> }
5569 [ { if | unless } <condition> ]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02005570
Cédric Dufour0d7712d2019-11-06 18:38:53 +01005571 This action sets the 32-bit unsigned GPT0 tag according to the sticky counter
5572 designated by <sc-id> and the value of <int>/<expr>. The expected result is a
5573 boolean. If an error occurs, this action silently fails and the actions
5574 evaluation continues.
Frédéric Lécaille6778b272018-01-29 15:22:53 +01005575
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005576http-response send-spoe-group [ { if | unless } <condition> ]
Willy Tarreau2d392c22015-08-24 01:43:45 +02005577
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005578 This action is used to trigger sending of a group of SPOE messages. To do so,
5579 the SPOE engine used to send messages must be defined, as well as the SPOE
5580 group to send. Of course, the SPOE engine must refer to an existing SPOE
5581 filter. If not engine name is provided on the SPOE filter line, the SPOE
5582 agent name must be used.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02005583
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005584 Arguments:
5585 <engine-name> The SPOE engine name.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02005586
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005587 <group-name> The SPOE group name as specified in the engine
5588 configuration.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02005589
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005590http-response set-header <name> <fmt> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02005591
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005592 This does the same as "add-header" except that the header name is first
5593 removed if it existed. This is useful when passing security information to
5594 the server, where the header must not be manipulated by external users.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02005595
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005596http-response set-log-level <level> [ { if | unless } <condition> ]
5597
5598 This is used to change the log level of the current request when a certain
5599 condition is met. Valid levels are the 8 syslog levels (see the "log"
5600 keyword) plus the special level "silent" which disables logging for this
5601 request. This rule is not final so the last matching rule wins. This rule can
5602 be useful to disable health checks coming from another equipment.
5603
5604http-response set-map(<file-name>) <key fmt> <value fmt>
5605
5606 This is used to add a new entry into a MAP. The MAP must be loaded from a
5607 file (even a dummy empty file). The file name of the MAP to be updated is
5608 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
5609 log-format rules, used to collect MAP key, and <value fmt>, which follows
5610 log-format rules, used to collect content for the new entry. It performs a
5611 lookup in the MAP before insertion, to avoid duplicated (or more) values.
5612 This lookup is done by a linear search and can be expensive with large lists!
5613 It is the equivalent of the "set map" command from the stats socket, but can
5614 be triggered by an HTTP response.
5615
5616http-response set-mark <mark> [ { if | unless } <condition> ]
5617
5618 This is used to set the Netfilter MARK on all packets sent to the client to
5619 the value passed in <mark> on platforms which support it. This value is an
5620 unsigned 32 bit value which can be matched by netfilter and by the routing
5621 table. It can be expressed both in decimal or hexadecimal format (prefixed
5622 by "0x"). This can be useful to force certain packets to take a different
5623 route (for example a cheaper network path for bulk downloads). This works on
5624 Linux kernels 2.6.32 and above and requires admin privileges.
5625
5626http-response set-nice <nice> [ { if | unless } <condition> ]
5627
5628 This sets the "nice" factor of the current request being processed.
5629 It only has effect against the other requests being processed at the same
5630 time. The default value is 0, unless altered by the "nice" setting on the
5631 "bind" line. The accepted range is -1024..1024. The higher the value, the
5632 nicest the request will be. Lower values will make the request more important
5633 than other ones. This can be useful to improve the speed of some requests, or
5634 lower the priority of non-important requests. Using this setting without
5635 prior experimentation can cause some major slowdown.
5636
5637http-response set-status <status> [reason <str>]
5638 [ { if | unless } <condition> ]
5639
5640 This replaces the response status code with <status> which must be an integer
5641 between 100 and 999. Optionally, a custom reason text can be provided defined
5642 by <str>, or the default reason for the specified code will be used as a
5643 fallback.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08005644
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005645 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005646 # return "431 Request Header Fields Too Large"
5647 http-response set-status 431
5648 # return "503 Slow Down", custom reason
5649 http-response set-status 503 reason "Slow Down".
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005650
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005651http-response set-tos <tos> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005652
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005653 This is used to set the TOS or DSCP field value of packets sent to the client
5654 to the value passed in <tos> on platforms which support this.
5655 This value represents the whole 8 bits of the IP TOS field, and can be
5656 expressed both in decimal or hexadecimal format (prefixed by "0x"). Note that
5657 only the 6 higher bits are used in DSCP or TOS, and the two lower bits are
5658 always 0. This can be used to adjust some routing behavior on border routers
5659 based on some information from the request.
5660
5661 See RFC 2474, 2597, 3260 and 4594 for more information.
5662
5663http-response set-var(<var-name>) <expr> [ { if | unless } <condition> ]
5664
5665 This is used to set the contents of a variable. The variable is declared
5666 inline.
5667
5668 Arguments:
5669 <var-name> The name of the variable starts with an indication about its
5670 scope. The scopes allowed are:
5671 "proc" : the variable is shared with the whole process
5672 "sess" : the variable is shared with the whole session
5673 "txn" : the variable is shared with the transaction
5674 (request and response)
5675 "req" : the variable is shared only during request
5676 processing
5677 "res" : the variable is shared only during response
5678 processing
5679 This prefix is followed by a name. The separator is a '.'.
5680 The name may only contain characters 'a-z', 'A-Z', '0-9', '.'
5681 and '_'.
5682
5683 <expr> Is a standard HAProxy expression formed by a sample-fetch
5684 followed by some converters.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005685
5686 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005687 http-response set-var(sess.last_redir) res.hdr(location)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005688
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005689http-response silent-drop [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005690
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005691 This stops the evaluation of the rules and makes the client-facing connection
5692 suddenly disappear using a system-dependent way that tries to prevent the
5693 client from being notified. The effect it then that the client still sees an
5694 established connection while there's none on HAProxy. The purpose is to
5695 achieve a comparable effect to "tarpit" except that it doesn't use any local
5696 resource at all on the machine running HAProxy. It can resist much higher
5697 loads than "tarpit", and slow down stronger attackers. It is important to
5698 understand the impact of using this mechanism. All stateful equipment placed
5699 between the client and HAProxy (firewalls, proxies, load balancers) will also
5700 keep the established connection for a long time and may suffer from this
5701 action.
5702 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
5703 option is used to block the emission of a TCP reset. On other systems, the
5704 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
5705 router, though it's still delivered to local networks. Do not use it unless
5706 you fully understand how it works.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005707
Christopher Faulet46f95542019-12-20 10:07:22 +01005708http-response strict-mode { on | off }
5709
5710 This enables or disables the strict rewriting mode for following rules. It
5711 does not affect rules declared before it and it is only applicable on rules
5712 performing a rewrite on the responses. When the strict mode is enabled, any
5713 rewrite failure triggers an internal error. Otherwise, such errors are
5714 silently ignored. The purpose of the strict rewriting mode is to make some
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05005715 rewrites optional while others must be performed to continue the response
Christopher Faulet46f95542019-12-20 10:07:22 +01005716 processing.
5717
Christopher Faulet1aea50e2020-01-17 16:03:53 +01005718 By default, the strict rewriting mode is enabled. Its value is also reset
Christopher Faulet46f95542019-12-20 10:07:22 +01005719 when a ruleset evaluation ends. So, for instance, if you change the mode on
5720 the bacnkend, the default mode is restored when HAProxy starts the frontend
5721 rules evaluation.
5722
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005723http-response track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
5724http-response track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
5725http-response track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02005726
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005727 This enables tracking of sticky counters from current response. Please refer
5728 to "http-request track-sc" for a complete description. The only difference
5729 from "http-request track-sc" is the <key> sample expression can only make use
5730 of samples in response (e.g. res.*, status etc.) and samples below Layer 6
5731 (e.g. SSL-related samples, see section 7.3.4). If the sample is not
5732 supported, haproxy will fail and warn while parsing the config.
5733
5734http-response unset-var(<var-name>) [ { if | unless } <condition> ]
5735
5736 This is used to unset a variable. See "http-response set-var" for details
5737 about <var-name>.
5738
5739 Example:
5740 http-response unset-var(sess.last_redir)
5741
Baptiste Assmann5ecb77f2013-10-06 23:24:13 +02005742
Willy Tarreau30631952015-08-06 15:05:24 +02005743http-reuse { never | safe | aggressive | always }
5744 Declare how idle HTTP connections may be shared between requests
5745
5746 May be used in sections: defaults | frontend | listen | backend
5747 yes | no | yes | yes
5748
5749 By default, a connection established between haproxy and the backend server
Olivier Houchard86006a52018-12-14 19:37:49 +01005750 which is considered safe for reuse is moved back to the server's idle
5751 connections pool so that any other request can make use of it. This is the
5752 "safe" strategy below.
Willy Tarreau30631952015-08-06 15:05:24 +02005753
5754 The argument indicates the desired connection reuse strategy :
5755
Olivier Houchard86006a52018-12-14 19:37:49 +01005756 - "never" : idle connections are never shared between sessions. This mode
5757 may be enforced to cancel a different strategy inherited from
5758 a defaults section or for troubleshooting. For example, if an
5759 old bogus application considers that multiple requests over
5760 the same connection come from the same client and it is not
5761 possible to fix the application, it may be desirable to
5762 disable connection sharing in a single backend. An example of
5763 such an application could be an old haproxy using cookie
5764 insertion in tunnel mode and not checking any request past the
5765 first one.
Willy Tarreau30631952015-08-06 15:05:24 +02005766
Olivier Houchard86006a52018-12-14 19:37:49 +01005767 - "safe" : this is the default and the recommended strategy. The first
5768 request of a session is always sent over its own connection,
5769 and only subsequent requests may be dispatched over other
5770 existing connections. This ensures that in case the server
5771 closes the connection when the request is being sent, the
5772 browser can decide to silently retry it. Since it is exactly
5773 equivalent to regular keep-alive, there should be no side
5774 effects.
Willy Tarreau30631952015-08-06 15:05:24 +02005775
5776 - "aggressive" : this mode may be useful in webservices environments where
5777 all servers are not necessarily known and where it would be
5778 appreciable to deliver most first requests over existing
5779 connections. In this case, first requests are only delivered
5780 over existing connections that have been reused at least once,
5781 proving that the server correctly supports connection reuse.
5782 It should only be used when it's sure that the client can
5783 retry a failed request once in a while and where the benefit
Michael Prokop4438c602019-05-24 10:25:45 +02005784 of aggressive connection reuse significantly outweighs the
Willy Tarreau30631952015-08-06 15:05:24 +02005785 downsides of rare connection failures.
5786
5787 - "always" : this mode is only recommended when the path to the server is
5788 known for never breaking existing connections quickly after
5789 releasing them. It allows the first request of a session to be
5790 sent to an existing connection. This can provide a significant
5791 performance increase over the "safe" strategy when the backend
5792 is a cache farm, since such components tend to show a
Davor Ocelice9ed2812017-12-25 17:49:28 +01005793 consistent behavior and will benefit from the connection
Willy Tarreau30631952015-08-06 15:05:24 +02005794 sharing. It is recommended that the "http-keep-alive" timeout
5795 remains low in this mode so that no dead connections remain
5796 usable. In most cases, this will lead to the same performance
5797 gains as "aggressive" but with more risks. It should only be
5798 used when it improves the situation over "aggressive".
5799
5800 When http connection sharing is enabled, a great care is taken to respect the
Davor Ocelice9ed2812017-12-25 17:49:28 +01005801 connection properties and compatibility. Specifically :
5802 - connections made with "usesrc" followed by a client-dependent value
5803 ("client", "clientip", "hdr_ip") are marked private and never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02005804
5805 - connections sent to a server with a TLS SNI extension are marked private
Davor Ocelice9ed2812017-12-25 17:49:28 +01005806 and are never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02005807
Lukas Tribusfd9b68c2018-10-27 20:06:59 +02005808 - connections with certain bogus authentication schemes (relying on the
5809 connection) like NTLM are detected, marked private and are never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02005810
Lukas Tribuse8adfeb2019-11-06 11:50:25 +01005811 A connection pool is involved and configurable with "pool-max-conn".
Willy Tarreau30631952015-08-06 15:05:24 +02005812
5813 Note: connection reuse improves the accuracy of the "server maxconn" setting,
5814 because almost no new connection will be established while idle connections
5815 remain available. This is particularly true with the "always" strategy.
5816
5817 See also : "option http-keep-alive", "server maxconn"
5818
5819
Mark Lamourinec2247f02012-01-04 13:02:01 -05005820http-send-name-header [<header>]
5821 Add the server name to a request. Use the header string given by <header>
Mark Lamourinec2247f02012-01-04 13:02:01 -05005822 May be used in sections: defaults | frontend | listen | backend
5823 yes | no | yes | yes
Mark Lamourinec2247f02012-01-04 13:02:01 -05005824 Arguments :
Mark Lamourinec2247f02012-01-04 13:02:01 -05005825 <header> The header string to use to send the server name
5826
Willy Tarreau81bef7e2019-10-07 14:58:02 +02005827 The "http-send-name-header" statement causes the header field named <header>
5828 to be set to the name of the target server at the moment the request is about
5829 to be sent on the wire. Any existing occurrences of this header are removed.
5830 Upon retries and redispatches, the header field is updated to always reflect
5831 the server being attempted to connect to. Given that this header is modified
5832 very late in the connection setup, it may have unexpected effects on already
5833 modified headers. For example using it with transport-level header such as
5834 connection, content-length, transfer-encoding and so on will likely result in
5835 invalid requests being sent to the server. Additionally it has been reported
5836 that this directive is currently being used as a way to overwrite the Host
5837 header field in outgoing requests; while this trick has been known to work
5838 as a side effect of the feature for some time, it is not officially supported
5839 and might possibly not work anymore in a future version depending on the
5840 technical difficulties this feature induces. A long-term solution instead
5841 consists in fixing the application which required this trick so that it binds
5842 to the correct host name.
Mark Lamourinec2247f02012-01-04 13:02:01 -05005843
5844 See also : "server"
5845
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01005846id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +02005847 Set a persistent ID to a proxy.
5848 May be used in sections : defaults | frontend | listen | backend
5849 no | yes | yes | yes
5850 Arguments : none
5851
5852 Set a persistent ID for the proxy. This ID must be unique and positive.
5853 An unused ID will automatically be assigned if unset. The first assigned
5854 value will be 1. This ID is currently only returned in statistics.
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01005855
5856
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005857ignore-persist { if | unless } <condition>
5858 Declare a condition to ignore persistence
5859 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01005860 no | no | yes | yes
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005861
5862 By default, when cookie persistence is enabled, every requests containing
5863 the cookie are unconditionally persistent (assuming the target server is up
5864 and running).
5865
5866 The "ignore-persist" statement allows one to declare various ACL-based
5867 conditions which, when met, will cause a request to ignore persistence.
5868 This is sometimes useful to load balance requests for static files, which
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005869 often don't require persistence. This can also be used to fully disable
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005870 persistence for a specific User-Agent (for example, some web crawler bots).
5871
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005872 The persistence is ignored when an "if" condition is met, or unless an
5873 "unless" condition is met.
5874
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03005875 Example:
5876 acl url_static path_beg /static /images /img /css
5877 acl url_static path_end .gif .png .jpg .css .js
5878 ignore-persist if url_static
5879
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005880 See also : "force-persist", "cookie", and section 7 about ACL usage.
5881
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005882load-server-state-from-file { global | local | none }
5883 Allow seamless reload of HAProxy
5884 May be used in sections: defaults | frontend | listen | backend
5885 yes | no | yes | yes
5886
5887 This directive points HAProxy to a file where server state from previous
5888 running process has been saved. That way, when starting up, before handling
5889 traffic, the new process can apply old states to servers exactly has if no
Davor Ocelice9ed2812017-12-25 17:49:28 +01005890 reload occurred. The purpose of the "load-server-state-from-file" directive is
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005891 to tell haproxy which file to use. For now, only 2 arguments to either prevent
5892 loading state or load states from a file containing all backends and servers.
5893 The state file can be generated by running the command "show servers state"
5894 over the stats socket and redirect output.
5895
Davor Ocelice9ed2812017-12-25 17:49:28 +01005896 The format of the file is versioned and is very specific. To understand it,
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005897 please read the documentation of the "show servers state" command (chapter
Willy Tarreau1af20c72017-06-23 16:01:14 +02005898 9.3 of Management Guide).
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005899
5900 Arguments:
5901 global load the content of the file pointed by the global directive
5902 named "server-state-file".
5903
5904 local load the content of the file pointed by the directive
5905 "server-state-file-name" if set. If not set, then the backend
5906 name is used as a file name.
5907
5908 none don't load any stat for this backend
5909
5910 Notes:
Willy Tarreaue5a60682016-11-09 14:54:53 +01005911 - server's IP address is preserved across reloads by default, but the
5912 order can be changed thanks to the server's "init-addr" setting. This
5913 means that an IP address change performed on the CLI at run time will
Davor Ocelice9ed2812017-12-25 17:49:28 +01005914 be preserved, and that any change to the local resolver (e.g. /etc/hosts)
Willy Tarreaue5a60682016-11-09 14:54:53 +01005915 will possibly not have any effect if the state file is in use.
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005916
5917 - server's weight is applied from previous running process unless it has
5918 has changed between previous and new configuration files.
5919
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005920 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005921
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005922 global
5923 stats socket /tmp/socket
5924 server-state-file /tmp/server_state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005925
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005926 defaults
5927 load-server-state-from-file global
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005928
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005929 backend bk
5930 server s1 127.0.0.1:22 check weight 11
5931 server s2 127.0.0.1:22 check weight 12
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005932
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005933
5934 Then one can run :
5935
5936 socat /tmp/socket - <<< "show servers state" > /tmp/server_state
5937
5938 Content of the file /tmp/server_state would be like this:
5939
5940 1
5941 # <field names skipped for the doc example>
5942 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
5943 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
5944
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005945 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005946
5947 global
5948 stats socket /tmp/socket
5949 server-state-base /etc/haproxy/states
5950
5951 defaults
5952 load-server-state-from-file local
5953
5954 backend bk
5955 server s1 127.0.0.1:22 check weight 11
5956 server s2 127.0.0.1:22 check weight 12
5957
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005958
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005959 Then one can run :
5960
5961 socat /tmp/socket - <<< "show servers state bk" > /etc/haproxy/states/bk
5962
5963 Content of the file /etc/haproxy/states/bk would be like this:
5964
5965 1
5966 # <field names skipped for the doc example>
5967 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
5968 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
5969
5970 See also: "server-state-file", "server-state-file-name", and
5971 "show servers state"
5972
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005973
Willy Tarreau2769aa02007-12-27 18:26:09 +01005974log global
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02005975log <address> [len <length>] [format <format>] [sample <ranges>:<smp_size>]
5976 <facility> [<level> [<minlevel>]]
William Lallemand0f99e342011-10-12 17:50:54 +02005977no log
Willy Tarreau2769aa02007-12-27 18:26:09 +01005978 Enable per-instance logging of events and traffic.
5979 May be used in sections : defaults | frontend | listen | backend
5980 yes | yes | yes | yes
William Lallemand0f99e342011-10-12 17:50:54 +02005981
5982 Prefix :
5983 no should be used when the logger list must be flushed. For example,
5984 if you don't want to inherit from the default logger list. This
5985 prefix does not allow arguments.
5986
Willy Tarreau2769aa02007-12-27 18:26:09 +01005987 Arguments :
5988 global should be used when the instance's logging parameters are the
5989 same as the global ones. This is the most common usage. "global"
5990 replaces <address>, <facility> and <level> with those of the log
5991 entries found in the "global" section. Only one "log global"
5992 statement may be used per instance, and this form takes no other
5993 parameter.
5994
5995 <address> indicates where to send the logs. It takes the same format as
5996 for the "global" section's logs, and can be one of :
5997
5998 - An IPv4 address optionally followed by a colon (':') and a UDP
5999 port. If no port is specified, 514 is used by default (the
6000 standard syslog port).
6001
David du Colombier24bb5f52011-03-17 10:40:23 +01006002 - An IPv6 address followed by a colon (':') and optionally a UDP
6003 port. If no port is specified, 514 is used by default (the
6004 standard syslog port).
6005
Willy Tarreau2769aa02007-12-27 18:26:09 +01006006 - A filesystem path to a UNIX domain socket, keeping in mind
6007 considerations for chroot (be sure the path is accessible
6008 inside the chroot) and uid/gid (be sure the path is
Davor Ocelice9ed2812017-12-25 17:49:28 +01006009 appropriately writable).
Willy Tarreau2769aa02007-12-27 18:26:09 +01006010
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01006011 - A file descriptor number in the form "fd@<number>", which may
6012 point to a pipe, terminal, or socket. In this case unbuffered
6013 logs are used and one writev() call per log is performed. This
6014 is a bit expensive but acceptable for most workloads. Messages
6015 sent this way will not be truncated but may be dropped, in
6016 which case the DroppedLogs counter will be incremented. The
6017 writev() call is atomic even on pipes for messages up to
6018 PIPE_BUF size, which POSIX recommends to be at least 512 and
6019 which is 4096 bytes on most modern operating systems. Any
6020 larger message may be interleaved with messages from other
6021 processes. Exceptionally for debugging purposes the file
6022 descriptor may also be directed to a file, but doing so will
6023 significantly slow haproxy down as non-blocking calls will be
6024 ignored. Also there will be no way to purge nor rotate this
6025 file without restarting the process. Note that the configured
6026 syslog format is preserved, so the output is suitable for use
Willy Tarreauc1b06452018-11-12 11:57:56 +01006027 with a TCP syslog server. See also the "short" and "raw"
6028 formats below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01006029
6030 - "stdout" / "stderr", which are respectively aliases for "fd@1"
6031 and "fd@2", see above.
6032
Willy Tarreauc046d162019-08-30 15:24:59 +02006033 - A ring buffer in the form "ring@<name>", which will correspond
6034 to an in-memory ring buffer accessible over the CLI using the
6035 "show events" command, which will also list existing rings and
6036 their sizes. Such buffers are lost on reload or restart but
6037 when used as a complement this can help troubleshooting by
6038 having the logs instantly available.
6039
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01006040 You may want to reference some environment variables in the
6041 address parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01006042
Willy Tarreau18324f52014-06-27 18:10:07 +02006043 <length> is an optional maximum line length. Log lines larger than this
6044 value will be truncated before being sent. The reason is that
6045 syslog servers act differently on log line length. All servers
6046 support the default value of 1024, but some servers simply drop
6047 larger lines while others do log them. If a server supports long
6048 lines, it may make sense to set this value here in order to avoid
6049 truncating long lines. Similarly, if a server drops long lines,
6050 it is preferable to truncate them before sending them. Accepted
6051 values are 80 to 65535 inclusive. The default value of 1024 is
6052 generally fine for all standard usages. Some specific cases of
Davor Ocelice9ed2812017-12-25 17:49:28 +01006053 long captures or JSON-formatted logs may require larger values.
Willy Tarreau18324f52014-06-27 18:10:07 +02006054
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02006055 <ranges> A list of comma-separated ranges to identify the logs to sample.
6056 This is used to balance the load of the logs to send to the log
6057 server. The limits of the ranges cannot be null. They are numbered
6058 from 1. The size or period (in number of logs) of the sample must
6059 be set with <sample_size> parameter.
6060
6061 <sample_size>
6062 The size of the sample in number of logs to consider when balancing
6063 their logging loads. It is used to balance the load of the logs to
6064 send to the syslog server. This size must be greater or equal to the
6065 maximum of the high limits of the ranges.
6066 (see also <ranges> parameter).
6067
Willy Tarreauadb345d2018-11-12 07:56:13 +01006068 <format> is the log format used when generating syslog messages. It may be
6069 one of the following :
6070
6071 rfc3164 The RFC3164 syslog message format. This is the default.
6072 (https://tools.ietf.org/html/rfc3164)
6073
6074 rfc5424 The RFC5424 syslog message format.
6075 (https://tools.ietf.org/html/rfc5424)
6076
Willy Tarreaue8746a02018-11-12 08:45:00 +01006077 short A message containing only a level between angle brackets such as
6078 '<3>', followed by the text. The PID, date, time, process name
6079 and system name are omitted. This is designed to be used with a
6080 local log server. This format is compatible with what the
6081 systemd logger consumes.
6082
Willy Tarreauc1b06452018-11-12 11:57:56 +01006083 raw A message containing only the text. The level, PID, date, time,
6084 process name and system name are omitted. This is designed to
6085 be used in containers or during development, where the severity
6086 only depends on the file descriptor used (stdout/stderr).
6087
Willy Tarreau2769aa02007-12-27 18:26:09 +01006088 <facility> must be one of the 24 standard syslog facilities :
6089
Willy Tarreaue8746a02018-11-12 08:45:00 +01006090 kern user mail daemon auth syslog lpr news
6091 uucp cron auth2 ftp ntp audit alert cron2
6092 local0 local1 local2 local3 local4 local5 local6 local7
6093
Willy Tarreauc1b06452018-11-12 11:57:56 +01006094 Note that the facility is ignored for the "short" and "raw"
6095 formats, but still required as a positional field. It is
6096 recommended to use "daemon" in this case to make it clear that
6097 it's only supposed to be used locally.
Willy Tarreau2769aa02007-12-27 18:26:09 +01006098
6099 <level> is optional and can be specified to filter outgoing messages. By
6100 default, all messages are sent. If a level is specified, only
6101 messages with a severity at least as important as this level
Willy Tarreauf7edefa2009-05-10 17:20:05 +02006102 will be sent. An optional minimum level can be specified. If it
6103 is set, logs emitted with a more severe level than this one will
6104 be capped to this level. This is used to avoid sending "emerg"
6105 messages on all terminals on some default syslog configurations.
6106 Eight levels are known :
Willy Tarreau2769aa02007-12-27 18:26:09 +01006107
6108 emerg alert crit err warning notice info debug
6109
William Lallemand0f99e342011-10-12 17:50:54 +02006110 It is important to keep in mind that it is the frontend which decides what to
6111 log from a connection, and that in case of content switching, the log entries
6112 from the backend will be ignored. Connections are logged at level "info".
Willy Tarreaucc6c8912009-02-22 10:53:55 +01006113
6114 However, backend log declaration define how and where servers status changes
6115 will be logged. Level "notice" will be used to indicate a server going up,
6116 "warning" will be used for termination signals and definitive service
6117 termination, and "alert" will be used for when a server goes down.
6118
6119 Note : According to RFC3164, messages are truncated to 1024 bytes before
6120 being emitted.
Willy Tarreau2769aa02007-12-27 18:26:09 +01006121
6122 Example :
6123 log global
Willy Tarreauc1b06452018-11-12 11:57:56 +01006124 log stdout format short daemon # send log to systemd
6125 log stdout format raw daemon # send everything to stdout
6126 log stderr format raw daemon notice # send important events to stderr
Willy Tarreauf7edefa2009-05-10 17:20:05 +02006127 log 127.0.0.1:514 local0 notice # only send important events
6128 log 127.0.0.1:514 local0 notice notice # same but limit output level
William Lallemandb2f07452015-05-12 14:27:13 +02006129 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
Willy Tarreaudad36a32013-03-11 01:20:04 +01006130
Willy Tarreau2769aa02007-12-27 18:26:09 +01006131
William Lallemand48940402012-01-30 16:47:22 +01006132log-format <string>
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01006133 Specifies the log format string to use for traffic logs
6134 May be used in sections: defaults | frontend | listen | backend
6135 yes | yes | yes | no
William Lallemand48940402012-01-30 16:47:22 +01006136
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01006137 This directive specifies the log format string that will be used for all logs
6138 resulting from traffic passing through the frontend using this line. If the
6139 directive is used in a defaults section, all subsequent frontends will use
6140 the same log format. Please see section 8.2.4 which covers the log format
6141 string in depth.
William Lallemand48940402012-01-30 16:47:22 +01006142
Guillaume de Lafond29f45602017-03-31 19:52:15 +02006143 "log-format" directive overrides previous "option tcplog", "log-format" and
6144 "option httplog" directives.
6145
Dragan Dosen7ad31542015-09-28 17:16:47 +02006146log-format-sd <string>
6147 Specifies the RFC5424 structured-data log format string
6148 May be used in sections: defaults | frontend | listen | backend
6149 yes | yes | yes | no
6150
6151 This directive specifies the RFC5424 structured-data log format string that
6152 will be used for all logs resulting from traffic passing through the frontend
6153 using this line. If the directive is used in a defaults section, all
6154 subsequent frontends will use the same log format. Please see section 8.2.4
6155 which covers the log format string in depth.
6156
6157 See https://tools.ietf.org/html/rfc5424#section-6.3 for more information
6158 about the RFC5424 structured-data part.
6159
6160 Note : This log format string will be used only for loggers that have set
6161 log format to "rfc5424".
6162
6163 Example :
6164 log-format-sd [exampleSDID@1234\ bytes=\"%B\"\ status=\"%ST\"]
6165
6166
Willy Tarreau094af4e2015-01-07 15:03:42 +01006167log-tag <string>
6168 Specifies the log tag to use for all outgoing logs
6169 May be used in sections: defaults | frontend | listen | backend
6170 yes | yes | yes | yes
6171
6172 Sets the tag field in the syslog header to this string. It defaults to the
6173 log-tag set in the global section, otherwise the program name as launched
6174 from the command line, which usually is "haproxy". Sometimes it can be useful
6175 to differentiate between multiple processes running on the same host, or to
6176 differentiate customer instances running in the same process. In the backend,
6177 logs about servers up/down will use this tag. As a hint, it can be convenient
6178 to set a log-tag related to a hosted customer in a defaults section then put
6179 all the frontends and backends for that customer, then start another customer
6180 in a new defaults section. See also the global "log-tag" directive.
Willy Tarreau2769aa02007-12-27 18:26:09 +01006181
Willy Tarreauc35362a2014-04-25 13:58:37 +02006182max-keep-alive-queue <value>
6183 Set the maximum server queue size for maintaining keep-alive connections
6184 May be used in sections: defaults | frontend | listen | backend
6185 yes | no | yes | yes
6186
6187 HTTP keep-alive tries to reuse the same server connection whenever possible,
6188 but sometimes it can be counter-productive, for example if a server has a lot
6189 of connections while other ones are idle. This is especially true for static
6190 servers.
6191
6192 The purpose of this setting is to set a threshold on the number of queued
6193 connections at which haproxy stops trying to reuse the same server and prefers
6194 to find another one. The default value, -1, means there is no limit. A value
6195 of zero means that keep-alive requests will never be queued. For very close
6196 servers which can be reached with a low latency and which are not sensible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01006197 breaking keep-alive, a low value is recommended (e.g. local static server can
Willy Tarreauc35362a2014-04-25 13:58:37 +02006198 use a value of 10 or less). For remote servers suffering from a high latency,
6199 higher values might be needed to cover for the latency and/or the cost of
6200 picking a different server.
6201
6202 Note that this has no impact on responses which are maintained to the same
6203 server consecutively to a 401 response. They will still go to the same server
6204 even if they have to be queued.
6205
6206 See also : "option http-server-close", "option prefer-last-server", server
6207 "maxconn" and cookie persistence.
6208
Olivier Houcharda4d4fdf2018-12-14 19:27:06 +01006209max-session-srv-conns <nb>
6210 Set the maximum number of outgoing connections we can keep idling for a given
6211 client session. The default is 5 (it precisely equals MAX_SRV_LIST which is
6212 defined at build time).
Willy Tarreauc35362a2014-04-25 13:58:37 +02006213
Willy Tarreau2769aa02007-12-27 18:26:09 +01006214maxconn <conns>
6215 Fix the maximum number of concurrent connections on a frontend
6216 May be used in sections : defaults | frontend | listen | backend
6217 yes | yes | yes | no
6218 Arguments :
6219 <conns> is the maximum number of concurrent connections the frontend will
6220 accept to serve. Excess connections will be queued by the system
6221 in the socket's listen queue and will be served once a connection
6222 closes.
6223
6224 If the system supports it, it can be useful on big sites to raise this limit
6225 very high so that haproxy manages connection queues, instead of leaving the
6226 clients with unanswered connection attempts. This value should not exceed the
6227 global maxconn. Also, keep in mind that a connection contains two buffers
Baptiste Assmann79fb45d2016-03-06 23:34:31 +01006228 of tune.bufsize (16kB by default) each, as well as some other data resulting
6229 in about 33 kB of RAM being consumed per established connection. That means
6230 that a medium system equipped with 1GB of RAM can withstand around
6231 20000-25000 concurrent connections if properly tuned.
Willy Tarreau2769aa02007-12-27 18:26:09 +01006232
6233 Also, when <conns> is set to large values, it is possible that the servers
6234 are not sized to accept such loads, and for this reason it is generally wise
6235 to assign them some reasonable connection limits.
6236
Willy Tarreauc8d5b952019-02-27 17:25:52 +01006237 When this value is set to zero, which is the default, the global "maxconn"
6238 value is used.
Vincent Bernat6341be52012-06-27 17:18:30 +02006239
Willy Tarreau2769aa02007-12-27 18:26:09 +01006240 See also : "server", global section's "maxconn", "fullconn"
6241
6242
6243mode { tcp|http|health }
6244 Set the running mode or protocol of the instance
6245 May be used in sections : defaults | frontend | listen | backend
6246 yes | yes | yes | yes
6247 Arguments :
6248 tcp The instance will work in pure TCP mode. A full-duplex connection
6249 will be established between clients and servers, and no layer 7
6250 examination will be performed. This is the default mode. It
6251 should be used for SSL, SSH, SMTP, ...
6252
6253 http The instance will work in HTTP mode. The client request will be
6254 analyzed in depth before connecting to any server. Any request
6255 which is not RFC-compliant will be rejected. Layer 7 filtering,
6256 processing and switching will be possible. This is the mode which
6257 brings HAProxy most of its value.
6258
6259 health The instance will work in "health" mode. It will just reply "OK"
Willy Tarreau82569f92012-09-27 23:48:56 +02006260 to incoming connections and close the connection. Alternatively,
6261 If the "httpchk" option is set, "HTTP/1.0 200 OK" will be sent
6262 instead. Nothing will be logged in either case. This mode is used
6263 to reply to external components health checks. This mode is
6264 deprecated and should not be used anymore as it is possible to do
6265 the same and even better by combining TCP or HTTP modes with the
6266 "monitor" keyword.
Willy Tarreau2769aa02007-12-27 18:26:09 +01006267
Cyril Bonté108cf6e2012-04-21 23:30:29 +02006268 When doing content switching, it is mandatory that the frontend and the
6269 backend are in the same mode (generally HTTP), otherwise the configuration
6270 will be refused.
Willy Tarreau2769aa02007-12-27 18:26:09 +01006271
Cyril Bonté108cf6e2012-04-21 23:30:29 +02006272 Example :
Willy Tarreau2769aa02007-12-27 18:26:09 +01006273 defaults http_instances
6274 mode http
6275
Cyril Bonté108cf6e2012-04-21 23:30:29 +02006276 See also : "monitor", "monitor-net"
Willy Tarreau2769aa02007-12-27 18:26:09 +01006277
Willy Tarreau0ba27502007-12-24 16:55:16 +01006278
Cyril Bontéf0c60612010-02-06 14:44:47 +01006279monitor fail { if | unless } <condition>
Willy Tarreau2769aa02007-12-27 18:26:09 +01006280 Add a condition to report a failure to a monitor HTTP request.
Willy Tarreau0ba27502007-12-24 16:55:16 +01006281 May be used in sections : defaults | frontend | listen | backend
6282 no | yes | yes | no
Willy Tarreau0ba27502007-12-24 16:55:16 +01006283 Arguments :
6284 if <cond> the monitor request will fail if the condition is satisfied,
6285 and will succeed otherwise. The condition should describe a
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006286 combined test which must induce a failure if all conditions
Willy Tarreau0ba27502007-12-24 16:55:16 +01006287 are met, for instance a low number of servers both in a
6288 backend and its backup.
6289
6290 unless <cond> the monitor request will succeed only if the condition is
6291 satisfied, and will fail otherwise. Such a condition may be
6292 based on a test on the presence of a minimum number of active
6293 servers in a list of backends.
6294
6295 This statement adds a condition which can force the response to a monitor
6296 request to report a failure. By default, when an external component queries
6297 the URI dedicated to monitoring, a 200 response is returned. When one of the
6298 conditions above is met, haproxy will return 503 instead of 200. This is
6299 very useful to report a site failure to an external component which may base
6300 routing advertisements between multiple sites on the availability reported by
6301 haproxy. In this case, one would rely on an ACL involving the "nbsrv"
Willy Tarreauae94d4d2011-05-11 16:28:49 +02006302 criterion. Note that "monitor fail" only works in HTTP mode. Both status
6303 messages may be tweaked using "errorfile" or "errorloc" if needed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01006304
6305 Example:
6306 frontend www
Willy Tarreau2769aa02007-12-27 18:26:09 +01006307 mode http
Willy Tarreau0ba27502007-12-24 16:55:16 +01006308 acl site_dead nbsrv(dynamic) lt 2
6309 acl site_dead nbsrv(static) lt 2
6310 monitor-uri /site_alive
6311 monitor fail if site_dead
6312
Willy Tarreauae94d4d2011-05-11 16:28:49 +02006313 See also : "monitor-net", "monitor-uri", "errorfile", "errorloc"
Willy Tarreau2769aa02007-12-27 18:26:09 +01006314
6315
6316monitor-net <source>
6317 Declare a source network which is limited to monitor requests
6318 May be used in sections : defaults | frontend | listen | backend
6319 yes | yes | yes | no
6320 Arguments :
6321 <source> is the source IPv4 address or network which will only be able to
6322 get monitor responses to any request. It can be either an IPv4
6323 address, a host name, or an address followed by a slash ('/')
6324 followed by a mask.
6325
6326 In TCP mode, any connection coming from a source matching <source> will cause
6327 the connection to be immediately closed without any log. This allows another
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006328 equipment to probe the port and verify that it is still listening, without
Willy Tarreau2769aa02007-12-27 18:26:09 +01006329 forwarding the connection to a remote server.
6330
6331 In HTTP mode, a connection coming from a source matching <source> will be
6332 accepted, the following response will be sent without waiting for a request,
6333 then the connection will be closed : "HTTP/1.0 200 OK". This is normally
6334 enough for any front-end HTTP probe to detect that the service is UP and
Willy Tarreau82569f92012-09-27 23:48:56 +02006335 running without forwarding the request to a backend server. Note that this
6336 response is sent in raw format, without any transformation. This is important
6337 as it means that it will not be SSL-encrypted on SSL listeners.
Willy Tarreau2769aa02007-12-27 18:26:09 +01006338
Willy Tarreau82569f92012-09-27 23:48:56 +02006339 Monitor requests are processed very early, just after tcp-request connection
6340 ACLs which are the only ones able to block them. These connections are short
6341 lived and never wait for any data from the client. They cannot be logged, and
6342 it is the intended purpose. They are only used to report HAProxy's health to
6343 an upper component, nothing more. Please note that "monitor fail" rules do
6344 not apply to connections intercepted by "monitor-net".
Willy Tarreau2769aa02007-12-27 18:26:09 +01006345
Willy Tarreau95cd2832010-03-04 23:36:33 +01006346 Last, please note that only one "monitor-net" statement can be specified in
6347 a frontend. If more than one is found, only the last one will be considered.
Cyril Bonté108cf6e2012-04-21 23:30:29 +02006348
Willy Tarreau2769aa02007-12-27 18:26:09 +01006349 Example :
6350 # addresses .252 and .253 are just probing us.
6351 frontend www
6352 monitor-net 192.168.0.252/31
6353
6354 See also : "monitor fail", "monitor-uri"
6355
6356
6357monitor-uri <uri>
6358 Intercept a URI used by external components' monitor requests
6359 May be used in sections : defaults | frontend | listen | backend
6360 yes | yes | yes | no
6361 Arguments :
6362 <uri> is the exact URI which we want to intercept to return HAProxy's
6363 health status instead of forwarding the request.
6364
6365 When an HTTP request referencing <uri> will be received on a frontend,
6366 HAProxy will not forward it nor log it, but instead will return either
6367 "HTTP/1.0 200 OK" or "HTTP/1.0 503 Service unavailable", depending on failure
6368 conditions defined with "monitor fail". This is normally enough for any
6369 front-end HTTP probe to detect that the service is UP and running without
6370 forwarding the request to a backend server. Note that the HTTP method, the
6371 version and all headers are ignored, but the request must at least be valid
6372 at the HTTP level. This keyword may only be used with an HTTP-mode frontend.
6373
Willy Tarreau721d8e02017-12-01 18:25:08 +01006374 Monitor requests are processed very early, just after the request is parsed
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02006375 and even before any "http-request". The only rulesets applied before are the
6376 tcp-request ones. They cannot be logged either, and it is the intended
6377 purpose. They are only used to report HAProxy's health to an upper component,
6378 nothing more. However, it is possible to add any number of conditions using
6379 "monitor fail" and ACLs so that the result can be adjusted to whatever check
6380 can be imagined (most often the number of available servers in a backend).
Willy Tarreau2769aa02007-12-27 18:26:09 +01006381
Christopher Faulet6072beb2020-02-18 15:34:58 +01006382 Note: if <uri> starts by a slash ('/'), the matching is performed against the
6383 request's path instead of the request's uri. It is a workaround to let
6384 the HTTP/2 requests match the monitor-uri. Indeed, in HTTP/2, clients
6385 are encouraged to send absolute URIs only.
6386
Willy Tarreau2769aa02007-12-27 18:26:09 +01006387 Example :
6388 # Use /haproxy_test to report haproxy's status
6389 frontend www
6390 mode http
6391 monitor-uri /haproxy_test
6392
6393 See also : "monitor fail", "monitor-net"
6394
Willy Tarreau0ba27502007-12-24 16:55:16 +01006395
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006396option abortonclose
6397no option abortonclose
6398 Enable or disable early dropping of aborted requests pending in queues.
6399 May be used in sections : defaults | frontend | listen | backend
6400 yes | no | yes | yes
6401 Arguments : none
6402
6403 In presence of very high loads, the servers will take some time to respond.
6404 The per-instance connection queue will inflate, and the response time will
6405 increase respective to the size of the queue times the average per-session
6406 response time. When clients will wait for more than a few seconds, they will
Willy Tarreau198a7442008-01-17 12:05:32 +01006407 often hit the "STOP" button on their browser, leaving a useless request in
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006408 the queue, and slowing down other users, and the servers as well, because the
6409 request will eventually be served, then aborted at the first error
6410 encountered while delivering the response.
6411
6412 As there is no way to distinguish between a full STOP and a simple output
6413 close on the client side, HTTP agents should be conservative and consider
6414 that the client might only have closed its output channel while waiting for
6415 the response. However, this introduces risks of congestion when lots of users
6416 do the same, and is completely useless nowadays because probably no client at
6417 all will close the session while waiting for the response. Some HTTP agents
Davor Ocelice9ed2812017-12-25 17:49:28 +01006418 support this behavior (Squid, Apache, HAProxy), and others do not (TUX, most
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006419 hardware-based load balancers). So the probability for a closed input channel
Willy Tarreau198a7442008-01-17 12:05:32 +01006420 to represent a user hitting the "STOP" button is close to 100%, and the risk
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006421 of being the single component to break rare but valid traffic is extremely
6422 low, which adds to the temptation to be able to abort a session early while
6423 still not served and not pollute the servers.
6424
Davor Ocelice9ed2812017-12-25 17:49:28 +01006425 In HAProxy, the user can choose the desired behavior using the option
6426 "abortonclose". By default (without the option) the behavior is HTTP
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006427 compliant and aborted requests will be served. But when the option is
6428 specified, a session with an incoming channel closed will be aborted while
6429 it is still possible, either pending in the queue for a connection slot, or
6430 during the connection establishment if the server has not yet acknowledged
6431 the connection request. This considerably reduces the queue size and the load
6432 on saturated servers when users are tempted to click on STOP, which in turn
Willy Tarreaud72758d2010-01-12 10:42:19 +01006433 reduces the response time for other users.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006434
6435 If this option has been enabled in a "defaults" section, it can be disabled
6436 in a specific instance by prepending the "no" keyword before it.
6437
6438 See also : "timeout queue" and server's "maxconn" and "maxqueue" parameters
6439
6440
Willy Tarreau4076a152009-04-02 15:18:36 +02006441option accept-invalid-http-request
6442no option accept-invalid-http-request
6443 Enable or disable relaxing of HTTP request parsing
6444 May be used in sections : defaults | frontend | listen | backend
6445 yes | yes | yes | no
6446 Arguments : none
6447
Willy Tarreau91852eb2015-05-01 13:26:00 +02006448 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02006449 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01006450 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02006451 forbidden characters are essentially used to build attacks exploiting server
6452 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
6453 server will emit invalid header names for whatever reason (configuration,
6454 implementation) and the issue will not be immediately fixed. In such a case,
6455 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau422246e2012-01-07 23:54:13 +01006456 even if that does not make sense, by specifying this option. Similarly, the
6457 list of characters allowed to appear in a URI is well defined by RFC3986, and
6458 chars 0-31, 32 (space), 34 ('"'), 60 ('<'), 62 ('>'), 92 ('\'), 94 ('^'), 96
6459 ('`'), 123 ('{'), 124 ('|'), 125 ('}'), 127 (delete) and anything above are
Davor Ocelice9ed2812017-12-25 17:49:28 +01006460 not allowed at all. HAProxy always blocks a number of them (0..32, 127). The
Willy Tarreau91852eb2015-05-01 13:26:00 +02006461 remaining ones are blocked by default unless this option is enabled. This
Willy Tarreau13317662015-05-01 13:47:08 +02006462 option also relaxes the test on the HTTP version, it allows HTTP/0.9 requests
6463 to pass through (no version specified) and multiple digits for both the major
6464 and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02006465
6466 This option should never be enabled by default as it hides application bugs
6467 and open security breaches. It should only be deployed after a problem has
6468 been confirmed.
6469
6470 When this option is enabled, erroneous header names will still be accepted in
6471 requests, but the complete request will be captured in order to permit later
Willy Tarreau422246e2012-01-07 23:54:13 +01006472 analysis using the "show errors" request on the UNIX stats socket. Similarly,
6473 requests containing invalid chars in the URI part will be logged. Doing this
Willy Tarreau4076a152009-04-02 15:18:36 +02006474 also helps confirming that the issue has been solved.
6475
6476 If this option has been enabled in a "defaults" section, it can be disabled
6477 in a specific instance by prepending the "no" keyword before it.
6478
6479 See also : "option accept-invalid-http-response" and "show errors" on the
6480 stats socket.
6481
6482
6483option accept-invalid-http-response
6484no option accept-invalid-http-response
6485 Enable or disable relaxing of HTTP response parsing
6486 May be used in sections : defaults | frontend | listen | backend
6487 yes | no | yes | yes
6488 Arguments : none
6489
Willy Tarreau91852eb2015-05-01 13:26:00 +02006490 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02006491 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01006492 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02006493 forbidden characters are essentially used to build attacks exploiting server
6494 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
6495 server will emit invalid header names for whatever reason (configuration,
6496 implementation) and the issue will not be immediately fixed. In such a case,
6497 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau91852eb2015-05-01 13:26:00 +02006498 even if that does not make sense, by specifying this option. This option also
6499 relaxes the test on the HTTP version format, it allows multiple digits for
6500 both the major and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02006501
6502 This option should never be enabled by default as it hides application bugs
6503 and open security breaches. It should only be deployed after a problem has
6504 been confirmed.
6505
6506 When this option is enabled, erroneous header names will still be accepted in
6507 responses, but the complete response will be captured in order to permit
6508 later analysis using the "show errors" request on the UNIX stats socket.
6509 Doing this also helps confirming that the issue has been solved.
6510
6511 If this option has been enabled in a "defaults" section, it can be disabled
6512 in a specific instance by prepending the "no" keyword before it.
6513
6514 See also : "option accept-invalid-http-request" and "show errors" on the
6515 stats socket.
6516
6517
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006518option allbackups
6519no option allbackups
6520 Use either all backup servers at a time or only the first one
6521 May be used in sections : defaults | frontend | listen | backend
6522 yes | no | yes | yes
6523 Arguments : none
6524
6525 By default, the first operational backup server gets all traffic when normal
6526 servers are all down. Sometimes, it may be preferred to use multiple backups
6527 at once, because one will not be enough. When "option allbackups" is enabled,
6528 the load balancing will be performed among all backup servers when all normal
6529 ones are unavailable. The same load balancing algorithm will be used and the
6530 servers' weights will be respected. Thus, there will not be any priority
6531 order between the backup servers anymore.
6532
6533 This option is mostly used with static server farms dedicated to return a
6534 "sorry" page when an application is completely offline.
6535
6536 If this option has been enabled in a "defaults" section, it can be disabled
6537 in a specific instance by prepending the "no" keyword before it.
6538
6539
6540option checkcache
6541no option checkcache
Godbach7056a352013-12-11 20:01:07 +08006542 Analyze all server responses and block responses with cacheable cookies
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006543 May be used in sections : defaults | frontend | listen | backend
6544 yes | no | yes | yes
6545 Arguments : none
6546
6547 Some high-level frameworks set application cookies everywhere and do not
6548 always let enough control to the developer to manage how the responses should
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006549 be cached. When a session cookie is returned on a cacheable object, there is a
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006550 high risk of session crossing or stealing between users traversing the same
6551 caches. In some situations, it is better to block the response than to let
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02006552 some sensitive session information go in the wild.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006553
6554 The option "checkcache" enables deep inspection of all server responses for
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006555 strict compliance with HTTP specification in terms of cacheability. It
Willy Tarreau198a7442008-01-17 12:05:32 +01006556 carefully checks "Cache-control", "Pragma" and "Set-cookie" headers in server
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006557 response to check if there's a risk of caching a cookie on a client-side
6558 proxy. When this option is enabled, the only responses which can be delivered
Willy Tarreau198a7442008-01-17 12:05:32 +01006559 to the client are :
Davor Ocelice9ed2812017-12-25 17:49:28 +01006560 - all those without "Set-Cookie" header;
Willy Tarreauc55ddce2017-12-21 11:41:38 +01006561 - all those with a return code other than 200, 203, 204, 206, 300, 301,
6562 404, 405, 410, 414, 501, provided that the server has not set a
Davor Ocelice9ed2812017-12-25 17:49:28 +01006563 "Cache-control: public" header field;
Willy Tarreau24ea0bc2017-12-21 11:32:55 +01006564 - all those that result from a request using a method other than GET, HEAD,
6565 OPTIONS, TRACE, provided that the server has not set a 'Cache-Control:
Davor Ocelice9ed2812017-12-25 17:49:28 +01006566 public' header field;
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006567 - those with a 'Pragma: no-cache' header
6568 - those with a 'Cache-control: private' header
6569 - those with a 'Cache-control: no-store' header
6570 - those with a 'Cache-control: max-age=0' header
6571 - those with a 'Cache-control: s-maxage=0' header
6572 - those with a 'Cache-control: no-cache' header
6573 - those with a 'Cache-control: no-cache="set-cookie"' header
6574 - those with a 'Cache-control: no-cache="set-cookie,' header
6575 (allowing other fields after set-cookie)
6576
6577 If a response doesn't respect these requirements, then it will be blocked
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02006578 just as if it was from an "http-response deny" rule, with an "HTTP 502 bad
6579 gateway". The session state shows "PH--" meaning that the proxy blocked the
6580 response during headers processing. Additionally, an alert will be sent in
6581 the logs so that admins are informed that there's something to be fixed.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006582
6583 Due to the high impact on the application, the application should be tested
6584 in depth with the option enabled before going to production. It is also a
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01006585 good practice to always activate it during tests, even if it is not used in
Davor Ocelice9ed2812017-12-25 17:49:28 +01006586 production, as it will report potentially dangerous application behaviors.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006587
6588 If this option has been enabled in a "defaults" section, it can be disabled
6589 in a specific instance by prepending the "no" keyword before it.
6590
6591
6592option clitcpka
6593no option clitcpka
6594 Enable or disable the sending of TCP keepalive packets on the client side
6595 May be used in sections : defaults | frontend | listen | backend
6596 yes | yes | yes | no
6597 Arguments : none
6598
6599 When there is a firewall or any session-aware component between a client and
6600 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01006601 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006602 components decides to expire a session which has remained idle for too long.
6603
6604 Enabling socket-level TCP keep-alives makes the system regularly send packets
6605 to the other end of the connection, leaving it active. The delay between
6606 keep-alive probes is controlled by the system only and depends both on the
6607 operating system and its tuning parameters.
6608
6609 It is important to understand that keep-alive packets are neither emitted nor
6610 received at the application level. It is only the network stacks which sees
6611 them. For this reason, even if one side of the proxy already uses keep-alives
6612 to maintain its connection alive, those keep-alive packets will not be
6613 forwarded to the other side of the proxy.
6614
6615 Please note that this has nothing to do with HTTP keep-alive.
6616
6617 Using option "clitcpka" enables the emission of TCP keep-alive probes on the
6618 client side of a connection, which should help when session expirations are
6619 noticed between HAProxy and a client.
6620
6621 If this option has been enabled in a "defaults" section, it can be disabled
6622 in a specific instance by prepending the "no" keyword before it.
6623
6624 See also : "option srvtcpka", "option tcpka"
6625
6626
Willy Tarreau0ba27502007-12-24 16:55:16 +01006627option contstats
6628 Enable continuous traffic statistics updates
6629 May be used in sections : defaults | frontend | listen | backend
6630 yes | yes | yes | no
6631 Arguments : none
6632
6633 By default, counters used for statistics calculation are incremented
6634 only when a session finishes. It works quite well when serving small
6635 objects, but with big ones (for example large images or archives) or
6636 with A/V streaming, a graph generated from haproxy counters looks like
Willy Tarreaudef0d222016-11-08 22:03:00 +01006637 a hedgehog. With this option enabled counters get incremented frequently
6638 along the session, typically every 5 seconds, which is often enough to
6639 produce clean graphs. Recounting touches a hotpath directly so it is not
6640 not enabled by default, as it can cause a lot of wakeups for very large
6641 session counts and cause a small performance drop.
Willy Tarreau0ba27502007-12-24 16:55:16 +01006642
6643
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006644option dontlog-normal
6645no option dontlog-normal
6646 Enable or disable logging of normal, successful connections
6647 May be used in sections : defaults | frontend | listen | backend
6648 yes | yes | yes | no
6649 Arguments : none
6650
6651 There are large sites dealing with several thousand connections per second
6652 and for which logging is a major pain. Some of them are even forced to turn
6653 logs off and cannot debug production issues. Setting this option ensures that
6654 normal connections, those which experience no error, no timeout, no retry nor
6655 redispatch, will not be logged. This leaves disk space for anomalies. In HTTP
6656 mode, the response status code is checked and return codes 5xx will still be
6657 logged.
6658
6659 It is strongly discouraged to use this option as most of the time, the key to
6660 complex issues is in the normal logs which will not be logged here. If you
6661 need to separate logs, see the "log-separate-errors" option instead.
6662
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006663 See also : "log", "dontlognull", "log-separate-errors" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006664 logging.
6665
6666
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006667option dontlognull
6668no option dontlognull
6669 Enable or disable logging of null connections
6670 May be used in sections : defaults | frontend | listen | backend
6671 yes | yes | yes | no
6672 Arguments : none
6673
6674 In certain environments, there are components which will regularly connect to
6675 various systems to ensure that they are still alive. It can be the case from
6676 another load balancer as well as from monitoring systems. By default, even a
6677 simple port probe or scan will produce a log. If those connections pollute
6678 the logs too much, it is possible to enable option "dontlognull" to indicate
6679 that a connection on which no data has been transferred will not be logged,
Willy Tarreau0f228a02015-05-01 15:37:53 +02006680 which typically corresponds to those probes. Note that errors will still be
6681 returned to the client and accounted for in the stats. If this is not what is
6682 desired, option http-ignore-probes can be used instead.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006683
6684 It is generally recommended not to use this option in uncontrolled
Davor Ocelice9ed2812017-12-25 17:49:28 +01006685 environments (e.g. internet), otherwise scans and other malicious activities
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006686 would not be logged.
6687
6688 If this option has been enabled in a "defaults" section, it can be disabled
6689 in a specific instance by prepending the "no" keyword before it.
6690
Willy Tarreau0f228a02015-05-01 15:37:53 +02006691 See also : "log", "http-ignore-probes", "monitor-net", "monitor-uri", and
6692 section 8 about logging.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006693
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006694
Willy Tarreau87cf5142011-08-19 22:57:24 +02006695option forwardfor [ except <network> ] [ header <name> ] [ if-none ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01006696 Enable insertion of the X-Forwarded-For header to requests sent to servers
6697 May be used in sections : defaults | frontend | listen | backend
6698 yes | yes | yes | yes
6699 Arguments :
6700 <network> is an optional argument used to disable this option for sources
6701 matching <network>
Ross Westaf72a1d2008-08-03 10:51:45 +02006702 <name> an optional argument to specify a different "X-Forwarded-For"
Willy Tarreaud72758d2010-01-12 10:42:19 +01006703 header name.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006704
6705 Since HAProxy works in reverse-proxy mode, the servers see its IP address as
6706 their client address. This is sometimes annoying when the client's IP address
6707 is expected in server logs. To solve this problem, the well-known HTTP header
6708 "X-Forwarded-For" may be added by HAProxy to all requests sent to the server.
6709 This header contains a value representing the client's IP address. Since this
6710 header is always appended at the end of the existing header list, the server
6711 must be configured to always use the last occurrence of this header only. See
Ross Westaf72a1d2008-08-03 10:51:45 +02006712 the server's manual to find how to enable use of this standard header. Note
6713 that only the last occurrence of the header must be used, since it is really
6714 possible that the client has already brought one.
6715
Willy Tarreaud72758d2010-01-12 10:42:19 +01006716 The keyword "header" may be used to supply a different header name to replace
Ross Westaf72a1d2008-08-03 10:51:45 +02006717 the default "X-Forwarded-For". This can be useful where you might already
Davor Ocelice9ed2812017-12-25 17:49:28 +01006718 have a "X-Forwarded-For" header from a different application (e.g. stunnel),
Willy Tarreaud72758d2010-01-12 10:42:19 +01006719 and you need preserve it. Also if your backend server doesn't use the
Davor Ocelice9ed2812017-12-25 17:49:28 +01006720 "X-Forwarded-For" header and requires different one (e.g. Zeus Web Servers
Ross Westaf72a1d2008-08-03 10:51:45 +02006721 require "X-Cluster-Client-IP").
Willy Tarreauc27debf2008-01-06 08:57:02 +01006722
6723 Sometimes, a same HAProxy instance may be shared between a direct client
6724 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
6725 used to decrypt HTTPS traffic). It is possible to disable the addition of the
6726 header for a known source address or network by adding the "except" keyword
6727 followed by the network address. In this case, any source IP matching the
6728 network will not cause an addition of this header. Most common uses are with
6729 private networks or 127.0.0.1.
6730
Willy Tarreau87cf5142011-08-19 22:57:24 +02006731 Alternatively, the keyword "if-none" states that the header will only be
6732 added if it is not present. This should only be used in perfectly trusted
6733 environment, as this might cause a security issue if headers reaching haproxy
6734 are under the control of the end-user.
6735
Willy Tarreauc27debf2008-01-06 08:57:02 +01006736 This option may be specified either in the frontend or in the backend. If at
Ross Westaf72a1d2008-08-03 10:51:45 +02006737 least one of them uses it, the header will be added. Note that the backend's
6738 setting of the header subargument takes precedence over the frontend's if
Willy Tarreau87cf5142011-08-19 22:57:24 +02006739 both are defined. In the case of the "if-none" argument, if at least one of
6740 the frontend or the backend does not specify it, it wants the addition to be
6741 mandatory, so it wins.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006742
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02006743 Example :
Willy Tarreauc27debf2008-01-06 08:57:02 +01006744 # Public HTTP address also used by stunnel on the same machine
6745 frontend www
6746 mode http
6747 option forwardfor except 127.0.0.1 # stunnel already adds the header
6748
Ross Westaf72a1d2008-08-03 10:51:45 +02006749 # Those servers want the IP Address in X-Client
6750 backend www
6751 mode http
6752 option forwardfor header X-Client
6753
Willy Tarreau87cf5142011-08-19 22:57:24 +02006754 See also : "option httpclose", "option http-server-close",
Christopher Faulet315b39c2018-09-21 16:26:19 +02006755 "option http-keep-alive"
Willy Tarreauc27debf2008-01-06 08:57:02 +01006756
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006757
Christopher Faulet98fbe952019-07-22 16:18:24 +02006758option h1-case-adjust-bogus-client
6759no option h1-case-adjust-bogus-client
6760 Enable or disable the case adjustment of HTTP/1 headers sent to bogus clients
6761 May be used in sections : defaults | frontend | listen | backend
6762 yes | yes | yes | no
6763 Arguments : none
6764
6765 There is no standard case for header names because, as stated in RFC7230,
6766 they are case-insensitive. So applications must handle them in a case-
6767 insensitive manner. But some bogus applications violate the standards and
6768 erroneously rely on the cases most commonly used by browsers. This problem
6769 becomes critical with HTTP/2 because all header names must be exchanged in
6770 lower case, and HAProxy follows the same convention. All header names are
6771 sent in lower case to clients and servers, regardless of the HTTP version.
6772
6773 When HAProxy receives an HTTP/1 response, its header names are converted to
6774 lower case and manipulated and sent this way to the clients. If a client is
6775 known to violate the HTTP standards and to fail to process a response coming
6776 from HAProxy, it is possible to transform the lower case header names to a
6777 different format when the response is formatted and sent to the client, by
6778 enabling this option and specifying the list of headers to be reformatted
6779 using the global directives "h1-case-adjust" or "h1-case-adjust-file". This
6780 must only be a temporary workaround for the time it takes the client to be
6781 fixed, because clients which require such workarounds might be vulnerable to
6782 content smuggling attacks and must absolutely be fixed.
6783
6784 Please note that this option will not affect standards-compliant clients.
6785
6786 If this option has been enabled in a "defaults" section, it can be disabled
6787 in a specific instance by prepending the "no" keyword before it.
6788
6789 See also: "option h1-case-adjust-bogus-server", "h1-case-adjust",
6790 "h1-case-adjust-file".
6791
6792
6793option h1-case-adjust-bogus-server
6794no option h1-case-adjust-bogus-server
6795 Enable or disable the case adjustment of HTTP/1 headers sent to bogus servers
6796 May be used in sections : defaults | frontend | listen | backend
6797 yes | no | yes | yes
6798 Arguments : none
6799
6800 There is no standard case for header names because, as stated in RFC7230,
6801 they are case-insensitive. So applications must handle them in a case-
6802 insensitive manner. But some bogus applications violate the standards and
6803 erroneously rely on the cases most commonly used by browsers. This problem
6804 becomes critical with HTTP/2 because all header names must be exchanged in
6805 lower case, and HAProxy follows the same convention. All header names are
6806 sent in lower case to clients and servers, regardless of the HTTP version.
6807
6808 When HAProxy receives an HTTP/1 request, its header names are converted to
6809 lower case and manipulated and sent this way to the servers. If a server is
6810 known to violate the HTTP standards and to fail to process a request coming
6811 from HAProxy, it is possible to transform the lower case header names to a
6812 different format when the request is formatted and sent to the server, by
6813 enabling this option and specifying the list of headers to be reformatted
6814 using the global directives "h1-case-adjust" or "h1-case-adjust-file". This
6815 must only be a temporary workaround for the time it takes the server to be
6816 fixed, because servers which require such workarounds might be vulnerable to
6817 content smuggling attacks and must absolutely be fixed.
6818
6819 Please note that this option will not affect standards-compliant servers.
6820
6821 If this option has been enabled in a "defaults" section, it can be disabled
6822 in a specific instance by prepending the "no" keyword before it.
6823
6824 See also: "option h1-case-adjust-bogus-client", "h1-case-adjust",
6825 "h1-case-adjust-file".
6826
6827
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02006828option http-buffer-request
6829no option http-buffer-request
6830 Enable or disable waiting for whole HTTP request body before proceeding
6831 May be used in sections : defaults | frontend | listen | backend
6832 yes | yes | yes | yes
6833 Arguments : none
6834
6835 It is sometimes desirable to wait for the body of an HTTP request before
6836 taking a decision. This is what is being done by "balance url_param" for
6837 example. The first use case is to buffer requests from slow clients before
6838 connecting to the server. Another use case consists in taking the routing
6839 decision based on the request body's contents. This option placed in a
6840 frontend or backend forces the HTTP processing to wait until either the whole
Christopher Faulet6db8a2e2019-11-19 16:27:25 +01006841 body is received or the request buffer is full. It can have undesired side
6842 effects with some applications abusing HTTP by expecting unbuffered
6843 transmissions between the frontend and the backend, so this should definitely
6844 not be used by default.
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02006845
Baptiste Assmanneccdf432015-10-28 13:49:01 +01006846 See also : "option http-no-delay", "timeout http-request"
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02006847
6848
Willy Tarreau0f228a02015-05-01 15:37:53 +02006849option http-ignore-probes
6850no option http-ignore-probes
6851 Enable or disable logging of null connections and request timeouts
6852 May be used in sections : defaults | frontend | listen | backend
6853 yes | yes | yes | no
6854 Arguments : none
6855
6856 Recently some browsers started to implement a "pre-connect" feature
6857 consisting in speculatively connecting to some recently visited web sites
6858 just in case the user would like to visit them. This results in many
6859 connections being established to web sites, which end up in 408 Request
6860 Timeout if the timeout strikes first, or 400 Bad Request when the browser
6861 decides to close them first. These ones pollute the log and feed the error
6862 counters. There was already "option dontlognull" but it's insufficient in
6863 this case. Instead, this option does the following things :
6864 - prevent any 400/408 message from being sent to the client if nothing
Davor Ocelice9ed2812017-12-25 17:49:28 +01006865 was received over a connection before it was closed;
6866 - prevent any log from being emitted in this situation;
Willy Tarreau0f228a02015-05-01 15:37:53 +02006867 - prevent any error counter from being incremented
6868
6869 That way the empty connection is silently ignored. Note that it is better
6870 not to use this unless it is clear that it is needed, because it will hide
6871 real problems. The most common reason for not receiving a request and seeing
6872 a 408 is due to an MTU inconsistency between the client and an intermediary
6873 element such as a VPN, which blocks too large packets. These issues are
6874 generally seen with POST requests as well as GET with large cookies. The logs
6875 are often the only way to detect them.
6876
6877 If this option has been enabled in a "defaults" section, it can be disabled
6878 in a specific instance by prepending the "no" keyword before it.
6879
6880 See also : "log", "dontlognull", "errorfile", and section 8 about logging.
6881
6882
Willy Tarreau16bfb022010-01-16 19:48:41 +01006883option http-keep-alive
6884no option http-keep-alive
6885 Enable or disable HTTP keep-alive from client to server
6886 May be used in sections : defaults | frontend | listen | backend
6887 yes | yes | yes | yes
6888 Arguments : none
6889
Willy Tarreau70dffda2014-01-30 03:07:23 +01006890 By default HAProxy operates in keep-alive mode with regards to persistent
6891 connections: for each connection it processes each request and response, and
Christopher Faulet315b39c2018-09-21 16:26:19 +02006892 leaves the connection idle on both sides between the end of a response and
6893 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02006894 as "option http-server-close" or "option httpclose". This option allows to
6895 set back the keep-alive mode, which can be useful when another mode was used
6896 in a defaults section.
Willy Tarreau70dffda2014-01-30 03:07:23 +01006897
6898 Setting "option http-keep-alive" enables HTTP keep-alive mode on the client-
6899 and server- sides. This provides the lowest latency on the client side (slow
Willy Tarreau16bfb022010-01-16 19:48:41 +01006900 network) and the fastest session reuse on the server side at the expense
6901 of maintaining idle connections to the servers. In general, it is possible
6902 with this option to achieve approximately twice the request rate that the
6903 "http-server-close" option achieves on small objects. There are mainly two
6904 situations where this option may be useful :
6905
6906 - when the server is non-HTTP compliant and authenticates the connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01006907 instead of requests (e.g. NTLM authentication)
Willy Tarreau16bfb022010-01-16 19:48:41 +01006908
6909 - when the cost of establishing the connection to the server is significant
6910 compared to the cost of retrieving the associated object from the server.
6911
6912 This last case can happen when the server is a fast static server of cache.
6913 In this case, the server will need to be properly tuned to support high enough
6914 connection counts because connections will last until the client sends another
6915 request.
6916
6917 If the client request has to go to another backend or another server due to
6918 content switching or the load balancing algorithm, the idle connection will
Willy Tarreau9420b122013-12-15 18:58:25 +01006919 immediately be closed and a new one re-opened. Option "prefer-last-server" is
6920 available to try optimize server selection so that if the server currently
6921 attached to an idle connection is usable, it will be used.
Willy Tarreau16bfb022010-01-16 19:48:41 +01006922
Willy Tarreau16bfb022010-01-16 19:48:41 +01006923 At the moment, logs will not indicate whether requests came from the same
6924 session or not. The accept date reported in the logs corresponds to the end
6925 of the previous request, and the request time corresponds to the time spent
6926 waiting for a new request. The keep-alive request time is still bound to the
6927 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
6928 not set.
6929
Christopher Faulet159e6672019-07-16 15:09:52 +02006930 This option disables and replaces any previous "option httpclose" or "option
6931 http-server-close". When backend and frontend options differ, all of these 4
6932 options have precedence over "option http-keep-alive".
Willy Tarreau16bfb022010-01-16 19:48:41 +01006933
Christopher Faulet315b39c2018-09-21 16:26:19 +02006934 See also : "option httpclose",, "option http-server-close",
Willy Tarreau9420b122013-12-15 18:58:25 +01006935 "option prefer-last-server", "option http-pretend-keepalive",
Frédéric Lécaille93d33162019-03-06 09:35:59 +01006936 and "1.1. The HTTP transaction model".
Willy Tarreau16bfb022010-01-16 19:48:41 +01006937
6938
Willy Tarreau96e31212011-05-30 18:10:30 +02006939option http-no-delay
6940no option http-no-delay
6941 Instruct the system to favor low interactive delays over performance in HTTP
6942 May be used in sections : defaults | frontend | listen | backend
6943 yes | yes | yes | yes
6944 Arguments : none
6945
6946 In HTTP, each payload is unidirectional and has no notion of interactivity.
6947 Any agent is expected to queue data somewhat for a reasonably low delay.
6948 There are some very rare server-to-server applications that abuse the HTTP
6949 protocol and expect the payload phase to be highly interactive, with many
6950 interleaved data chunks in both directions within a single request. This is
6951 absolutely not supported by the HTTP specification and will not work across
6952 most proxies or servers. When such applications attempt to do this through
6953 haproxy, it works but they will experience high delays due to the network
6954 optimizations which favor performance by instructing the system to wait for
6955 enough data to be available in order to only send full packets. Typical
6956 delays are around 200 ms per round trip. Note that this only happens with
6957 abnormal uses. Normal uses such as CONNECT requests nor WebSockets are not
6958 affected.
6959
6960 When "option http-no-delay" is present in either the frontend or the backend
6961 used by a connection, all such optimizations will be disabled in order to
6962 make the exchanges as fast as possible. Of course this offers no guarantee on
6963 the functionality, as it may break at any other place. But if it works via
6964 HAProxy, it will work as fast as possible. This option should never be used
6965 by default, and should never be used at all unless such a buggy application
6966 is discovered. The impact of using this option is an increase of bandwidth
6967 usage and CPU usage, which may significantly lower performance in high
6968 latency environments.
6969
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02006970 See also : "option http-buffer-request"
6971
Willy Tarreau96e31212011-05-30 18:10:30 +02006972
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006973option http-pretend-keepalive
6974no option http-pretend-keepalive
6975 Define whether haproxy will announce keepalive to the server or not
6976 May be used in sections : defaults | frontend | listen | backend
Christopher Faulet98db9762018-09-21 10:25:19 +02006977 yes | no | yes | yes
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006978 Arguments : none
6979
Christopher Faulet315b39c2018-09-21 16:26:19 +02006980 When running with "option http-server-close" or "option httpclose", haproxy
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006981 adds a "Connection: close" header to the request forwarded to the server.
6982 Unfortunately, when some servers see this header, they automatically refrain
6983 from using the chunked encoding for responses of unknown length, while this
6984 is totally unrelated. The immediate effect is that this prevents haproxy from
6985 maintaining the client connection alive. A second effect is that a client or
6986 a cache could receive an incomplete response without being aware of it, and
6987 consider the response complete.
6988
6989 By setting "option http-pretend-keepalive", haproxy will make the server
6990 believe it will keep the connection alive. The server will then not fall back
6991 to the abnormal undesired above. When haproxy gets the whole response, it
6992 will close the connection with the server just as it would do with the
Christopher Faulet315b39c2018-09-21 16:26:19 +02006993 "option httpclose". That way the client gets a normal response and the
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006994 connection is correctly closed on the server side.
6995
6996 It is recommended not to enable this option by default, because most servers
6997 will more efficiently close the connection themselves after the last packet,
6998 and release its buffers slightly earlier. Also, the added packet on the
6999 network could slightly reduce the overall peak performance. However it is
7000 worth noting that when this option is enabled, haproxy will have slightly
7001 less work to do. So if haproxy is the bottleneck on the whole architecture,
7002 enabling this option might save a few CPU cycles.
7003
Christopher Faulet98db9762018-09-21 10:25:19 +02007004 This option may be set in backend and listen sections. Using it in a frontend
7005 section will be ignored and a warning will be reported during startup. It is
7006 a backend related option, so there is no real reason to set it on a
7007 frontend. This option may be combined with "option httpclose", which will
7008 cause keepalive to be announced to the server and close to be announced to
7009 the client. This practice is discouraged though.
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02007010
7011 If this option has been enabled in a "defaults" section, it can be disabled
7012 in a specific instance by prepending the "no" keyword before it.
7013
Christopher Faulet315b39c2018-09-21 16:26:19 +02007014 See also : "option httpclose", "option http-server-close", and
Willy Tarreau16bfb022010-01-16 19:48:41 +01007015 "option http-keep-alive"
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02007016
Willy Tarreauc27debf2008-01-06 08:57:02 +01007017
Willy Tarreaub608feb2010-01-02 22:47:18 +01007018option http-server-close
7019no option http-server-close
7020 Enable or disable HTTP connection closing on the server side
7021 May be used in sections : defaults | frontend | listen | backend
7022 yes | yes | yes | yes
7023 Arguments : none
7024
Willy Tarreau70dffda2014-01-30 03:07:23 +01007025 By default HAProxy operates in keep-alive mode with regards to persistent
7026 connections: for each connection it processes each request and response, and
7027 leaves the connection idle on both sides between the end of a response and
7028 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02007029 as "option http-server-close" or "option httpclose". Setting "option
7030 http-server-close" enables HTTP connection-close mode on the server side
7031 while keeping the ability to support HTTP keep-alive and pipelining on the
7032 client side. This provides the lowest latency on the client side (slow
7033 network) and the fastest session reuse on the server side to save server
7034 resources, similarly to "option httpclose". It also permits non-keepalive
7035 capable servers to be served in keep-alive mode to the clients if they
7036 conform to the requirements of RFC7230. Please note that some servers do not
7037 always conform to those requirements when they see "Connection: close" in the
7038 request. The effect will be that keep-alive will never be used. A workaround
7039 consists in enabling "option http-pretend-keepalive".
Willy Tarreaub608feb2010-01-02 22:47:18 +01007040
7041 At the moment, logs will not indicate whether requests came from the same
7042 session or not. The accept date reported in the logs corresponds to the end
7043 of the previous request, and the request time corresponds to the time spent
7044 waiting for a new request. The keep-alive request time is still bound to the
Willy Tarreaub16a5742010-01-10 14:46:16 +01007045 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
7046 not set.
Willy Tarreaub608feb2010-01-02 22:47:18 +01007047
7048 This option may be set both in a frontend and in a backend. It is enabled if
7049 at least one of the frontend or backend holding a connection has it enabled.
Christopher Faulet159e6672019-07-16 15:09:52 +02007050 It disables and replaces any previous "option httpclose" or "option
7051 http-keep-alive". Please check section 4 ("Proxies") to see how this option
7052 combines with others when frontend and backend options differ.
Willy Tarreaub608feb2010-01-02 22:47:18 +01007053
7054 If this option has been enabled in a "defaults" section, it can be disabled
7055 in a specific instance by prepending the "no" keyword before it.
7056
Christopher Faulet315b39c2018-09-21 16:26:19 +02007057 See also : "option httpclose", "option http-pretend-keepalive",
7058 "option http-keep-alive", and "1.1. The HTTP transaction model".
Willy Tarreaub608feb2010-01-02 22:47:18 +01007059
Willy Tarreau88d349d2010-01-25 12:15:43 +01007060option http-use-proxy-header
Cyril Bontéf0c60612010-02-06 14:44:47 +01007061no option http-use-proxy-header
Willy Tarreau88d349d2010-01-25 12:15:43 +01007062 Make use of non-standard Proxy-Connection header instead of Connection
7063 May be used in sections : defaults | frontend | listen | backend
7064 yes | yes | yes | no
7065 Arguments : none
7066
Lukas Tribus23953682017-04-28 13:24:30 +00007067 While RFC7230 explicitly states that HTTP/1.1 agents must use the
Willy Tarreau88d349d2010-01-25 12:15:43 +01007068 Connection header to indicate their wish of persistent or non-persistent
7069 connections, both browsers and proxies ignore this header for proxied
7070 connections and make use of the undocumented, non-standard Proxy-Connection
7071 header instead. The issue begins when trying to put a load balancer between
7072 browsers and such proxies, because there will be a difference between what
7073 haproxy understands and what the client and the proxy agree on.
7074
7075 By setting this option in a frontend, haproxy can automatically switch to use
7076 that non-standard header if it sees proxied requests. A proxied request is
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01007077 defined here as one where the URI begins with neither a '/' nor a '*'. This
7078 is incompatible with the HTTP tunnel mode. Note that this option can only be
7079 specified in a frontend and will affect the request along its whole life.
Willy Tarreau88d349d2010-01-25 12:15:43 +01007080
Willy Tarreau844a7e72010-01-31 21:46:18 +01007081 Also, when this option is set, a request which requires authentication will
7082 automatically switch to use proxy authentication headers if it is itself a
7083 proxied request. That makes it possible to check or enforce authentication in
7084 front of an existing proxy.
7085
Willy Tarreau88d349d2010-01-25 12:15:43 +01007086 This option should normally never be used, except in front of a proxy.
7087
Christopher Faulet315b39c2018-09-21 16:26:19 +02007088 See also : "option httpclose", and "option http-server-close".
Willy Tarreau88d349d2010-01-25 12:15:43 +01007089
Willy Tarreaud63335a2010-02-26 12:56:52 +01007090option httpchk
7091option httpchk <uri>
7092option httpchk <method> <uri>
7093option httpchk <method> <uri> <version>
7094 Enable HTTP protocol to check on the servers health
7095 May be used in sections : defaults | frontend | listen | backend
7096 yes | no | yes | yes
7097 Arguments :
7098 <method> is the optional HTTP method used with the requests. When not set,
7099 the "OPTIONS" method is used, as it generally requires low server
7100 processing and is easy to filter out from the logs. Any method
7101 may be used, though it is not recommended to invent non-standard
7102 ones.
7103
7104 <uri> is the URI referenced in the HTTP requests. It defaults to " / "
7105 which is accessible by default on almost any server, but may be
7106 changed to any other URI. Query strings are permitted.
7107
7108 <version> is the optional HTTP version string. It defaults to "HTTP/1.0"
7109 but some servers might behave incorrectly in HTTP 1.0, so turning
7110 it to HTTP/1.1 may sometimes help. Note that the Host field is
7111 mandatory in HTTP/1.1, and as a trick, it is possible to pass it
7112 after "\r\n" following the version string.
7113
7114 By default, server health checks only consist in trying to establish a TCP
7115 connection. When "option httpchk" is specified, a complete HTTP request is
7116 sent once the TCP connection is established, and responses 2xx and 3xx are
7117 considered valid, while all other ones indicate a server failure, including
7118 the lack of any response.
7119
7120 The port and interval are specified in the server configuration.
7121
7122 This option does not necessarily require an HTTP backend, it also works with
7123 plain TCP backends. This is particularly useful to check simple scripts bound
7124 to some dedicated ports using the inetd daemon.
7125
7126 Examples :
7127 # Relay HTTPS traffic to Apache instance and check service availability
7128 # using HTTP request "OPTIONS * HTTP/1.1" on port 80.
7129 backend https_relay
7130 mode tcp
7131 option httpchk OPTIONS * HTTP/1.1\r\nHost:\ www
7132 server apache1 192.168.1.1:443 check port 80
7133
Simon Hormanafc47ee2013-11-25 10:46:35 +09007134 See also : "option ssl-hello-chk", "option smtpchk", "option mysql-check",
7135 "option pgsql-check", "http-check" and the "check", "port" and
7136 "inter" server options.
Willy Tarreaud63335a2010-02-26 12:56:52 +01007137
7138
Willy Tarreauc27debf2008-01-06 08:57:02 +01007139option httpclose
7140no option httpclose
Christopher Faulet315b39c2018-09-21 16:26:19 +02007141 Enable or disable HTTP connection closing
Willy Tarreauc27debf2008-01-06 08:57:02 +01007142 May be used in sections : defaults | frontend | listen | backend
7143 yes | yes | yes | yes
7144 Arguments : none
7145
Willy Tarreau70dffda2014-01-30 03:07:23 +01007146 By default HAProxy operates in keep-alive mode with regards to persistent
7147 connections: for each connection it processes each request and response, and
7148 leaves the connection idle on both sides between the end of a response and
7149 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02007150 as "option http-server-close" or "option httpclose".
Willy Tarreau70dffda2014-01-30 03:07:23 +01007151
Christopher Faulet315b39c2018-09-21 16:26:19 +02007152 If "option httpclose" is set, HAProxy will close connections with the server
7153 and the client as soon as the request and the response are received. It will
John Roeslerfb2fce12019-07-10 15:45:51 -05007154 also check if a "Connection: close" header is already set in each direction,
Christopher Faulet315b39c2018-09-21 16:26:19 +02007155 and will add one if missing. Any "Connection" header different from "close"
7156 will also be removed.
Willy Tarreauc27debf2008-01-06 08:57:02 +01007157
Christopher Faulet315b39c2018-09-21 16:26:19 +02007158 This option may also be combined with "option http-pretend-keepalive", which
7159 will disable sending of the "Connection: close" header, but will still cause
7160 the connection to be closed once the whole response is received.
Willy Tarreauc27debf2008-01-06 08:57:02 +01007161
7162 This option may be set both in a frontend and in a backend. It is enabled if
7163 at least one of the frontend or backend holding a connection has it enabled.
Christopher Faulet159e6672019-07-16 15:09:52 +02007164 It disables and replaces any previous "option http-server-close" or "option
7165 http-keep-alive". Please check section 4 ("Proxies") to see how this option
7166 combines with others when frontend and backend options differ.
Willy Tarreauc27debf2008-01-06 08:57:02 +01007167
7168 If this option has been enabled in a "defaults" section, it can be disabled
7169 in a specific instance by prepending the "no" keyword before it.
7170
Christopher Faulet315b39c2018-09-21 16:26:19 +02007171 See also : "option http-server-close" and "1.1. The HTTP transaction model".
Willy Tarreauc27debf2008-01-06 08:57:02 +01007172
7173
Emeric Brun3a058f32009-06-30 18:26:00 +02007174option httplog [ clf ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01007175 Enable logging of HTTP request, session state and timers
7176 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01007177 yes | yes | yes | no
Emeric Brun3a058f32009-06-30 18:26:00 +02007178 Arguments :
7179 clf if the "clf" argument is added, then the output format will be
7180 the CLF format instead of HAProxy's default HTTP format. You can
7181 use this when you need to feed HAProxy's logs through a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01007182 log analyzer which only support the CLF format and which is not
Emeric Brun3a058f32009-06-30 18:26:00 +02007183 extensible.
Willy Tarreauc27debf2008-01-06 08:57:02 +01007184
7185 By default, the log output format is very poor, as it only contains the
7186 source and destination addresses, and the instance name. By specifying
7187 "option httplog", each log line turns into a much richer format including,
7188 but not limited to, the HTTP request, the connection timers, the session
7189 status, the connections numbers, the captured headers and cookies, the
7190 frontend, backend and server name, and of course the source address and
7191 ports.
7192
PiBa-NLbd556bf2014-12-11 21:31:54 +01007193 Specifying only "option httplog" will automatically clear the 'clf' mode
7194 if it was set by default.
Emeric Brun3a058f32009-06-30 18:26:00 +02007195
Guillaume de Lafond29f45602017-03-31 19:52:15 +02007196 "option httplog" overrides any previous "log-format" directive.
7197
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007198 See also : section 8 about logging.
Willy Tarreauc27debf2008-01-06 08:57:02 +01007199
Willy Tarreau55165fe2009-05-10 12:02:55 +02007200
7201option http_proxy
7202no option http_proxy
7203 Enable or disable plain HTTP proxy mode
7204 May be used in sections : defaults | frontend | listen | backend
7205 yes | yes | yes | yes
7206 Arguments : none
7207
7208 It sometimes happens that people need a pure HTTP proxy which understands
7209 basic proxy requests without caching nor any fancy feature. In this case,
7210 it may be worth setting up an HAProxy instance with the "option http_proxy"
7211 set. In this mode, no server is declared, and the connection is forwarded to
7212 the IP address and port found in the URL after the "http://" scheme.
7213
7214 No host address resolution is performed, so this only works when pure IP
7215 addresses are passed. Since this option's usage perimeter is rather limited,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01007216 it will probably be used only by experts who know they need exactly it. This
7217 is incompatible with the HTTP tunnel mode.
Willy Tarreau55165fe2009-05-10 12:02:55 +02007218
7219 If this option has been enabled in a "defaults" section, it can be disabled
7220 in a specific instance by prepending the "no" keyword before it.
7221
7222 Example :
7223 # this backend understands HTTP proxy requests and forwards them directly.
7224 backend direct_forward
7225 option httpclose
7226 option http_proxy
7227
7228 See also : "option httpclose"
7229
Willy Tarreau211ad242009-10-03 21:45:07 +02007230
Jamie Gloudon801a0a32012-08-25 00:18:33 -04007231option independent-streams
7232no option independent-streams
7233 Enable or disable independent timeout processing for both directions
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02007234 May be used in sections : defaults | frontend | listen | backend
7235 yes | yes | yes | yes
7236 Arguments : none
7237
7238 By default, when data is sent over a socket, both the write timeout and the
7239 read timeout for that socket are refreshed, because we consider that there is
7240 activity on that socket, and we have no other means of guessing if we should
7241 receive data or not.
7242
Davor Ocelice9ed2812017-12-25 17:49:28 +01007243 While this default behavior is desirable for almost all applications, there
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02007244 exists a situation where it is desirable to disable it, and only refresh the
7245 read timeout if there are incoming data. This happens on sessions with large
7246 timeouts and low amounts of exchanged data such as telnet session. If the
7247 server suddenly disappears, the output data accumulates in the system's
7248 socket buffers, both timeouts are correctly refreshed, and there is no way
7249 to know the server does not receive them, so we don't timeout. However, when
7250 the underlying protocol always echoes sent data, it would be enough by itself
7251 to detect the issue using the read timeout. Note that this problem does not
7252 happen with more verbose protocols because data won't accumulate long in the
7253 socket buffers.
7254
7255 When this option is set on the frontend, it will disable read timeout updates
7256 on data sent to the client. There probably is little use of this case. When
7257 the option is set on the backend, it will disable read timeout updates on
7258 data sent to the server. Doing so will typically break large HTTP posts from
7259 slow lines, so use it with caution.
7260
Willy Tarreauce887fd2012-05-12 12:50:00 +02007261 See also : "timeout client", "timeout server" and "timeout tunnel"
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02007262
7263
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02007264option ldap-check
7265 Use LDAPv3 health checks for server testing
7266 May be used in sections : defaults | frontend | listen | backend
7267 yes | no | yes | yes
7268 Arguments : none
7269
7270 It is possible to test that the server correctly talks LDAPv3 instead of just
7271 testing that it accepts the TCP connection. When this option is set, an
7272 LDAPv3 anonymous simple bind message is sent to the server, and the response
7273 is analyzed to find an LDAPv3 bind response message.
7274
7275 The server is considered valid only when the LDAP response contains success
7276 resultCode (http://tools.ietf.org/html/rfc4511#section-4.1.9).
7277
7278 Logging of bind requests is server dependent see your documentation how to
7279 configure it.
7280
7281 Example :
7282 option ldap-check
7283
7284 See also : "option httpchk"
7285
7286
Simon Horman98637e52014-06-20 12:30:16 +09007287option external-check
7288 Use external processes for server health checks
7289 May be used in sections : defaults | frontend | listen | backend
7290 yes | no | yes | yes
7291
7292 It is possible to test the health of a server using an external command.
7293 This is achieved by running the executable set using "external-check
7294 command".
7295
7296 Requires the "external-check" global to be set.
7297
7298 See also : "external-check", "external-check command", "external-check path"
7299
7300
Willy Tarreau211ad242009-10-03 21:45:07 +02007301option log-health-checks
7302no option log-health-checks
Willy Tarreaubef1b322014-05-13 21:01:39 +02007303 Enable or disable logging of health checks status updates
Willy Tarreau211ad242009-10-03 21:45:07 +02007304 May be used in sections : defaults | frontend | listen | backend
7305 yes | no | yes | yes
7306 Arguments : none
7307
Willy Tarreaubef1b322014-05-13 21:01:39 +02007308 By default, failed health check are logged if server is UP and successful
7309 health checks are logged if server is DOWN, so the amount of additional
7310 information is limited.
Willy Tarreau211ad242009-10-03 21:45:07 +02007311
Willy Tarreaubef1b322014-05-13 21:01:39 +02007312 When this option is enabled, any change of the health check status or to
7313 the server's health will be logged, so that it becomes possible to know
7314 that a server was failing occasional checks before crashing, or exactly when
7315 it failed to respond a valid HTTP status, then when the port started to
7316 reject connections, then when the server stopped responding at all.
7317
Davor Ocelice9ed2812017-12-25 17:49:28 +01007318 Note that status changes not caused by health checks (e.g. enable/disable on
Willy Tarreaubef1b322014-05-13 21:01:39 +02007319 the CLI) are intentionally not logged by this option.
Willy Tarreau211ad242009-10-03 21:45:07 +02007320
Willy Tarreaubef1b322014-05-13 21:01:39 +02007321 See also: "option httpchk", "option ldap-check", "option mysql-check",
7322 "option pgsql-check", "option redis-check", "option smtpchk",
7323 "option tcp-check", "log" and section 8 about logging.
Willy Tarreau211ad242009-10-03 21:45:07 +02007324
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02007325
7326option log-separate-errors
7327no option log-separate-errors
7328 Change log level for non-completely successful connections
7329 May be used in sections : defaults | frontend | listen | backend
7330 yes | yes | yes | no
7331 Arguments : none
7332
7333 Sometimes looking for errors in logs is not easy. This option makes haproxy
7334 raise the level of logs containing potentially interesting information such
7335 as errors, timeouts, retries, redispatches, or HTTP status codes 5xx. The
7336 level changes from "info" to "err". This makes it possible to log them
7337 separately to a different file with most syslog daemons. Be careful not to
7338 remove them from the original file, otherwise you would lose ordering which
7339 provides very important information.
7340
7341 Using this option, large sites dealing with several thousand connections per
7342 second may log normal traffic to a rotating buffer and only archive smaller
7343 error logs.
7344
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007345 See also : "log", "dontlognull", "dontlog-normal" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02007346 logging.
7347
Willy Tarreauc27debf2008-01-06 08:57:02 +01007348
7349option logasap
7350no option logasap
7351 Enable or disable early logging of HTTP requests
7352 May be used in sections : defaults | frontend | listen | backend
7353 yes | yes | yes | no
7354 Arguments : none
7355
7356 By default, HTTP requests are logged upon termination so that the total
7357 transfer time and the number of bytes appear in the logs. When large objects
7358 are being transferred, it may take a while before the request appears in the
7359 logs. Using "option logasap", the request gets logged as soon as the server
7360 sends the complete headers. The only missing information in the logs will be
7361 the total number of bytes which will indicate everything except the amount
7362 of data transferred, and the total time which will not take the transfer
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01007363 time into account. In such a situation, it's a good practice to capture the
Willy Tarreauc27debf2008-01-06 08:57:02 +01007364 "Content-Length" response header so that the logs at least indicate how many
7365 bytes are expected to be transferred.
7366
Willy Tarreaucc6c8912009-02-22 10:53:55 +01007367 Examples :
7368 listen http_proxy 0.0.0.0:80
7369 mode http
7370 option httplog
7371 option logasap
7372 log 192.168.2.200 local3
7373
7374 >>> Feb 6 12:14:14 localhost \
7375 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
7376 static/srv1 9/10/7/14/+30 200 +243 - - ---- 3/1/1/1/0 1/0 \
7377 "GET /image.iso HTTP/1.0"
7378
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007379 See also : "option httplog", "capture response header", and section 8 about
Willy Tarreauc27debf2008-01-06 08:57:02 +01007380 logging.
7381
7382
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02007383option mysql-check [ user <username> [ post-41 ] ]
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02007384 Use MySQL health checks for server testing
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01007385 May be used in sections : defaults | frontend | listen | backend
7386 yes | no | yes | yes
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02007387 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +02007388 <username> This is the username which will be used when connecting to MySQL
7389 server.
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02007390 post-41 Send post v4.1 client compatible checks
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02007391
7392 If you specify a username, the check consists of sending two MySQL packet,
7393 one Client Authentication packet, and one QUIT packet, to correctly close
Davor Ocelice9ed2812017-12-25 17:49:28 +01007394 MySQL session. We then parse the MySQL Handshake Initialization packet and/or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02007395 Error packet. It is a basic but useful test which does not produce error nor
7396 aborted connect on the server. However, it requires adding an authorization
7397 in the MySQL table, like this :
7398
7399 USE mysql;
7400 INSERT INTO user (Host,User) values ('<ip_of_haproxy>','<username>');
7401 FLUSH PRIVILEGES;
7402
7403 If you don't specify a username (it is deprecated and not recommended), the
Davor Ocelice9ed2812017-12-25 17:49:28 +01007404 check only consists in parsing the Mysql Handshake Initialization packet or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02007405 Error packet, we don't send anything in this mode. It was reported that it
7406 can generate lockout if check is too frequent and/or if there is not enough
7407 traffic. In fact, you need in this case to check MySQL "max_connect_errors"
7408 value as if a connection is established successfully within fewer than MySQL
7409 "max_connect_errors" attempts after a previous connection was interrupted,
7410 the error count for the host is cleared to zero. If HAProxy's server get
7411 blocked, the "FLUSH HOSTS" statement is the only way to unblock it.
7412
7413 Remember that this does not check database presence nor database consistency.
7414 To do this, you can use an external check with xinetd for example.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01007415
Hervé COMMOWICK212f7782011-06-10 14:05:59 +02007416 The check requires MySQL >=3.22, for older version, please use TCP check.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01007417
7418 Most often, an incoming MySQL server needs to see the client's IP address for
7419 various purposes, including IP privilege matching and connection logging.
7420 When possible, it is often wise to masquerade the client's IP address when
7421 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02007422 which requires the transparent proxy feature to be compiled in, and the MySQL
7423 server to route the client via the machine hosting haproxy.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01007424
7425 See also: "option httpchk"
7426
7427
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007428option nolinger
7429no option nolinger
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007430 Enable or disable immediate session resource cleaning after close
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007431 May be used in sections: defaults | frontend | listen | backend
7432 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007433 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007434
Davor Ocelice9ed2812017-12-25 17:49:28 +01007435 When clients or servers abort connections in a dirty way (e.g. they are
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007436 physically disconnected), the session timeouts triggers and the session is
7437 closed. But it will remain in FIN_WAIT1 state for some time in the system,
7438 using some resources and possibly limiting the ability to establish newer
7439 connections.
7440
7441 When this happens, it is possible to activate "option nolinger" which forces
7442 the system to immediately remove any socket's pending data on close. Thus,
7443 the session is instantly purged from the system's tables. This usually has
7444 side effects such as increased number of TCP resets due to old retransmits
7445 getting immediately rejected. Some firewalls may sometimes complain about
7446 this too.
7447
7448 For this reason, it is not recommended to use this option when not absolutely
7449 needed. You know that you need it when you have thousands of FIN_WAIT1
7450 sessions on your system (TIME_WAIT ones do not count).
7451
7452 This option may be used both on frontends and backends, depending on the side
7453 where it is required. Use it on the frontend for clients, and on the backend
7454 for servers.
7455
7456 If this option has been enabled in a "defaults" section, it can be disabled
7457 in a specific instance by prepending the "no" keyword before it.
7458
7459
Willy Tarreau55165fe2009-05-10 12:02:55 +02007460option originalto [ except <network> ] [ header <name> ]
7461 Enable insertion of the X-Original-To header to requests sent to servers
7462 May be used in sections : defaults | frontend | listen | backend
7463 yes | yes | yes | yes
7464 Arguments :
7465 <network> is an optional argument used to disable this option for sources
7466 matching <network>
7467 <name> an optional argument to specify a different "X-Original-To"
7468 header name.
7469
7470 Since HAProxy can work in transparent mode, every request from a client can
7471 be redirected to the proxy and HAProxy itself can proxy every request to a
7472 complex SQUID environment and the destination host from SO_ORIGINAL_DST will
7473 be lost. This is annoying when you want access rules based on destination ip
7474 addresses. To solve this problem, a new HTTP header "X-Original-To" may be
7475 added by HAProxy to all requests sent to the server. This header contains a
7476 value representing the original destination IP address. Since this must be
7477 configured to always use the last occurrence of this header only. Note that
7478 only the last occurrence of the header must be used, since it is really
7479 possible that the client has already brought one.
7480
7481 The keyword "header" may be used to supply a different header name to replace
7482 the default "X-Original-To". This can be useful where you might already
7483 have a "X-Original-To" header from a different application, and you need
7484 preserve it. Also if your backend server doesn't use the "X-Original-To"
7485 header and requires different one.
7486
7487 Sometimes, a same HAProxy instance may be shared between a direct client
7488 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
7489 used to decrypt HTTPS traffic). It is possible to disable the addition of the
7490 header for a known source address or network by adding the "except" keyword
7491 followed by the network address. In this case, any source IP matching the
7492 network will not cause an addition of this header. Most common uses are with
7493 private networks or 127.0.0.1.
7494
7495 This option may be specified either in the frontend or in the backend. If at
7496 least one of them uses it, the header will be added. Note that the backend's
7497 setting of the header subargument takes precedence over the frontend's if
7498 both are defined.
7499
Willy Tarreau55165fe2009-05-10 12:02:55 +02007500 Examples :
7501 # Original Destination address
7502 frontend www
7503 mode http
7504 option originalto except 127.0.0.1
7505
7506 # Those servers want the IP Address in X-Client-Dst
7507 backend www
7508 mode http
7509 option originalto header X-Client-Dst
7510
Christopher Faulet315b39c2018-09-21 16:26:19 +02007511 See also : "option httpclose", "option http-server-close".
Willy Tarreau55165fe2009-05-10 12:02:55 +02007512
7513
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007514option persist
7515no option persist
7516 Enable or disable forced persistence on down servers
7517 May be used in sections: defaults | frontend | listen | backend
7518 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007519 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007520
7521 When an HTTP request reaches a backend with a cookie which references a dead
7522 server, by default it is redispatched to another server. It is possible to
7523 force the request to be sent to the dead server first using "option persist"
7524 if absolutely needed. A common use case is when servers are under extreme
7525 load and spend their time flapping. In this case, the users would still be
7526 directed to the server they opened the session on, in the hope they would be
7527 correctly served. It is recommended to use "option redispatch" in conjunction
7528 with this option so that in the event it would not be possible to connect to
7529 the server at all (server definitely dead), the client would finally be
7530 redirected to another valid server.
7531
7532 If this option has been enabled in a "defaults" section, it can be disabled
7533 in a specific instance by prepending the "no" keyword before it.
7534
Willy Tarreau4de91492010-01-22 19:10:05 +01007535 See also : "option redispatch", "retries", "force-persist"
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007536
7537
Willy Tarreau0c122822013-12-15 18:49:01 +01007538option pgsql-check [ user <username> ]
7539 Use PostgreSQL health checks for server testing
7540 May be used in sections : defaults | frontend | listen | backend
7541 yes | no | yes | yes
7542 Arguments :
7543 <username> This is the username which will be used when connecting to
7544 PostgreSQL server.
7545
7546 The check sends a PostgreSQL StartupMessage and waits for either
7547 Authentication request or ErrorResponse message. It is a basic but useful
7548 test which does not produce error nor aborted connect on the server.
7549 This check is identical with the "mysql-check".
7550
7551 See also: "option httpchk"
7552
7553
Willy Tarreau9420b122013-12-15 18:58:25 +01007554option prefer-last-server
7555no option prefer-last-server
7556 Allow multiple load balanced requests to remain on the same server
7557 May be used in sections: defaults | frontend | listen | backend
7558 yes | no | yes | yes
7559 Arguments : none
7560
7561 When the load balancing algorithm in use is not deterministic, and a previous
7562 request was sent to a server to which haproxy still holds a connection, it is
7563 sometimes desirable that subsequent requests on a same session go to the same
7564 server as much as possible. Note that this is different from persistence, as
7565 we only indicate a preference which haproxy tries to apply without any form
7566 of warranty. The real use is for keep-alive connections sent to servers. When
7567 this option is used, haproxy will try to reuse the same connection that is
7568 attached to the server instead of rebalancing to another server, causing a
7569 close of the connection. This can make sense for static file servers. It does
Willy Tarreau068621e2013-12-23 15:11:25 +01007570 not make much sense to use this in combination with hashing algorithms. Note,
7571 haproxy already automatically tries to stick to a server which sends a 401 or
Lukas Tribus80512b12018-10-27 20:07:40 +02007572 to a proxy which sends a 407 (authentication required), when the load
7573 balancing algorithm is not deterministic. This is mandatory for use with the
7574 broken NTLM authentication challenge, and significantly helps in
Willy Tarreau068621e2013-12-23 15:11:25 +01007575 troubleshooting some faulty applications. Option prefer-last-server might be
7576 desirable in these environments as well, to avoid redistributing the traffic
7577 after every other response.
Willy Tarreau9420b122013-12-15 18:58:25 +01007578
7579 If this option has been enabled in a "defaults" section, it can be disabled
7580 in a specific instance by prepending the "no" keyword before it.
7581
7582 See also: "option http-keep-alive"
7583
7584
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007585option redispatch
Joseph Lynch726ab712015-05-11 23:25:34 -07007586option redispatch <interval>
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007587no option redispatch
7588 Enable or disable session redistribution in case of connection failure
7589 May be used in sections: defaults | frontend | listen | backend
7590 yes | no | yes | yes
Joseph Lynch726ab712015-05-11 23:25:34 -07007591 Arguments :
7592 <interval> The optional integer value that controls how often redispatches
7593 occur when retrying connections. Positive value P indicates a
7594 redispatch is desired on every Pth retry, and negative value
Davor Ocelice9ed2812017-12-25 17:49:28 +01007595 N indicate a redispatch is desired on the Nth retry prior to the
Joseph Lynch726ab712015-05-11 23:25:34 -07007596 last retry. For example, the default of -1 preserves the
Davor Ocelice9ed2812017-12-25 17:49:28 +01007597 historical behavior of redispatching on the last retry, a
Joseph Lynch726ab712015-05-11 23:25:34 -07007598 positive value of 1 would indicate a redispatch on every retry,
7599 and a positive value of 3 would indicate a redispatch on every
7600 third retry. You can disable redispatches with a value of 0.
7601
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007602
7603 In HTTP mode, if a server designated by a cookie is down, clients may
7604 definitely stick to it because they cannot flush the cookie, so they will not
7605 be able to access the service anymore.
7606
Willy Tarreau59884a62019-01-02 14:48:31 +01007607 Specifying "option redispatch" will allow the proxy to break cookie or
7608 consistent hash based persistence and redistribute them to a working server.
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007609
Joseph Lynch726ab712015-05-11 23:25:34 -07007610 It also allows to retry connections to another server in case of multiple
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007611 connection failures. Of course, it requires having "retries" set to a nonzero
7612 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01007613
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007614 If this option has been enabled in a "defaults" section, it can be disabled
7615 in a specific instance by prepending the "no" keyword before it.
7616
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02007617 See also : "retries", "force-persist"
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007618
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007619
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02007620option redis-check
7621 Use redis health checks for server testing
7622 May be used in sections : defaults | frontend | listen | backend
7623 yes | no | yes | yes
7624 Arguments : none
7625
7626 It is possible to test that the server correctly talks REDIS protocol instead
7627 of just testing that it accepts the TCP connection. When this option is set,
7628 a PING redis command is sent to the server, and the response is analyzed to
7629 find the "+PONG" response message.
7630
7631 Example :
7632 option redis-check
7633
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03007634 See also : "option httpchk", "option tcp-check", "tcp-check expect"
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02007635
7636
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007637option smtpchk
7638option smtpchk <hello> <domain>
7639 Use SMTP health checks for server testing
7640 May be used in sections : defaults | frontend | listen | backend
7641 yes | no | yes | yes
Willy Tarreaud72758d2010-01-12 10:42:19 +01007642 Arguments :
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007643 <hello> is an optional argument. It is the "hello" command to use. It can
Lukas Tribus27935782018-10-01 02:00:16 +02007644 be either "HELO" (for SMTP) or "EHLO" (for ESMTP). All other
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007645 values will be turned into the default command ("HELO").
7646
7647 <domain> is the domain name to present to the server. It may only be
7648 specified (and is mandatory) if the hello command has been
7649 specified. By default, "localhost" is used.
7650
7651 When "option smtpchk" is set, the health checks will consist in TCP
7652 connections followed by an SMTP command. By default, this command is
7653 "HELO localhost". The server's return code is analyzed and only return codes
7654 starting with a "2" will be considered as valid. All other responses,
7655 including a lack of response will constitute an error and will indicate a
7656 dead server.
7657
7658 This test is meant to be used with SMTP servers or relays. Depending on the
7659 request, it is possible that some servers do not log each connection attempt,
Davor Ocelice9ed2812017-12-25 17:49:28 +01007660 so you may want to experiment to improve the behavior. Using telnet on port
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007661 25 is often easier than adjusting the configuration.
7662
7663 Most often, an incoming SMTP server needs to see the client's IP address for
7664 various purposes, including spam filtering, anti-spoofing and logging. When
7665 possible, it is often wise to masquerade the client's IP address when
7666 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02007667 which requires the transparent proxy feature to be compiled in.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007668
7669 Example :
7670 option smtpchk HELO mydomain.org
7671
7672 See also : "option httpchk", "source"
7673
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007674
Krzysztof Piotr Oledzkiaeebf9b2009-10-04 15:43:17 +02007675option socket-stats
7676no option socket-stats
7677
7678 Enable or disable collecting & providing separate statistics for each socket.
7679 May be used in sections : defaults | frontend | listen | backend
7680 yes | yes | yes | no
7681
7682 Arguments : none
7683
7684
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007685option splice-auto
7686no option splice-auto
7687 Enable or disable automatic kernel acceleration on sockets in both directions
7688 May be used in sections : defaults | frontend | listen | backend
7689 yes | yes | yes | yes
7690 Arguments : none
7691
7692 When this option is enabled either on a frontend or on a backend, haproxy
7693 will automatically evaluate the opportunity to use kernel tcp splicing to
Davor Ocelice9ed2812017-12-25 17:49:28 +01007694 forward data between the client and the server, in either direction. HAProxy
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007695 uses heuristics to estimate if kernel splicing might improve performance or
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007696 not. Both directions are handled independently. Note that the heuristics used
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007697 are not much aggressive in order to limit excessive use of splicing. This
7698 option requires splicing to be enabled at compile time, and may be globally
7699 disabled with the global option "nosplice". Since splice uses pipes, using it
7700 requires that there are enough spare pipes.
7701
7702 Important note: kernel-based TCP splicing is a Linux-specific feature which
7703 first appeared in kernel 2.6.25. It offers kernel-based acceleration to
7704 transfer data between sockets without copying these data to user-space, thus
7705 providing noticeable performance gains and CPU cycles savings. Since many
7706 early implementations are buggy, corrupt data and/or are inefficient, this
7707 feature is not enabled by default, and it should be used with extreme care.
7708 While it is not possible to detect the correctness of an implementation,
7709 2.6.29 is the first version offering a properly working implementation. In
7710 case of doubt, splicing may be globally disabled using the global "nosplice"
7711 keyword.
7712
7713 Example :
7714 option splice-auto
7715
7716 If this option has been enabled in a "defaults" section, it can be disabled
7717 in a specific instance by prepending the "no" keyword before it.
7718
7719 See also : "option splice-request", "option splice-response", and global
7720 options "nosplice" and "maxpipes"
7721
7722
7723option splice-request
7724no option splice-request
7725 Enable or disable automatic kernel acceleration on sockets for requests
7726 May be used in sections : defaults | frontend | listen | backend
7727 yes | yes | yes | yes
7728 Arguments : none
7729
7730 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04007731 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007732 the client to the server. It might still use the recv/send scheme if there
7733 are no spare pipes left. This option requires splicing to be enabled at
7734 compile time, and may be globally disabled with the global option "nosplice".
7735 Since splice uses pipes, using it requires that there are enough spare pipes.
7736
7737 Important note: see "option splice-auto" for usage limitations.
7738
7739 Example :
7740 option splice-request
7741
7742 If this option has been enabled in a "defaults" section, it can be disabled
7743 in a specific instance by prepending the "no" keyword before it.
7744
7745 See also : "option splice-auto", "option splice-response", and global options
7746 "nosplice" and "maxpipes"
7747
7748
7749option splice-response
7750no option splice-response
7751 Enable or disable automatic kernel acceleration on sockets for responses
7752 May be used in sections : defaults | frontend | listen | backend
7753 yes | yes | yes | yes
7754 Arguments : none
7755
7756 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04007757 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007758 the server to the client. It might still use the recv/send scheme if there
7759 are no spare pipes left. This option requires splicing to be enabled at
7760 compile time, and may be globally disabled with the global option "nosplice".
7761 Since splice uses pipes, using it requires that there are enough spare pipes.
7762
7763 Important note: see "option splice-auto" for usage limitations.
7764
7765 Example :
7766 option splice-response
7767
7768 If this option has been enabled in a "defaults" section, it can be disabled
7769 in a specific instance by prepending the "no" keyword before it.
7770
7771 See also : "option splice-auto", "option splice-request", and global options
7772 "nosplice" and "maxpipes"
7773
7774
Christopher Fauletba7bc162016-11-07 21:07:38 +01007775option spop-check
7776 Use SPOP health checks for server testing
7777 May be used in sections : defaults | frontend | listen | backend
7778 no | no | no | yes
7779 Arguments : none
7780
7781 It is possible to test that the server correctly talks SPOP protocol instead
7782 of just testing that it accepts the TCP connection. When this option is set,
7783 a HELLO handshake is performed between HAProxy and the server, and the
7784 response is analyzed to check no error is reported.
7785
7786 Example :
7787 option spop-check
7788
7789 See also : "option httpchk"
7790
7791
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007792option srvtcpka
7793no option srvtcpka
7794 Enable or disable the sending of TCP keepalive packets on the server side
7795 May be used in sections : defaults | frontend | listen | backend
7796 yes | no | yes | yes
7797 Arguments : none
7798
7799 When there is a firewall or any session-aware component between a client and
7800 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01007801 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007802 components decides to expire a session which has remained idle for too long.
7803
7804 Enabling socket-level TCP keep-alives makes the system regularly send packets
7805 to the other end of the connection, leaving it active. The delay between
7806 keep-alive probes is controlled by the system only and depends both on the
7807 operating system and its tuning parameters.
7808
7809 It is important to understand that keep-alive packets are neither emitted nor
7810 received at the application level. It is only the network stacks which sees
7811 them. For this reason, even if one side of the proxy already uses keep-alives
7812 to maintain its connection alive, those keep-alive packets will not be
7813 forwarded to the other side of the proxy.
7814
7815 Please note that this has nothing to do with HTTP keep-alive.
7816
7817 Using option "srvtcpka" enables the emission of TCP keep-alive probes on the
7818 server side of a connection, which should help when session expirations are
7819 noticed between HAProxy and a server.
7820
7821 If this option has been enabled in a "defaults" section, it can be disabled
7822 in a specific instance by prepending the "no" keyword before it.
7823
7824 See also : "option clitcpka", "option tcpka"
7825
7826
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007827option ssl-hello-chk
7828 Use SSLv3 client hello health checks for server testing
7829 May be used in sections : defaults | frontend | listen | backend
7830 yes | no | yes | yes
7831 Arguments : none
7832
7833 When some SSL-based protocols are relayed in TCP mode through HAProxy, it is
7834 possible to test that the server correctly talks SSL instead of just testing
7835 that it accepts the TCP connection. When "option ssl-hello-chk" is set, pure
7836 SSLv3 client hello messages are sent once the connection is established to
7837 the server, and the response is analyzed to find an SSL server hello message.
7838 The server is considered valid only when the response contains this server
7839 hello message.
7840
7841 All servers tested till there correctly reply to SSLv3 client hello messages,
7842 and most servers tested do not even log the requests containing only hello
7843 messages, which is appreciable.
7844
Willy Tarreau763a95b2012-10-04 23:15:39 +02007845 Note that this check works even when SSL support was not built into haproxy
7846 because it forges the SSL message. When SSL support is available, it is best
7847 to use native SSL health checks instead of this one.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007848
Willy Tarreau763a95b2012-10-04 23:15:39 +02007849 See also: "option httpchk", "check-ssl"
7850
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007851
Willy Tarreaued179852013-12-16 01:07:00 +01007852option tcp-check
7853 Perform health checks using tcp-check send/expect sequences
7854 May be used in sections: defaults | frontend | listen | backend
7855 yes | no | yes | yes
7856
7857 This health check method is intended to be combined with "tcp-check" command
7858 lists in order to support send/expect types of health check sequences.
7859
7860 TCP checks currently support 4 modes of operations :
7861 - no "tcp-check" directive : the health check only consists in a connection
7862 attempt, which remains the default mode.
7863
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007864 - "tcp-check send" or "tcp-check send-binary" only is mentioned : this is
Willy Tarreaued179852013-12-16 01:07:00 +01007865 used to send a string along with a connection opening. With some
7866 protocols, it helps sending a "QUIT" message for example that prevents
7867 the server from logging a connection error for each health check. The
7868 check result will still be based on the ability to open the connection
7869 only.
7870
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007871 - "tcp-check expect" only is mentioned : this is used to test a banner.
Willy Tarreaued179852013-12-16 01:07:00 +01007872 The connection is opened and haproxy waits for the server to present some
7873 contents which must validate some rules. The check result will be based
7874 on the matching between the contents and the rules. This is suited for
7875 POP, IMAP, SMTP, FTP, SSH, TELNET.
7876
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007877 - both "tcp-check send" and "tcp-check expect" are mentioned : this is
Davor Ocelice9ed2812017-12-25 17:49:28 +01007878 used to test a hello-type protocol. HAProxy sends a message, the server
7879 responds and its response is analyzed. the check result will be based on
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007880 the matching between the response contents and the rules. This is often
Willy Tarreaued179852013-12-16 01:07:00 +01007881 suited for protocols which require a binding or a request/response model.
7882 LDAP, MySQL, Redis and SSL are example of such protocols, though they
7883 already all have their dedicated checks with a deeper understanding of
7884 the respective protocols.
7885 In this mode, many questions may be sent and many answers may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01007886 analyzed.
Willy Tarreaued179852013-12-16 01:07:00 +01007887
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007888 A fifth mode can be used to insert comments in different steps of the
7889 script.
7890
7891 For each tcp-check rule you create, you can add a "comment" directive,
7892 followed by a string. This string will be reported in the log and stderr
7893 in debug mode. It is useful to make user-friendly error reporting.
7894 The "comment" is of course optional.
7895
7896
Willy Tarreaued179852013-12-16 01:07:00 +01007897 Examples :
Davor Ocelice9ed2812017-12-25 17:49:28 +01007898 # perform a POP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01007899 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007900 tcp-check expect string +OK\ POP3\ ready comment POP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01007901
Davor Ocelice9ed2812017-12-25 17:49:28 +01007902 # perform an IMAP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01007903 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007904 tcp-check expect string *\ OK\ IMAP4\ ready comment IMAP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01007905
7906 # look for the redis master server after ensuring it speaks well
7907 # redis protocol, then it exits properly.
Davor Ocelice9ed2812017-12-25 17:49:28 +01007908 # (send a command then analyze the response 3 times)
Willy Tarreaued179852013-12-16 01:07:00 +01007909 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007910 tcp-check comment PING\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01007911 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02007912 tcp-check expect string +PONG
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007913 tcp-check comment role\ check
Willy Tarreaued179852013-12-16 01:07:00 +01007914 tcp-check send info\ replication\r\n
7915 tcp-check expect string role:master
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007916 tcp-check comment QUIT\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01007917 tcp-check send QUIT\r\n
7918 tcp-check expect string +OK
7919
Davor Ocelice9ed2812017-12-25 17:49:28 +01007920 forge a HTTP request, then analyze the response
Willy Tarreaued179852013-12-16 01:07:00 +01007921 (send many headers before analyzing)
7922 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007923 tcp-check comment forge\ and\ send\ HTTP\ request
Willy Tarreaued179852013-12-16 01:07:00 +01007924 tcp-check send HEAD\ /\ HTTP/1.1\r\n
7925 tcp-check send Host:\ www.mydomain.com\r\n
7926 tcp-check send User-Agent:\ HAProxy\ tcpcheck\r\n
7927 tcp-check send \r\n
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007928 tcp-check expect rstring HTTP/1\..\ (2..|3..) comment check\ HTTP\ response
Willy Tarreaued179852013-12-16 01:07:00 +01007929
7930
7931 See also : "tcp-check expect", "tcp-check send"
7932
7933
Willy Tarreau9ea05a72009-06-14 12:07:01 +02007934option tcp-smart-accept
7935no option tcp-smart-accept
7936 Enable or disable the saving of one ACK packet during the accept sequence
7937 May be used in sections : defaults | frontend | listen | backend
7938 yes | yes | yes | no
7939 Arguments : none
7940
7941 When an HTTP connection request comes in, the system acknowledges it on
7942 behalf of HAProxy, then the client immediately sends its request, and the
7943 system acknowledges it too while it is notifying HAProxy about the new
7944 connection. HAProxy then reads the request and responds. This means that we
7945 have one TCP ACK sent by the system for nothing, because the request could
7946 very well be acknowledged by HAProxy when it sends its response.
7947
7948 For this reason, in HTTP mode, HAProxy automatically asks the system to avoid
7949 sending this useless ACK on platforms which support it (currently at least
7950 Linux). It must not cause any problem, because the system will send it anyway
7951 after 40 ms if the response takes more time than expected to come.
7952
7953 During complex network debugging sessions, it may be desirable to disable
7954 this optimization because delayed ACKs can make troubleshooting more complex
7955 when trying to identify where packets are delayed. It is then possible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01007956 fall back to normal behavior by specifying "no option tcp-smart-accept".
Willy Tarreau9ea05a72009-06-14 12:07:01 +02007957
7958 It is also possible to force it for non-HTTP proxies by simply specifying
7959 "option tcp-smart-accept". For instance, it can make sense with some services
7960 such as SMTP where the server speaks first.
7961
7962 It is recommended to avoid forcing this option in a defaults section. In case
7963 of doubt, consider setting it back to automatic values by prepending the
7964 "default" keyword before it, or disabling it using the "no" keyword.
7965
Willy Tarreaud88edf22009-06-14 15:48:17 +02007966 See also : "option tcp-smart-connect"
7967
7968
7969option tcp-smart-connect
7970no option tcp-smart-connect
7971 Enable or disable the saving of one ACK packet during the connect sequence
7972 May be used in sections : defaults | frontend | listen | backend
7973 yes | no | yes | yes
7974 Arguments : none
7975
7976 On certain systems (at least Linux), HAProxy can ask the kernel not to
7977 immediately send an empty ACK upon a connection request, but to directly
7978 send the buffer request instead. This saves one packet on the network and
7979 thus boosts performance. It can also be useful for some servers, because they
7980 immediately get the request along with the incoming connection.
7981
7982 This feature is enabled when "option tcp-smart-connect" is set in a backend.
7983 It is not enabled by default because it makes network troubleshooting more
7984 complex.
7985
7986 It only makes sense to enable it with protocols where the client speaks first
7987 such as HTTP. In other situations, if there is no data to send in place of
7988 the ACK, a normal ACK is sent.
7989
7990 If this option has been enabled in a "defaults" section, it can be disabled
7991 in a specific instance by prepending the "no" keyword before it.
7992
7993 See also : "option tcp-smart-accept"
7994
Willy Tarreau9ea05a72009-06-14 12:07:01 +02007995
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007996option tcpka
7997 Enable or disable the sending of TCP keepalive packets on both sides
7998 May be used in sections : defaults | frontend | listen | backend
7999 yes | yes | yes | yes
8000 Arguments : none
8001
8002 When there is a firewall or any session-aware component between a client and
8003 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01008004 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008005 components decides to expire a session which has remained idle for too long.
8006
8007 Enabling socket-level TCP keep-alives makes the system regularly send packets
8008 to the other end of the connection, leaving it active. The delay between
8009 keep-alive probes is controlled by the system only and depends both on the
8010 operating system and its tuning parameters.
8011
8012 It is important to understand that keep-alive packets are neither emitted nor
8013 received at the application level. It is only the network stacks which sees
8014 them. For this reason, even if one side of the proxy already uses keep-alives
8015 to maintain its connection alive, those keep-alive packets will not be
8016 forwarded to the other side of the proxy.
8017
8018 Please note that this has nothing to do with HTTP keep-alive.
8019
8020 Using option "tcpka" enables the emission of TCP keep-alive probes on both
8021 the client and server sides of a connection. Note that this is meaningful
8022 only in "defaults" or "listen" sections. If this option is used in a
8023 frontend, only the client side will get keep-alives, and if this option is
8024 used in a backend, only the server side will get keep-alives. For this
8025 reason, it is strongly recommended to explicitly use "option clitcpka" and
8026 "option srvtcpka" when the configuration is split between frontends and
8027 backends.
8028
8029 See also : "option clitcpka", "option srvtcpka"
8030
Willy Tarreau844e3c52008-01-11 16:28:18 +01008031
8032option tcplog
8033 Enable advanced logging of TCP connections with session state and timers
8034 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01008035 yes | yes | yes | no
Willy Tarreau844e3c52008-01-11 16:28:18 +01008036 Arguments : none
8037
8038 By default, the log output format is very poor, as it only contains the
8039 source and destination addresses, and the instance name. By specifying
8040 "option tcplog", each log line turns into a much richer format including, but
8041 not limited to, the connection timers, the session status, the connections
8042 numbers, the frontend, backend and server name, and of course the source
8043 address and ports. This option is useful for pure TCP proxies in order to
8044 find which of the client or server disconnects or times out. For normal HTTP
8045 proxies, it's better to use "option httplog" which is even more complete.
8046
Guillaume de Lafond29f45602017-03-31 19:52:15 +02008047 "option tcplog" overrides any previous "log-format" directive.
8048
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008049 See also : "option httplog", and section 8 about logging.
Willy Tarreau844e3c52008-01-11 16:28:18 +01008050
8051
Willy Tarreau844e3c52008-01-11 16:28:18 +01008052option transparent
8053no option transparent
8054 Enable client-side transparent proxying
8055 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +01008056 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +01008057 Arguments : none
8058
8059 This option was introduced in order to provide layer 7 persistence to layer 3
8060 load balancers. The idea is to use the OS's ability to redirect an incoming
8061 connection for a remote address to a local process (here HAProxy), and let
8062 this process know what address was initially requested. When this option is
8063 used, sessions without cookies will be forwarded to the original destination
8064 IP address of the incoming request (which should match that of another
8065 equipment), while requests with cookies will still be forwarded to the
8066 appropriate server.
8067
8068 Note that contrary to a common belief, this option does NOT make HAProxy
8069 present the client's IP to the server when establishing the connection.
8070
Willy Tarreaua1146052011-03-01 09:51:54 +01008071 See also: the "usesrc" argument of the "source" keyword, and the
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008072 "transparent" option of the "bind" keyword.
Willy Tarreau844e3c52008-01-11 16:28:18 +01008073
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008074
Simon Horman98637e52014-06-20 12:30:16 +09008075external-check command <command>
8076 Executable to run when performing an external-check
8077 May be used in sections : defaults | frontend | listen | backend
8078 yes | no | yes | yes
8079
8080 Arguments :
8081 <command> is the external command to run
8082
Simon Horman98637e52014-06-20 12:30:16 +09008083 The arguments passed to the to the command are:
8084
Cyril Bonté777be862014-12-02 21:21:35 +01008085 <proxy_address> <proxy_port> <server_address> <server_port>
Simon Horman98637e52014-06-20 12:30:16 +09008086
Cyril Bonté777be862014-12-02 21:21:35 +01008087 The <proxy_address> and <proxy_port> are derived from the first listener
8088 that is either IPv4, IPv6 or a UNIX socket. In the case of a UNIX socket
8089 listener the proxy_address will be the path of the socket and the
8090 <proxy_port> will be the string "NOT_USED". In a backend section, it's not
8091 possible to determine a listener, and both <proxy_address> and <proxy_port>
8092 will have the string value "NOT_USED".
Simon Horman98637e52014-06-20 12:30:16 +09008093
Cyril Bonté72cda2a2014-12-27 22:28:39 +01008094 Some values are also provided through environment variables.
8095
8096 Environment variables :
8097 HAPROXY_PROXY_ADDR The first bind address if available (or empty if not
8098 applicable, for example in a "backend" section).
8099
8100 HAPROXY_PROXY_ID The backend id.
8101
8102 HAPROXY_PROXY_NAME The backend name.
8103
8104 HAPROXY_PROXY_PORT The first bind port if available (or empty if not
8105 applicable, for example in a "backend" section or
8106 for a UNIX socket).
8107
8108 HAPROXY_SERVER_ADDR The server address.
8109
8110 HAPROXY_SERVER_CURCONN The current number of connections on the server.
8111
8112 HAPROXY_SERVER_ID The server id.
8113
8114 HAPROXY_SERVER_MAXCONN The server max connections.
8115
8116 HAPROXY_SERVER_NAME The server name.
8117
8118 HAPROXY_SERVER_PORT The server port if available (or empty for a UNIX
8119 socket).
8120
8121 PATH The PATH environment variable used when executing
8122 the command may be set using "external-check path".
8123
William Lallemand4d03e432019-06-14 15:35:37 +02008124 See also "2.3. Environment variables" for other variables.
8125
Simon Horman98637e52014-06-20 12:30:16 +09008126 If the command executed and exits with a zero status then the check is
8127 considered to have passed, otherwise the check is considered to have
8128 failed.
8129
8130 Example :
8131 external-check command /bin/true
8132
8133 See also : "external-check", "option external-check", "external-check path"
8134
8135
8136external-check path <path>
8137 The value of the PATH environment variable used when running an external-check
8138 May be used in sections : defaults | frontend | listen | backend
8139 yes | no | yes | yes
8140
8141 Arguments :
8142 <path> is the path used when executing external command to run
8143
8144 The default path is "".
8145
8146 Example :
8147 external-check path "/usr/bin:/bin"
8148
8149 See also : "external-check", "option external-check",
8150 "external-check command"
8151
8152
Emeric Brun647caf12009-06-30 17:57:00 +02008153persist rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02008154persist rdp-cookie(<name>)
Emeric Brun647caf12009-06-30 17:57:00 +02008155 Enable RDP cookie-based persistence
8156 May be used in sections : defaults | frontend | listen | backend
8157 yes | no | yes | yes
8158 Arguments :
8159 <name> is the optional name of the RDP cookie to check. If omitted, the
Willy Tarreau61e28f22010-05-16 22:31:05 +02008160 default cookie name "msts" will be used. There currently is no
8161 valid reason to change this name.
Emeric Brun647caf12009-06-30 17:57:00 +02008162
8163 This statement enables persistence based on an RDP cookie. The RDP cookie
8164 contains all information required to find the server in the list of known
Davor Ocelice9ed2812017-12-25 17:49:28 +01008165 servers. So when this option is set in the backend, the request is analyzed
Emeric Brun647caf12009-06-30 17:57:00 +02008166 and if an RDP cookie is found, it is decoded. If it matches a known server
8167 which is still UP (or if "option persist" is set), then the connection is
8168 forwarded to this server.
8169
8170 Note that this only makes sense in a TCP backend, but for this to work, the
8171 frontend must have waited long enough to ensure that an RDP cookie is present
8172 in the request buffer. This is the same requirement as with the "rdp-cookie"
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01008173 load-balancing method. Thus it is highly recommended to put all statements in
Emeric Brun647caf12009-06-30 17:57:00 +02008174 a single "listen" section.
8175
Willy Tarreau61e28f22010-05-16 22:31:05 +02008176 Also, it is important to understand that the terminal server will emit this
8177 RDP cookie only if it is configured for "token redirection mode", which means
8178 that the "IP address redirection" option is disabled.
8179
Emeric Brun647caf12009-06-30 17:57:00 +02008180 Example :
8181 listen tse-farm
8182 bind :3389
8183 # wait up to 5s for an RDP cookie in the request
8184 tcp-request inspect-delay 5s
8185 tcp-request content accept if RDP_COOKIE
8186 # apply RDP cookie persistence
8187 persist rdp-cookie
8188 # if server is unknown, let's balance on the same cookie.
Cyril Bontédc4d9032012-04-08 21:57:39 +02008189 # alternatively, "balance leastconn" may be useful too.
Emeric Brun647caf12009-06-30 17:57:00 +02008190 balance rdp-cookie
8191 server srv1 1.1.1.1:3389
8192 server srv2 1.1.1.2:3389
8193
Simon Hormanab814e02011-06-24 14:50:20 +09008194 See also : "balance rdp-cookie", "tcp-request", the "req_rdp_cookie" ACL and
8195 the rdp_cookie pattern fetch function.
Emeric Brun647caf12009-06-30 17:57:00 +02008196
8197
Willy Tarreau3a7d2072009-03-05 23:48:25 +01008198rate-limit sessions <rate>
8199 Set a limit on the number of new sessions accepted per second on a frontend
8200 May be used in sections : defaults | frontend | listen | backend
8201 yes | yes | yes | no
8202 Arguments :
8203 <rate> The <rate> parameter is an integer designating the maximum number
8204 of new sessions per second to accept on the frontend.
8205
8206 When the frontend reaches the specified number of new sessions per second, it
8207 stops accepting new connections until the rate drops below the limit again.
8208 During this time, the pending sessions will be kept in the socket's backlog
8209 (in system buffers) and haproxy will not even be aware that sessions are
8210 pending. When applying very low limit on a highly loaded service, it may make
8211 sense to increase the socket's backlog using the "backlog" keyword.
8212
8213 This feature is particularly efficient at blocking connection-based attacks
8214 or service abuse on fragile servers. Since the session rate is measured every
8215 millisecond, it is extremely accurate. Also, the limit applies immediately,
8216 no delay is needed at all to detect the threshold.
8217
8218 Example : limit the connection rate on SMTP to 10 per second max
8219 listen smtp
8220 mode tcp
8221 bind :25
8222 rate-limit sessions 10
Panagiotis Panagiotopoulos7282d8e2016-02-11 16:37:15 +02008223 server smtp1 127.0.0.1:1025
Willy Tarreau3a7d2072009-03-05 23:48:25 +01008224
Willy Tarreaua17c2d92011-07-25 08:16:20 +02008225 Note : when the maximum rate is reached, the frontend's status is not changed
8226 but its sockets appear as "WAITING" in the statistics if the
8227 "socket-stats" option is enabled.
Willy Tarreau3a7d2072009-03-05 23:48:25 +01008228
8229 See also : the "backlog" keyword and the "fe_sess_rate" ACL criterion.
8230
8231
Willy Tarreau2e1dca82012-09-12 08:43:15 +02008232redirect location <loc> [code <code>] <option> [{if | unless} <condition>]
8233redirect prefix <pfx> [code <code>] <option> [{if | unless} <condition>]
8234redirect scheme <sch> [code <code>] <option> [{if | unless} <condition>]
Willy Tarreaub463dfb2008-06-07 23:08:56 +02008235 Return an HTTP redirection if/unless a condition is matched
8236 May be used in sections : defaults | frontend | listen | backend
8237 no | yes | yes | yes
8238
8239 If/unless the condition is matched, the HTTP request will lead to a redirect
Willy Tarreauf285f542010-01-03 20:03:03 +01008240 response. If no condition is specified, the redirect applies unconditionally.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02008241
Willy Tarreau0140f252008-11-19 21:07:09 +01008242 Arguments :
Willy Tarreau2e1dca82012-09-12 08:43:15 +02008243 <loc> With "redirect location", the exact value in <loc> is placed into
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01008244 the HTTP "Location" header. When used in an "http-request" rule,
8245 <loc> value follows the log-format rules and can include some
8246 dynamic values (see Custom Log Format in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02008247
8248 <pfx> With "redirect prefix", the "Location" header is built from the
8249 concatenation of <pfx> and the complete URI path, including the
8250 query string, unless the "drop-query" option is specified (see
8251 below). As a special case, if <pfx> equals exactly "/", then
8252 nothing is inserted before the original URI. It allows one to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01008253 redirect to the same URL (for instance, to insert a cookie). When
8254 used in an "http-request" rule, <pfx> value follows the log-format
8255 rules and can include some dynamic values (see Custom Log Format
8256 in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02008257
8258 <sch> With "redirect scheme", then the "Location" header is built by
8259 concatenating <sch> with "://" then the first occurrence of the
8260 "Host" header, and then the URI path, including the query string
8261 unless the "drop-query" option is specified (see below). If no
8262 path is found or if the path is "*", then "/" is used instead. If
8263 no "Host" header is found, then an empty host component will be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008264 returned, which most recent browsers interpret as redirecting to
Willy Tarreau2e1dca82012-09-12 08:43:15 +02008265 the same host. This directive is mostly used to redirect HTTP to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01008266 HTTPS. When used in an "http-request" rule, <sch> value follows
8267 the log-format rules and can include some dynamic values (see
8268 Custom Log Format in section 8.2.4).
Willy Tarreau0140f252008-11-19 21:07:09 +01008269
8270 <code> The code is optional. It indicates which type of HTTP redirection
Willy Tarreaub67fdc42013-03-29 19:28:11 +01008271 is desired. Only codes 301, 302, 303, 307 and 308 are supported,
8272 with 302 used by default if no code is specified. 301 means
8273 "Moved permanently", and a browser may cache the Location. 302
Baptiste Assmannea849c02015-08-03 11:42:50 +02008274 means "Moved temporarily" and means that the browser should not
Willy Tarreaub67fdc42013-03-29 19:28:11 +01008275 cache the redirection. 303 is equivalent to 302 except that the
8276 browser will fetch the location with a GET method. 307 is just
8277 like 302 but makes it clear that the same method must be reused.
8278 Likewise, 308 replaces 301 if the same method must be used.
Willy Tarreau0140f252008-11-19 21:07:09 +01008279
8280 <option> There are several options which can be specified to adjust the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008281 expected behavior of a redirection :
Willy Tarreau0140f252008-11-19 21:07:09 +01008282
8283 - "drop-query"
8284 When this keyword is used in a prefix-based redirection, then the
8285 location will be set without any possible query-string, which is useful
8286 for directing users to a non-secure page for instance. It has no effect
8287 with a location-type redirect.
8288
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01008289 - "append-slash"
8290 This keyword may be used in conjunction with "drop-query" to redirect
8291 users who use a URL not ending with a '/' to the same one with the '/'.
8292 It can be useful to ensure that search engines will only see one URL.
8293 For this, a return code 301 is preferred.
8294
Willy Tarreau0140f252008-11-19 21:07:09 +01008295 - "set-cookie NAME[=value]"
8296 A "Set-Cookie" header will be added with NAME (and optionally "=value")
8297 to the response. This is sometimes used to indicate that a user has
8298 been seen, for instance to protect against some types of DoS. No other
8299 cookie option is added, so the cookie will be a session cookie. Note
8300 that for a browser, a sole cookie name without an equal sign is
8301 different from a cookie with an equal sign.
8302
8303 - "clear-cookie NAME[=]"
8304 A "Set-Cookie" header will be added with NAME (and optionally "="), but
8305 with the "Max-Age" attribute set to zero. This will tell the browser to
8306 delete this cookie. It is useful for instance on logout pages. It is
8307 important to note that clearing the cookie "NAME" will not remove a
8308 cookie set with "NAME=value". You have to clear the cookie "NAME=" for
8309 that, because the browser makes the difference.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02008310
8311 Example: move the login URL only to HTTPS.
8312 acl clear dst_port 80
8313 acl secure dst_port 8080
8314 acl login_page url_beg /login
Willy Tarreau0140f252008-11-19 21:07:09 +01008315 acl logout url_beg /logout
Willy Tarreau79da4692008-11-19 20:03:04 +01008316 acl uid_given url_reg /login?userid=[^&]+
Willy Tarreau0140f252008-11-19 21:07:09 +01008317 acl cookie_set hdr_sub(cookie) SEEN=1
8318
8319 redirect prefix https://mysite.com set-cookie SEEN=1 if !cookie_set
Willy Tarreau79da4692008-11-19 20:03:04 +01008320 redirect prefix https://mysite.com if login_page !secure
8321 redirect prefix http://mysite.com drop-query if login_page !uid_given
8322 redirect location http://mysite.com/ if !login_page secure
Willy Tarreau0140f252008-11-19 21:07:09 +01008323 redirect location / clear-cookie USERID= if logout
Willy Tarreaub463dfb2008-06-07 23:08:56 +02008324
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01008325 Example: send redirects for request for articles without a '/'.
8326 acl missing_slash path_reg ^/article/[^/]*$
8327 redirect code 301 prefix / drop-query append-slash if missing_slash
8328
Willy Tarreau2e1dca82012-09-12 08:43:15 +02008329 Example: redirect all HTTP traffic to HTTPS when SSL is handled by haproxy.
David BERARDe7153042012-11-03 00:11:31 +01008330 redirect scheme https if !{ ssl_fc }
Willy Tarreau2e1dca82012-09-12 08:43:15 +02008331
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01008332 Example: append 'www.' prefix in front of all hosts not having it
Coen Rosdorff596659b2016-04-11 11:33:49 +02008333 http-request redirect code 301 location \
8334 http://www.%[hdr(host)]%[capture.req.uri] \
8335 unless { hdr_beg(host) -i www }
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01008336
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008337 See section 7 about ACL usage.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02008338
Willy Tarreau303c0352008-01-17 19:01:39 +01008339
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02008340retries <value>
8341 Set the number of retries to perform on a server after a connection failure
8342 May be used in sections: defaults | frontend | listen | backend
8343 yes | no | yes | yes
8344 Arguments :
8345 <value> is the number of times a connection attempt should be retried on
8346 a server when a connection either is refused or times out. The
8347 default value is 3.
8348
8349 It is important to understand that this value applies to the number of
8350 connection attempts, not full requests. When a connection has effectively
8351 been established to a server, there will be no more retry.
8352
8353 In order to avoid immediate reconnections to a server which is restarting,
Joseph Lynch726ab712015-05-11 23:25:34 -07008354 a turn-around timer of min("timeout connect", one second) is applied before
8355 a retry occurs.
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02008356
8357 When "option redispatch" is set, the last retry may be performed on another
8358 server even if a cookie references a different server.
8359
8360 See also : "option redispatch"
8361
8362
Olivier Houcharda254a372019-04-05 15:30:12 +02008363retry-on [list of keywords]
8364 Specify when to attempt to automatically retry a failed request
8365 May be used in sections: defaults | frontend | listen | backend
8366 yes | no | yes | yes
8367 Arguments :
8368 <keywords> is a list of keywords or HTTP status codes, each representing a
8369 type of failure event on which an attempt to retry the request
8370 is desired. Please read the notes at the bottom before changing
8371 this setting. The following keywords are supported :
8372
8373 none never retry
8374
8375 conn-failure retry when the connection or the SSL handshake failed
8376 and the request could not be sent. This is the default.
8377
8378 empty-response retry when the server connection was closed after part
8379 of the request was sent, and nothing was received from
8380 the server. This type of failure may be caused by the
8381 request timeout on the server side, poor network
8382 condition, or a server crash or restart while
8383 processing the request.
8384
Olivier Houcharde3249a92019-05-03 23:01:47 +02008385 junk-response retry when the server returned something not looking
8386 like a complete HTTP response. This includes partial
8387 responses headers as well as non-HTTP contents. It
8388 usually is a bad idea to retry on such events, which
8389 may be caused a configuration issue (wrong server port)
8390 or by the request being harmful to the server (buffer
8391 overflow attack for example).
8392
Olivier Houcharda254a372019-04-05 15:30:12 +02008393 response-timeout the server timeout stroke while waiting for the server
8394 to respond to the request. This may be caused by poor
8395 network condition, the reuse of an idle connection
8396 which has expired on the path, or by the request being
8397 extremely expensive to process. It generally is a bad
8398 idea to retry on such events on servers dealing with
8399 heavy database processing (full scans, etc) as it may
8400 amplify denial of service attacks.
8401
Olivier Houchard865d8392019-05-03 22:46:27 +02008402 0rtt-rejected retry requests which were sent over early data and were
8403 rejected by the server. These requests are generally
8404 considered to be safe to retry.
8405
Olivier Houcharda254a372019-04-05 15:30:12 +02008406 <status> any HTTP status code among "404" (Not Found), "408"
8407 (Request Timeout), "425" (Too Early), "500" (Server
8408 Error), "501" (Not Implemented), "502" (Bad Gateway),
8409 "503" (Service Unavailable), "504" (Gateway Timeout).
8410
Olivier Houchardddf0e032019-05-10 18:05:40 +02008411 all-retryable-errors
8412 retry request for any error that are considered
8413 retryable. This currently activates "conn-failure",
8414 "empty-response", "junk-response", "response-timeout",
8415 "0rtt-rejected", "500", "502", "503", and "504".
8416
Olivier Houcharda254a372019-04-05 15:30:12 +02008417 Using this directive replaces any previous settings with the new ones; it is
8418 not cumulative.
8419
8420 Please note that using anything other than "none" and "conn-failure" requires
8421 to allocate a buffer and copy the whole request into it, so it has memory and
8422 performance impacts. Requests not fitting in a single buffer will never be
8423 retried (see the global tune.bufsize setting).
8424
8425 You have to make sure the application has a replay protection mechanism built
8426 in such as a unique transaction IDs passed in requests, or that replaying the
8427 same request has no consequence, or it is very dangerous to use any retry-on
8428 value beside "conn-failure" and "none". Static file servers and caches are
8429 generally considered safe against any type of retry. Using a status code can
8430 be useful to quickly leave a server showing an abnormal behavior (out of
8431 memory, file system issues, etc), but in this case it may be a good idea to
8432 immediately redispatch the connection to another server (please see "option
8433 redispatch" for this). Last, it is important to understand that most causes
8434 of failures are the requests themselves and that retrying a request causing a
8435 server to misbehave will often make the situation even worse for this server,
8436 or for the whole service in case of redispatch.
8437
8438 Unless you know exactly how the application deals with replayed requests, you
8439 should not use this directive.
8440
8441 The default is "conn-failure".
8442
8443 See also: "retries", "option redispatch", "tune.bufsize"
8444
David du Colombier486df472011-03-17 10:40:26 +01008445server <name> <address>[:[port]] [param*]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008446 Declare a server in a backend
8447 May be used in sections : defaults | frontend | listen | backend
8448 no | no | yes | yes
8449 Arguments :
8450 <name> is the internal name assigned to this server. This name will
Davor Ocelice9ed2812017-12-25 17:49:28 +01008451 appear in logs and alerts. If "http-send-name-header" is
Mark Lamourinec2247f02012-01-04 13:02:01 -05008452 set, it will be added to the request header sent to the server.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008453
David du Colombier486df472011-03-17 10:40:26 +01008454 <address> is the IPv4 or IPv6 address of the server. Alternatively, a
8455 resolvable hostname is supported, but this name will be resolved
8456 during start-up. Address "0.0.0.0" or "*" has a special meaning.
8457 It indicates that the connection will be forwarded to the same IP
Willy Tarreaud669a4f2010-07-13 14:49:50 +02008458 address as the one from the client connection. This is useful in
8459 transparent proxy architectures where the client's connection is
8460 intercepted and haproxy must forward to the original destination
8461 address. This is more or less what the "transparent" keyword does
8462 except that with a server it's possible to limit concurrency and
Willy Tarreau24709282013-03-10 21:32:12 +01008463 to report statistics. Optionally, an address family prefix may be
8464 used before the address to force the family regardless of the
8465 address format, which can be useful to specify a path to a unix
8466 socket with no slash ('/'). Currently supported prefixes are :
8467 - 'ipv4@' -> address is always IPv4
8468 - 'ipv6@' -> address is always IPv6
8469 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02008470 - 'abns@' -> address is in abstract namespace (Linux only)
William Lallemand2fe7dd02018-09-11 16:51:29 +02008471 - 'sockpair@' -> address is the FD of a connected unix
8472 socket or of a socketpair. During a connection, the
8473 backend creates a pair of connected sockets, and passes
8474 one of them over the FD. The bind part will use the
8475 received socket as the client FD. Should be used
8476 carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02008477 You may want to reference some environment variables in the
8478 address parameter, see section 2.3 about environment
Willy Tarreau6a031d12016-11-07 19:42:35 +01008479 variables. The "init-addr" setting can be used to modify the way
8480 IP addresses should be resolved upon startup.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008481
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008482 <port> is an optional port specification. If set, all connections will
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008483 be sent to this port. If unset, the same port the client
8484 connected to will be used. The port may also be prefixed by a "+"
8485 or a "-". In this case, the server's port will be determined by
8486 adding this value to the client's port.
8487
8488 <param*> is a list of parameters for this server. The "server" keywords
8489 accepts an important number of options and has a complete section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008490 dedicated to it. Please refer to section 5 for more details.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008491
8492 Examples :
8493 server first 10.1.1.1:1080 cookie first check inter 1000
8494 server second 10.1.1.2:1080 cookie second check inter 1000
Willy Tarreau24709282013-03-10 21:32:12 +01008495 server transp ipv4@
William Lallemandb2f07452015-05-12 14:27:13 +02008496 server backup "${SRV_BACKUP}:1080" backup
8497 server www1_dc1 "${LAN_DC1}.101:80"
8498 server www1_dc2 "${LAN_DC2}.101:80"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008499
Willy Tarreau55dcaf62015-09-27 15:03:15 +02008500 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
8501 sun_path length is used for the address length. Some other programs
8502 such as socat use the string length only by default. Pass the option
8503 ",unix-tightsocklen=0" to any abstract socket definition in socat to
8504 make it compatible with HAProxy's.
8505
Mark Lamourinec2247f02012-01-04 13:02:01 -05008506 See also: "default-server", "http-send-name-header" and section 5 about
8507 server options
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008508
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02008509server-state-file-name [<file>]
8510 Set the server state file to read, load and apply to servers available in
8511 this backend. It only applies when the directive "load-server-state-from-file"
8512 is set to "local". When <file> is not provided or if this directive is not
8513 set, then backend name is used. If <file> starts with a slash '/', then it is
8514 considered as an absolute path. Otherwise, <file> is concatenated to the
8515 global directive "server-state-file-base".
8516
8517 Example: the minimal configuration below would make HAProxy look for the
8518 state server file '/etc/haproxy/states/bk':
8519
8520 global
8521 server-state-file-base /etc/haproxy/states
8522
Willy Tarreauc9c6cdb2020-03-05 16:03:58 +01008523 backend bk
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02008524 load-server-state-from-file
8525
8526 See also: "server-state-file-base", "load-server-state-from-file", and
8527 "show servers state"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008528
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02008529server-template <prefix> <num | range> <fqdn>[:<port>] [params*]
8530 Set a template to initialize servers with shared parameters.
8531 The names of these servers are built from <prefix> and <num | range> parameters.
8532 May be used in sections : defaults | frontend | listen | backend
8533 no | no | yes | yes
8534
8535 Arguments:
8536 <prefix> A prefix for the server names to be built.
8537
8538 <num | range>
8539 If <num> is provided, this template initializes <num> servers
8540 with 1 up to <num> as server name suffixes. A range of numbers
8541 <num_low>-<num_high> may also be used to use <num_low> up to
8542 <num_high> as server name suffixes.
8543
8544 <fqdn> A FQDN for all the servers this template initializes.
8545
8546 <port> Same meaning as "server" <port> argument (see "server" keyword).
8547
8548 <params*>
8549 Remaining server parameters among all those supported by "server"
8550 keyword.
8551
8552 Examples:
8553 # Initializes 3 servers with srv1, srv2 and srv3 as names,
8554 # google.com as FQDN, and health-check enabled.
8555 server-template srv 1-3 google.com:80 check
8556
8557 # or
8558 server-template srv 3 google.com:80 check
8559
8560 # would be equivalent to:
8561 server srv1 google.com:80 check
8562 server srv2 google.com:80 check
8563 server srv3 google.com:80 check
8564
8565
8566
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008567source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +02008568source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreaud53f96b2009-02-04 18:46:54 +01008569source <addr>[:<port>] [interface <name>]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008570 Set the source address for outgoing connections
8571 May be used in sections : defaults | frontend | listen | backend
8572 yes | no | yes | yes
8573 Arguments :
8574 <addr> is the IPv4 address HAProxy will bind to before connecting to a
8575 server. This address is also used as a source for health checks.
Willy Tarreau24709282013-03-10 21:32:12 +01008576
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008577 The default value of 0.0.0.0 means that the system will select
Willy Tarreau24709282013-03-10 21:32:12 +01008578 the most appropriate address to reach its destination. Optionally
8579 an address family prefix may be used before the address to force
8580 the family regardless of the address format, which can be useful
8581 to specify a path to a unix socket with no slash ('/'). Currently
8582 supported prefixes are :
8583 - 'ipv4@' -> address is always IPv4
8584 - 'ipv6@' -> address is always IPv6
8585 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02008586 - 'abns@' -> address is in abstract namespace (Linux only)
Cyril Bonté307ee1e2015-09-28 23:16:06 +02008587 You may want to reference some environment variables in the
8588 address parameter, see section 2.3 about environment variables.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008589
8590 <port> is an optional port. It is normally not needed but may be useful
8591 in some very specific contexts. The default value of zero means
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02008592 the system will select a free port. Note that port ranges are not
8593 supported in the backend. If you want to force port ranges, you
8594 have to specify them on each "server" line.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008595
8596 <addr2> is the IP address to present to the server when connections are
8597 forwarded in full transparent proxy mode. This is currently only
8598 supported on some patched Linux kernels. When this address is
8599 specified, clients connecting to the server will be presented
8600 with this address, while health checks will still use the address
8601 <addr>.
8602
8603 <port2> is the optional port to present to the server when connections
8604 are forwarded in full transparent proxy mode (see <addr2> above).
8605 The default value of zero means the system will select a free
8606 port.
8607
Willy Tarreaubce70882009-09-07 11:51:47 +02008608 <hdr> is the name of a HTTP header in which to fetch the IP to bind to.
8609 This is the name of a comma-separated header list which can
8610 contain multiple IP addresses. By default, the last occurrence is
8611 used. This is designed to work with the X-Forwarded-For header
Baptiste Assmannea3e73b2013-02-02 23:47:49 +01008612 and to automatically bind to the client's IP address as seen
Willy Tarreaubce70882009-09-07 11:51:47 +02008613 by previous proxy, typically Stunnel. In order to use another
8614 occurrence from the last one, please see the <occ> parameter
8615 below. When the header (or occurrence) is not found, no binding
8616 is performed so that the proxy's default IP address is used. Also
8617 keep in mind that the header name is case insensitive, as for any
8618 HTTP header.
8619
8620 <occ> is the occurrence number of a value to be used in a multi-value
8621 header. This is to be used in conjunction with "hdr_ip(<hdr>)",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04008622 in order to specify which occurrence to use for the source IP
Willy Tarreaubce70882009-09-07 11:51:47 +02008623 address. Positive values indicate a position from the first
8624 occurrence, 1 being the first one. Negative values indicate
8625 positions relative to the last one, -1 being the last one. This
8626 is helpful for situations where an X-Forwarded-For header is set
8627 at the entry point of an infrastructure and must be used several
8628 proxy layers away. When this value is not specified, -1 is
8629 assumed. Passing a zero here disables the feature.
8630
Willy Tarreaud53f96b2009-02-04 18:46:54 +01008631 <name> is an optional interface name to which to bind to for outgoing
8632 traffic. On systems supporting this features (currently, only
8633 Linux), this allows one to bind all traffic to the server to
8634 this interface even if it is not the one the system would select
8635 based on routing tables. This should be used with extreme care.
8636 Note that using this option requires root privileges.
8637
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008638 The "source" keyword is useful in complex environments where a specific
8639 address only is allowed to connect to the servers. It may be needed when a
8640 private address must be used through a public gateway for instance, and it is
8641 known that the system cannot determine the adequate source address by itself.
8642
8643 An extension which is available on certain patched Linux kernels may be used
8644 through the "usesrc" optional keyword. It makes it possible to connect to the
8645 servers with an IP address which does not belong to the system itself. This
8646 is called "full transparent proxy mode". For this to work, the destination
8647 servers have to route their traffic back to this address through the machine
8648 running HAProxy, and IP forwarding must generally be enabled on this machine.
8649
8650 In this "full transparent proxy" mode, it is possible to force a specific IP
8651 address to be presented to the servers. This is not much used in fact. A more
8652 common use is to tell HAProxy to present the client's IP address. For this,
8653 there are two methods :
8654
8655 - present the client's IP and port addresses. This is the most transparent
8656 mode, but it can cause problems when IP connection tracking is enabled on
8657 the machine, because a same connection may be seen twice with different
8658 states. However, this solution presents the huge advantage of not
8659 limiting the system to the 64k outgoing address+port couples, because all
8660 of the client ranges may be used.
8661
8662 - present only the client's IP address and select a spare port. This
8663 solution is still quite elegant but slightly less transparent (downstream
8664 firewalls logs will not match upstream's). It also presents the downside
8665 of limiting the number of concurrent connections to the usual 64k ports.
8666 However, since the upstream and downstream ports are different, local IP
8667 connection tracking on the machine will not be upset by the reuse of the
8668 same session.
8669
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008670 This option sets the default source for all servers in the backend. It may
8671 also be specified in a "defaults" section. Finer source address specification
8672 is possible at the server level using the "source" server option. Refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008673 section 5 for more information.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008674
Baptiste Assmann91bd3372015-07-17 21:59:42 +02008675 In order to work, "usesrc" requires root privileges.
8676
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008677 Examples :
8678 backend private
8679 # Connect to the servers using our 192.168.1.200 source address
8680 source 192.168.1.200
8681
8682 backend transparent_ssl1
8683 # Connect to the SSL farm from the client's source address
8684 source 192.168.1.200 usesrc clientip
8685
8686 backend transparent_ssl2
8687 # Connect to the SSL farm from the client's source address and port
8688 # not recommended if IP conntrack is present on the local machine.
8689 source 192.168.1.200 usesrc client
8690
8691 backend transparent_ssl3
8692 # Connect to the SSL farm from the client's source address. It
8693 # is more conntrack-friendly.
8694 source 192.168.1.200 usesrc clientip
8695
8696 backend transparent_smtp
8697 # Connect to the SMTP farm from the client's source address/port
8698 # with Tproxy version 4.
8699 source 0.0.0.0 usesrc clientip
8700
Willy Tarreaubce70882009-09-07 11:51:47 +02008701 backend transparent_http
8702 # Connect to the servers using the client's IP as seen by previous
8703 # proxy.
8704 source 0.0.0.0 usesrc hdr_ip(x-forwarded-for,-1)
8705
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008706 See also : the "source" server option in section 5, the Tproxy patches for
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008707 the Linux kernel on www.balabit.com, the "bind" keyword.
8708
Willy Tarreau844e3c52008-01-11 16:28:18 +01008709
Cyril Bonté66c327d2010-10-12 00:14:37 +02008710stats admin { if | unless } <cond>
8711 Enable statistics admin level if/unless a condition is matched
8712 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008713 no | yes | yes | yes
Cyril Bonté66c327d2010-10-12 00:14:37 +02008714
8715 This statement enables the statistics admin level if/unless a condition is
8716 matched.
8717
8718 The admin level allows to enable/disable servers from the web interface. By
8719 default, statistics page is read-only for security reasons.
8720
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008721 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8722 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008723 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008724
Cyril Bonté23b39d92011-02-10 22:54:44 +01008725 Currently, the POST request is limited to the buffer size minus the reserved
8726 buffer space, which means that if the list of servers is too long, the
8727 request won't be processed. It is recommended to alter few servers at a
8728 time.
Cyril Bonté66c327d2010-10-12 00:14:37 +02008729
8730 Example :
8731 # statistics admin level only for localhost
8732 backend stats_localhost
8733 stats enable
8734 stats admin if LOCALHOST
8735
8736 Example :
8737 # statistics admin level always enabled because of the authentication
8738 backend stats_auth
8739 stats enable
8740 stats auth admin:AdMiN123
8741 stats admin if TRUE
8742
8743 Example :
8744 # statistics admin level depends on the authenticated user
8745 userlist stats-auth
8746 group admin users admin
8747 user admin insecure-password AdMiN123
8748 group readonly users haproxy
8749 user haproxy insecure-password haproxy
8750
8751 backend stats_auth
8752 stats enable
8753 acl AUTH http_auth(stats-auth)
8754 acl AUTH_ADMIN http_auth_group(stats-auth) admin
8755 stats http-request auth unless AUTH
8756 stats admin if AUTH_ADMIN
8757
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008758 See also : "stats enable", "stats auth", "stats http-request", "nbproc",
8759 "bind-process", section 3.4 about userlists and section 7 about
8760 ACL usage.
Cyril Bonté66c327d2010-10-12 00:14:37 +02008761
8762
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008763stats auth <user>:<passwd>
8764 Enable statistics with authentication and grant access to an account
8765 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008766 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008767 Arguments :
8768 <user> is a user name to grant access to
8769
8770 <passwd> is the cleartext password associated to this user
8771
8772 This statement enables statistics with default settings, and restricts access
8773 to declared users only. It may be repeated as many times as necessary to
8774 allow as many users as desired. When a user tries to access the statistics
8775 without a valid account, a "401 Forbidden" response will be returned so that
8776 the browser asks the user to provide a valid user and password. The real
8777 which will be returned to the browser is configurable using "stats realm".
8778
8779 Since the authentication method is HTTP Basic Authentication, the passwords
8780 circulate in cleartext on the network. Thus, it was decided that the
8781 configuration file would also use cleartext passwords to remind the users
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02008782 that those ones should not be sensitive and not shared with any other account.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008783
8784 It is also possible to reduce the scope of the proxies which appear in the
8785 report using "stats scope".
8786
8787 Though this statement alone is enough to enable statistics reporting, it is
8788 recommended to set all other settings in order to avoid relying on default
8789 unobvious parameters.
8790
8791 Example :
8792 # public access (limited to this backend only)
8793 backend public_www
8794 server srv1 192.168.0.1:80
8795 stats enable
8796 stats hide-version
8797 stats scope .
8798 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008799 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008800 stats auth admin1:AdMiN123
8801 stats auth admin2:AdMiN321
8802
8803 # internal monitoring access (unlimited)
8804 backend private_monitoring
8805 stats enable
8806 stats uri /admin?stats
8807 stats refresh 5s
8808
8809 See also : "stats enable", "stats realm", "stats scope", "stats uri"
8810
8811
8812stats enable
8813 Enable statistics reporting with default settings
8814 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008815 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008816 Arguments : none
8817
8818 This statement enables statistics reporting with default settings defined
8819 at build time. Unless stated otherwise, these settings are used :
8820 - stats uri : /haproxy?stats
8821 - stats realm : "HAProxy Statistics"
8822 - stats auth : no authentication
8823 - stats scope : no restriction
8824
8825 Though this statement alone is enough to enable statistics reporting, it is
8826 recommended to set all other settings in order to avoid relying on default
8827 unobvious parameters.
8828
8829 Example :
8830 # public access (limited to this backend only)
8831 backend public_www
8832 server srv1 192.168.0.1:80
8833 stats enable
8834 stats hide-version
8835 stats scope .
8836 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008837 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008838 stats auth admin1:AdMiN123
8839 stats auth admin2:AdMiN321
8840
8841 # internal monitoring access (unlimited)
8842 backend private_monitoring
8843 stats enable
8844 stats uri /admin?stats
8845 stats refresh 5s
8846
8847 See also : "stats auth", "stats realm", "stats uri"
8848
8849
Willy Tarreaud63335a2010-02-26 12:56:52 +01008850stats hide-version
8851 Enable statistics and hide HAProxy version reporting
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008852 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008853 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008854 Arguments : none
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008855
Willy Tarreaud63335a2010-02-26 12:56:52 +01008856 By default, the stats page reports some useful status information along with
8857 the statistics. Among them is HAProxy's version. However, it is generally
8858 considered dangerous to report precise version to anyone, as it can help them
8859 target known weaknesses with specific attacks. The "stats hide-version"
8860 statement removes the version from the statistics report. This is recommended
8861 for public sites or any site with a weak login/password.
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008862
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02008863 Though this statement alone is enough to enable statistics reporting, it is
8864 recommended to set all other settings in order to avoid relying on default
8865 unobvious parameters.
8866
Willy Tarreaud63335a2010-02-26 12:56:52 +01008867 Example :
8868 # public access (limited to this backend only)
8869 backend public_www
8870 server srv1 192.168.0.1:80
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02008871 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01008872 stats hide-version
8873 stats scope .
8874 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008875 stats realm HAProxy\ Statistics
Willy Tarreaud63335a2010-02-26 12:56:52 +01008876 stats auth admin1:AdMiN123
8877 stats auth admin2:AdMiN321
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008878
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008879 # internal monitoring access (unlimited)
8880 backend private_monitoring
8881 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01008882 stats uri /admin?stats
8883 stats refresh 5s
Krzysztof Piotr Oledzki15514c22010-01-04 16:03:09 +01008884
Willy Tarreaud63335a2010-02-26 12:56:52 +01008885 See also : "stats auth", "stats enable", "stats realm", "stats uri"
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008886
Willy Tarreau983e01e2010-01-11 18:42:06 +01008887
Cyril Bonté2be1b3f2010-09-30 23:46:30 +02008888stats http-request { allow | deny | auth [realm <realm>] }
8889 [ { if | unless } <condition> ]
8890 Access control for statistics
8891
8892 May be used in sections: defaults | frontend | listen | backend
8893 no | no | yes | yes
8894
8895 As "http-request", these set of options allow to fine control access to
8896 statistics. Each option may be followed by if/unless and acl.
8897 First option with matched condition (or option without condition) is final.
8898 For "deny" a 403 error will be returned, for "allow" normal processing is
8899 performed, for "auth" a 401/407 error code is returned so the client
8900 should be asked to enter a username and password.
8901
8902 There is no fixed limit to the number of http-request statements per
8903 instance.
8904
8905 See also : "http-request", section 3.4 about userlists and section 7
8906 about ACL usage.
8907
8908
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008909stats realm <realm>
8910 Enable statistics and set authentication realm
8911 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008912 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008913 Arguments :
8914 <realm> is the name of the HTTP Basic Authentication realm reported to
8915 the browser. The browser uses it to display it in the pop-up
8916 inviting the user to enter a valid username and password.
8917
8918 The realm is read as a single word, so any spaces in it should be escaped
8919 using a backslash ('\').
8920
8921 This statement is useful only in conjunction with "stats auth" since it is
8922 only related to authentication.
8923
8924 Though this statement alone is enough to enable statistics reporting, it is
8925 recommended to set all other settings in order to avoid relying on default
8926 unobvious parameters.
8927
8928 Example :
8929 # public access (limited to this backend only)
8930 backend public_www
8931 server srv1 192.168.0.1:80
8932 stats enable
8933 stats hide-version
8934 stats scope .
8935 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008936 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008937 stats auth admin1:AdMiN123
8938 stats auth admin2:AdMiN321
8939
8940 # internal monitoring access (unlimited)
8941 backend private_monitoring
8942 stats enable
8943 stats uri /admin?stats
8944 stats refresh 5s
8945
8946 See also : "stats auth", "stats enable", "stats uri"
8947
8948
8949stats refresh <delay>
8950 Enable statistics with automatic refresh
8951 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008952 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008953 Arguments :
8954 <delay> is the suggested refresh delay, specified in seconds, which will
8955 be returned to the browser consulting the report page. While the
8956 browser is free to apply any delay, it will generally respect it
8957 and refresh the page this every seconds. The refresh interval may
8958 be specified in any other non-default time unit, by suffixing the
8959 unit after the value, as explained at the top of this document.
8960
8961 This statement is useful on monitoring displays with a permanent page
8962 reporting the load balancer's activity. When set, the HTML report page will
8963 include a link "refresh"/"stop refresh" so that the user can select whether
8964 he wants automatic refresh of the page or not.
8965
8966 Though this statement alone is enough to enable statistics reporting, it is
8967 recommended to set all other settings in order to avoid relying on default
8968 unobvious parameters.
8969
8970 Example :
8971 # public access (limited to this backend only)
8972 backend public_www
8973 server srv1 192.168.0.1:80
8974 stats enable
8975 stats hide-version
8976 stats scope .
8977 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008978 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008979 stats auth admin1:AdMiN123
8980 stats auth admin2:AdMiN321
8981
8982 # internal monitoring access (unlimited)
8983 backend private_monitoring
8984 stats enable
8985 stats uri /admin?stats
8986 stats refresh 5s
8987
8988 See also : "stats auth", "stats enable", "stats realm", "stats uri"
8989
8990
8991stats scope { <name> | "." }
8992 Enable statistics and limit access scope
8993 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008994 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008995 Arguments :
8996 <name> is the name of a listen, frontend or backend section to be
8997 reported. The special name "." (a single dot) designates the
8998 section in which the statement appears.
8999
9000 When this statement is specified, only the sections enumerated with this
9001 statement will appear in the report. All other ones will be hidden. This
9002 statement may appear as many times as needed if multiple sections need to be
9003 reported. Please note that the name checking is performed as simple string
9004 comparisons, and that it is never checked that a give section name really
9005 exists.
9006
9007 Though this statement alone is enough to enable statistics reporting, it is
9008 recommended to set all other settings in order to avoid relying on default
9009 unobvious parameters.
9010
9011 Example :
9012 # public access (limited to this backend only)
9013 backend public_www
9014 server srv1 192.168.0.1:80
9015 stats enable
9016 stats hide-version
9017 stats scope .
9018 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01009019 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009020 stats auth admin1:AdMiN123
9021 stats auth admin2:AdMiN321
9022
9023 # internal monitoring access (unlimited)
9024 backend private_monitoring
9025 stats enable
9026 stats uri /admin?stats
9027 stats refresh 5s
9028
9029 See also : "stats auth", "stats enable", "stats realm", "stats uri"
9030
Willy Tarreaud63335a2010-02-26 12:56:52 +01009031
Willy Tarreauc9705a12010-07-27 20:05:50 +02009032stats show-desc [ <desc> ]
Willy Tarreaud63335a2010-02-26 12:56:52 +01009033 Enable reporting of a description on the statistics page.
9034 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02009035 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01009036
Willy Tarreauc9705a12010-07-27 20:05:50 +02009037 <desc> is an optional description to be reported. If unspecified, the
Willy Tarreaud63335a2010-02-26 12:56:52 +01009038 description from global section is automatically used instead.
9039
9040 This statement is useful for users that offer shared services to their
9041 customers, where node or description should be different for each customer.
9042
9043 Though this statement alone is enough to enable statistics reporting, it is
9044 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +01009045 unobvious parameters. By default description is not shown.
Willy Tarreaud63335a2010-02-26 12:56:52 +01009046
9047 Example :
9048 # internal monitoring access (unlimited)
9049 backend private_monitoring
9050 stats enable
9051 stats show-desc Master node for Europe, Asia, Africa
9052 stats uri /admin?stats
9053 stats refresh 5s
9054
9055 See also: "show-node", "stats enable", "stats uri" and "description" in
9056 global section.
9057
9058
9059stats show-legends
Willy Tarreaued2119c2014-04-24 22:10:39 +02009060 Enable reporting additional information on the statistics page
9061 May be used in sections : defaults | frontend | listen | backend
9062 yes | yes | yes | yes
9063 Arguments : none
9064
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009065 Enable reporting additional information on the statistics page :
Willy Tarreaud63335a2010-02-26 12:56:52 +01009066 - cap: capabilities (proxy)
9067 - mode: one of tcp, http or health (proxy)
9068 - id: SNMP ID (proxy, socket, server)
9069 - IP (socket, server)
9070 - cookie (backend, server)
9071
9072 Though this statement alone is enough to enable statistics reporting, it is
9073 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +01009074 unobvious parameters. Default behavior is not to show this information.
Willy Tarreaud63335a2010-02-26 12:56:52 +01009075
9076 See also: "stats enable", "stats uri".
9077
9078
9079stats show-node [ <name> ]
9080 Enable reporting of a host name on the statistics page.
9081 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02009082 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01009083 Arguments:
9084 <name> is an optional name to be reported. If unspecified, the
9085 node name from global section is automatically used instead.
9086
9087 This statement is useful for users that offer shared services to their
9088 customers, where node or description might be different on a stats page
Davor Ocelice9ed2812017-12-25 17:49:28 +01009089 provided for each customer. Default behavior is not to show host name.
Willy Tarreaud63335a2010-02-26 12:56:52 +01009090
9091 Though this statement alone is enough to enable statistics reporting, it is
9092 recommended to set all other settings in order to avoid relying on default
9093 unobvious parameters.
9094
9095 Example:
9096 # internal monitoring access (unlimited)
9097 backend private_monitoring
9098 stats enable
9099 stats show-node Europe-1
9100 stats uri /admin?stats
9101 stats refresh 5s
9102
9103 See also: "show-desc", "stats enable", "stats uri", and "node" in global
9104 section.
9105
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009106
9107stats uri <prefix>
9108 Enable statistics and define the URI prefix to access them
9109 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02009110 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009111 Arguments :
9112 <prefix> is the prefix of any URI which will be redirected to stats. This
9113 prefix may contain a question mark ('?') to indicate part of a
9114 query string.
9115
9116 The statistics URI is intercepted on the relayed traffic, so it appears as a
9117 page within the normal application. It is strongly advised to ensure that the
9118 selected URI will never appear in the application, otherwise it will never be
9119 possible to reach it in the application.
9120
9121 The default URI compiled in haproxy is "/haproxy?stats", but this may be
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009122 changed at build time, so it's better to always explicitly specify it here.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009123 It is generally a good idea to include a question mark in the URI so that
9124 intermediate proxies refrain from caching the results. Also, since any string
9125 beginning with the prefix will be accepted as a stats request, the question
9126 mark helps ensuring that no valid URI will begin with the same words.
9127
9128 It is sometimes very convenient to use "/" as the URI prefix, and put that
9129 statement in a "listen" instance of its own. That makes it easy to dedicate
9130 an address or a port to statistics only.
9131
9132 Though this statement alone is enough to enable statistics reporting, it is
9133 recommended to set all other settings in order to avoid relying on default
9134 unobvious parameters.
9135
9136 Example :
9137 # public access (limited to this backend only)
9138 backend public_www
9139 server srv1 192.168.0.1:80
9140 stats enable
9141 stats hide-version
9142 stats scope .
9143 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01009144 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009145 stats auth admin1:AdMiN123
9146 stats auth admin2:AdMiN321
9147
9148 # internal monitoring access (unlimited)
9149 backend private_monitoring
9150 stats enable
9151 stats uri /admin?stats
9152 stats refresh 5s
9153
9154 See also : "stats auth", "stats enable", "stats realm"
9155
9156
Willy Tarreaud63335a2010-02-26 12:56:52 +01009157stick match <pattern> [table <table>] [{if | unless} <cond>]
9158 Define a request pattern matching condition to stick a user to a server
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009159 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaud63335a2010-02-26 12:56:52 +01009160 no | no | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009161
9162 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02009163 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009164 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01009165 will be analyzed in the hope to find a matching entry in a
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009166 stickiness table. This rule is mandatory.
9167
9168 <table> is an optional stickiness table name. If unspecified, the same
9169 backend's table is used. A stickiness table is declared using
9170 the "stick-table" statement.
9171
9172 <cond> is an optional matching condition. It makes it possible to match
9173 on a certain criterion only when other conditions are met (or
9174 not met). For instance, it could be used to match on a source IP
9175 address except when a request passes through a known proxy, in
9176 which case we'd match on a header containing that IP address.
9177
9178 Some protocols or applications require complex stickiness rules and cannot
9179 always simply rely on cookies nor hashing. The "stick match" statement
9180 describes a rule to extract the stickiness criterion from an incoming request
9181 or connection. See section 7 for a complete list of possible patterns and
9182 transformation rules.
9183
9184 The table has to be declared using the "stick-table" statement. It must be of
9185 a type compatible with the pattern. By default it is the one which is present
9186 in the same backend. It is possible to share a table with other backends by
9187 referencing it using the "table" keyword. If another table is referenced,
9188 the server's ID inside the backends are used. By default, all server IDs
9189 start at 1 in each backend, so the server ordering is enough. But in case of
9190 doubt, it is highly recommended to force server IDs using their "id" setting.
9191
9192 It is possible to restrict the conditions where a "stick match" statement
9193 will apply, using "if" or "unless" followed by a condition. See section 7 for
9194 ACL based conditions.
9195
9196 There is no limit on the number of "stick match" statements. The first that
9197 applies and matches will cause the request to be directed to the same server
9198 as was used for the request which created the entry. That way, multiple
9199 matches can be used as fallbacks.
9200
9201 The stick rules are checked after the persistence cookies, so they will not
9202 affect stickiness if a cookie has already been used to select a server. That
9203 way, it becomes very easy to insert cookies and match on IP addresses in
9204 order to maintain stickiness between HTTP and HTTPS.
9205
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009206 Note : Consider not using this feature in multi-process mode (nbproc > 1)
9207 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01009208 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009209
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009210 Example :
9211 # forward SMTP users to the same server they just used for POP in the
9212 # last 30 minutes
9213 backend pop
9214 mode tcp
9215 balance roundrobin
9216 stick store-request src
9217 stick-table type ip size 200k expire 30m
9218 server s1 192.168.1.1:110
9219 server s2 192.168.1.1:110
9220
9221 backend smtp
9222 mode tcp
9223 balance roundrobin
9224 stick match src table pop
9225 server s1 192.168.1.1:25
9226 server s2 192.168.1.1:25
9227
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009228 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02009229 about ACLs and samples fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009230
9231
9232stick on <pattern> [table <table>] [{if | unless} <condition>]
9233 Define a request pattern to associate a user to a server
9234 May be used in sections : defaults | frontend | listen | backend
9235 no | no | yes | yes
9236
9237 Note : This form is exactly equivalent to "stick match" followed by
9238 "stick store-request", all with the same arguments. Please refer
9239 to both keywords for details. It is only provided as a convenience
9240 for writing more maintainable configurations.
9241
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009242 Note : Consider not using this feature in multi-process mode (nbproc > 1)
9243 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01009244 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009245
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009246 Examples :
9247 # The following form ...
Willy Tarreauec579d82010-02-26 19:15:04 +01009248 stick on src table pop if !localhost
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009249
9250 # ...is strictly equivalent to this one :
9251 stick match src table pop if !localhost
9252 stick store-request src table pop if !localhost
9253
9254
9255 # Use cookie persistence for HTTP, and stick on source address for HTTPS as
9256 # well as HTTP without cookie. Share the same table between both accesses.
9257 backend http
9258 mode http
9259 balance roundrobin
9260 stick on src table https
9261 cookie SRV insert indirect nocache
9262 server s1 192.168.1.1:80 cookie s1
9263 server s2 192.168.1.1:80 cookie s2
9264
9265 backend https
9266 mode tcp
9267 balance roundrobin
9268 stick-table type ip size 200k expire 30m
9269 stick on src
9270 server s1 192.168.1.1:443
9271 server s2 192.168.1.1:443
9272
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009273 See also : "stick match", "stick store-request", "nbproc" and "bind-process".
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009274
9275
9276stick store-request <pattern> [table <table>] [{if | unless} <condition>]
9277 Define a request pattern used to create an entry in a stickiness table
9278 May be used in sections : defaults | frontend | listen | backend
9279 no | no | yes | yes
9280
9281 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02009282 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009283 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01009284 will be analyzed, extracted and stored in the table once a
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009285 server is selected.
9286
9287 <table> is an optional stickiness table name. If unspecified, the same
9288 backend's table is used. A stickiness table is declared using
9289 the "stick-table" statement.
9290
9291 <cond> is an optional storage condition. It makes it possible to store
9292 certain criteria only when some conditions are met (or not met).
9293 For instance, it could be used to store the source IP address
9294 except when the request passes through a known proxy, in which
9295 case we'd store a converted form of a header containing that IP
9296 address.
9297
9298 Some protocols or applications require complex stickiness rules and cannot
9299 always simply rely on cookies nor hashing. The "stick store-request" statement
9300 describes a rule to decide what to extract from the request and when to do
9301 it, in order to store it into a stickiness table for further requests to
9302 match it using the "stick match" statement. Obviously the extracted part must
9303 make sense and have a chance to be matched in a further request. Storing a
9304 client's IP address for instance often makes sense. Storing an ID found in a
9305 URL parameter also makes sense. Storing a source port will almost never make
9306 any sense because it will be randomly matched. See section 7 for a complete
9307 list of possible patterns and transformation rules.
9308
9309 The table has to be declared using the "stick-table" statement. It must be of
9310 a type compatible with the pattern. By default it is the one which is present
9311 in the same backend. It is possible to share a table with other backends by
9312 referencing it using the "table" keyword. If another table is referenced,
9313 the server's ID inside the backends are used. By default, all server IDs
9314 start at 1 in each backend, so the server ordering is enough. But in case of
9315 doubt, it is highly recommended to force server IDs using their "id" setting.
9316
9317 It is possible to restrict the conditions where a "stick store-request"
9318 statement will apply, using "if" or "unless" followed by a condition. This
9319 condition will be evaluated while parsing the request, so any criteria can be
9320 used. See section 7 for ACL based conditions.
9321
9322 There is no limit on the number of "stick store-request" statements, but
9323 there is a limit of 8 simultaneous stores per request or response. This
9324 makes it possible to store up to 8 criteria, all extracted from either the
9325 request or the response, regardless of the number of rules. Only the 8 first
9326 ones which match will be kept. Using this, it is possible to feed multiple
9327 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01009328 another protocol or access method. Using multiple store-request rules with
9329 the same table is possible and may be used to find the best criterion to rely
9330 on, by arranging the rules by decreasing preference order. Only the first
9331 extracted criterion for a given table will be stored. All subsequent store-
9332 request rules referencing the same table will be skipped and their ACLs will
9333 not be evaluated.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009334
9335 The "store-request" rules are evaluated once the server connection has been
9336 established, so that the table will contain the real server that processed
9337 the request.
9338
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009339 Note : Consider not using this feature in multi-process mode (nbproc > 1)
9340 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01009341 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009342
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009343 Example :
9344 # forward SMTP users to the same server they just used for POP in the
9345 # last 30 minutes
9346 backend pop
9347 mode tcp
9348 balance roundrobin
9349 stick store-request src
9350 stick-table type ip size 200k expire 30m
9351 server s1 192.168.1.1:110
9352 server s2 192.168.1.1:110
9353
9354 backend smtp
9355 mode tcp
9356 balance roundrobin
9357 stick match src table pop
9358 server s1 192.168.1.1:25
9359 server s2 192.168.1.1:25
9360
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009361 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02009362 about ACLs and sample fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009363
9364
Emeric Brun7c6b82e2010-09-24 16:34:28 +02009365stick-table type {ip | integer | string [len <length>] | binary [len <length>]}
Emeric Brunf099e792010-09-27 12:05:28 +02009366 size <size> [expire <expire>] [nopurge] [peers <peersect>]
9367 [store <data_type>]*
Godbach64cef792013-12-04 16:08:22 +08009368 Configure the stickiness table for the current section
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009369 May be used in sections : defaults | frontend | listen | backend
Willy Tarreauc00cdc22010-06-06 16:48:26 +02009370 no | yes | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009371
9372 Arguments :
9373 ip a table declared with "type ip" will only store IPv4 addresses.
9374 This form is very compact (about 50 bytes per entry) and allows
9375 very fast entry lookup and stores with almost no overhead. This
9376 is mainly used to store client source IP addresses.
9377
David du Colombier9a6d3c92011-03-17 10:40:24 +01009378 ipv6 a table declared with "type ipv6" will only store IPv6 addresses.
9379 This form is very compact (about 60 bytes per entry) and allows
9380 very fast entry lookup and stores with almost no overhead. This
9381 is mainly used to store client source IP addresses.
9382
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009383 integer a table declared with "type integer" will store 32bit integers
9384 which can represent a client identifier found in a request for
9385 instance.
9386
9387 string a table declared with "type string" will store substrings of up
9388 to <len> characters. If the string provided by the pattern
9389 extractor is larger than <len>, it will be truncated before
9390 being stored. During matching, at most <len> characters will be
9391 compared between the string in the table and the extracted
9392 pattern. When not specified, the string is automatically limited
Emeric Brun7c6b82e2010-09-24 16:34:28 +02009393 to 32 characters.
9394
9395 binary a table declared with "type binary" will store binary blocks
9396 of <len> bytes. If the block provided by the pattern
9397 extractor is larger than <len>, it will be truncated before
Willy Tarreaube722a22014-06-13 16:31:59 +02009398 being stored. If the block provided by the sample expression
Emeric Brun7c6b82e2010-09-24 16:34:28 +02009399 is shorter than <len>, it will be padded by 0. When not
9400 specified, the block is automatically limited to 32 bytes.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009401
9402 <length> is the maximum number of characters that will be stored in a
Emeric Brun7c6b82e2010-09-24 16:34:28 +02009403 "string" type table (See type "string" above). Or the number
9404 of bytes of the block in "binary" type table. Be careful when
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009405 changing this parameter as memory usage will proportionally
9406 increase.
9407
9408 <size> is the maximum number of entries that can fit in the table. This
Cyril Bonté78caf842010-03-10 22:41:43 +01009409 value directly impacts memory usage. Count approximately
9410 50 bytes per entry, plus the size of a string if any. The size
9411 supports suffixes "k", "m", "g" for 2^10, 2^20 and 2^30 factors.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009412
9413 [nopurge] indicates that we refuse to purge older entries when the table
9414 is full. When not specified and the table is full when haproxy
9415 wants to store an entry in it, it will flush a few of the oldest
9416 entries in order to release some space for the new ones. This is
Davor Ocelice9ed2812017-12-25 17:49:28 +01009417 most often the desired behavior. In some specific cases, it
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009418 be desirable to refuse new entries instead of purging the older
9419 ones. That may be the case when the amount of data to store is
9420 far above the hardware limits and we prefer not to offer access
9421 to new clients than to reject the ones already connected. When
9422 using this parameter, be sure to properly set the "expire"
9423 parameter (see below).
9424
Emeric Brunf099e792010-09-27 12:05:28 +02009425 <peersect> is the name of the peers section to use for replication. Entries
9426 which associate keys to server IDs are kept synchronized with
9427 the remote peers declared in this section. All entries are also
9428 automatically learned from the local peer (old process) during a
9429 soft restart.
9430
Willy Tarreau1abc6732015-05-01 19:21:02 +02009431 NOTE : each peers section may be referenced only by tables
9432 belonging to the same unique process.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009433
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009434 <expire> defines the maximum duration of an entry in the table since it
9435 was last created, refreshed or matched. The expiration delay is
9436 defined using the standard time format, similarly as the various
9437 timeouts. The maximum duration is slightly above 24 days. See
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +03009438 section 2.4 for more information. If this delay is not specified,
Cyril Bontédc4d9032012-04-08 21:57:39 +02009439 the session won't automatically expire, but older entries will
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009440 be removed once full. Be sure not to use the "nopurge" parameter
9441 if not expiration delay is specified.
9442
Willy Tarreau08d5f982010-06-06 13:34:54 +02009443 <data_type> is used to store additional information in the stick-table. This
9444 may be used by ACLs in order to control various criteria related
9445 to the activity of the client matching the stick-table. For each
9446 item specified here, the size of each entry will be inflated so
Willy Tarreauc9705a12010-07-27 20:05:50 +02009447 that the additional data can fit. Several data types may be
9448 stored with an entry. Multiple data types may be specified after
9449 the "store" keyword, as a comma-separated list. Alternatively,
9450 it is possible to repeat the "store" keyword followed by one or
9451 several data types. Except for the "server_id" type which is
9452 automatically detected and enabled, all data types must be
9453 explicitly declared to be stored. If an ACL references a data
9454 type which is not stored, the ACL will simply not match. Some
9455 data types require an argument which must be passed just after
9456 the type between parenthesis. See below for the supported data
9457 types and their arguments.
9458
9459 The data types that can be stored with an entry are the following :
9460 - server_id : this is an integer which holds the numeric ID of the server a
9461 request was assigned to. It is used by the "stick match", "stick store",
9462 and "stick on" rules. It is automatically enabled when referenced.
9463
9464 - gpc0 : first General Purpose Counter. It is a positive 32-bit integer
9465 integer which may be used for anything. Most of the time it will be used
9466 to put a special tag on some entries, for instance to note that a
Davor Ocelice9ed2812017-12-25 17:49:28 +01009467 specific behavior was detected and must be known for future matches.
Willy Tarreauc9705a12010-07-27 20:05:50 +02009468
Willy Tarreauba2ffd12013-05-29 15:54:14 +02009469 - gpc0_rate(<period>) : increment rate of the first General Purpose Counter
9470 over a period. It is a positive 32-bit integer integer which may be used
9471 for anything. Just like <gpc0>, it counts events, but instead of keeping
Davor Ocelice9ed2812017-12-25 17:49:28 +01009472 a cumulative number, it maintains the rate at which the counter is
Willy Tarreauba2ffd12013-05-29 15:54:14 +02009473 incremented. Most of the time it will be used to measure the frequency of
Davor Ocelice9ed2812017-12-25 17:49:28 +01009474 occurrence of certain events (e.g. requests to a specific URL).
Willy Tarreauba2ffd12013-05-29 15:54:14 +02009475
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009476 - gpc1 : second General Purpose Counter. It is a positive 32-bit integer
9477 integer which may be used for anything. Most of the time it will be used
9478 to put a special tag on some entries, for instance to note that a
9479 specific behavior was detected and must be known for future matches.
9480
9481 - gpc1_rate(<period>) : increment rate of the second General Purpose Counter
9482 over a period. It is a positive 32-bit integer integer which may be used
9483 for anything. Just like <gpc1>, it counts events, but instead of keeping
9484 a cumulative number, it maintains the rate at which the counter is
9485 incremented. Most of the time it will be used to measure the frequency of
9486 occurrence of certain events (e.g. requests to a specific URL).
9487
Willy Tarreauc9705a12010-07-27 20:05:50 +02009488 - conn_cnt : Connection Count. It is a positive 32-bit integer which counts
9489 the absolute number of connections received from clients which matched
9490 this entry. It does not mean the connections were accepted, just that
9491 they were received.
9492
9493 - conn_cur : Current Connections. It is a positive 32-bit integer which
9494 stores the concurrent connection counts for the entry. It is incremented
9495 once an incoming connection matches the entry, and decremented once the
9496 connection leaves. That way it is possible to know at any time the exact
9497 number of concurrent connections for an entry.
9498
9499 - conn_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9500 integer parameter <period> which indicates in milliseconds the length
9501 of the period over which the average is measured. It reports the average
9502 incoming connection rate over that period, in connections per period. The
9503 result is an integer which can be matched using ACLs.
9504
9505 - sess_cnt : Session Count. It is a positive 32-bit integer which counts
9506 the absolute number of sessions received from clients which matched this
9507 entry. A session is a connection that was accepted by the layer 4 rules.
9508
9509 - sess_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9510 integer parameter <period> which indicates in milliseconds the length
9511 of the period over which the average is measured. It reports the average
9512 incoming session rate over that period, in sessions per period. The
9513 result is an integer which can be matched using ACLs.
9514
9515 - http_req_cnt : HTTP request Count. It is a positive 32-bit integer which
9516 counts the absolute number of HTTP requests received from clients which
9517 matched this entry. It does not matter whether they are valid requests or
9518 not. Note that this is different from sessions when keep-alive is used on
9519 the client side.
9520
9521 - http_req_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9522 integer parameter <period> which indicates in milliseconds the length
9523 of the period over which the average is measured. It reports the average
9524 HTTP request rate over that period, in requests per period. The result is
9525 an integer which can be matched using ACLs. It does not matter whether
9526 they are valid requests or not. Note that this is different from sessions
9527 when keep-alive is used on the client side.
9528
9529 - http_err_cnt : HTTP Error Count. It is a positive 32-bit integer which
9530 counts the absolute number of HTTP requests errors induced by clients
9531 which matched this entry. Errors are counted on invalid and truncated
9532 requests, as well as on denied or tarpitted requests, and on failed
9533 authentications. If the server responds with 4xx, then the request is
9534 also counted as an error since it's an error triggered by the client
Davor Ocelice9ed2812017-12-25 17:49:28 +01009535 (e.g. vulnerability scan).
Willy Tarreauc9705a12010-07-27 20:05:50 +02009536
9537 - http_err_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9538 integer parameter <period> which indicates in milliseconds the length
9539 of the period over which the average is measured. It reports the average
9540 HTTP request error rate over that period, in requests per period (see
9541 http_err_cnt above for what is accounted as an error). The result is an
9542 integer which can be matched using ACLs.
9543
9544 - bytes_in_cnt : client to server byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +01009545 integer which counts the cumulative number of bytes received from clients
Willy Tarreauc9705a12010-07-27 20:05:50 +02009546 which matched this entry. Headers are included in the count. This may be
9547 used to limit abuse of upload features on photo or video servers.
9548
9549 - bytes_in_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9550 integer parameter <period> which indicates in milliseconds the length
9551 of the period over which the average is measured. It reports the average
9552 incoming bytes rate over that period, in bytes per period. It may be used
9553 to detect users which upload too much and too fast. Warning: with large
9554 uploads, it is possible that the amount of uploaded data will be counted
9555 once upon termination, thus causing spikes in the average transfer speed
9556 instead of having a smooth one. This may partially be smoothed with
9557 "option contstats" though this is not perfect yet. Use of byte_in_cnt is
9558 recommended for better fairness.
9559
9560 - bytes_out_cnt : server to client byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +01009561 integer which counts the cumulative number of bytes sent to clients which
Willy Tarreauc9705a12010-07-27 20:05:50 +02009562 matched this entry. Headers are included in the count. This may be used
9563 to limit abuse of bots sucking the whole site.
9564
9565 - bytes_out_rate(<period>) : frequency counter (takes 12 bytes). It takes
9566 an integer parameter <period> which indicates in milliseconds the length
9567 of the period over which the average is measured. It reports the average
9568 outgoing bytes rate over that period, in bytes per period. It may be used
9569 to detect users which download too much and too fast. Warning: with large
9570 transfers, it is possible that the amount of transferred data will be
9571 counted once upon termination, thus causing spikes in the average
9572 transfer speed instead of having a smooth one. This may partially be
9573 smoothed with "option contstats" though this is not perfect yet. Use of
9574 byte_out_cnt is recommended for better fairness.
Willy Tarreau08d5f982010-06-06 13:34:54 +02009575
Willy Tarreauc00cdc22010-06-06 16:48:26 +02009576 There is only one stick-table per proxy. At the moment of writing this doc,
9577 it does not seem useful to have multiple tables per proxy. If this happens
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009578 to be required, simply create a dummy backend with a stick-table in it and
9579 reference it.
9580
9581 It is important to understand that stickiness based on learning information
9582 has some limitations, including the fact that all learned associations are
Baptiste Assmann123ff042016-03-06 23:29:28 +01009583 lost upon restart unless peers are properly configured to transfer such
9584 information upon restart (recommended). In general it can be good as a
9585 complement but not always as an exclusive stickiness.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009586
Willy Tarreauc9705a12010-07-27 20:05:50 +02009587 Last, memory requirements may be important when storing many data types.
9588 Indeed, storing all indicators above at once in each entry requires 116 bytes
9589 per entry, or 116 MB for a 1-million entries table. This is definitely not
9590 something that can be ignored.
9591
9592 Example:
9593 # Keep track of counters of up to 1 million IP addresses over 5 minutes
9594 # and store a general purpose counter and the average connection rate
9595 # computed over a sliding window of 30 seconds.
9596 stick-table type ip size 1m expire 5m store gpc0,conn_rate(30s)
9597
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +03009598 See also : "stick match", "stick on", "stick store-request", section 2.4
David du Colombiera13d1b92011-03-17 10:40:22 +01009599 about time format and section 7 about ACLs.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009600
9601
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009602stick store-response <pattern> [table <table>] [{if | unless} <condition>]
Baptiste Assmann2f2d2ec2016-03-06 23:27:24 +01009603 Define a response pattern used to create an entry in a stickiness table
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009604 May be used in sections : defaults | frontend | listen | backend
9605 no | no | yes | yes
9606
9607 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02009608 <pattern> is a sample expression rule as described in section 7.3. It
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009609 describes what elements of the response or connection will
Davor Ocelice9ed2812017-12-25 17:49:28 +01009610 be analyzed, extracted and stored in the table once a
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009611 server is selected.
9612
9613 <table> is an optional stickiness table name. If unspecified, the same
9614 backend's table is used. A stickiness table is declared using
9615 the "stick-table" statement.
9616
9617 <cond> is an optional storage condition. It makes it possible to store
9618 certain criteria only when some conditions are met (or not met).
9619 For instance, it could be used to store the SSL session ID only
9620 when the response is a SSL server hello.
9621
9622 Some protocols or applications require complex stickiness rules and cannot
9623 always simply rely on cookies nor hashing. The "stick store-response"
9624 statement describes a rule to decide what to extract from the response and
9625 when to do it, in order to store it into a stickiness table for further
9626 requests to match it using the "stick match" statement. Obviously the
9627 extracted part must make sense and have a chance to be matched in a further
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009628 request. Storing an ID found in a header of a response makes sense.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009629 See section 7 for a complete list of possible patterns and transformation
9630 rules.
9631
9632 The table has to be declared using the "stick-table" statement. It must be of
9633 a type compatible with the pattern. By default it is the one which is present
9634 in the same backend. It is possible to share a table with other backends by
9635 referencing it using the "table" keyword. If another table is referenced,
9636 the server's ID inside the backends are used. By default, all server IDs
9637 start at 1 in each backend, so the server ordering is enough. But in case of
9638 doubt, it is highly recommended to force server IDs using their "id" setting.
9639
9640 It is possible to restrict the conditions where a "stick store-response"
9641 statement will apply, using "if" or "unless" followed by a condition. This
9642 condition will be evaluated while parsing the response, so any criteria can
9643 be used. See section 7 for ACL based conditions.
9644
9645 There is no limit on the number of "stick store-response" statements, but
9646 there is a limit of 8 simultaneous stores per request or response. This
9647 makes it possible to store up to 8 criteria, all extracted from either the
9648 request or the response, regardless of the number of rules. Only the 8 first
9649 ones which match will be kept. Using this, it is possible to feed multiple
9650 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01009651 another protocol or access method. Using multiple store-response rules with
9652 the same table is possible and may be used to find the best criterion to rely
9653 on, by arranging the rules by decreasing preference order. Only the first
9654 extracted criterion for a given table will be stored. All subsequent store-
9655 response rules referencing the same table will be skipped and their ACLs will
9656 not be evaluated. However, even if a store-request rule references a table, a
9657 store-response rule may also use the same table. This means that each table
9658 may learn exactly one element from the request and one element from the
9659 response at once.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009660
9661 The table will contain the real server that processed the request.
9662
9663 Example :
9664 # Learn SSL session ID from both request and response and create affinity.
9665 backend https
9666 mode tcp
9667 balance roundrobin
Cyril Bontédc4d9032012-04-08 21:57:39 +02009668 # maximum SSL session ID length is 32 bytes.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009669 stick-table type binary len 32 size 30k expire 30m
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009670
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009671 acl clienthello req_ssl_hello_type 1
9672 acl serverhello rep_ssl_hello_type 2
9673
9674 # use tcp content accepts to detects ssl client and server hello.
9675 tcp-request inspect-delay 5s
9676 tcp-request content accept if clienthello
9677
9678 # no timeout on response inspect delay by default.
9679 tcp-response content accept if serverhello
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009680
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009681 # SSL session ID (SSLID) may be present on a client or server hello.
9682 # Its length is coded on 1 byte at offset 43 and its value starts
9683 # at offset 44.
9684
9685 # Match and learn on request if client hello.
9686 stick on payload_lv(43,1) if clienthello
9687
9688 # Learn on response if server hello.
9689 stick store-response payload_lv(43,1) if serverhello
Cyril Bontédc4d9032012-04-08 21:57:39 +02009690
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009691 server s1 192.168.1.1:443
9692 server s2 192.168.1.1:443
9693
9694 See also : "stick-table", "stick on", and section 7 about ACLs and pattern
9695 extraction.
9696
9697
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009698tcp-check connect [params*]
9699 Opens a new connection
9700 May be used in sections: defaults | frontend | listen | backend
9701 no | no | yes | yes
9702
9703 When an application lies on more than a single TCP port or when HAProxy
9704 load-balance many services in a single backend, it makes sense to probe all
9705 the services individually before considering a server as operational.
9706
9707 When there are no TCP port configured on the server line neither server port
9708 directive, then the 'tcp-check connect port <port>' must be the first step
9709 of the sequence.
9710
9711 In a tcp-check ruleset a 'connect' is required, it is also mandatory to start
9712 the ruleset with a 'connect' rule. Purpose is to ensure admin know what they
9713 do.
9714
9715 Parameters :
9716 They are optional and can be used to describe how HAProxy should open and
9717 use the TCP connection.
9718
9719 port if not set, check port or server port is used.
9720 It tells HAProxy where to open the connection to.
9721 <port> must be a valid TCP port source integer, from 1 to 65535.
9722
9723 send-proxy send a PROXY protocol string
9724
9725 ssl opens a ciphered connection
9726
9727 Examples:
9728 # check HTTP and HTTPs services on a server.
9729 # first open port 80 thanks to server line port directive, then
9730 # tcp-check opens port 443, ciphered and run a request on it:
9731 option tcp-check
9732 tcp-check connect
9733 tcp-check send GET\ /\ HTTP/1.0\r\n
9734 tcp-check send Host:\ haproxy.1wt.eu\r\n
9735 tcp-check send \r\n
9736 tcp-check expect rstring (2..|3..)
9737 tcp-check connect port 443 ssl
9738 tcp-check send GET\ /\ HTTP/1.0\r\n
9739 tcp-check send Host:\ haproxy.1wt.eu\r\n
9740 tcp-check send \r\n
9741 tcp-check expect rstring (2..|3..)
9742 server www 10.0.0.1 check port 80
9743
9744 # check both POP and IMAP from a single server:
9745 option tcp-check
9746 tcp-check connect port 110
9747 tcp-check expect string +OK\ POP3\ ready
9748 tcp-check connect port 143
9749 tcp-check expect string *\ OK\ IMAP4\ ready
9750 server mail 10.0.0.1 check
9751
9752 See also : "option tcp-check", "tcp-check send", "tcp-check expect"
9753
9754
9755tcp-check expect [!] <match> <pattern>
Davor Ocelice9ed2812017-12-25 17:49:28 +01009756 Specify data to be collected and analyzed during a generic health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009757 May be used in sections: defaults | frontend | listen | backend
9758 no | no | yes | yes
9759
9760 Arguments :
9761 <match> is a keyword indicating how to look for a specific pattern in the
9762 response. The keyword may be one of "string", "rstring" or
9763 binary.
9764 The keyword may be preceded by an exclamation mark ("!") to negate
9765 the match. Spaces are allowed between the exclamation mark and the
9766 keyword. See below for more details on the supported keywords.
9767
9768 <pattern> is the pattern to look for. It may be a string or a regular
9769 expression. If the pattern contains spaces, they must be escaped
9770 with the usual backslash ('\').
9771 If the match is set to binary, then the pattern must be passed as
Davor Ocelice9ed2812017-12-25 17:49:28 +01009772 a series of hexadecimal digits in an even number. Each sequence of
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009773 two digits will represent a byte. The hexadecimal digits may be
9774 used upper or lower case.
9775
9776
9777 The available matches are intentionally similar to their http-check cousins :
9778
9779 string <string> : test the exact string matches in the response buffer.
9780 A health check response will be considered valid if the
9781 response's buffer contains this exact string. If the
9782 "string" keyword is prefixed with "!", then the response
9783 will be considered invalid if the body contains this
9784 string. This can be used to look for a mandatory pattern
9785 in a protocol response, or to detect a failure when a
9786 specific error appears in a protocol banner.
9787
9788 rstring <regex> : test a regular expression on the response buffer.
9789 A health check response will be considered valid if the
9790 response's buffer matches this expression. If the
9791 "rstring" keyword is prefixed with "!", then the response
9792 will be considered invalid if the body matches the
9793 expression.
9794
9795 binary <hexstring> : test the exact string in its hexadecimal form matches
9796 in the response buffer. A health check response will
9797 be considered valid if the response's buffer contains
9798 this exact hexadecimal string.
9799 Purpose is to match data on binary protocols.
9800
9801 It is important to note that the responses will be limited to a certain size
9802 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
9803 Thus, too large responses may not contain the mandatory pattern when using
9804 "string", "rstring" or binary. If a large response is absolutely required, it
9805 is possible to change the default max size by setting the global variable.
9806 However, it is worth keeping in mind that parsing very large responses can
9807 waste some CPU cycles, especially when regular expressions are used, and that
9808 it is always better to focus the checks on smaller resources. Also, in its
9809 current state, the check will not find any string nor regex past a null
9810 character in the response. Similarly it is not possible to request matching
9811 the null character.
9812
9813 Examples :
9814 # perform a POP check
9815 option tcp-check
9816 tcp-check expect string +OK\ POP3\ ready
9817
9818 # perform an IMAP check
9819 option tcp-check
9820 tcp-check expect string *\ OK\ IMAP4\ ready
9821
9822 # look for the redis master server
9823 option tcp-check
9824 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02009825 tcp-check expect string +PONG
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009826 tcp-check send info\ replication\r\n
9827 tcp-check expect string role:master
9828 tcp-check send QUIT\r\n
9829 tcp-check expect string +OK
9830
9831
9832 See also : "option tcp-check", "tcp-check connect", "tcp-check send",
9833 "tcp-check send-binary", "http-check expect", tune.chksize
9834
9835
9836tcp-check send <data>
9837 Specify a string to be sent as a question during a generic health check
9838 May be used in sections: defaults | frontend | listen | backend
9839 no | no | yes | yes
9840
9841 <data> : the data to be sent as a question during a generic health check
9842 session. For now, <data> must be a string.
9843
9844 Examples :
9845 # look for the redis master server
9846 option tcp-check
9847 tcp-check send info\ replication\r\n
9848 tcp-check expect string role:master
9849
9850 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
9851 "tcp-check send-binary", tune.chksize
9852
9853
Davor Ocelice9ed2812017-12-25 17:49:28 +01009854tcp-check send-binary <hexstring>
9855 Specify a hex digits string to be sent as a binary question during a raw
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009856 tcp health check
9857 May be used in sections: defaults | frontend | listen | backend
9858 no | no | yes | yes
9859
9860 <data> : the data to be sent as a question during a generic health check
9861 session. For now, <data> must be a string.
Davor Ocelice9ed2812017-12-25 17:49:28 +01009862 <hexstring> : test the exact string in its hexadecimal form matches in the
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009863 response buffer. A health check response will be considered
9864 valid if the response's buffer contains this exact
9865 hexadecimal string.
9866 Purpose is to send binary data to ask on binary protocols.
9867
9868 Examples :
9869 # redis check in binary
9870 option tcp-check
9871 tcp-check send-binary 50494e470d0a # PING\r\n
9872 tcp-check expect binary 2b504F4e47 # +PONG
9873
9874
9875 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
9876 "tcp-check send", tune.chksize
9877
9878
Willy Tarreaue9656522010-08-17 15:40:09 +02009879tcp-request connection <action> [{if | unless} <condition>]
9880 Perform an action on an incoming connection depending on a layer 4 condition
Willy Tarreau1a687942010-05-23 22:40:30 +02009881 May be used in sections : defaults | frontend | listen | backend
9882 no | yes | yes | no
Willy Tarreaue9656522010-08-17 15:40:09 +02009883 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02009884 <action> defines the action to perform if the condition applies. See
9885 below.
Willy Tarreau1a687942010-05-23 22:40:30 +02009886
Willy Tarreaue9656522010-08-17 15:40:09 +02009887 <condition> is a standard layer4-only ACL-based condition (see section 7).
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009888
9889 Immediately after acceptance of a new incoming connection, it is possible to
9890 evaluate some conditions to decide whether this connection must be accepted
Willy Tarreaue9656522010-08-17 15:40:09 +02009891 or dropped or have its counters tracked. Those conditions cannot make use of
9892 any data contents because the connection has not been read from yet, and the
9893 buffers are not yet allocated. This is used to selectively and very quickly
9894 accept or drop connections from various sources with a very low overhead. If
9895 some contents need to be inspected in order to take the decision, the
9896 "tcp-request content" statements must be used instead.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009897
Willy Tarreaue9656522010-08-17 15:40:09 +02009898 The "tcp-request connection" rules are evaluated in their exact declaration
9899 order. If no rule matches or if there is no rule, the default action is to
9900 accept the incoming connection. There is no specific limit to the number of
9901 rules which may be inserted.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009902
Willy Tarreaua9083d02015-05-08 15:27:59 +02009903 Four types of actions are supported :
Willy Tarreaue9656522010-08-17 15:40:09 +02009904 - accept :
9905 accepts the connection if the condition is true (when used with "if")
9906 or false (when used with "unless"). The first such rule executed ends
9907 the rules evaluation.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009908
Willy Tarreaue9656522010-08-17 15:40:09 +02009909 - reject :
9910 rejects the connection if the condition is true (when used with "if")
9911 or false (when used with "unless"). The first such rule executed ends
9912 the rules evaluation. Rejected connections do not even become a
9913 session, which is why they are accounted separately for in the stats,
9914 as "denied connections". They are not considered for the session
9915 rate-limit and are not logged either. The reason is that these rules
9916 should only be used to filter extremely high connection rates such as
9917 the ones encountered during a massive DDoS attack. Under these extreme
9918 conditions, the simple action of logging each event would make the
9919 system collapse and would considerably lower the filtering capacity. If
9920 logging is absolutely desired, then "tcp-request content" rules should
Willy Tarreau4f614292016-10-21 17:49:36 +02009921 be used instead, as "tcp-request session" rules will not log either.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009922
Willy Tarreau4f0d9192013-06-11 20:40:55 +02009923 - expect-proxy layer4 :
9924 configures the client-facing connection to receive a PROXY protocol
9925 header before any byte is read from the socket. This is equivalent to
9926 having the "accept-proxy" keyword on the "bind" line, except that using
9927 the TCP rule allows the PROXY protocol to be accepted only for certain
9928 IP address ranges using an ACL. This is convenient when multiple layers
9929 of load balancers are passed through by traffic coming from public
9930 hosts.
9931
Bertrand Jacquin90759682016-06-06 15:35:39 +01009932 - expect-netscaler-cip layer4 :
9933 configures the client-facing connection to receive a NetScaler Client
9934 IP insertion protocol header before any byte is read from the socket.
9935 This is equivalent to having the "accept-netscaler-cip" keyword on the
9936 "bind" line, except that using the TCP rule allows the PROXY protocol
9937 to be accepted only for certain IP address ranges using an ACL. This
9938 is convenient when multiple layers of load balancers are passed
9939 through by traffic coming from public hosts.
9940
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009941 - capture <sample> len <length> :
9942 This only applies to "tcp-request content" rules. It captures sample
9943 expression <sample> from the request buffer, and converts it to a
9944 string of at most <len> characters. The resulting string is stored into
9945 the next request "capture" slot, so it will possibly appear next to
9946 some captured HTTP headers. It will then automatically appear in the
9947 logs, and it will be possible to extract it using sample fetch rules to
9948 feed it into headers or anything. The length should be limited given
9949 that this size will be allocated for each capture during the whole
Willy Tarreaua9083d02015-05-08 15:27:59 +02009950 session life. Please check section 7.3 (Fetching samples) and "capture
9951 request header" for more information.
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009952
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009953 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] :
Willy Tarreaue9656522010-08-17 15:40:09 +02009954 enables tracking of sticky counters from current connection. These
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +02009955 rules do not stop evaluation and do not change default action. The
9956 number of counters that may be simultaneously tracked by the same
9957 connection is set in MAX_SESS_STKCTR at build time (reported in
John Roeslerfb2fce12019-07-10 15:45:51 -05009958 haproxy -vv) which defaults to 3, so the track-sc number is between 0
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +02009959 and (MAX_SESS_STCKTR-1). The first "track-sc0" rule executed enables
9960 tracking of the counters of the specified table as the first set. The
9961 first "track-sc1" rule executed enables tracking of the counters of the
9962 specified table as the second set. The first "track-sc2" rule executed
9963 enables tracking of the counters of the specified table as the third
9964 set. It is a recommended practice to use the first set of counters for
9965 the per-frontend counters and the second set for the per-backend ones.
9966 But this is just a guideline, all may be used everywhere.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009967
Willy Tarreaue9656522010-08-17 15:40:09 +02009968 These actions take one or two arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02009969 <key> is mandatory, and is a sample expression rule as described
Willy Tarreau74ca5042013-06-11 23:12:07 +02009970 in section 7.3. It describes what elements of the incoming
Davor Ocelice9ed2812017-12-25 17:49:28 +01009971 request or connection will be analyzed, extracted, combined,
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009972 and used to select which table entry to update the counters.
9973 Note that "tcp-request connection" cannot use content-based
9974 fetches.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009975
Willy Tarreaue9656522010-08-17 15:40:09 +02009976 <table> is an optional table to be used instead of the default one,
9977 which is the stick-table declared in the current proxy. All
9978 the counters for the matches and updates for the key will
9979 then be performed in that table until the session ends.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009980
Willy Tarreaue9656522010-08-17 15:40:09 +02009981 Once a "track-sc*" rule is executed, the key is looked up in the table
9982 and if it is not found, an entry is allocated for it. Then a pointer to
9983 that entry is kept during all the session's life, and this entry's
9984 counters are updated as often as possible, every time the session's
9985 counters are updated, and also systematically when the session ends.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009986 Counters are only updated for events that happen after the tracking has
9987 been started. For example, connection counters will not be updated when
9988 tracking layer 7 information, since the connection event happens before
9989 layer7 information is extracted.
9990
Willy Tarreaue9656522010-08-17 15:40:09 +02009991 If the entry tracks concurrent connection counters, one connection is
9992 counted for as long as the entry is tracked, and the entry will not
9993 expire during that time. Tracking counters also provides a performance
9994 advantage over just checking the keys, because only one table lookup is
9995 performed for all ACL checks that make use of it.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009996
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02009997 - sc-inc-gpc0(<sc-id>):
9998 The "sc-inc-gpc0" increments the GPC0 counter according to the sticky
9999 counter designated by <sc-id>. If an error occurs, this action silently
10000 fails and the actions evaluation continues.
10001
Frédéric Lécaille6778b272018-01-29 15:22:53 +010010002 - sc-inc-gpc1(<sc-id>):
10003 The "sc-inc-gpc1" increments the GPC1 counter according to the sticky
10004 counter designated by <sc-id>. If an error occurs, this action silently
10005 fails and the actions evaluation continues.
10006
Cédric Dufour0d7712d2019-11-06 18:38:53 +010010007 - sc-set-gpt0(<sc-id>) { <int> | <expr> }:
10008 This action sets the 32-bit unsigned GPT0 tag according to the sticky
10009 counter designated by <sc-id> and the value of <int>/<expr>. The
10010 expected result is a boolean. If an error occurs, this action silently
10011 fails and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +020010012
William Lallemand2e785f22016-05-25 01:48:42 +020010013 - set-src <expr> :
10014 Is used to set the source IP address to the value of specified
10015 expression. Useful if you want to mask source IP for privacy.
10016 If you want to provide an IP from a HTTP header use "http-request
Cyril Bonté6c81d5f2018-10-17 00:14:51 +020010017 set-src".
William Lallemand2e785f22016-05-25 01:48:42 +020010018
Cyril Bonté6c81d5f2018-10-17 00:14:51 +020010019 Arguments:
10020 <expr> Is a standard HAProxy expression formed by a sample-fetch
10021 followed by some converters.
William Lallemand2e785f22016-05-25 01:48:42 +020010022
10023 Example:
William Lallemand2e785f22016-05-25 01:48:42 +020010024 tcp-request connection set-src src,ipmask(24)
10025
Willy Tarreau0c630532016-10-21 17:52:58 +020010026 When possible, set-src preserves the original source port as long as the
10027 address family allows it, otherwise the source port is set to 0.
William Lallemand2e785f22016-05-25 01:48:42 +020010028
William Lallemand44be6402016-05-25 01:51:35 +020010029 - set-src-port <expr> :
10030 Is used to set the source port address to the value of specified
10031 expression.
10032
Cyril Bonté6c81d5f2018-10-17 00:14:51 +020010033 Arguments:
10034 <expr> Is a standard HAProxy expression formed by a sample-fetch
10035 followed by some converters.
William Lallemand44be6402016-05-25 01:51:35 +020010036
10037 Example:
William Lallemand44be6402016-05-25 01:51:35 +020010038 tcp-request connection set-src-port int(4000)
10039
Willy Tarreau0c630532016-10-21 17:52:58 +020010040 When possible, set-src-port preserves the original source address as long
10041 as the address family supports a port, otherwise it forces the source
10042 address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +020010043
William Lallemand13e9b0c2016-05-25 02:34:07 +020010044 - set-dst <expr> :
10045 Is used to set the destination IP address to the value of specified
10046 expression. Useful if you want to mask IP for privacy in log.
10047 If you want to provide an IP from a HTTP header use "http-request
10048 set-dst". If you want to connect to the new address/port, use
10049 '0.0.0.0:0' as a server address in the backend.
10050
10051 <expr> Is a standard HAProxy expression formed by a sample-fetch
10052 followed by some converters.
10053
10054 Example:
10055
10056 tcp-request connection set-dst dst,ipmask(24)
10057 tcp-request connection set-dst ipv4(10.0.0.1)
10058
Willy Tarreau0c630532016-10-21 17:52:58 +020010059 When possible, set-dst preserves the original destination port as long as
10060 the address family allows it, otherwise the destination port is set to 0.
10061
William Lallemand13e9b0c2016-05-25 02:34:07 +020010062 - set-dst-port <expr> :
10063 Is used to set the destination port address to the value of specified
10064 expression. If you want to connect to the new address/port, use
10065 '0.0.0.0:0' as a server address in the backend.
10066
10067
10068 <expr> Is a standard HAProxy expression formed by a sample-fetch
10069 followed by some converters.
10070
10071 Example:
10072
10073 tcp-request connection set-dst-port int(4000)
10074
Willy Tarreau0c630532016-10-21 17:52:58 +020010075 When possible, set-dst-port preserves the original destination address as
10076 long as the address family supports a port, otherwise it forces the
10077 destination address to IPv4 "0.0.0.0" before rewriting the port.
10078
Willy Tarreau2d392c22015-08-24 01:43:45 +020010079 - "silent-drop" :
10080 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +010010081 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +020010082 to prevent the client from being notified. The effect it then that the
10083 client still sees an established connection while there's none on
10084 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
10085 except that it doesn't use any local resource at all on the machine
10086 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +010010087 slow down stronger attackers. It is important to understand the impact
10088 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +020010089 client and HAProxy (firewalls, proxies, load balancers) will also keep
10090 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +010010091 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +020010092 TCP_REPAIR socket option is used to block the emission of a TCP
10093 reset. On other systems, the socket's TTL is reduced to 1 so that the
10094 TCP reset doesn't pass the first router, though it's still delivered to
10095 local networks. Do not use it unless you fully understand how it works.
10096
Willy Tarreaue9656522010-08-17 15:40:09 +020010097 Note that the "if/unless" condition is optional. If no condition is set on
10098 the action, it is simply performed unconditionally. That can be useful for
10099 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010100
Willy Tarreaue9656522010-08-17 15:40:09 +020010101 Example: accept all connections from white-listed hosts, reject too fast
10102 connection without counting them, and track accepted connections.
10103 This results in connection rate being capped from abusive sources.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010104
Willy Tarreaue9656522010-08-17 15:40:09 +020010105 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010106 tcp-request connection reject if { src_conn_rate gt 10 }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010107 tcp-request connection track-sc0 src
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010108
Willy Tarreaue9656522010-08-17 15:40:09 +020010109 Example: accept all connections from white-listed hosts, count all other
10110 connections and reject too fast ones. This results in abusive ones
10111 being blocked as long as they don't slow down.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010112
Willy Tarreaue9656522010-08-17 15:40:09 +020010113 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010114 tcp-request connection track-sc0 src
10115 tcp-request connection reject if { sc0_conn_rate gt 10 }
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010116
Willy Tarreau4f0d9192013-06-11 20:40:55 +020010117 Example: enable the PROXY protocol for traffic coming from all known proxies.
10118
10119 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
10120
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010121 See section 7 about ACL usage.
10122
Willy Tarreau4f614292016-10-21 17:49:36 +020010123 See also : "tcp-request session", "tcp-request content", "stick-table"
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010124
10125
Willy Tarreaue9656522010-08-17 15:40:09 +020010126tcp-request content <action> [{if | unless} <condition>]
10127 Perform an action on a new session depending on a layer 4-7 condition
Willy Tarreau62644772008-07-16 18:36:06 +020010128 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +020010129 no | yes | yes | yes
Willy Tarreaue9656522010-08-17 15:40:09 +020010130 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +020010131 <action> defines the action to perform if the condition applies. See
10132 below.
Willy Tarreau62644772008-07-16 18:36:06 +020010133
Willy Tarreaue9656522010-08-17 15:40:09 +020010134 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
Willy Tarreau62644772008-07-16 18:36:06 +020010135
Davor Ocelice9ed2812017-12-25 17:49:28 +010010136 A request's contents can be analyzed at an early stage of request processing
Willy Tarreaue9656522010-08-17 15:40:09 +020010137 called "TCP content inspection". During this stage, ACL-based rules are
10138 evaluated every time the request contents are updated, until either an
10139 "accept" or a "reject" rule matches, or the TCP request inspection delay
10140 expires with no matching rule.
Willy Tarreau62644772008-07-16 18:36:06 +020010141
Willy Tarreaue9656522010-08-17 15:40:09 +020010142 The first difference between these rules and "tcp-request connection" rules
10143 is that "tcp-request content" rules can make use of contents to take a
10144 decision. Most often, these decisions will consider a protocol recognition or
10145 validity. The second difference is that content-based rules can be used in
Willy Tarreauf3338342014-01-28 21:40:28 +010010146 both frontends and backends. In case of HTTP keep-alive with the client, all
10147 tcp-request content rules are evaluated again, so haproxy keeps a record of
10148 what sticky counters were assigned by a "tcp-request connection" versus a
10149 "tcp-request content" rule, and flushes all the content-related ones after
10150 processing an HTTP request, so that they may be evaluated again by the rules
10151 being evaluated again for the next request. This is of particular importance
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010152 when the rule tracks some L7 information or when it is conditioned by an
Willy Tarreauf3338342014-01-28 21:40:28 +010010153 L7-based ACL, since tracking may change between requests.
Willy Tarreau62644772008-07-16 18:36:06 +020010154
Willy Tarreaue9656522010-08-17 15:40:09 +020010155 Content-based rules are evaluated in their exact declaration order. If no
10156 rule matches or if there is no rule, the default action is to accept the
10157 contents. There is no specific limit to the number of rules which may be
10158 inserted.
Willy Tarreau62644772008-07-16 18:36:06 +020010159
Thierry FOURNIER236657b2015-08-19 08:25:14 +020010160 Several types of actions are supported :
Willy Tarreau18bf01e2014-06-13 16:18:52 +020010161 - accept : the request is accepted
Baptiste Assmann333939c2019-01-21 08:34:50 +010010162 - do-resolve: perform a DNS resolution
Willy Tarreau18bf01e2014-06-13 16:18:52 +020010163 - reject : the request is rejected and the connection is closed
10164 - capture : the specified sample expression is captured
Patrick Hemmer268a7072018-05-11 12:52:31 -040010165 - set-priority-class <expr> | set-priority-offset <expr>
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010166 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +020010167 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +010010168 - sc-inc-gpc1(<sc-id>)
Cédric Dufour0d7712d2019-11-06 18:38:53 +010010169 - sc-set-gpt0(<sc-id>) { <int> | <expr> }
Baptiste Assmanne1afd4f2019-04-18 16:21:13 +020010170 - set-dst <expr>
10171 - set-dst-port <expr>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010172 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +010010173 - unset-var(<var-name>)
Willy Tarreau2d392c22015-08-24 01:43:45 +020010174 - silent-drop
Davor Ocelice9ed2812017-12-25 17:49:28 +010010175 - send-spoe-group <engine-name> <group-name>
Christopher Faulet579d83b2019-11-22 15:34:17 +010010176 - use-service <service-name>
Willy Tarreau62644772008-07-16 18:36:06 +020010177
Willy Tarreaue9656522010-08-17 15:40:09 +020010178 They have the same meaning as their counter-parts in "tcp-request connection"
10179 so please refer to that section for a complete description.
Baptiste Assmann333939c2019-01-21 08:34:50 +010010180 For "do-resolve" action, please check the "http-request do-resolve"
10181 configuration section.
Willy Tarreau62644772008-07-16 18:36:06 +020010182
Willy Tarreauf3338342014-01-28 21:40:28 +010010183 While there is nothing mandatory about it, it is recommended to use the
10184 track-sc0 in "tcp-request connection" rules, track-sc1 for "tcp-request
10185 content" rules in the frontend, and track-sc2 for "tcp-request content"
10186 rules in the backend, because that makes the configuration more readable
10187 and easier to troubleshoot, but this is just a guideline and all counters
10188 may be used everywhere.
Willy Tarreau62644772008-07-16 18:36:06 +020010189
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010010190 Note that the "if/unless" condition is optional. If no condition is set on
Willy Tarreaue9656522010-08-17 15:40:09 +020010191 the action, it is simply performed unconditionally. That can be useful for
10192 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau62644772008-07-16 18:36:06 +020010193
Willy Tarreaue9656522010-08-17 15:40:09 +020010194 It is perfectly possible to match layer 7 contents with "tcp-request content"
Willy Tarreauc0239e02012-04-16 14:42:55 +020010195 rules, since HTTP-specific ACL matches are able to preliminarily parse the
10196 contents of a buffer before extracting the required data. If the buffered
10197 contents do not parse as a valid HTTP message, then the ACL does not match.
10198 The parser which is involved there is exactly the same as for all other HTTP
Willy Tarreauf3338342014-01-28 21:40:28 +010010199 processing, so there is no risk of parsing something differently. In an HTTP
10200 backend connected to from an HTTP frontend, it is guaranteed that HTTP
10201 contents will always be immediately present when the rule is evaluated first.
Willy Tarreau62644772008-07-16 18:36:06 +020010202
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010010203 Tracking layer7 information is also possible provided that the information
Willy Tarreau4d54c7c2014-09-16 15:48:15 +020010204 are present when the rule is processed. The rule processing engine is able to
10205 wait until the inspect delay expires when the data to be tracked is not yet
10206 available.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010010207
Baptiste Assmanne1afd4f2019-04-18 16:21:13 +020010208 The "set-dst" and "set-dst-port" are used to set respectively the destination
10209 IP and port. More information on how to use it at "http-request set-dst".
10210
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010211 The "set-var" is used to set the content of a variable. The variable is
Willy Tarreau4f614292016-10-21 17:49:36 +020010212 declared inline. For "tcp-request session" rules, only session-level
10213 variables can be used, without any layer7 contents.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010214
Daniel Schneller0b547052016-03-21 20:46:57 +010010215 <var-name> The name of the variable starts with an indication about
10216 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010010217 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010010218 "sess" : the variable is shared with the whole session
10219 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010220 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +010010221 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010222 processing
Daniel Schneller0b547052016-03-21 20:46:57 +010010223 "res" : the variable is shared only during response
10224 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010225 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +010010226 The name may only contain characters 'a-z', 'A-Z', '0-9',
10227 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010228
10229 <expr> Is a standard HAProxy expression formed by a sample-fetch
10230 followed by some converters.
10231
Christopher Faulet85d79c92016-11-09 16:54:56 +010010232 The "unset-var" is used to unset a variable. See above for details about
10233 <var-name>.
10234
Patrick Hemmer268a7072018-05-11 12:52:31 -040010235 The "set-priority-class" is used to set the queue priority class of the
10236 current request. The value must be a sample expression which converts to an
10237 integer in the range -2047..2047. Results outside this range will be
10238 truncated. The priority class determines the order in which queued requests
10239 are processed. Lower values have higher priority.
10240
10241 The "set-priority-offset" is used to set the queue priority timestamp offset
10242 of the current request. The value must be a sample expression which converts
10243 to an integer in the range -524287..524287. Results outside this range will be
10244 truncated. When a request is queued, it is ordered first by the priority
10245 class, then by the current timestamp adjusted by the given offset in
10246 milliseconds. Lower values have higher priority.
10247 Note that the resulting timestamp is is only tracked with enough precision for
10248 524,287ms (8m44s287ms). If the request is queued long enough to where the
10249 adjusted timestamp exceeds this value, it will be misidentified as highest
10250 priority. Thus it is important to set "timeout queue" to a value, where when
10251 combined with the offset, does not exceed this limit.
10252
Christopher Faulet76c09ef2017-09-21 11:03:52 +020010253 The "send-spoe-group" is used to trigger sending of a group of SPOE
10254 messages. To do so, the SPOE engine used to send messages must be defined, as
10255 well as the SPOE group to send. Of course, the SPOE engine must refer to an
10256 existing SPOE filter. If not engine name is provided on the SPOE filter line,
10257 the SPOE agent name must be used.
10258
10259 <engine-name> The SPOE engine name.
10260
10261 <group-name> The SPOE group name as specified in the engine configuration.
10262
Christopher Faulet579d83b2019-11-22 15:34:17 +010010263 The "use-service" is used to executes a TCP service which will reply to the
10264 request and stop the evaluation of the rules. This service may choose to
10265 reply by sending any valid response or it may immediately close the
10266 connection without sending anything. Outside natives services, it is possible
10267 to write your own services in Lua. No further "tcp-request" rules are
10268 evaluated.
10269
10270 Example:
10271 tcp-request content use-service lua.deny { src -f /etc/haproxy/blacklist.lst }
10272
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010273 Example:
10274
10275 tcp-request content set-var(sess.my_var) src
Christopher Faulet85d79c92016-11-09 16:54:56 +010010276 tcp-request content unset-var(sess.my_var2)
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010277
Willy Tarreau62644772008-07-16 18:36:06 +020010278 Example:
Willy Tarreaue9656522010-08-17 15:40:09 +020010279 # Accept HTTP requests containing a Host header saying "example.com"
10280 # and reject everything else.
10281 acl is_host_com hdr(Host) -i example.com
10282 tcp-request inspect-delay 30s
Willy Tarreauc0239e02012-04-16 14:42:55 +020010283 tcp-request content accept if is_host_com
Willy Tarreaue9656522010-08-17 15:40:09 +020010284 tcp-request content reject
10285
10286 Example:
Willy Tarreau62644772008-07-16 18:36:06 +020010287 # reject SMTP connection if client speaks first
10288 tcp-request inspect-delay 30s
10289 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010290 tcp-request content reject if content_present
Willy Tarreau62644772008-07-16 18:36:06 +020010291
10292 # Forward HTTPS connection only if client speaks
10293 tcp-request inspect-delay 30s
10294 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010295 tcp-request content accept if content_present
Willy Tarreaue9656522010-08-17 15:40:09 +020010296 tcp-request content reject
10297
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010010298 Example:
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030010299 # Track the last IP(stick-table type string) from X-Forwarded-For
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010010300 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +020010301 tcp-request content track-sc0 hdr(x-forwarded-for,-1)
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030010302 # Or track the last IP(stick-table type ip|ipv6) from X-Forwarded-For
10303 tcp-request content track-sc0 req.hdr_ip(x-forwarded-for,-1)
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010010304
10305 Example:
10306 # track request counts per "base" (concatenation of Host+URL)
10307 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +020010308 tcp-request content track-sc0 base table req-rate
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010010309
Willy Tarreaue9656522010-08-17 15:40:09 +020010310 Example: track per-frontend and per-backend counters, block abusers at the
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030010311 frontend when the backend detects abuse(and marks gpc0).
Willy Tarreaue9656522010-08-17 15:40:09 +020010312
10313 frontend http
Davor Ocelice9ed2812017-12-25 17:49:28 +010010314 # Use General Purpose Counter 0 in SC0 as a global abuse counter
Willy Tarreaue9656522010-08-17 15:40:09 +020010315 # protecting all our sites
10316 stick-table type ip size 1m expire 5m store gpc0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010317 tcp-request connection track-sc0 src
10318 tcp-request connection reject if { sc0_get_gpc0 gt 0 }
Willy Tarreaue9656522010-08-17 15:40:09 +020010319 ...
10320 use_backend http_dynamic if { path_end .php }
10321
10322 backend http_dynamic
10323 # if a source makes too fast requests to this dynamic site (tracked
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010324 # by SC1), block it globally in the frontend.
Willy Tarreaue9656522010-08-17 15:40:09 +020010325 stick-table type ip size 1m expire 5m store http_req_rate(10s)
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010326 acl click_too_fast sc1_http_req_rate gt 10
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030010327 acl mark_as_abuser sc0_inc_gpc0(http) gt 0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010328 tcp-request content track-sc1 src
Willy Tarreaue9656522010-08-17 15:40:09 +020010329 tcp-request content reject if click_too_fast mark_as_abuser
Willy Tarreau62644772008-07-16 18:36:06 +020010330
Willy Tarreauc57f0e22009-05-10 13:12:33 +020010331 See section 7 about ACL usage.
Willy Tarreau62644772008-07-16 18:36:06 +020010332
Jarno Huuskonen95b012b2017-04-06 13:59:14 +030010333 See also : "tcp-request connection", "tcp-request session",
10334 "tcp-request inspect-delay", and "http-request".
Willy Tarreau62644772008-07-16 18:36:06 +020010335
10336
10337tcp-request inspect-delay <timeout>
10338 Set the maximum allowed time to wait for data during content inspection
10339 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +020010340 no | yes | yes | yes
Willy Tarreau62644772008-07-16 18:36:06 +020010341 Arguments :
10342 <timeout> is the timeout value specified in milliseconds by default, but
10343 can be in any other unit if the number is suffixed by the unit,
10344 as explained at the top of this document.
10345
10346 People using haproxy primarily as a TCP relay are often worried about the
10347 risk of passing any type of protocol to a server without any analysis. In
10348 order to be able to analyze the request contents, we must first withhold
10349 the data then analyze them. This statement simply enables withholding of
10350 data for at most the specified amount of time.
10351
Willy Tarreaufb356202010-08-03 14:02:05 +020010352 TCP content inspection applies very early when a connection reaches a
10353 frontend, then very early when the connection is forwarded to a backend. This
10354 means that a connection may experience a first delay in the frontend and a
10355 second delay in the backend if both have tcp-request rules.
10356
Willy Tarreau62644772008-07-16 18:36:06 +020010357 Note that when performing content inspection, haproxy will evaluate the whole
10358 rules for every new chunk which gets in, taking into account the fact that
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010010359 those data are partial. If no rule matches before the aforementioned delay,
Willy Tarreau62644772008-07-16 18:36:06 +020010360 a last check is performed upon expiration, this time considering that the
Willy Tarreaud869b242009-03-15 14:43:58 +010010361 contents are definitive. If no delay is set, haproxy will not wait at all
10362 and will immediately apply a verdict based on the available information.
10363 Obviously this is unlikely to be very useful and might even be racy, so such
10364 setups are not recommended.
Willy Tarreau62644772008-07-16 18:36:06 +020010365
10366 As soon as a rule matches, the request is released and continues as usual. If
10367 the timeout is reached and no rule matches, the default policy will be to let
10368 it pass through unaffected.
10369
10370 For most protocols, it is enough to set it to a few seconds, as most clients
10371 send the full request immediately upon connection. Add 3 or more seconds to
10372 cover TCP retransmits but that's all. For some protocols, it may make sense
Willy Tarreaud72758d2010-01-12 10:42:19 +010010373 to use large values, for instance to ensure that the client never talks
Davor Ocelice9ed2812017-12-25 17:49:28 +010010374 before the server (e.g. SMTP), or to wait for a client to talk before passing
10375 data to the server (e.g. SSL). Note that the client timeout must cover at
Willy Tarreaub824b002010-09-29 16:36:16 +020010376 least the inspection delay, otherwise it will expire first. If the client
10377 closes the connection or if the buffer is full, the delay immediately expires
10378 since the contents will not be able to change anymore.
Willy Tarreau62644772008-07-16 18:36:06 +020010379
Willy Tarreau55165fe2009-05-10 12:02:55 +020010380 See also : "tcp-request content accept", "tcp-request content reject",
Willy Tarreau62644772008-07-16 18:36:06 +020010381 "timeout client".
10382
10383
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010384tcp-response content <action> [{if | unless} <condition>]
10385 Perform an action on a session response depending on a layer 4-7 condition
10386 May be used in sections : defaults | frontend | listen | backend
10387 no | no | yes | yes
10388 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +020010389 <action> defines the action to perform if the condition applies. See
10390 below.
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010391
10392 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
10393
Davor Ocelice9ed2812017-12-25 17:49:28 +010010394 Response contents can be analyzed at an early stage of response processing
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010395 called "TCP content inspection". During this stage, ACL-based rules are
10396 evaluated every time the response contents are updated, until either an
Willy Tarreaucc1e04b2013-09-11 23:20:29 +020010397 "accept", "close" or a "reject" rule matches, or a TCP response inspection
10398 delay is set and expires with no matching rule.
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010399
10400 Most often, these decisions will consider a protocol recognition or validity.
10401
10402 Content-based rules are evaluated in their exact declaration order. If no
10403 rule matches or if there is no rule, the default action is to accept the
10404 contents. There is no specific limit to the number of rules which may be
10405 inserted.
10406
Thierry FOURNIER236657b2015-08-19 08:25:14 +020010407 Several types of actions are supported :
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010408 - accept :
10409 accepts the response if the condition is true (when used with "if")
10410 or false (when used with "unless"). The first such rule executed ends
10411 the rules evaluation.
10412
Willy Tarreaucc1e04b2013-09-11 23:20:29 +020010413 - close :
10414 immediately closes the connection with the server if the condition is
10415 true (when used with "if"), or false (when used with "unless"). The
10416 first such rule executed ends the rules evaluation. The main purpose of
10417 this action is to force a connection to be finished between a client
10418 and a server after an exchange when the application protocol expects
10419 some long time outs to elapse first. The goal is to eliminate idle
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010420 connections which take significant resources on servers with certain
Willy Tarreaucc1e04b2013-09-11 23:20:29 +020010421 protocols.
10422
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010423 - reject :
10424 rejects the response if the condition is true (when used with "if")
10425 or false (when used with "unless"). The first such rule executed ends
Jamie Gloudonaaa21002012-08-25 00:18:33 -040010426 the rules evaluation. Rejected session are immediately closed.
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010427
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010428 - set-var(<var-name>) <expr>
10429 Sets a variable.
10430
Christopher Faulet85d79c92016-11-09 16:54:56 +010010431 - unset-var(<var-name>)
10432 Unsets a variable.
10433
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +020010434 - sc-inc-gpc0(<sc-id>):
10435 This action increments the GPC0 counter according to the sticky
10436 counter designated by <sc-id>. If an error occurs, this action fails
10437 silently and the actions evaluation continues.
10438
Frédéric Lécaille6778b272018-01-29 15:22:53 +010010439 - sc-inc-gpc1(<sc-id>):
10440 This action increments the GPC1 counter according to the sticky
10441 counter designated by <sc-id>. If an error occurs, this action fails
10442 silently and the actions evaluation continues.
10443
Cédric Dufour0d7712d2019-11-06 18:38:53 +010010444 - sc-set-gpt0(<sc-id>) { <int> | <expr> }
10445 This action sets the 32-bit unsigned GPT0 tag according to the sticky
10446 counter designated by <sc-id> and the value of <int>/<expr>. The
10447 expected result is a boolean. If an error occurs, this action silently
10448 fails and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +020010449
Willy Tarreau2d392c22015-08-24 01:43:45 +020010450 - "silent-drop" :
10451 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +010010452 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +020010453 to prevent the client from being notified. The effect it then that the
10454 client still sees an established connection while there's none on
10455 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
10456 except that it doesn't use any local resource at all on the machine
10457 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +010010458 slow down stronger attackers. It is important to understand the impact
10459 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +020010460 client and HAProxy (firewalls, proxies, load balancers) will also keep
10461 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +010010462 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +020010463 TCP_REPAIR socket option is used to block the emission of a TCP
10464 reset. On other systems, the socket's TTL is reduced to 1 so that the
10465 TCP reset doesn't pass the first router, though it's still delivered to
10466 local networks. Do not use it unless you fully understand how it works.
10467
Christopher Faulet76c09ef2017-09-21 11:03:52 +020010468 - send-spoe-group <engine-name> <group-name>
10469 Send a group of SPOE messages.
10470
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010471 Note that the "if/unless" condition is optional. If no condition is set on
10472 the action, it is simply performed unconditionally. That can be useful for
10473 for changing the default action to a reject.
10474
Jamie Gloudonaaa21002012-08-25 00:18:33 -040010475 It is perfectly possible to match layer 7 contents with "tcp-response
10476 content" rules, but then it is important to ensure that a full response has
10477 been buffered, otherwise no contents will match. In order to achieve this,
10478 the best solution involves detecting the HTTP protocol during the inspection
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010479 period.
10480
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010481 The "set-var" is used to set the content of a variable. The variable is
10482 declared inline.
10483
Daniel Schneller0b547052016-03-21 20:46:57 +010010484 <var-name> The name of the variable starts with an indication about
10485 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010010486 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010010487 "sess" : the variable is shared with the whole session
10488 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010489 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +010010490 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010491 processing
Daniel Schneller0b547052016-03-21 20:46:57 +010010492 "res" : the variable is shared only during response
10493 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010494 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +010010495 The name may only contain characters 'a-z', 'A-Z', '0-9',
10496 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010497
10498 <expr> Is a standard HAProxy expression formed by a sample-fetch
10499 followed by some converters.
10500
10501 Example:
10502
10503 tcp-request content set-var(sess.my_var) src
10504
Christopher Faulet85d79c92016-11-09 16:54:56 +010010505 The "unset-var" is used to unset a variable. See above for details about
10506 <var-name>.
10507
10508 Example:
10509
10510 tcp-request content unset-var(sess.my_var)
10511
Christopher Faulet76c09ef2017-09-21 11:03:52 +020010512 The "send-spoe-group" is used to trigger sending of a group of SPOE
10513 messages. To do so, the SPOE engine used to send messages must be defined, as
10514 well as the SPOE group to send. Of course, the SPOE engine must refer to an
10515 existing SPOE filter. If not engine name is provided on the SPOE filter line,
10516 the SPOE agent name must be used.
10517
10518 <engine-name> The SPOE engine name.
10519
10520 <group-name> The SPOE group name as specified in the engine configuration.
10521
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010522 See section 7 about ACL usage.
10523
10524 See also : "tcp-request content", "tcp-response inspect-delay"
10525
10526
Willy Tarreau4f614292016-10-21 17:49:36 +020010527tcp-request session <action> [{if | unless} <condition>]
10528 Perform an action on a validated session depending on a layer 5 condition
10529 May be used in sections : defaults | frontend | listen | backend
10530 no | yes | yes | no
10531 Arguments :
10532 <action> defines the action to perform if the condition applies. See
10533 below.
10534
10535 <condition> is a standard layer5-only ACL-based condition (see section 7).
10536
Davor Ocelice9ed2812017-12-25 17:49:28 +010010537 Once a session is validated, (i.e. after all handshakes have been completed),
Willy Tarreau4f614292016-10-21 17:49:36 +020010538 it is possible to evaluate some conditions to decide whether this session
10539 must be accepted or dropped or have its counters tracked. Those conditions
10540 cannot make use of any data contents because no buffers are allocated yet and
10541 the processing cannot wait at this stage. The main use case it to copy some
10542 early information into variables (since variables are accessible in the
10543 session), or to keep track of some information collected after the handshake,
10544 such as SSL-level elements (SNI, ciphers, client cert's CN) or information
Davor Ocelice9ed2812017-12-25 17:49:28 +010010545 from the PROXY protocol header (e.g. track a source forwarded this way). The
Willy Tarreau4f614292016-10-21 17:49:36 +020010546 extracted information can thus be copied to a variable or tracked using
10547 "track-sc" rules. Of course it is also possible to decide to accept/reject as
10548 with other rulesets. Most operations performed here could also be performed
10549 in "tcp-request content" rules, except that in HTTP these rules are evaluated
10550 for each new request, and that might not always be acceptable. For example a
10551 rule might increment a counter on each evaluation. It would also be possible
10552 that a country is resolved by geolocation from the source IP address,
10553 assigned to a session-wide variable, then the source address rewritten from
10554 an HTTP header for all requests. If some contents need to be inspected in
10555 order to take the decision, the "tcp-request content" statements must be used
10556 instead.
10557
10558 The "tcp-request session" rules are evaluated in their exact declaration
10559 order. If no rule matches or if there is no rule, the default action is to
10560 accept the incoming session. There is no specific limit to the number of
10561 rules which may be inserted.
10562
10563 Several types of actions are supported :
10564 - accept : the request is accepted
10565 - reject : the request is rejected and the connection is closed
10566 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
10567 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +010010568 - sc-inc-gpc1(<sc-id>)
Cédric Dufour0d7712d2019-11-06 18:38:53 +010010569 - sc-set-gpt0(<sc-id>) { <int> | <expr> }
Willy Tarreau4f614292016-10-21 17:49:36 +020010570 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +010010571 - unset-var(<var-name>)
Willy Tarreau4f614292016-10-21 17:49:36 +020010572 - silent-drop
10573
10574 These actions have the same meaning as their respective counter-parts in
10575 "tcp-request connection" and "tcp-request content", so please refer to these
10576 sections for a complete description.
10577
10578 Note that the "if/unless" condition is optional. If no condition is set on
10579 the action, it is simply performed unconditionally. That can be useful for
10580 "track-sc*" actions as well as for changing the default action to a reject.
10581
10582 Example: track the original source address by default, or the one advertised
10583 in the PROXY protocol header for connection coming from the local
10584 proxies. The first connection-level rule enables receipt of the
10585 PROXY protocol for these ones, the second rule tracks whatever
10586 address we decide to keep after optional decoding.
10587
10588 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
10589 tcp-request session track-sc0 src
10590
10591 Example: accept all sessions from white-listed hosts, reject too fast
10592 sessions without counting them, and track accepted sessions.
10593 This results in session rate being capped from abusive sources.
10594
10595 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
10596 tcp-request session reject if { src_sess_rate gt 10 }
10597 tcp-request session track-sc0 src
10598
10599 Example: accept all sessions from white-listed hosts, count all other
10600 sessions and reject too fast ones. This results in abusive ones
10601 being blocked as long as they don't slow down.
10602
10603 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
10604 tcp-request session track-sc0 src
10605 tcp-request session reject if { sc0_sess_rate gt 10 }
10606
10607 See section 7 about ACL usage.
10608
10609 See also : "tcp-request connection", "tcp-request content", "stick-table"
10610
10611
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010612tcp-response inspect-delay <timeout>
10613 Set the maximum allowed time to wait for a response during content inspection
10614 May be used in sections : defaults | frontend | listen | backend
10615 no | no | yes | yes
10616 Arguments :
10617 <timeout> is the timeout value specified in milliseconds by default, but
10618 can be in any other unit if the number is suffixed by the unit,
10619 as explained at the top of this document.
10620
10621 See also : "tcp-response content", "tcp-request inspect-delay".
10622
10623
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010624timeout check <timeout>
10625 Set additional check timeout, but only after a connection has been already
10626 established.
10627
10628 May be used in sections: defaults | frontend | listen | backend
10629 yes | no | yes | yes
10630 Arguments:
10631 <timeout> is the timeout value specified in milliseconds by default, but
10632 can be in any other unit if the number is suffixed by the unit,
10633 as explained at the top of this document.
10634
10635 If set, haproxy uses min("timeout connect", "inter") as a connect timeout
10636 for check and "timeout check" as an additional read timeout. The "min" is
Davor Ocelice9ed2812017-12-25 17:49:28 +010010637 used so that people running with *very* long "timeout connect" (e.g. those
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010638 who needed this due to the queue or tarpit) do not slow down their checks.
Willy Tarreaud7550a22010-02-10 05:10:19 +010010639 (Please also note that there is no valid reason to have such long connect
10640 timeouts, because "timeout queue" and "timeout tarpit" can always be used to
10641 avoid that).
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010642
10643 If "timeout check" is not set haproxy uses "inter" for complete check
10644 timeout (connect + read) exactly like all <1.3.15 version.
10645
10646 In most cases check request is much simpler and faster to handle than normal
10647 requests and people may want to kick out laggy servers so this timeout should
Willy Tarreau41a340d2008-01-22 12:25:31 +010010648 be smaller than "timeout server".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010649
10650 This parameter is specific to backends, but can be specified once for all in
10651 "defaults" sections. This is in fact one of the easiest solutions not to
10652 forget about it.
10653
Willy Tarreau41a340d2008-01-22 12:25:31 +010010654 See also: "timeout connect", "timeout queue", "timeout server",
10655 "timeout tarpit".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010656
10657
Willy Tarreau0ba27502007-12-24 16:55:16 +010010658timeout client <timeout>
Willy Tarreau0ba27502007-12-24 16:55:16 +010010659 Set the maximum inactivity time on the client side.
10660 May be used in sections : defaults | frontend | listen | backend
10661 yes | yes | yes | no
10662 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010663 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010010664 can be in any other unit if the number is suffixed by the unit,
10665 as explained at the top of this document.
10666
10667 The inactivity timeout applies when the client is expected to acknowledge or
10668 send data. In HTTP mode, this timeout is particularly important to consider
10669 during the first phase, when the client sends the request, and during the
Baptiste Assmann2e1941e2016-03-06 23:24:12 +010010670 response while it is reading data sent by the server. That said, for the
10671 first phase, it is preferable to set the "timeout http-request" to better
10672 protect HAProxy from Slowloris like attacks. The value is specified in
10673 milliseconds by default, but can be in any other unit if the number is
Willy Tarreau0ba27502007-12-24 16:55:16 +010010674 suffixed by the unit, as specified at the top of this document. In TCP mode
10675 (and to a lesser extent, in HTTP mode), it is highly recommended that the
10676 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010677 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +010010678 losses by specifying timeouts that are slightly above multiples of 3 seconds
Davor Ocelice9ed2812017-12-25 17:49:28 +010010679 (e.g. 4 or 5 seconds). If some long-lived sessions are mixed with short-lived
10680 sessions (e.g. WebSocket and HTTP), it's worth considering "timeout tunnel",
Willy Tarreau05cdd962014-05-10 14:30:07 +020010681 which overrides "timeout client" and "timeout server" for tunnels, as well as
10682 "timeout client-fin" for half-closed connections.
Willy Tarreau0ba27502007-12-24 16:55:16 +010010683
10684 This parameter is specific to frontends, but can be specified once for all in
10685 "defaults" sections. This is in fact one of the easiest solutions not to
10686 forget about it. An unspecified timeout results in an infinite timeout, which
10687 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050010688 during startup because it may result in accumulation of expired sessions in
Willy Tarreau0ba27502007-12-24 16:55:16 +010010689 the system if the system's timeouts are not configured either.
10690
Willy Tarreau95c4e142017-11-26 12:18:55 +010010691 This also applies to HTTP/2 connections, which will be closed with GOAWAY.
Lukas Tribus75df9d72017-11-24 19:05:12 +010010692
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020010693 See also : "timeout server", "timeout tunnel", "timeout http-request".
Willy Tarreau0ba27502007-12-24 16:55:16 +010010694
Willy Tarreau0ba27502007-12-24 16:55:16 +010010695
Willy Tarreau05cdd962014-05-10 14:30:07 +020010696timeout client-fin <timeout>
10697 Set the inactivity timeout on the client side for half-closed connections.
10698 May be used in sections : defaults | frontend | listen | backend
10699 yes | yes | yes | no
10700 Arguments :
10701 <timeout> is the timeout value specified in milliseconds by default, but
10702 can be in any other unit if the number is suffixed by the unit,
10703 as explained at the top of this document.
10704
10705 The inactivity timeout applies when the client is expected to acknowledge or
10706 send data while one direction is already shut down. This timeout is different
10707 from "timeout client" in that it only applies to connections which are closed
10708 in one direction. This is particularly useful to avoid keeping connections in
10709 FIN_WAIT state for too long when clients do not disconnect cleanly. This
10710 problem is particularly common long connections such as RDP or WebSocket.
10711 Note that this timeout can override "timeout tunnel" when a connection shuts
Willy Tarreau599391a2017-11-24 10:16:00 +010010712 down in one direction. It is applied to idle HTTP/2 connections once a GOAWAY
10713 frame was sent, often indicating an expectation that the connection quickly
10714 ends.
Willy Tarreau05cdd962014-05-10 14:30:07 +020010715
10716 This parameter is specific to frontends, but can be specified once for all in
10717 "defaults" sections. By default it is not set, so half-closed connections
10718 will use the other timeouts (timeout.client or timeout.tunnel).
10719
10720 See also : "timeout client", "timeout server-fin", and "timeout tunnel".
10721
10722
Willy Tarreau0ba27502007-12-24 16:55:16 +010010723timeout connect <timeout>
Willy Tarreau0ba27502007-12-24 16:55:16 +010010724 Set the maximum time to wait for a connection attempt to a server to succeed.
10725 May be used in sections : defaults | frontend | listen | backend
10726 yes | no | yes | yes
10727 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010728 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010010729 can be in any other unit if the number is suffixed by the unit,
10730 as explained at the top of this document.
10731
10732 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010733 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +010010734 cover one or several TCP packet losses by specifying timeouts that are
Davor Ocelice9ed2812017-12-25 17:49:28 +010010735 slightly above multiples of 3 seconds (e.g. 4 or 5 seconds). By default, the
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010736 connect timeout also presets both queue and tarpit timeouts to the same value
10737 if these have not been specified.
Willy Tarreau0ba27502007-12-24 16:55:16 +010010738
10739 This parameter is specific to backends, but can be specified once for all in
10740 "defaults" sections. This is in fact one of the easiest solutions not to
10741 forget about it. An unspecified timeout results in an infinite timeout, which
10742 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050010743 during startup because it may result in accumulation of failed sessions in
Willy Tarreau0ba27502007-12-24 16:55:16 +010010744 the system if the system's timeouts are not configured either.
10745
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020010746 See also: "timeout check", "timeout queue", "timeout server", "timeout tarpit".
Willy Tarreau0ba27502007-12-24 16:55:16 +010010747
Willy Tarreau0ba27502007-12-24 16:55:16 +010010748
Willy Tarreaub16a5742010-01-10 14:46:16 +010010749timeout http-keep-alive <timeout>
10750 Set the maximum allowed time to wait for a new HTTP request to appear
10751 May be used in sections : defaults | frontend | listen | backend
10752 yes | yes | yes | yes
10753 Arguments :
10754 <timeout> is the timeout value specified in milliseconds by default, but
10755 can be in any other unit if the number is suffixed by the unit,
10756 as explained at the top of this document.
10757
10758 By default, the time to wait for a new request in case of keep-alive is set
10759 by "timeout http-request". However this is not always convenient because some
10760 people want very short keep-alive timeouts in order to release connections
10761 faster, and others prefer to have larger ones but still have short timeouts
10762 once the request has started to present itself.
10763
10764 The "http-keep-alive" timeout covers these needs. It will define how long to
10765 wait for a new HTTP request to start coming after a response was sent. Once
10766 the first byte of request has been seen, the "http-request" timeout is used
10767 to wait for the complete request to come. Note that empty lines prior to a
10768 new request do not refresh the timeout and are not counted as a new request.
10769
10770 There is also another difference between the two timeouts : when a connection
10771 expires during timeout http-keep-alive, no error is returned, the connection
10772 just closes. If the connection expires in "http-request" while waiting for a
10773 connection to complete, a HTTP 408 error is returned.
10774
10775 In general it is optimal to set this value to a few tens to hundreds of
10776 milliseconds, to allow users to fetch all objects of a page at once but
Davor Ocelice9ed2812017-12-25 17:49:28 +010010777 without waiting for further clicks. Also, if set to a very small value (e.g.
Willy Tarreaub16a5742010-01-10 14:46:16 +010010778 1 millisecond) it will probably only accept pipelined requests but not the
10779 non-pipelined ones. It may be a nice trade-off for very large sites running
Patrick Mézard2382ad62010-05-09 10:43:32 +020010780 with tens to hundreds of thousands of clients.
Willy Tarreaub16a5742010-01-10 14:46:16 +010010781
10782 If this parameter is not set, the "http-request" timeout applies, and if both
10783 are not set, "timeout client" still applies at the lower level. It should be
10784 set in the frontend to take effect, unless the frontend is in TCP mode, in
10785 which case the HTTP backend's timeout will be used.
10786
Willy Tarreau95c4e142017-11-26 12:18:55 +010010787 When using HTTP/2 "timeout client" is applied instead. This is so we can keep
10788 using short keep-alive timeouts in HTTP/1.1 while using longer ones in HTTP/2
Lukas Tribus75df9d72017-11-24 19:05:12 +010010789 (where we only have one connection per client and a connection setup).
10790
Willy Tarreaub16a5742010-01-10 14:46:16 +010010791 See also : "timeout http-request", "timeout client".
10792
10793
Willy Tarreau036fae02008-01-06 13:24:40 +010010794timeout http-request <timeout>
10795 Set the maximum allowed time to wait for a complete HTTP request
10796 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaucd7afc02009-07-12 10:03:17 +020010797 yes | yes | yes | yes
Willy Tarreau036fae02008-01-06 13:24:40 +010010798 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010799 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau036fae02008-01-06 13:24:40 +010010800 can be in any other unit if the number is suffixed by the unit,
10801 as explained at the top of this document.
10802
10803 In order to offer DoS protection, it may be required to lower the maximum
10804 accepted time to receive a complete HTTP request without affecting the client
10805 timeout. This helps protecting against established connections on which
10806 nothing is sent. The client timeout cannot offer a good protection against
10807 this abuse because it is an inactivity timeout, which means that if the
10808 attacker sends one character every now and then, the timeout will not
10809 trigger. With the HTTP request timeout, no matter what speed the client
Willy Tarreau2705a612014-05-23 17:38:34 +020010810 types, the request will be aborted if it does not complete in time. When the
10811 timeout expires, an HTTP 408 response is sent to the client to inform it
10812 about the problem, and the connection is closed. The logs will report
10813 termination codes "cR". Some recent browsers are having problems with this
Davor Ocelice9ed2812017-12-25 17:49:28 +010010814 standard, well-documented behavior, so it might be needed to hide the 408
Willy Tarreau0f228a02015-05-01 15:37:53 +020010815 code using "option http-ignore-probes" or "errorfile 408 /dev/null". See
10816 more details in the explanations of the "cR" termination code in section 8.5.
Willy Tarreau036fae02008-01-06 13:24:40 +010010817
Baptiste Assmanneccdf432015-10-28 13:49:01 +010010818 By default, this timeout only applies to the header part of the request,
10819 and not to any data. As soon as the empty line is received, this timeout is
10820 not used anymore. When combined with "option http-buffer-request", this
10821 timeout also applies to the body of the request..
10822 It is used again on keep-alive connections to wait for a second
Willy Tarreaub16a5742010-01-10 14:46:16 +010010823 request if "timeout http-keep-alive" is not set.
Willy Tarreau036fae02008-01-06 13:24:40 +010010824
10825 Generally it is enough to set it to a few seconds, as most clients send the
10826 full request immediately upon connection. Add 3 or more seconds to cover TCP
Davor Ocelice9ed2812017-12-25 17:49:28 +010010827 retransmits but that's all. Setting it to very low values (e.g. 50 ms) will
Willy Tarreau036fae02008-01-06 13:24:40 +010010828 generally work on local networks as long as there are no packet losses. This
10829 will prevent people from sending bare HTTP requests using telnet.
10830
10831 If this parameter is not set, the client timeout still applies between each
Willy Tarreaucd7afc02009-07-12 10:03:17 +020010832 chunk of the incoming request. It should be set in the frontend to take
10833 effect, unless the frontend is in TCP mode, in which case the HTTP backend's
10834 timeout will be used.
Willy Tarreau036fae02008-01-06 13:24:40 +010010835
Willy Tarreau0f228a02015-05-01 15:37:53 +020010836 See also : "errorfile", "http-ignore-probes", "timeout http-keep-alive", and
Baptiste Assmanneccdf432015-10-28 13:49:01 +010010837 "timeout client", "option http-buffer-request".
Willy Tarreau036fae02008-01-06 13:24:40 +010010838
Willy Tarreau844e3c52008-01-11 16:28:18 +010010839
10840timeout queue <timeout>
10841 Set the maximum time to wait in the queue for a connection slot to be free
10842 May be used in sections : defaults | frontend | listen | backend
10843 yes | no | yes | yes
10844 Arguments :
10845 <timeout> is the timeout value specified in milliseconds by default, but
10846 can be in any other unit if the number is suffixed by the unit,
10847 as explained at the top of this document.
10848
10849 When a server's maxconn is reached, connections are left pending in a queue
10850 which may be server-specific or global to the backend. In order not to wait
10851 indefinitely, a timeout is applied to requests pending in the queue. If the
10852 timeout is reached, it is considered that the request will almost never be
10853 served, so it is dropped and a 503 error is returned to the client.
10854
10855 The "timeout queue" statement allows to fix the maximum time for a request to
10856 be left pending in a queue. If unspecified, the same value as the backend's
10857 connection timeout ("timeout connect") is used, for backwards compatibility
10858 with older versions with no "timeout queue" parameter.
10859
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020010860 See also : "timeout connect".
Willy Tarreau844e3c52008-01-11 16:28:18 +010010861
10862
10863timeout server <timeout>
Willy Tarreau844e3c52008-01-11 16:28:18 +010010864 Set the maximum inactivity time on the server side.
10865 May be used in sections : defaults | frontend | listen | backend
10866 yes | no | yes | yes
10867 Arguments :
10868 <timeout> is the timeout value specified in milliseconds by default, but
10869 can be in any other unit if the number is suffixed by the unit,
10870 as explained at the top of this document.
10871
10872 The inactivity timeout applies when the server is expected to acknowledge or
10873 send data. In HTTP mode, this timeout is particularly important to consider
10874 during the first phase of the server's response, when it has to send the
10875 headers, as it directly represents the server's processing time for the
10876 request. To find out what value to put there, it's often good to start with
10877 what would be considered as unacceptable response times, then check the logs
10878 to observe the response time distribution, and adjust the value accordingly.
10879
10880 The value is specified in milliseconds by default, but can be in any other
10881 unit if the number is suffixed by the unit, as specified at the top of this
10882 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
10883 recommended that the client timeout remains equal to the server timeout in
10884 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010885 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +010010886 packet losses by specifying timeouts that are slightly above multiples of 3
Davor Ocelice9ed2812017-12-25 17:49:28 +010010887 seconds (e.g. 4 or 5 seconds minimum). If some long-lived sessions are mixed
10888 with short-lived sessions (e.g. WebSocket and HTTP), it's worth considering
Willy Tarreauce887fd2012-05-12 12:50:00 +020010889 "timeout tunnel", which overrides "timeout client" and "timeout server" for
10890 tunnels.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010891
10892 This parameter is specific to backends, but can be specified once for all in
10893 "defaults" sections. This is in fact one of the easiest solutions not to
10894 forget about it. An unspecified timeout results in an infinite timeout, which
10895 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050010896 during startup because it may result in accumulation of expired sessions in
Willy Tarreau844e3c52008-01-11 16:28:18 +010010897 the system if the system's timeouts are not configured either.
10898
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020010899 See also : "timeout client" and "timeout tunnel".
Willy Tarreau844e3c52008-01-11 16:28:18 +010010900
Willy Tarreau05cdd962014-05-10 14:30:07 +020010901
10902timeout server-fin <timeout>
10903 Set the inactivity timeout on the server side for half-closed connections.
10904 May be used in sections : defaults | frontend | listen | backend
10905 yes | no | yes | yes
10906 Arguments :
10907 <timeout> is the timeout value specified in milliseconds by default, but
10908 can be in any other unit if the number is suffixed by the unit,
10909 as explained at the top of this document.
10910
10911 The inactivity timeout applies when the server is expected to acknowledge or
10912 send data while one direction is already shut down. This timeout is different
10913 from "timeout server" in that it only applies to connections which are closed
10914 in one direction. This is particularly useful to avoid keeping connections in
10915 FIN_WAIT state for too long when a remote server does not disconnect cleanly.
10916 This problem is particularly common long connections such as RDP or WebSocket.
10917 Note that this timeout can override "timeout tunnel" when a connection shuts
10918 down in one direction. This setting was provided for completeness, but in most
10919 situations, it should not be needed.
10920
10921 This parameter is specific to backends, but can be specified once for all in
10922 "defaults" sections. By default it is not set, so half-closed connections
10923 will use the other timeouts (timeout.server or timeout.tunnel).
10924
10925 See also : "timeout client-fin", "timeout server", and "timeout tunnel".
10926
Willy Tarreau844e3c52008-01-11 16:28:18 +010010927
10928timeout tarpit <timeout>
Cyril Bonté78caf842010-03-10 22:41:43 +010010929 Set the duration for which tarpitted connections will be maintained
Willy Tarreau844e3c52008-01-11 16:28:18 +010010930 May be used in sections : defaults | frontend | listen | backend
10931 yes | yes | yes | yes
10932 Arguments :
10933 <timeout> is the tarpit duration specified in milliseconds by default, but
10934 can be in any other unit if the number is suffixed by the unit,
10935 as explained at the top of this document.
10936
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020010937 When a connection is tarpitted using "http-request tarpit", it is maintained
10938 open with no activity for a certain amount of time, then closed. "timeout
10939 tarpit" defines how long it will be maintained open.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010940
10941 The value is specified in milliseconds by default, but can be in any other
10942 unit if the number is suffixed by the unit, as specified at the top of this
10943 document. If unspecified, the same value as the backend's connection timeout
10944 ("timeout connect") is used, for backwards compatibility with older versions
Cyril Bonté78caf842010-03-10 22:41:43 +010010945 with no "timeout tarpit" parameter.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010946
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020010947 See also : "timeout connect".
Willy Tarreau844e3c52008-01-11 16:28:18 +010010948
10949
Willy Tarreauce887fd2012-05-12 12:50:00 +020010950timeout tunnel <timeout>
10951 Set the maximum inactivity time on the client and server side for tunnels.
10952 May be used in sections : defaults | frontend | listen | backend
10953 yes | no | yes | yes
10954 Arguments :
10955 <timeout> is the timeout value specified in milliseconds by default, but
10956 can be in any other unit if the number is suffixed by the unit,
10957 as explained at the top of this document.
10958
Jamie Gloudonaaa21002012-08-25 00:18:33 -040010959 The tunnel timeout applies when a bidirectional connection is established
Willy Tarreauce887fd2012-05-12 12:50:00 +020010960 between a client and a server, and the connection remains inactive in both
10961 directions. This timeout supersedes both the client and server timeouts once
10962 the connection becomes a tunnel. In TCP, this timeout is used as soon as no
Davor Ocelice9ed2812017-12-25 17:49:28 +010010963 analyzer remains attached to either connection (e.g. tcp content rules are
10964 accepted). In HTTP, this timeout is used when a connection is upgraded (e.g.
Willy Tarreauce887fd2012-05-12 12:50:00 +020010965 when switching to the WebSocket protocol, or forwarding a CONNECT request
10966 to a proxy), or after the first response when no keepalive/close option is
10967 specified.
10968
Willy Tarreau05cdd962014-05-10 14:30:07 +020010969 Since this timeout is usually used in conjunction with long-lived connections,
10970 it usually is a good idea to also set "timeout client-fin" to handle the
10971 situation where a client suddenly disappears from the net and does not
10972 acknowledge a close, or sends a shutdown and does not acknowledge pending
10973 data anymore. This can happen in lossy networks where firewalls are present,
10974 and is detected by the presence of large amounts of sessions in a FIN_WAIT
10975 state.
10976
Willy Tarreauce887fd2012-05-12 12:50:00 +020010977 The value is specified in milliseconds by default, but can be in any other
10978 unit if the number is suffixed by the unit, as specified at the top of this
10979 document. Whatever the expected normal idle time, it is a good practice to
10980 cover at least one or several TCP packet losses by specifying timeouts that
Davor Ocelice9ed2812017-12-25 17:49:28 +010010981 are slightly above multiples of 3 seconds (e.g. 4 or 5 seconds minimum).
Willy Tarreauce887fd2012-05-12 12:50:00 +020010982
10983 This parameter is specific to backends, but can be specified once for all in
10984 "defaults" sections. This is in fact one of the easiest solutions not to
10985 forget about it.
10986
10987 Example :
10988 defaults http
10989 option http-server-close
10990 timeout connect 5s
10991 timeout client 30s
Willy Tarreau05cdd962014-05-10 14:30:07 +020010992 timeout client-fin 30s
Willy Tarreauce887fd2012-05-12 12:50:00 +020010993 timeout server 30s
10994 timeout tunnel 1h # timeout to use with WebSocket and CONNECT
10995
Willy Tarreau05cdd962014-05-10 14:30:07 +020010996 See also : "timeout client", "timeout client-fin", "timeout server".
Willy Tarreauce887fd2012-05-12 12:50:00 +020010997
10998
Willy Tarreau844e3c52008-01-11 16:28:18 +010010999transparent (deprecated)
11000 Enable client-side transparent proxying
11001 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +010011002 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +010011003 Arguments : none
11004
11005 This keyword was introduced in order to provide layer 7 persistence to layer
11006 3 load balancers. The idea is to use the OS's ability to redirect an incoming
11007 connection for a remote address to a local process (here HAProxy), and let
11008 this process know what address was initially requested. When this option is
11009 used, sessions without cookies will be forwarded to the original destination
11010 IP address of the incoming request (which should match that of another
11011 equipment), while requests with cookies will still be forwarded to the
11012 appropriate server.
11013
11014 The "transparent" keyword is deprecated, use "option transparent" instead.
11015
11016 Note that contrary to a common belief, this option does NOT make HAProxy
11017 present the client's IP to the server when establishing the connection.
11018
Willy Tarreau844e3c52008-01-11 16:28:18 +010011019 See also: "option transparent"
11020
William Lallemanda73203e2012-03-12 12:48:57 +010011021unique-id-format <string>
11022 Generate a unique ID for each request.
11023 May be used in sections : defaults | frontend | listen | backend
11024 yes | yes | yes | no
11025 Arguments :
11026 <string> is a log-format string.
11027
Cyril Bonté108cf6e2012-04-21 23:30:29 +020011028 This keyword creates a ID for each request using the custom log format. A
11029 unique ID is useful to trace a request passing through many components of
11030 a complex infrastructure. The newly created ID may also be logged using the
11031 %ID tag the log-format string.
William Lallemanda73203e2012-03-12 12:48:57 +010011032
Cyril Bonté108cf6e2012-04-21 23:30:29 +020011033 The format should be composed from elements that are guaranteed to be
11034 unique when combined together. For instance, if multiple haproxy instances
11035 are involved, it might be important to include the node name. It is often
11036 needed to log the incoming connection's source and destination addresses
11037 and ports. Note that since multiple requests may be performed over the same
11038 connection, including a request counter may help differentiate them.
11039 Similarly, a timestamp may protect against a rollover of the counter.
11040 Logging the process ID will avoid collisions after a service restart.
William Lallemanda73203e2012-03-12 12:48:57 +010011041
Cyril Bonté108cf6e2012-04-21 23:30:29 +020011042 It is recommended to use hexadecimal notation for many fields since it
11043 makes them more compact and saves space in logs.
William Lallemanda73203e2012-03-12 12:48:57 +010011044
Cyril Bonté108cf6e2012-04-21 23:30:29 +020011045 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010011046
Julien Vehentf21be322014-03-07 08:27:34 -050011047 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010011048
11049 will generate:
11050
11051 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
11052
11053 See also: "unique-id-header"
11054
11055unique-id-header <name>
11056 Add a unique ID header in the HTTP request.
11057 May be used in sections : defaults | frontend | listen | backend
11058 yes | yes | yes | no
11059 Arguments :
11060 <name> is the name of the header.
11061
Cyril Bonté108cf6e2012-04-21 23:30:29 +020011062 Add a unique-id header in the HTTP request sent to the server, using the
11063 unique-id-format. It can't work if the unique-id-format doesn't exist.
William Lallemanda73203e2012-03-12 12:48:57 +010011064
Cyril Bonté108cf6e2012-04-21 23:30:29 +020011065 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010011066
Julien Vehentf21be322014-03-07 08:27:34 -050011067 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010011068 unique-id-header X-Unique-ID
11069
11070 will generate:
11071
11072 X-Unique-ID: 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
11073
11074 See also: "unique-id-format"
Willy Tarreau844e3c52008-01-11 16:28:18 +010011075
Willy Tarreauf51658d2014-04-23 01:21:56 +020011076use_backend <backend> [{if | unless} <condition>]
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020011077 Switch to a specific backend if/unless an ACL-based condition is matched.
Willy Tarreau844e3c52008-01-11 16:28:18 +010011078 May be used in sections : defaults | frontend | listen | backend
11079 no | yes | yes | no
11080 Arguments :
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010011081 <backend> is the name of a valid backend or "listen" section, or a
11082 "log-format" string resolving to a backend name.
Willy Tarreau844e3c52008-01-11 16:28:18 +010011083
Willy Tarreauf51658d2014-04-23 01:21:56 +020011084 <condition> is a condition composed of ACLs, as described in section 7. If
11085 it is omitted, the rule is unconditionally applied.
Willy Tarreau844e3c52008-01-11 16:28:18 +010011086
11087 When doing content-switching, connections arrive on a frontend and are then
11088 dispatched to various backends depending on a number of conditions. The
11089 relation between the conditions and the backends is described with the
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020011090 "use_backend" keyword. While it is normally used with HTTP processing, it can
Davor Ocelice9ed2812017-12-25 17:49:28 +010011091 also be used in pure TCP, either without content using stateless ACLs (e.g.
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020011092 source address validation) or combined with a "tcp-request" rule to wait for
11093 some payload.
Willy Tarreau844e3c52008-01-11 16:28:18 +010011094
11095 There may be as many "use_backend" rules as desired. All of these rules are
11096 evaluated in their declaration order, and the first one which matches will
11097 assign the backend.
11098
11099 In the first form, the backend will be used if the condition is met. In the
11100 second form, the backend will be used if the condition is not met. If no
11101 condition is valid, the backend defined with "default_backend" will be used.
11102 If no default backend is defined, either the servers in the same section are
11103 used (in case of a "listen" section) or, in case of a frontend, no server is
11104 used and a 503 service unavailable response is returned.
11105
Willy Tarreau51aecc72009-07-12 09:47:04 +020011106 Note that it is possible to switch from a TCP frontend to an HTTP backend. In
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011107 this case, either the frontend has already checked that the protocol is HTTP,
Willy Tarreau51aecc72009-07-12 09:47:04 +020011108 and backend processing will immediately follow, or the backend will wait for
11109 a complete HTTP request to get in. This feature is useful when a frontend
11110 must decode several protocols on a unique port, one of them being HTTP.
11111
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010011112 When <backend> is a simple name, it is resolved at configuration time, and an
11113 error is reported if the specified backend does not exist. If <backend> is
11114 a log-format string instead, no check may be done at configuration time, so
11115 the backend name is resolved dynamically at run time. If the resulting
11116 backend name does not correspond to any valid backend, no other rule is
11117 evaluated, and the default_backend directive is applied instead. Note that
11118 when using dynamic backend names, it is highly recommended to use a prefix
11119 that no other backend uses in order to ensure that an unauthorized backend
11120 cannot be forced from the request.
11121
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011122 It is worth mentioning that "use_backend" rules with an explicit name are
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010011123 used to detect the association between frontends and backends to compute the
11124 backend's "fullconn" setting. This cannot be done for dynamic names.
11125
11126 See also: "default_backend", "tcp-request", "fullconn", "log-format", and
11127 section 7 about ACLs.
Willy Tarreaud72758d2010-01-12 10:42:19 +010011128
Christopher Fauletb30b3102019-09-12 23:03:09 +020011129use-fcgi-app <name>
11130 Defines the FastCGI application to use for the backend.
11131 May be used in sections : defaults | frontend | listen | backend
11132 no | no | yes | yes
11133 Arguments :
11134 <name> is the name of the FastCGI application to use.
11135
11136 See section 10.1 about FastCGI application setup for details.
Willy Tarreau036fae02008-01-06 13:24:40 +010011137
Willy Tarreau4a5cade2012-04-05 21:09:48 +020011138use-server <server> if <condition>
11139use-server <server> unless <condition>
11140 Only use a specific server if/unless an ACL-based condition is matched.
11141 May be used in sections : defaults | frontend | listen | backend
11142 no | no | yes | yes
11143 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +020011144 <server> is the name of a valid server in the same backend section.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020011145
11146 <condition> is a condition composed of ACLs, as described in section 7.
11147
11148 By default, connections which arrive to a backend are load-balanced across
11149 the available servers according to the configured algorithm, unless a
11150 persistence mechanism such as a cookie is used and found in the request.
11151
11152 Sometimes it is desirable to forward a particular request to a specific
11153 server without having to declare a dedicated backend for this server. This
11154 can be achieved using the "use-server" rules. These rules are evaluated after
11155 the "redirect" rules and before evaluating cookies, and they have precedence
11156 on them. There may be as many "use-server" rules as desired. All of these
11157 rules are evaluated in their declaration order, and the first one which
11158 matches will assign the server.
11159
11160 If a rule designates a server which is down, and "option persist" is not used
11161 and no force-persist rule was validated, it is ignored and evaluation goes on
11162 with the next rules until one matches.
11163
11164 In the first form, the server will be used if the condition is met. In the
11165 second form, the server will be used if the condition is not met. If no
11166 condition is valid, the processing continues and the server will be assigned
11167 according to other persistence mechanisms.
11168
11169 Note that even if a rule is matched, cookie processing is still performed but
11170 does not assign the server. This allows prefixed cookies to have their prefix
11171 stripped.
11172
11173 The "use-server" statement works both in HTTP and TCP mode. This makes it
11174 suitable for use with content-based inspection. For instance, a server could
11175 be selected in a farm according to the TLS SNI field. And if these servers
11176 have their weight set to zero, they will not be used for other traffic.
11177
11178 Example :
11179 # intercept incoming TLS requests based on the SNI field
11180 use-server www if { req_ssl_sni -i www.example.com }
11181 server www 192.168.0.1:443 weight 0
11182 use-server mail if { req_ssl_sni -i mail.example.com }
11183 server mail 192.168.0.1:587 weight 0
11184 use-server imap if { req_ssl_sni -i imap.example.com }
Lukas Tribus98a3e3f2017-03-26 12:55:35 +000011185 server imap 192.168.0.1:993 weight 0
Willy Tarreau4a5cade2012-04-05 21:09:48 +020011186 # all the rest is forwarded to this server
11187 server default 192.168.0.2:443 check
11188
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011189 See also: "use_backend", section 5 about server and section 7 about ACLs.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020011190
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011191
Davor Ocelice9ed2812017-12-25 17:49:28 +0100111925. Bind and server options
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011193--------------------------
11194
11195The "bind", "server" and "default-server" keywords support a number of settings
11196depending on some build options and on the system HAProxy was built on. These
11197settings generally each consist in one word sometimes followed by a value,
11198written on the same line as the "bind" or "server" line. All these options are
11199described in this section.
11200
11201
112025.1. Bind options
11203-----------------
11204
11205The "bind" keyword supports a certain number of settings which are all passed
11206as arguments on the same line. The order in which those arguments appear makes
11207no importance, provided that they appear after the bind address. All of these
11208parameters are optional. Some of them consist in a single words (booleans),
11209while other ones expect a value after them. In this case, the value must be
11210provided immediately after the setting name.
11211
11212The currently supported settings are the following ones.
11213
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010011214accept-netscaler-cip <magic number>
11215 Enforces the use of the NetScaler Client IP insertion protocol over any
11216 connection accepted by any of the TCP sockets declared on the same line. The
11217 NetScaler Client IP insertion protocol dictates the layer 3/4 addresses of
11218 the incoming connection to be used everywhere an address is used, with the
11219 only exception of "tcp-request connection" rules which will only see the
11220 real connection address. Logs will reflect the addresses indicated in the
11221 protocol, unless it is violated, in which case the real address will still
11222 be used. This keyword combined with support from external components can be
11223 used as an efficient and reliable alternative to the X-Forwarded-For
Bertrand Jacquin90759682016-06-06 15:35:39 +010011224 mechanism which is not always reliable and not even always usable. See also
11225 "tcp-request connection expect-netscaler-cip" for a finer-grained setting of
11226 which client is allowed to use the protocol.
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010011227
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011228accept-proxy
11229 Enforces the use of the PROXY protocol over any connection accepted by any of
Willy Tarreau77992672014-06-14 11:06:17 +020011230 the sockets declared on the same line. Versions 1 and 2 of the PROXY protocol
11231 are supported and correctly detected. The PROXY protocol dictates the layer
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011232 3/4 addresses of the incoming connection to be used everywhere an address is
11233 used, with the only exception of "tcp-request connection" rules which will
11234 only see the real connection address. Logs will reflect the addresses
11235 indicated in the protocol, unless it is violated, in which case the real
Davor Ocelice9ed2812017-12-25 17:49:28 +010011236 address will still be used. This keyword combined with support from external
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011237 components can be used as an efficient and reliable alternative to the
11238 X-Forwarded-For mechanism which is not always reliable and not even always
Willy Tarreau4f0d9192013-06-11 20:40:55 +020011239 usable. See also "tcp-request connection expect-proxy" for a finer-grained
11240 setting of which client is allowed to use the protocol.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011241
Olivier Houchardc2aae742017-09-22 18:26:28 +020011242allow-0rtt
Bertrand Jacquina25282b2018-08-14 00:56:13 +010011243 Allow receiving early data when using TLSv1.3. This is disabled by default,
Olivier Houchard69752962019-01-08 15:35:32 +010011244 due to security considerations. Because it is vulnerable to replay attacks,
John Roeslerfb2fce12019-07-10 15:45:51 -050011245 you should only allow if for requests that are safe to replay, i.e. requests
Olivier Houchard69752962019-01-08 15:35:32 +010011246 that are idempotent. You can use the "wait-for-handshake" action for any
11247 request that wouldn't be safe with early data.
Olivier Houchardc2aae742017-09-22 18:26:28 +020011248
Willy Tarreauab861d32013-04-02 02:30:41 +020011249alpn <protocols>
11250 This enables the TLS ALPN extension and advertises the specified protocol
11251 list as supported on top of ALPN. The protocol list consists in a comma-
11252 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
John Roeslerfb2fce12019-07-10 15:45:51 -050011253 quotes). This requires that the SSL library is built with support for TLS
Willy Tarreauab861d32013-04-02 02:30:41 +020011254 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
Willy Tarreau95c4e142017-11-26 12:18:55 +010011255 initial NPN extension. ALPN is required to enable HTTP/2 on an HTTP frontend.
11256 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
11257 now obsolete NPN extension. At the time of writing this, most browsers still
11258 support both ALPN and NPN for HTTP/2 so a fallback to NPN may still work for
11259 a while. But ALPN must be used whenever possible. If both HTTP/2 and HTTP/1.1
11260 are expected to be supported, both versions can be advertised, in order of
11261 preference, like below :
11262
11263 bind :443 ssl crt pub.pem alpn h2,http/1.1
Willy Tarreauab861d32013-04-02 02:30:41 +020011264
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011265backlog <backlog>
Willy Tarreaue2711c72019-02-27 15:39:41 +010011266 Sets the socket's backlog to this value. If unspecified or 0, the frontend's
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011267 backlog is used instead, which generally defaults to the maxconn value.
11268
Emmanuel Hocdete7f2b732017-01-09 16:15:54 +010011269curves <curves>
11270 This setting is only available when support for OpenSSL was built in. It sets
11271 the string describing the list of elliptic curves algorithms ("curve suite")
11272 that are negotiated during the SSL/TLS handshake with ECDHE. The format of the
11273 string is a colon-delimited list of curve name.
11274 Example: "X25519:P-256" (without quote)
11275 When "curves" is set, "ecdhe" parameter is ignored.
11276
Emeric Brun7fb34422012-09-28 15:26:15 +020011277ecdhe <named curve>
11278 This setting is only available when support for OpenSSL was built in. It sets
Emeric Brun6924ef82013-03-06 14:08:53 +010011279 the named curve (RFC 4492) used to generate ECDH ephemeral keys. By default,
11280 used named curve is prime256v1.
Emeric Brun7fb34422012-09-28 15:26:15 +020011281
Emeric Brunfd33a262012-10-11 16:28:27 +020011282ca-file <cafile>
Emeric Brun1a073b42012-09-28 17:07:34 +020011283 This setting is only available when support for OpenSSL was built in. It
11284 designates a PEM file from which to load CA certificates used to verify
11285 client's certificate.
11286
Emeric Brunb6dc9342012-09-28 17:55:37 +020011287ca-ignore-err [all|<errorID>,...]
11288 This setting is only available when support for OpenSSL was built in.
11289 Sets a comma separated list of errorIDs to ignore during verify at depth > 0.
11290 If set to 'all', all errors are ignored. SSL handshake is not aborted if an
11291 error is ignored.
11292
Christopher Faulet31af49d2015-06-09 17:29:50 +020011293ca-sign-file <cafile>
11294 This setting is only available when support for OpenSSL was built in. It
11295 designates a PEM file containing both the CA certificate and the CA private
11296 key used to create and sign server's certificates. This is a mandatory
11297 setting when the dynamic generation of certificates is enabled. See
11298 'generate-certificates' for details.
11299
Bertrand Jacquind4d0a232016-11-13 16:37:12 +000011300ca-sign-pass <passphrase>
Christopher Faulet31af49d2015-06-09 17:29:50 +020011301 This setting is only available when support for OpenSSL was built in. It is
11302 the CA private key passphrase. This setting is optional and used only when
11303 the dynamic generation of certificates is enabled. See
11304 'generate-certificates' for details.
11305
Emmanuel Hocdet842e94e2019-12-16 16:39:17 +010011306ca-verify-file <cafile>
11307 This setting designates a PEM file from which to load CA certificates used to
11308 verify client's certificate. It designates CA certificates which must not be
11309 included in CA names sent in server hello message. Typically, "ca-file" must
11310 be defined with intermediate certificates, and "ca-verify-file" with
11311 certificates to ending the chain, like root CA.
11312
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011313ciphers <ciphers>
11314 This setting is only available when support for OpenSSL was built in. It sets
11315 the string describing the list of cipher algorithms ("cipher suite") that are
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +000011316 negotiated during the SSL/TLS handshake up to TLSv1.2. The format of the
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000011317 string is defined in "man 1 ciphers" from OpenSSL man pages. For background
Dirkjan Bussink415150f2018-09-14 11:14:21 +020011318 information and recommendations see e.g.
11319 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
11320 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
11321 cipher configuration, please check the "ciphersuites" keyword.
11322
11323ciphersuites <ciphersuites>
11324 This setting is only available when support for OpenSSL was built in and
11325 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the string describing
11326 the list of cipher algorithms ("cipher suite") that are negotiated during the
11327 TLSv1.3 handshake. The format of the string is defined in "man 1 ciphers" from
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000011328 OpenSSL man pages under the "ciphersuites" section. For cipher configuration
11329 for TLSv1.2 and earlier, please check the "ciphers" keyword.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011330
Emeric Brunfd33a262012-10-11 16:28:27 +020011331crl-file <crlfile>
Emeric Brun1a073b42012-09-28 17:07:34 +020011332 This setting is only available when support for OpenSSL was built in. It
11333 designates a PEM file from which to load certificate revocation list used
11334 to verify client's certificate.
11335
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011336crt <cert>
Alex Davies0fbf0162013-03-02 16:04:50 +000011337 This setting is only available when support for OpenSSL was built in. It
11338 designates a PEM file containing both the required certificates and any
11339 associated private keys. This file can be built by concatenating multiple
11340 PEM files into one (e.g. cat cert.pem key.pem > combined.pem). If your CA
11341 requires an intermediate certificate, this can also be concatenated into this
Emmanuel Hocdet70df7bf2019-01-04 11:08:20 +010011342 file. Intermediate certificate can also be shared in a directory via
11343 "issuers-chain-path" directive.
Alex Davies0fbf0162013-03-02 16:04:50 +000011344
William Lallemand4c5adbf2020-02-24 14:23:22 +010011345 If the file does not contain a private key, HAProxy will try to load
11346 the key at the same path suffixed by a ".key".
11347
Alex Davies0fbf0162013-03-02 16:04:50 +000011348 If the OpenSSL used supports Diffie-Hellman, parameters present in this file
11349 are loaded.
11350
11351 If a directory name is used instead of a PEM file, then all files found in
William Lallemand3f25ae32020-02-24 16:30:12 +010011352 that directory will be loaded in alphabetic order unless their name ends
11353 with '.key', '.issuer', '.ocsp' or '.sctl' (reserved extensions). This
11354 directive may be specified multiple times in order to load certificates from
11355 multiple files or directories. The certificates will be presented to clients
11356 who provide a valid TLS Server Name Indication field matching one of their
11357 CN or alt subjects. Wildcards are supported, where a wildcard character '*'
11358 is used instead of the first hostname component (e.g. *.example.org matches
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010011359 www.example.org but not www.sub.example.org).
Alex Davies0fbf0162013-03-02 16:04:50 +000011360
11361 If no SNI is provided by the client or if the SSL library does not support
11362 TLS extensions, or if the client provides an SNI hostname which does not
11363 match any certificate, then the first loaded certificate will be presented.
11364 This means that when loading certificates from a directory, it is highly
Cyril Bonté3180f7b2015-01-25 00:16:08 +010011365 recommended to load the default one first as a file or to ensure that it will
11366 always be the first one in the directory.
Alex Davies0fbf0162013-03-02 16:04:50 +000011367
Emeric Brune032bfa2012-09-28 13:01:45 +020011368 Note that the same cert may be loaded multiple times without side effects.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011369
Davor Ocelice9ed2812017-12-25 17:49:28 +010011370 Some CAs (such as GoDaddy) offer a drop down list of server types that do not
Alex Davies0fbf0162013-03-02 16:04:50 +000011371 include HAProxy when obtaining a certificate. If this happens be sure to
Davor Ocelice9ed2812017-12-25 17:49:28 +010011372 choose a web server that the CA believes requires an intermediate CA (for
11373 GoDaddy, selection Apache Tomcat will get the correct bundle, but many
Alex Davies0fbf0162013-03-02 16:04:50 +000011374 others, e.g. nginx, result in a wrong bundle that will not work for some
11375 clients).
11376
Emeric Brun4147b2e2014-06-16 18:36:30 +020011377 For each PEM file, haproxy checks for the presence of file at the same path
11378 suffixed by ".ocsp". If such file is found, support for the TLS Certificate
11379 Status Request extension (also known as "OCSP stapling") is automatically
11380 enabled. The content of this file is optional. If not empty, it must contain
11381 a valid OCSP Response in DER format. In order to be valid an OCSP Response
11382 must comply with the following rules: it has to indicate a good status,
11383 it has to be a single response for the certificate of the PEM file, and it
11384 has to be valid at the moment of addition. If these rules are not respected
11385 the OCSP Response is ignored and a warning is emitted. In order to identify
11386 which certificate an OCSP Response applies to, the issuer's certificate is
11387 necessary. If the issuer's certificate is not found in the PEM file, it will
11388 be loaded from a file at the same path as the PEM file suffixed by ".issuer"
11389 if it exists otherwise it will fail with an error.
11390
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010011391 For each PEM file, haproxy also checks for the presence of file at the same
11392 path suffixed by ".sctl". If such file is found, support for Certificate
11393 Transparency (RFC6962) TLS extension is enabled. The file must contain a
11394 valid Signed Certificate Timestamp List, as described in RFC. File is parsed
11395 to check basic syntax, but no signatures are verified.
11396
yanbzhu6c25e9e2016-01-05 12:52:02 -050011397 There are cases where it is desirable to support multiple key types, e.g. RSA
11398 and ECDSA in the cipher suites offered to the clients. This allows clients
11399 that support EC certificates to be able to use EC ciphers, while
11400 simultaneously supporting older, RSA only clients.
yanbzhud19630c2015-12-14 15:10:25 -050011401
11402 In order to provide this functionality, multiple PEM files, each with a
11403 different key type, are required. To associate these PEM files into a
11404 "cert bundle" that is recognized by haproxy, they must be named in the
11405 following way: All PEM files that are to be bundled must have the same base
11406 name, with a suffix indicating the key type. Currently, three suffixes are
11407 supported: rsa, dsa and ecdsa. For example, if www.example.com has two PEM
11408 files, an RSA file and an ECDSA file, they must be named: "example.pem.rsa"
11409 and "example.pem.ecdsa". The first part of the filename is arbitrary; only the
11410 suffix matters. To load this bundle into haproxy, specify the base name only:
11411
11412 Example : bind :8443 ssl crt example.pem
11413
yanbzhu6c25e9e2016-01-05 12:52:02 -050011414 Note that the suffix is not given to haproxy; this tells haproxy to look for
yanbzhud19630c2015-12-14 15:10:25 -050011415 a cert bundle.
11416
Davor Ocelice9ed2812017-12-25 17:49:28 +010011417 HAProxy will load all PEM files in the bundle at the same time to try to
yanbzhud19630c2015-12-14 15:10:25 -050011418 support multiple key types. PEM files are combined based on Common Name
11419 (CN) and Subject Alternative Name (SAN) to support SNI lookups. This means
11420 that even if you give haproxy a cert bundle, if there are no shared CN/SAN
11421 entries in the certificates in that bundle, haproxy will not be able to
11422 provide multi-cert support.
11423
11424 Assuming bundle in the example above contained the following:
11425
11426 Filename | CN | SAN
11427 -------------------+-----------------+-------------------
11428 example.pem.rsa | www.example.com | rsa.example.com
yanbzhu6c25e9e2016-01-05 12:52:02 -050011429 -------------------+-----------------+-------------------
yanbzhud19630c2015-12-14 15:10:25 -050011430 example.pem.ecdsa | www.example.com | ecdsa.example.com
11431 -------------------+-----------------+-------------------
11432
11433 Users connecting with an SNI of "www.example.com" will be able
11434 to use both RSA and ECDSA cipher suites. Users connecting with an SNI of
11435 "rsa.example.com" will only be able to use RSA cipher suites, and users
11436 connecting with "ecdsa.example.com" will only be able to use ECDSA cipher
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020011437 suites. With BoringSSL and Openssl >= 1.1.1 multi-cert is natively supported,
11438 no need to bundle certificates. ECDSA certificate will be preferred if client
11439 support it.
yanbzhud19630c2015-12-14 15:10:25 -050011440
11441 If a directory name is given as the <cert> argument, haproxy will
11442 automatically search and load bundled files in that directory.
11443
11444 OSCP files (.ocsp) and issuer files (.issuer) are supported with multi-cert
11445 bundling. Each certificate can have its own .ocsp and .issuer file. At this
11446 time, sctl is not supported in multi-certificate bundling.
11447
Emeric Brunb6dc9342012-09-28 17:55:37 +020011448crt-ignore-err <errors>
Alex Davies0fbf0162013-03-02 16:04:50 +000011449 This setting is only available when support for OpenSSL was built in. Sets a
Davor Ocelice9ed2812017-12-25 17:49:28 +010011450 comma separated list of errorIDs to ignore during verify at depth == 0. If
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011451 set to 'all', all errors are ignored. SSL handshake is not aborted if an error
Alex Davies0fbf0162013-03-02 16:04:50 +000011452 is ignored.
Emeric Brunb6dc9342012-09-28 17:55:37 +020011453
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010011454crt-list <file>
11455 This setting is only available when support for OpenSSL was built in. It
Emmanuel Hocdet98263292016-12-29 18:26:15 +010011456 designates a list of PEM file with an optional ssl configuration and a SNI
11457 filter per certificate, with the following format for each line :
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010011458
Emmanuel Hocdet98263292016-12-29 18:26:15 +010011459 <crtfile> [\[<sslbindconf> ...\]] [[!]<snifilter> ...]
11460
Emmanuel Hocdet842e94e2019-12-16 16:39:17 +010011461 sslbindconf support "npn", "alpn", "verify", "ca-file", "ca-verify-file",
11462 "no-ca-names", "crl-file", "ecdhe", "curves", "ciphers" configuration. With
11463 BoringSSL and Openssl >= 1.1.1 "ssl-min-ver" and "ssl-max-ver" are also
11464 supported. It override the configuration set in bind line for the certificate.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010011465
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +020011466 Wildcards are supported in the SNI filter. Negative filter are also supported,
11467 only useful in combination with a wildcard filter to exclude a particular SNI.
11468 The certificates will be presented to clients who provide a valid TLS Server
11469 Name Indication field matching one of the SNI filters. If no SNI filter is
11470 specified, the CN and alt subjects are used. This directive may be specified
11471 multiple times. See the "crt" option for more information. The default
11472 certificate is still needed to meet OpenSSL expectations. If it is not used,
11473 the 'strict-sni' option may be used.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010011474
yanbzhu6c25e9e2016-01-05 12:52:02 -050011475 Multi-cert bundling (see "crt") is supported with crt-list, as long as only
Emmanuel Hocdetd294aea2016-05-13 11:14:06 +020011476 the base name is given in the crt-list. SNI filter will do the same work on
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020011477 all bundled certificates. With BoringSSL and Openssl >= 1.1.1 multi-cert is
11478 natively supported, avoid multi-cert bundling. RSA and ECDSA certificates can
11479 be declared in a row, and set different ssl and filter parameter.
yanbzhud19630c2015-12-14 15:10:25 -050011480
Emmanuel Hocdet98263292016-12-29 18:26:15 +010011481 crt-list file example:
11482 cert1.pem
Emmanuel Hocdet05942112017-02-20 16:11:50 +010011483 cert2.pem [alpn h2,http/1.1]
Emmanuel Hocdet98263292016-12-29 18:26:15 +010011484 certW.pem *.domain.tld !secure.domain.tld
Emmanuel Hocdet05942112017-02-20 16:11:50 +010011485 certS.pem [curves X25519:P-256 ciphers ECDHE-ECDSA-AES256-GCM-SHA384] secure.domain.tld
Emmanuel Hocdet98263292016-12-29 18:26:15 +010011486
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011487defer-accept
11488 Is an optional keyword which is supported only on certain Linux kernels. It
11489 states that a connection will only be accepted once some data arrive on it,
11490 or at worst after the first retransmit. This should be used only on protocols
Davor Ocelice9ed2812017-12-25 17:49:28 +010011491 for which the client talks first (e.g. HTTP). It can slightly improve
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011492 performance by ensuring that most of the request is already available when
11493 the connection is accepted. On the other hand, it will not be able to detect
11494 connections which don't talk. It is important to note that this option is
11495 broken in all kernels up to 2.6.31, as the connection is never accepted until
11496 the client talks. This can cause issues with front firewalls which would see
11497 an established connection while the proxy will only see it in SYN_RECV. This
11498 option is only supported on TCPv4/TCPv6 sockets and ignored by other ones.
11499
William Lallemandf6975e92017-05-26 17:42:10 +020011500expose-fd listeners
11501 This option is only usable with the stats socket. It gives your stats socket
11502 the capability to pass listeners FD to another HAProxy process.
William Lallemande202b1e2017-06-01 17:38:56 +020011503 During a reload with the master-worker mode, the process is automatically
11504 reexecuted adding -x and one of the stats socket with this option.
Davor Ocelice9ed2812017-12-25 17:49:28 +010011505 See also "-x" in the management guide.
William Lallemandf6975e92017-05-26 17:42:10 +020011506
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011507force-sslv3
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011508 This option enforces use of SSLv3 only on SSL connections instantiated from
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011509 this listener. SSLv3 is generally less expensive than the TLS counterparts
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011510 for high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011511 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011512
11513force-tlsv10
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011514 This option enforces use of TLSv1.0 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011515 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011516 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011517
11518force-tlsv11
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011519 This option enforces use of TLSv1.1 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011520 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011521 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011522
11523force-tlsv12
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011524 This option enforces use of TLSv1.2 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011525 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011526 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011527
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011528force-tlsv13
11529 This option enforces use of TLSv1.3 only on SSL connections instantiated from
11530 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011531 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011532
Christopher Faulet31af49d2015-06-09 17:29:50 +020011533generate-certificates
11534 This setting is only available when support for OpenSSL was built in. It
11535 enables the dynamic SSL certificates generation. A CA certificate and its
11536 private key are necessary (see 'ca-sign-file'). When HAProxy is configured as
11537 a transparent forward proxy, SSL requests generate errors because of a common
11538 name mismatch on the certificate presented to the client. With this option
11539 enabled, HAProxy will try to forge a certificate using the SNI hostname
11540 indicated by the client. This is done only if no certificate matches the SNI
11541 hostname (see 'crt-list'). If an error occurs, the default certificate is
11542 used, else the 'strict-sni' option is set.
11543 It can also be used when HAProxy is configured as a reverse proxy to ease the
11544 deployment of an architecture with many backends.
11545
11546 Creating a SSL certificate is an expensive operation, so a LRU cache is used
11547 to store forged certificates (see 'tune.ssl.ssl-ctx-cache-size'). It
Davor Ocelice9ed2812017-12-25 17:49:28 +010011548 increases the HAProxy's memory footprint to reduce latency when the same
Christopher Faulet31af49d2015-06-09 17:29:50 +020011549 certificate is used many times.
11550
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011551gid <gid>
11552 Sets the group of the UNIX sockets to the designated system gid. It can also
11553 be set by default in the global section's "unix-bind" statement. Note that
11554 some platforms simply ignore this. This setting is equivalent to the "group"
11555 setting except that the group ID is used instead of its name. This setting is
11556 ignored by non UNIX sockets.
11557
11558group <group>
11559 Sets the group of the UNIX sockets to the designated system group. It can
11560 also be set by default in the global section's "unix-bind" statement. Note
11561 that some platforms simply ignore this. This setting is equivalent to the
11562 "gid" setting except that the group name is used instead of its gid. This
11563 setting is ignored by non UNIX sockets.
11564
11565id <id>
11566 Fixes the socket ID. By default, socket IDs are automatically assigned, but
11567 sometimes it is more convenient to fix them to ease monitoring. This value
11568 must be strictly positive and unique within the listener/frontend. This
11569 option can only be used when defining only a single socket.
11570
11571interface <interface>
Lukas Tribusfce2e962013-02-12 22:13:19 +010011572 Restricts the socket to a specific interface. When specified, only packets
11573 received from that particular interface are processed by the socket. This is
11574 currently only supported on Linux. The interface must be a primary system
11575 interface, not an aliased interface. It is also possible to bind multiple
11576 frontends to the same address if they are bound to different interfaces. Note
11577 that binding to a network interface requires root privileges. This parameter
Jérôme Magnin61275192018-02-07 11:39:58 +010011578 is only compatible with TCPv4/TCPv6 sockets. When specified, return traffic
11579 uses the same interface as inbound traffic, and its associated routing table,
11580 even if there are explicit routes through different interfaces configured.
11581 This can prove useful to address asymmetric routing issues when the same
11582 client IP addresses need to be able to reach frontends hosted on different
11583 interfaces.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011584
Willy Tarreauabb175f2012-09-24 12:43:26 +020011585level <level>
11586 This setting is used with the stats sockets only to restrict the nature of
11587 the commands that can be issued on the socket. It is ignored by other
11588 sockets. <level> can be one of :
Davor Ocelice9ed2812017-12-25 17:49:28 +010011589 - "user" is the least privileged level; only non-sensitive stats can be
Willy Tarreauabb175f2012-09-24 12:43:26 +020011590 read, and no change is allowed. It would make sense on systems where it
11591 is not easy to restrict access to the socket.
11592 - "operator" is the default level and fits most common uses. All data can
Davor Ocelice9ed2812017-12-25 17:49:28 +010011593 be read, and only non-sensitive changes are permitted (e.g. clear max
Willy Tarreauabb175f2012-09-24 12:43:26 +020011594 counters).
Davor Ocelice9ed2812017-12-25 17:49:28 +010011595 - "admin" should be used with care, as everything is permitted (e.g. clear
Willy Tarreauabb175f2012-09-24 12:43:26 +020011596 all counters).
11597
Andjelko Iharosc4df59e2017-07-20 11:59:48 +020011598severity-output <format>
11599 This setting is used with the stats sockets only to configure severity
11600 level output prepended to informational feedback messages. Severity
11601 level of messages can range between 0 and 7, conforming to syslog
11602 rfc5424. Valid and successful socket commands requesting data
11603 (i.e. "show map", "get acl foo" etc.) will never have a severity level
11604 prepended. It is ignored by other sockets. <format> can be one of :
11605 - "none" (default) no severity level is prepended to feedback messages.
11606 - "number" severity level is prepended as a number.
11607 - "string" severity level is prepended as a string following the
11608 rfc5424 convention.
11609
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011610maxconn <maxconn>
11611 Limits the sockets to this number of concurrent connections. Extraneous
11612 connections will remain in the system's backlog until a connection is
11613 released. If unspecified, the limit will be the same as the frontend's
11614 maxconn. Note that in case of port ranges or multiple addresses, the same
11615 value will be applied to each socket. This setting enables different
11616 limitations on expensive sockets, for instance SSL entries which may easily
11617 eat all memory.
11618
11619mode <mode>
11620 Sets the octal mode used to define access permissions on the UNIX socket. It
11621 can also be set by default in the global section's "unix-bind" statement.
11622 Note that some platforms simply ignore this. This setting is ignored by non
11623 UNIX sockets.
11624
11625mss <maxseg>
11626 Sets the TCP Maximum Segment Size (MSS) value to be advertised on incoming
11627 connections. This can be used to force a lower MSS for certain specific
11628 ports, for instance for connections passing through a VPN. Note that this
11629 relies on a kernel feature which is theoretically supported under Linux but
11630 was buggy in all versions prior to 2.6.28. It may or may not work on other
11631 operating systems. It may also not change the advertised value but change the
11632 effective size of outgoing segments. The commonly advertised value for TCPv4
11633 over Ethernet networks is 1460 = 1500(MTU) - 40(IP+TCP). If this value is
11634 positive, it will be used as the advertised MSS. If it is negative, it will
11635 indicate by how much to reduce the incoming connection's advertised MSS for
11636 outgoing segments. This parameter is only compatible with TCP v4/v6 sockets.
11637
11638name <name>
11639 Sets an optional name for these sockets, which will be reported on the stats
11640 page.
11641
Willy Tarreaud72f0f32015-10-13 14:50:22 +020011642namespace <name>
11643 On Linux, it is possible to specify which network namespace a socket will
11644 belong to. This directive makes it possible to explicitly bind a listener to
11645 a namespace different from the default one. Please refer to your operating
11646 system's documentation to find more details about network namespaces.
11647
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011648nice <nice>
11649 Sets the 'niceness' of connections initiated from the socket. Value must be
11650 in the range -1024..1024 inclusive, and defaults to zero. Positive values
11651 means that such connections are more friendly to others and easily offer
11652 their place in the scheduler. On the opposite, negative values mean that
11653 connections want to run with a higher priority than others. The difference
11654 only happens under high loads when the system is close to saturation.
11655 Negative values are appropriate for low-latency or administration services,
11656 and high values are generally recommended for CPU intensive tasks such as SSL
11657 processing or bulk transfers which are less sensible to latency. For example,
11658 it may make sense to use a positive value for an SMTP socket and a negative
11659 one for an RDP socket.
11660
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020011661no-ca-names
11662 This setting is only available when support for OpenSSL was built in. It
11663 prevents from send CA names in server hello message when ca-file is used.
Emmanuel Hocdet842e94e2019-12-16 16:39:17 +010011664 Use "ca-verify-file" instead of "ca-file" with "no-ca-names".
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020011665
Emeric Brun9b3009b2012-10-05 11:55:06 +020011666no-sslv3
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011667 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011668 disables support for SSLv3 on any sockets instantiated from the listener when
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011669 SSL is supported. Note that SSLv2 is forced disabled in the code and cannot
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011670 be enabled using any configuration option. This option is also available on
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011671 global statement "ssl-default-bind-options". Use "ssl-min-ver" and
11672 "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011673
Emeric Brun90ad8722012-10-02 14:00:59 +020011674no-tls-tickets
11675 This setting is only available when support for OpenSSL was built in. It
11676 disables the stateless session resumption (RFC 5077 TLS Ticket
11677 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011678 session resumption is more expensive in CPU usage. This option is also
11679 available on global statement "ssl-default-bind-options".
Lukas Tribusbdb386d2020-03-10 00:56:09 +010011680 The TLS ticket mechanism is only used up to TLS 1.2.
11681 Forward Secrecy is compromised with TLS tickets, unless ticket keys
11682 are periodically rotated (via reload or by using "tls-ticket-keys").
Emeric Brun90ad8722012-10-02 14:00:59 +020011683
Emeric Brun9b3009b2012-10-05 11:55:06 +020011684no-tlsv10
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011685 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011686 disables support for TLSv1.0 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011687 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011688 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011689 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11690 and "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011691
Emeric Brun9b3009b2012-10-05 11:55:06 +020011692no-tlsv11
Emeric Brunf5da4932012-09-28 19:42:54 +020011693 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011694 disables support for TLSv1.1 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011695 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011696 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011697 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11698 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020011699
Emeric Brun9b3009b2012-10-05 11:55:06 +020011700no-tlsv12
Emeric Brunf5da4932012-09-28 19:42:54 +020011701 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011702 disables support for TLSv1.2 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011703 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011704 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011705 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11706 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020011707
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011708no-tlsv13
11709 This setting is only available when support for OpenSSL was built in. It
11710 disables support for TLSv1.3 on any sockets instantiated from the listener
11711 when SSL is supported. Note that SSLv2 is forced disabled in the code and
11712 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011713 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11714 and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011715
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020011716npn <protocols>
11717 This enables the NPN TLS extension and advertises the specified protocol list
11718 as supported on top of NPN. The protocol list consists in a comma-delimited
11719 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
John Roeslerfb2fce12019-07-10 15:45:51 -050011720 This requires that the SSL library is built with support for TLS extensions
Willy Tarreauab861d32013-04-02 02:30:41 +020011721 enabled (check with haproxy -vv). Note that the NPN extension has been
Willy Tarreau95c4e142017-11-26 12:18:55 +010011722 replaced with the ALPN extension (see the "alpn" keyword), though this one is
11723 only available starting with OpenSSL 1.0.2. If HTTP/2 is desired on an older
11724 version of OpenSSL, NPN might still be used as most clients still support it
11725 at the time of writing this. It is possible to enable both NPN and ALPN
11726 though it probably doesn't make any sense out of testing.
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020011727
Lukas Tribus53ae85c2017-05-04 15:45:40 +000011728prefer-client-ciphers
11729 Use the client's preference when selecting the cipher suite, by default
11730 the server's preference is enforced. This option is also available on
11731 global statement "ssl-default-bind-options".
Lukas Tribus926594f2018-05-18 17:55:57 +020011732 Note that with OpenSSL >= 1.1.1 ChaCha20-Poly1305 is reprioritized anyway
11733 (without setting this option), if a ChaCha20-Poly1305 cipher is at the top of
11734 the client cipher list.
Lukas Tribus53ae85c2017-05-04 15:45:40 +000011735
Christopher Fauletc644fa92017-11-23 22:44:11 +010011736process <process-set>[/<thread-set>]
Willy Tarreaua36b3242019-02-02 13:14:34 +010011737 This restricts the list of processes or threads on which this listener is
Christopher Fauletc644fa92017-11-23 22:44:11 +010011738 allowed to run. It does not enforce any process but eliminates those which do
Davor Ocelice9ed2812017-12-25 17:49:28 +010011739 not match. If the frontend uses a "bind-process" setting, the intersection
Christopher Fauletc644fa92017-11-23 22:44:11 +010011740 between the two is applied. If in the end the listener is not allowed to run
11741 on any remaining process, a warning is emitted, and the listener will either
11742 run on the first process of the listener if a single process was specified,
11743 or on all of its processes if multiple processes were specified. If a thread
Davor Ocelice9ed2812017-12-25 17:49:28 +010011744 set is specified, it limits the threads allowed to process incoming
Willy Tarreaua36b3242019-02-02 13:14:34 +010011745 connections for this listener, for the the process set. If multiple processes
11746 and threads are configured, a warning is emitted, as it either results from a
11747 configuration error or a misunderstanding of these models. For the unlikely
11748 case where several ranges are needed, this directive may be repeated.
11749 <process-set> and <thread-set> must use the format
Christopher Fauletc644fa92017-11-23 22:44:11 +010011750
11751 all | odd | even | number[-[number]]
11752
11753 Ranges can be partially defined. The higher bound can be omitted. In such
11754 case, it is replaced by the corresponding maximum value. The main purpose of
11755 this directive is to be used with the stats sockets and have one different
11756 socket per process. The second purpose is to have multiple bind lines sharing
11757 the same IP:port but not the same process in a listener, so that the system
11758 can distribute the incoming connections into multiple queues and allow a
11759 smoother inter-process load balancing. Currently Linux 3.9 and above is known
11760 for supporting this. See also "bind-process" and "nbproc".
Willy Tarreau6ae1ba62014-05-07 19:01:58 +020011761
Christopher Fauleta717b992018-04-10 14:43:00 +020011762proto <name>
11763 Forces the multiplexer's protocol to use for the incoming connections. It
11764 must be compatible with the mode of the frontend (TCP or HTTP). It must also
11765 be usable on the frontend side. The list of available protocols is reported
11766 in haproxy -vv.
11767 Idea behind this optipon is to bypass the selection of the best multiplexer's
11768 protocol for all connections instantiated from this listening socket. For
Joseph Herlant71b4b152018-11-13 16:55:16 -080011769 instance, it is possible to force the http/2 on clear TCP by specifying "proto
Christopher Fauleta717b992018-04-10 14:43:00 +020011770 h2" on the bind line.
11771
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011772ssl
11773 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011774 enables SSL deciphering on connections instantiated from this listener. A
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011775 certificate is necessary (see "crt" above). All contents in the buffers will
11776 appear in clear text, so that ACLs and HTTP processing will only have access
Emmanuel Hocdetbd695fe2017-05-15 15:53:41 +020011777 to deciphered contents. SSLv3 is disabled per default, use "ssl-min-ver SSLv3"
11778 to enable it.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011779
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011780ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
11781 This option enforces use of <version> or lower on SSL connections instantiated
11782 from this listener. This option is also available on global statement
11783 "ssl-default-bind-options". See also "ssl-min-ver".
11784
11785ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
11786 This option enforces use of <version> or upper on SSL connections instantiated
11787 from this listener. This option is also available on global statement
11788 "ssl-default-bind-options". See also "ssl-max-ver".
11789
Emmanuel Hocdet65623372013-01-24 17:17:15 +010011790strict-sni
11791 This setting is only available when support for OpenSSL was built in. The
11792 SSL/TLS negotiation is allow only if the client provided an SNI which match
11793 a certificate. The default certificate is not used.
11794 See the "crt" option for more information.
11795
Willy Tarreau2af207a2015-02-04 00:45:58 +010011796tcp-ut <delay>
Tim Düsterhus4896c442016-11-29 02:15:19 +010011797 Sets the TCP User Timeout for all incoming connections instantiated from this
Willy Tarreau2af207a2015-02-04 00:45:58 +010011798 listening socket. This option is available on Linux since version 2.6.37. It
11799 allows haproxy to configure a timeout for sockets which contain data not
Davor Ocelice9ed2812017-12-25 17:49:28 +010011800 receiving an acknowledgment for the configured delay. This is especially
Willy Tarreau2af207a2015-02-04 00:45:58 +010011801 useful on long-lived connections experiencing long idle periods such as
11802 remote terminals or database connection pools, where the client and server
11803 timeouts must remain high to allow a long period of idle, but where it is
11804 important to detect that the client has disappeared in order to release all
11805 resources associated with its connection (and the server's session). The
11806 argument is a delay expressed in milliseconds by default. This only works
11807 for regular TCP connections, and is ignored for other protocols.
11808
Willy Tarreau1c862c52012-10-05 16:21:00 +020011809tfo
Lukas Tribus0defb902013-02-13 23:35:39 +010011810 Is an optional keyword which is supported only on Linux kernels >= 3.7. It
Willy Tarreau1c862c52012-10-05 16:21:00 +020011811 enables TCP Fast Open on the listening socket, which means that clients which
11812 support this feature will be able to send a request and receive a response
11813 during the 3-way handshake starting from second connection, thus saving one
11814 round-trip after the first connection. This only makes sense with protocols
11815 that use high connection rates and where each round trip matters. This can
11816 possibly cause issues with many firewalls which do not accept data on SYN
11817 packets, so this option should only be enabled once well tested. This option
Lukas Tribus0999f762013-04-02 16:43:24 +020011818 is only supported on TCPv4/TCPv6 sockets and ignored by other ones. You may
11819 need to build HAProxy with USE_TFO=1 if your libc doesn't define
11820 TCP_FASTOPEN.
Willy Tarreau1c862c52012-10-05 16:21:00 +020011821
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010011822tls-ticket-keys <keyfile>
11823 Sets the TLS ticket keys file to load the keys from. The keys need to be 48
Emeric Brun9e754772019-01-10 17:51:55 +010011824 or 80 bytes long, depending if aes128 or aes256 is used, encoded with base64
11825 with one line per key (ex. openssl rand 80 | openssl base64 -A | xargs echo).
11826 The first key determines the key length used for next keys: you can't mix
11827 aes128 and aes256 keys. Number of keys is specified by the TLS_TICKETS_NO
11828 build option (default 3) and at least as many keys need to be present in
11829 the file. Last TLS_TICKETS_NO keys will be used for decryption and the
11830 penultimate one for encryption. This enables easy key rotation by just
11831 appending new key to the file and reloading the process. Keys must be
11832 periodically rotated (ex. every 12h) or Perfect Forward Secrecy is
11833 compromised. It is also a good idea to keep the keys off any permanent
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010011834 storage such as hard drives (hint: use tmpfs and don't swap those files).
11835 Lifetime hint can be changed using tune.ssl.timeout.
11836
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011837transparent
11838 Is an optional keyword which is supported only on certain Linux kernels. It
11839 indicates that the addresses will be bound even if they do not belong to the
11840 local machine, and that packets targeting any of these addresses will be
11841 intercepted just as if the addresses were locally configured. This normally
11842 requires that IP forwarding is enabled. Caution! do not use this with the
11843 default address '*', as it would redirect any traffic for the specified port.
11844 This keyword is available only when HAProxy is built with USE_LINUX_TPROXY=1.
11845 This parameter is only compatible with TCPv4 and TCPv6 sockets, depending on
11846 kernel version. Some distribution kernels include backports of the feature,
11847 so check for support with your vendor.
11848
Willy Tarreau77e3af92012-11-24 15:07:23 +010011849v4v6
11850 Is an optional keyword which is supported only on most recent systems
11851 including Linux kernels >= 2.4.21. It is used to bind a socket to both IPv4
11852 and IPv6 when it uses the default address. Doing so is sometimes necessary
11853 on systems which bind to IPv6 only by default. It has no effect on non-IPv6
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011854 sockets, and is overridden by the "v6only" option.
Willy Tarreau77e3af92012-11-24 15:07:23 +010011855
Willy Tarreau9b6700f2012-11-24 11:55:28 +010011856v6only
11857 Is an optional keyword which is supported only on most recent systems
11858 including Linux kernels >= 2.4.21. It is used to bind a socket to IPv6 only
11859 when it uses the default address. Doing so is sometimes preferred to doing it
Willy Tarreau77e3af92012-11-24 15:07:23 +010011860 system-wide as it is per-listener. It has no effect on non-IPv6 sockets and
11861 has precedence over the "v4v6" option.
Willy Tarreau9b6700f2012-11-24 11:55:28 +010011862
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011863uid <uid>
11864 Sets the owner of the UNIX sockets to the designated system uid. It can also
11865 be set by default in the global section's "unix-bind" statement. Note that
11866 some platforms simply ignore this. This setting is equivalent to the "user"
11867 setting except that the user numeric ID is used instead of its name. This
11868 setting is ignored by non UNIX sockets.
11869
11870user <user>
11871 Sets the owner of the UNIX sockets to the designated system user. It can also
11872 be set by default in the global section's "unix-bind" statement. Note that
11873 some platforms simply ignore this. This setting is equivalent to the "uid"
11874 setting except that the user name is used instead of its uid. This setting is
11875 ignored by non UNIX sockets.
11876
Emeric Brun1a073b42012-09-28 17:07:34 +020011877verify [none|optional|required]
11878 This setting is only available when support for OpenSSL was built in. If set
11879 to 'none', client certificate is not requested. This is the default. In other
11880 cases, a client certificate is requested. If the client does not provide a
11881 certificate after the request and if 'verify' is set to 'required', then the
11882 handshake is aborted, while it would have succeeded if set to 'optional'. The
Emeric Brunfd33a262012-10-11 16:28:27 +020011883 certificate provided by the client is always verified using CAs from
11884 'ca-file' and optional CRLs from 'crl-file'. On verify failure the handshake
11885 is aborted, regardless of the 'verify' option, unless the error code exactly
11886 matches one of those listed with 'ca-ignore-err' or 'crt-ignore-err'.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020011887
Willy Tarreaub6205fd2012-09-24 12:27:33 +0200118885.2. Server and default-server options
Cyril Bontéf0c60612010-02-06 14:44:47 +010011889------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020011890
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010011891The "server" and "default-server" keywords support a certain number of settings
11892which are all passed as arguments on the server line. The order in which those
11893arguments appear does not count, and they are all optional. Some of those
11894settings are single words (booleans) while others expect one or several values
11895after them. In this case, the values must immediately follow the setting name.
11896Except default-server, all those settings must be specified after the server's
11897address if they are used:
Willy Tarreau6a06a402007-07-15 20:15:28 +020011898
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011899 server <name> <address>[:port] [settings ...]
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010011900 default-server [settings ...]
Willy Tarreau6a06a402007-07-15 20:15:28 +020011901
Frédéric Lécailled2376272017-03-21 18:52:12 +010011902Note that all these settings are supported both by "server" and "default-server"
11903keywords, except "id" which is only supported by "server".
11904
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011905The currently supported settings are the following ones.
Willy Tarreau0ba27502007-12-24 16:55:16 +010011906
Willy Tarreauceb4ac92012-04-28 00:41:46 +020011907addr <ipv4|ipv6>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011908 Using the "addr" parameter, it becomes possible to use a different IP address
Baptiste Assmann13f83532016-03-06 23:14:36 +010011909 to send health-checks or to probe the agent-check. On some servers, it may be
11910 desirable to dedicate an IP address to specific component able to perform
11911 complex tests which are more suitable to health-checks than the application.
11912 This parameter is ignored if the "check" parameter is not set. See also the
11913 "port" parameter.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011914
Simon Hormand60d6912013-11-25 10:46:36 +090011915agent-check
11916 Enable an auxiliary agent check which is run independently of a regular
Willy Tarreau81f5d942013-12-09 20:51:51 +010011917 health check. An agent health check is performed by making a TCP connection
Willy Tarreau7a0139e2018-12-16 08:42:56 +010011918 to the port set by the "agent-port" parameter and reading an ASCII string
11919 terminated by the first '\r' or '\n' met. The string is made of a series of
11920 words delimited by spaces, tabs or commas in any order, each consisting of :
Simon Hormand60d6912013-11-25 10:46:36 +090011921
Willy Tarreau81f5d942013-12-09 20:51:51 +010011922 - An ASCII representation of a positive integer percentage, e.g. "75%".
Simon Hormand60d6912013-11-25 10:46:36 +090011923 Values in this format will set the weight proportional to the initial
Willy Tarreauc5af3a62014-10-07 15:27:33 +020011924 weight of a server as configured when haproxy starts. Note that a zero
11925 weight is reported on the stats page as "DRAIN" since it has the same
11926 effect on the server (it's removed from the LB farm).
Simon Hormand60d6912013-11-25 10:46:36 +090011927
Davor Ocelice9ed2812017-12-25 17:49:28 +010011928 - The string "maxconn:" followed by an integer (no space between). Values
11929 in this format will set the maxconn of a server. The maximum number of
11930 connections advertised needs to be multiplied by the number of load
11931 balancers and different backends that use this health check to get the
11932 total number of connections the server might receive. Example: maxconn:30
Nenad Merdanovic174dd372016-04-24 23:10:06 +020011933
Willy Tarreau81f5d942013-12-09 20:51:51 +010011934 - The word "ready". This will turn the server's administrative state to the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011935 READY mode, thus canceling any DRAIN or MAINT state
Simon Hormand60d6912013-11-25 10:46:36 +090011936
Willy Tarreau81f5d942013-12-09 20:51:51 +010011937 - The word "drain". This will turn the server's administrative state to the
11938 DRAIN mode, thus it will not accept any new connections other than those
11939 that are accepted via persistence.
Simon Hormand60d6912013-11-25 10:46:36 +090011940
Willy Tarreau81f5d942013-12-09 20:51:51 +010011941 - The word "maint". This will turn the server's administrative state to the
11942 MAINT mode, thus it will not accept any new connections at all, and health
11943 checks will be stopped.
Simon Hormand60d6912013-11-25 10:46:36 +090011944
Willy Tarreau81f5d942013-12-09 20:51:51 +010011945 - The words "down", "failed", or "stopped", optionally followed by a
11946 description string after a sharp ('#'). All of these mark the server's
11947 operating state as DOWN, but since the word itself is reported on the stats
11948 page, the difference allows an administrator to know if the situation was
11949 expected or not : the service may intentionally be stopped, may appear up
Davor Ocelice9ed2812017-12-25 17:49:28 +010011950 but fail some validity tests, or may be seen as down (e.g. missing process,
Willy Tarreau81f5d942013-12-09 20:51:51 +010011951 or port not responding).
Simon Hormand60d6912013-11-25 10:46:36 +090011952
Willy Tarreau81f5d942013-12-09 20:51:51 +010011953 - The word "up" sets back the server's operating state as UP if health checks
11954 also report that the service is accessible.
Simon Hormand60d6912013-11-25 10:46:36 +090011955
Willy Tarreau81f5d942013-12-09 20:51:51 +010011956 Parameters which are not advertised by the agent are not changed. For
11957 example, an agent might be designed to monitor CPU usage and only report a
11958 relative weight and never interact with the operating status. Similarly, an
11959 agent could be designed as an end-user interface with 3 radio buttons
11960 allowing an administrator to change only the administrative state. However,
11961 it is important to consider that only the agent may revert its own actions,
11962 so if a server is set to DRAIN mode or to DOWN state using the agent, the
11963 agent must implement the other equivalent actions to bring the service into
11964 operations again.
Simon Hormand60d6912013-11-25 10:46:36 +090011965
Simon Horman2f1f9552013-11-25 10:46:37 +090011966 Failure to connect to the agent is not considered an error as connectivity
11967 is tested by the regular health check which is enabled by the "check"
Willy Tarreau81f5d942013-12-09 20:51:51 +010011968 parameter. Warning though, it is not a good idea to stop an agent after it
11969 reports "down", since only an agent reporting "up" will be able to turn the
11970 server up again. Note that the CLI on the Unix stats socket is also able to
Willy Tarreau989222a2016-01-15 10:26:26 +010011971 force an agent's result in order to work around a bogus agent if needed.
Simon Horman2f1f9552013-11-25 10:46:37 +090011972
Willy Tarreau81f5d942013-12-09 20:51:51 +010011973 Requires the "agent-port" parameter to be set. See also the "agent-inter"
Frédéric Lécailled2376272017-03-21 18:52:12 +010011974 and "no-agent-check" parameters.
Simon Hormand60d6912013-11-25 10:46:36 +090011975
James Brown55f9ff12015-10-21 18:19:05 -070011976agent-send <string>
11977 If this option is specified, haproxy will send the given string (verbatim)
11978 to the agent server upon connection. You could, for example, encode
11979 the backend name into this string, which would enable your agent to send
11980 different responses based on the backend. Make sure to include a '\n' if
11981 you want to terminate your request with a newline.
11982
Simon Hormand60d6912013-11-25 10:46:36 +090011983agent-inter <delay>
11984 The "agent-inter" parameter sets the interval between two agent checks
11985 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
11986
11987 Just as with every other time-based parameter, it may be entered in any
11988 other explicit unit among { us, ms, s, m, h, d }. The "agent-inter"
11989 parameter also serves as a timeout for agent checks "timeout check" is
11990 not set. In order to reduce "resonance" effects when multiple servers are
11991 hosted on the same hardware, the agent and health checks of all servers
11992 are started with a small time offset between them. It is also possible to
11993 add some random noise in the agent and health checks interval using the
11994 global "spread-checks" keyword. This makes sense for instance when a lot
11995 of backends use the same servers.
11996
11997 See also the "agent-check" and "agent-port" parameters.
11998
Misiek768d8602017-01-09 09:52:43 +010011999agent-addr <addr>
12000 The "agent-addr" parameter sets address for agent check.
12001
12002 You can offload agent-check to another target, so you can make single place
12003 managing status and weights of servers defined in haproxy in case you can't
12004 make self-aware and self-managing services. You can specify both IP or
12005 hostname, it will be resolved.
12006
Simon Hormand60d6912013-11-25 10:46:36 +090012007agent-port <port>
12008 The "agent-port" parameter sets the TCP port used for agent checks.
12009
12010 See also the "agent-check" and "agent-inter" parameters.
12011
Olivier Houchard8cb2d2e2019-05-06 18:58:48 +020012012allow-0rtt
12013 Allow sending early data to the server when using TLS 1.3.
Olivier Houchard22c9b442019-05-06 19:01:04 +020012014 Note that early data will be sent only if the client used early data, or
12015 if the backend uses "retry-on" with the "0rtt-rejected" keyword.
Olivier Houchard8cb2d2e2019-05-06 18:58:48 +020012016
Olivier Houchardc7566002018-11-20 23:33:50 +010012017alpn <protocols>
12018 This enables the TLS ALPN extension and advertises the specified protocol
12019 list as supported on top of ALPN. The protocol list consists in a comma-
12020 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
John Roeslerfb2fce12019-07-10 15:45:51 -050012021 quotes). This requires that the SSL library is built with support for TLS
Olivier Houchardc7566002018-11-20 23:33:50 +010012022 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
12023 initial NPN extension. ALPN is required to connect to HTTP/2 servers.
12024 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
12025 now obsolete NPN extension.
12026 If both HTTP/2 and HTTP/1.1 are expected to be supported, both versions can
12027 be advertised, in order of preference, like below :
12028
12029 server 127.0.0.1:443 ssl crt pub.pem alpn h2,http/1.1
12030
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012031backup
12032 When "backup" is present on a server line, the server is only used in load
12033 balancing when all other non-backup servers are unavailable. Requests coming
12034 with a persistence cookie referencing the server will always be served
12035 though. By default, only the first operational backup server is used, unless
Frédéric Lécailled2376272017-03-21 18:52:12 +010012036 the "allbackups" option is set in the backend. See also the "no-backup" and
12037 "allbackups" options.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012038
Emeric Brunef42d922012-10-11 16:11:36 +020012039ca-file <cafile>
12040 This setting is only available when support for OpenSSL was built in. It
12041 designates a PEM file from which to load CA certificates used to verify
12042 server's certificate.
12043
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012044check
12045 This option enables health checks on the server. By default, a server is
Patrick Mézardb7aeec62012-01-22 16:01:22 +010012046 always considered available. If "check" is set, the server is available when
12047 accepting periodic TCP connections, to ensure that it is really able to serve
12048 requests. The default address and port to send the tests to are those of the
12049 server, and the default source is the same as the one defined in the
12050 backend. It is possible to change the address using the "addr" parameter, the
12051 port using the "port" parameter, the source address using the "source"
12052 address, and the interval and timers using the "inter", "rise" and "fall"
Simon Hormanafc47ee2013-11-25 10:46:35 +090012053 parameters. The request method is define in the backend using the "httpchk",
12054 "smtpchk", "mysql-check", "pgsql-check" and "ssl-hello-chk" options. Please
Frédéric Lécailled2376272017-03-21 18:52:12 +010012055 refer to those options and parameters for more information. See also
12056 "no-check" option.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012057
Willy Tarreau6c16adc2012-10-05 00:04:16 +020012058check-send-proxy
12059 This option forces emission of a PROXY protocol line with outgoing health
12060 checks, regardless of whether the server uses send-proxy or not for the
12061 normal traffic. By default, the PROXY protocol is enabled for health checks
12062 if it is already enabled for normal traffic and if no "port" nor "addr"
12063 directive is present. However, if such a directive is present, the
12064 "check-send-proxy" option needs to be used to force the use of the
12065 protocol. See also the "send-proxy" option for more information.
12066
Olivier Houchard92150142018-12-21 19:47:01 +010012067check-alpn <protocols>
12068 Defines which protocols to advertise with ALPN. The protocol list consists in
12069 a comma-delimited list of protocol names, for instance: "http/1.1,http/1.0"
12070 (without quotes). If it is not set, the server ALPN is used.
12071
Jérôme Magninae9bb762018-12-09 16:08:26 +010012072check-sni <sni>
Olivier Houchard9130a962017-10-17 17:33:43 +020012073 This option allows you to specify the SNI to be used when doing health checks
Jérôme Magninae9bb762018-12-09 16:08:26 +010012074 over SSL. It is only possible to use a string to set <sni>. If you want to
12075 set a SNI for proxied traffic, see "sni".
Olivier Houchard9130a962017-10-17 17:33:43 +020012076
Willy Tarreau763a95b2012-10-04 23:15:39 +020012077check-ssl
12078 This option forces encryption of all health checks over SSL, regardless of
12079 whether the server uses SSL or not for the normal traffic. This is generally
12080 used when an explicit "port" or "addr" directive is specified and SSL health
12081 checks are not inherited. It is important to understand that this option
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012082 inserts an SSL transport layer below the checks, so that a simple TCP connect
Willy Tarreau763a95b2012-10-04 23:15:39 +020012083 check becomes an SSL connect, which replaces the old ssl-hello-chk. The most
12084 common use is to send HTTPS checks by combining "httpchk" with SSL checks.
Davor Ocelice9ed2812017-12-25 17:49:28 +010012085 All SSL settings are common to health checks and traffic (e.g. ciphers).
Frédéric Lécailled2376272017-03-21 18:52:12 +010012086 See the "ssl" option for more information and "no-check-ssl" to disable
12087 this option.
Willy Tarreau763a95b2012-10-04 23:15:39 +020012088
Alexander Liu2a54bb72019-05-22 19:44:48 +080012089check-via-socks4
John Roeslerfb2fce12019-07-10 15:45:51 -050012090 This option enables outgoing health checks using upstream socks4 proxy. By
Alexander Liu2a54bb72019-05-22 19:44:48 +080012091 default, the health checks won't go through socks tunnel even it was enabled
12092 for normal traffic.
12093
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012094ciphers <ciphers>
Dirkjan Bussink415150f2018-09-14 11:14:21 +020012095 This setting is only available when support for OpenSSL was built in. This
12096 option sets the string describing the list of cipher algorithms that is
12097 negotiated during the SSL/TLS handshake with the server. The format of the
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000012098 string is defined in "man 1 ciphers" from OpenSSL man pages. For background
12099 information and recommendations see e.g.
12100 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
12101 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
12102 cipher configuration, please check the "ciphersuites" keyword.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012103
Dirkjan Bussink415150f2018-09-14 11:14:21 +020012104ciphersuites <ciphersuites>
12105 This setting is only available when support for OpenSSL was built in and
12106 OpenSSL 1.1.1 or later was used to build HAProxy. This option sets the string
12107 describing the list of cipher algorithms that is negotiated during the TLS
12108 1.3 handshake with the server. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000012109 "man 1 ciphers" from OpenSSL man pages under the "ciphersuites" section.
12110 For cipher configuration for TLSv1.2 and earlier, please check the "ciphers"
12111 keyword.
Dirkjan Bussink415150f2018-09-14 11:14:21 +020012112
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012113cookie <value>
12114 The "cookie" parameter sets the cookie value assigned to the server to
12115 <value>. This value will be checked in incoming requests, and the first
12116 operational server possessing the same value will be selected. In return, in
12117 cookie insertion or rewrite modes, this value will be assigned to the cookie
12118 sent to the client. There is nothing wrong in having several servers sharing
12119 the same cookie value, and it is in fact somewhat common between normal and
12120 backup servers. See also the "cookie" keyword in backend section.
12121
Emeric Brunef42d922012-10-11 16:11:36 +020012122crl-file <crlfile>
12123 This setting is only available when support for OpenSSL was built in. It
12124 designates a PEM file from which to load certificate revocation list used
12125 to verify server's certificate.
12126
Emeric Bruna7aa3092012-10-26 12:58:00 +020012127crt <cert>
12128 This setting is only available when support for OpenSSL was built in.
12129 It designates a PEM file from which to load both a certificate and the
12130 associated private key. This file can be built by concatenating both PEM
12131 files into one. This certificate will be sent if the server send a client
12132 certificate request.
12133
Willy Tarreau96839092010-03-29 10:02:24 +020012134disabled
12135 The "disabled" keyword starts the server in the "disabled" state. That means
12136 that it is marked down in maintenance mode, and no connection other than the
12137 ones allowed by persist mode will reach it. It is very well suited to setup
12138 new servers, because normal traffic will never reach them, while it is still
12139 possible to test the service by making use of the force-persist mechanism.
Frédéric Lécailled2376272017-03-21 18:52:12 +010012140 See also "enabled" setting.
Willy Tarreau96839092010-03-29 10:02:24 +020012141
Frédéric Lécailled2376272017-03-21 18:52:12 +010012142enabled
12143 This option may be used as 'server' setting to reset any 'disabled'
12144 setting which would have been inherited from 'default-server' directive as
12145 default value.
12146 It may also be used as 'default-server' setting to reset any previous
12147 'default-server' 'disabled' setting.
Willy Tarreau96839092010-03-29 10:02:24 +020012148
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012149error-limit <count>
Willy Tarreau983e01e2010-01-11 18:42:06 +010012150 If health observing is enabled, the "error-limit" parameter specifies the
12151 number of consecutive errors that triggers event selected by the "on-error"
12152 option. By default it is set to 10 consecutive errors.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010012153
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012154 See also the "check", "error-limit" and "on-error".
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010012155
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012156fall <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012157 The "fall" parameter states that a server will be considered as dead after
12158 <count> consecutive unsuccessful health checks. This value defaults to 3 if
12159 unspecified. See also the "check", "inter" and "rise" parameters.
12160
Emeric Brun8694b9a2012-10-05 14:39:07 +020012161force-sslv3
12162 This option enforces use of SSLv3 only when SSL is used to communicate with
12163 the server. SSLv3 is generally less expensive than the TLS counterparts for
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012164 high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012165 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020012166
12167force-tlsv10
12168 This option enforces use of TLSv1.0 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012169 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012170 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020012171
12172force-tlsv11
12173 This option enforces use of TLSv1.1 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012174 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012175 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020012176
12177force-tlsv12
12178 This option enforces use of TLSv1.2 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012179 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012180 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020012181
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020012182force-tlsv13
12183 This option enforces use of TLSv1.3 only when SSL is used to communicate with
12184 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012185 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020012186
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012187id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +020012188 Set a persistent ID for the server. This ID must be positive and unique for
12189 the proxy. An unused ID will automatically be assigned if unset. The first
12190 assigned value will be 1. This ID is currently only returned in statistics.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012191
Willy Tarreau6a031d12016-11-07 19:42:35 +010012192init-addr {last | libc | none | <ip>},[...]*
12193 Indicate in what order the server's address should be resolved upon startup
12194 if it uses an FQDN. Attempts are made to resolve the address by applying in
Davor Ocelice9ed2812017-12-25 17:49:28 +010012195 turn each of the methods mentioned in the comma-delimited list. The first
Willy Tarreau6a031d12016-11-07 19:42:35 +010012196 method which succeeds is used. If the end of the list is reached without
12197 finding a working method, an error is thrown. Method "last" suggests to pick
12198 the address which appears in the state file (see "server-state-file"). Method
12199 "libc" uses the libc's internal resolver (gethostbyname() or getaddrinfo()
12200 depending on the operating system and build options). Method "none"
12201 specifically indicates that the server should start without any valid IP
12202 address in a down state. It can be useful to ignore some DNS issues upon
12203 startup, waiting for the situation to get fixed later. Finally, an IP address
12204 (IPv4 or IPv6) may be provided. It can be the currently known address of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012205 server (e.g. filled by a configuration generator), or the address of a dummy
Willy Tarreau6a031d12016-11-07 19:42:35 +010012206 server used to catch old sessions and present them with a decent error
12207 message for example. When the "first" load balancing algorithm is used, this
12208 IP address could point to a fake server used to trigger the creation of new
12209 instances on the fly. This option defaults to "last,libc" indicating that the
12210 previous address found in the state file (if any) is used first, otherwise
12211 the libc's resolver is used. This ensures continued compatibility with the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012212 historic behavior.
Willy Tarreau6a031d12016-11-07 19:42:35 +010012213
12214 Example:
12215 defaults
12216 # never fail on address resolution
12217 default-server init-addr last,libc,none
12218
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012219inter <delay>
12220fastinter <delay>
12221downinter <delay>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012222 The "inter" parameter sets the interval between two consecutive health checks
12223 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
12224 It is also possible to use "fastinter" and "downinter" to optimize delays
12225 between checks depending on the server state :
12226
Pieter Baauw44fc9df2015-09-17 21:30:46 +020012227 Server state | Interval used
12228 ----------------------------------------+----------------------------------
12229 UP 100% (non-transitional) | "inter"
12230 ----------------------------------------+----------------------------------
12231 Transitionally UP (going down "fall"), | "fastinter" if set,
12232 Transitionally DOWN (going up "rise"), | "inter" otherwise.
12233 or yet unchecked. |
12234 ----------------------------------------+----------------------------------
12235 DOWN 100% (non-transitional) | "downinter" if set,
12236 | "inter" otherwise.
12237 ----------------------------------------+----------------------------------
Willy Tarreaud72758d2010-01-12 10:42:19 +010012238
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012239 Just as with every other time-based parameter, they can be entered in any
12240 other explicit unit among { us, ms, s, m, h, d }. The "inter" parameter also
12241 serves as a timeout for health checks sent to servers if "timeout check" is
12242 not set. In order to reduce "resonance" effects when multiple servers are
Simon Hormand60d6912013-11-25 10:46:36 +090012243 hosted on the same hardware, the agent and health checks of all servers
12244 are started with a small time offset between them. It is also possible to
12245 add some random noise in the agent and health checks interval using the
12246 global "spread-checks" keyword. This makes sense for instance when a lot
12247 of backends use the same servers.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012248
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012249maxconn <maxconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012250 The "maxconn" parameter specifies the maximal number of concurrent
12251 connections that will be sent to this server. If the number of incoming
Tim Duesterhuscefbbd92019-11-27 22:35:27 +010012252 concurrent connections goes higher than this value, they will be queued,
12253 waiting for a slot to be released. This parameter is very important as it can
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012254 save fragile servers from going down under extreme loads. If a "minconn"
12255 parameter is specified, the limit becomes dynamic. The default value is "0"
12256 which means unlimited. See also the "minconn" and "maxqueue" parameters, and
12257 the backend's "fullconn" keyword.
12258
Tim Duesterhuscefbbd92019-11-27 22:35:27 +010012259 In HTTP mode this parameter limits the number of concurrent requests instead
12260 of the number of connections. Multiple requests might be multiplexed over a
12261 single TCP connection to the server. As an example if you specify a maxconn
12262 of 50 you might see between 1 and 50 actual server connections, but no more
12263 than 50 concurrent requests.
12264
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012265maxqueue <maxqueue>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012266 The "maxqueue" parameter specifies the maximal number of connections which
12267 will wait in the queue for this server. If this limit is reached, next
12268 requests will be redispatched to other servers instead of indefinitely
12269 waiting to be served. This will break persistence but may allow people to
12270 quickly re-log in when the server they try to connect to is dying. The
12271 default value is "0" which means the queue is unlimited. See also the
12272 "maxconn" and "minconn" parameters.
12273
Willy Tarreau9c538e02019-01-23 10:21:49 +010012274max-reuse <count>
12275 The "max-reuse" argument indicates the HTTP connection processors that they
12276 should not reuse a server connection more than this number of times to send
12277 new requests. Permitted values are -1 (the default), which disables this
12278 limit, or any positive value. Value zero will effectively disable keep-alive.
12279 This is only used to work around certain server bugs which cause them to leak
12280 resources over time. The argument is not necessarily respected by the lower
12281 layers as there might be technical limitations making it impossible to
12282 enforce. At least HTTP/2 connections to servers will respect it.
12283
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012284minconn <minconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012285 When the "minconn" parameter is set, the maxconn limit becomes a dynamic
12286 limit following the backend's load. The server will always accept at least
12287 <minconn> connections, never more than <maxconn>, and the limit will be on
12288 the ramp between both values when the backend has less than <fullconn>
12289 concurrent connections. This makes it possible to limit the load on the
12290 server during normal loads, but push it further for important loads without
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012291 overloading the server during exceptional loads. See also the "maxconn"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012292 and "maxqueue" parameters, as well as the "fullconn" backend keyword.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010012293
Willy Tarreaud72f0f32015-10-13 14:50:22 +020012294namespace <name>
12295 On Linux, it is possible to specify which network namespace a socket will
12296 belong to. This directive makes it possible to explicitly bind a server to
12297 a namespace different from the default one. Please refer to your operating
12298 system's documentation to find more details about network namespaces.
12299
Frédéric Lécailled2376272017-03-21 18:52:12 +010012300no-agent-check
12301 This option may be used as "server" setting to reset any "agent-check"
12302 setting which would have been inherited from "default-server" directive as
12303 default value.
12304 It may also be used as "default-server" setting to reset any previous
12305 "default-server" "agent-check" setting.
12306
12307no-backup
12308 This option may be used as "server" setting to reset any "backup"
12309 setting which would have been inherited from "default-server" directive as
12310 default value.
12311 It may also be used as "default-server" setting to reset any previous
12312 "default-server" "backup" setting.
12313
12314no-check
12315 This option may be used as "server" setting to reset any "check"
12316 setting which would have been inherited from "default-server" directive as
12317 default value.
12318 It may also be used as "default-server" setting to reset any previous
12319 "default-server" "check" setting.
12320
12321no-check-ssl
12322 This option may be used as "server" setting to reset any "check-ssl"
12323 setting which would have been inherited from "default-server" directive as
12324 default value.
12325 It may also be used as "default-server" setting to reset any previous
12326 "default-server" "check-ssl" setting.
12327
Frédéric Lécailled2376272017-03-21 18:52:12 +010012328no-send-proxy
12329 This option may be used as "server" setting to reset any "send-proxy"
12330 setting which would have been inherited from "default-server" directive as
12331 default value.
12332 It may also be used as "default-server" setting to reset any previous
12333 "default-server" "send-proxy" setting.
12334
12335no-send-proxy-v2
12336 This option may be used as "server" setting to reset any "send-proxy-v2"
12337 setting which would have been inherited from "default-server" directive as
12338 default value.
12339 It may also be used as "default-server" setting to reset any previous
12340 "default-server" "send-proxy-v2" setting.
12341
12342no-send-proxy-v2-ssl
12343 This option may be used as "server" setting to reset any "send-proxy-v2-ssl"
12344 setting which would have been inherited from "default-server" directive as
12345 default value.
12346 It may also be used as "default-server" setting to reset any previous
12347 "default-server" "send-proxy-v2-ssl" setting.
12348
12349no-send-proxy-v2-ssl-cn
12350 This option may be used as "server" setting to reset any "send-proxy-v2-ssl-cn"
12351 setting which would have been inherited from "default-server" directive as
12352 default value.
12353 It may also be used as "default-server" setting to reset any previous
12354 "default-server" "send-proxy-v2-ssl-cn" setting.
12355
12356no-ssl
12357 This option may be used as "server" setting to reset any "ssl"
12358 setting which would have been inherited from "default-server" directive as
12359 default value.
12360 It may also be used as "default-server" setting to reset any previous
12361 "default-server" "ssl" setting.
12362
Willy Tarreau2a3fb1c2015-02-05 16:47:07 +010012363no-ssl-reuse
12364 This option disables SSL session reuse when SSL is used to communicate with
12365 the server. It will force the server to perform a full handshake for every
12366 new connection. It's probably only useful for benchmarking, troubleshooting,
12367 and for paranoid users.
12368
Emeric Brun9b3009b2012-10-05 11:55:06 +020012369no-sslv3
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012370 This option disables support for SSLv3 when SSL is used to communicate with
12371 the server. Note that SSLv2 is disabled in the code and cannot be enabled
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012372 using any configuration option. Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012373
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020012374 Supported in default-server: No
12375
Emeric Brunf9c5c472012-10-11 15:28:34 +020012376no-tls-tickets
12377 This setting is only available when support for OpenSSL was built in. It
12378 disables the stateless session resumption (RFC 5077 TLS Ticket
12379 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012380 session resumption is more expensive in CPU usage for servers. This option
12381 is also available on global statement "ssl-default-server-options".
Lukas Tribusbdb386d2020-03-10 00:56:09 +010012382 The TLS ticket mechanism is only used up to TLS 1.2.
12383 Forward Secrecy is compromised with TLS tickets, unless ticket keys
12384 are periodically rotated (via reload or by using "tls-ticket-keys").
Frédéric Lécailled2376272017-03-21 18:52:12 +010012385 See also "tls-tickets".
Emeric Brunf9c5c472012-10-11 15:28:34 +020012386
Emeric Brun9b3009b2012-10-05 11:55:06 +020012387no-tlsv10
Emeric Brun8694b9a2012-10-05 14:39:07 +020012388 This option disables support for TLSv1.0 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020012389 the server. Note that SSLv2 is disabled in the code and cannot be enabled
12390 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012391 often makes sense to disable it when communicating with local servers. This
12392 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012393 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020012394
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020012395 Supported in default-server: No
12396
Emeric Brun9b3009b2012-10-05 11:55:06 +020012397no-tlsv11
Emeric Brun8694b9a2012-10-05 14:39:07 +020012398 This option disables support for TLSv1.1 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020012399 the server. Note that SSLv2 is disabled in the code and cannot be enabled
12400 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012401 often makes sense to disable it when communicating with local servers. This
12402 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012403 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020012404
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020012405 Supported in default-server: No
12406
Emeric Brun9b3009b2012-10-05 11:55:06 +020012407no-tlsv12
Emeric Brun8694b9a2012-10-05 14:39:07 +020012408 This option disables support for TLSv1.2 when SSL is used to communicate with
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012409 the server. Note that SSLv2 is disabled in the code and cannot be enabled
12410 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012411 often makes sense to disable it when communicating with local servers. This
12412 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012413 Use "ssl-min-ver" and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020012414
12415 Supported in default-server: No
12416
12417no-tlsv13
12418 This option disables support for TLSv1.3 when SSL is used to communicate with
12419 the server. Note that SSLv2 is disabled in the code and cannot be enabled
12420 using any configuration option. TLSv1 is more expensive than SSLv3 so it
12421 often makes sense to disable it when communicating with local servers. This
12422 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012423 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012424
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020012425 Supported in default-server: No
12426
Frédéric Lécailled2376272017-03-21 18:52:12 +010012427no-verifyhost
12428 This option may be used as "server" setting to reset any "verifyhost"
12429 setting which would have been inherited from "default-server" directive as
12430 default value.
12431 It may also be used as "default-server" setting to reset any previous
12432 "default-server" "verifyhost" setting.
Willy Tarreau763a95b2012-10-04 23:15:39 +020012433
Frédéric Lécaille1b9423d2019-07-04 14:19:06 +020012434no-tfo
12435 This option may be used as "server" setting to reset any "tfo"
12436 setting which would have been inherited from "default-server" directive as
12437 default value.
12438 It may also be used as "default-server" setting to reset any previous
12439 "default-server" "tfo" setting.
12440
Simon Hormanfa461682011-06-25 09:39:49 +090012441non-stick
12442 Never add connections allocated to this sever to a stick-table.
12443 This may be used in conjunction with backup to ensure that
12444 stick-table persistence is disabled for backup servers.
12445
Olivier Houchardc7566002018-11-20 23:33:50 +010012446npn <protocols>
12447 This enables the NPN TLS extension and advertises the specified protocol list
12448 as supported on top of NPN. The protocol list consists in a comma-delimited
12449 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
John Roeslerfb2fce12019-07-10 15:45:51 -050012450 This requires that the SSL library is built with support for TLS extensions
Olivier Houchardc7566002018-11-20 23:33:50 +010012451 enabled (check with haproxy -vv). Note that the NPN extension has been
12452 replaced with the ALPN extension (see the "alpn" keyword), though this one is
12453 only available starting with OpenSSL 1.0.2.
12454
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010012455observe <mode>
12456 This option enables health adjusting based on observing communication with
12457 the server. By default this functionality is disabled and enabling it also
12458 requires to enable health checks. There are two supported modes: "layer4" and
12459 "layer7". In layer4 mode, only successful/unsuccessful tcp connections are
12460 significant. In layer7, which is only allowed for http proxies, responses
12461 received from server are verified, like valid/wrong http code, unparsable
Willy Tarreau150d1462012-03-10 08:19:02 +010012462 headers, a timeout, etc. Valid status codes include 100 to 499, 501 and 505.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010012463
12464 See also the "check", "on-error" and "error-limit".
12465
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012466on-error <mode>
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010012467 Select what should happen when enough consecutive errors are detected.
12468 Currently, four modes are available:
12469 - fastinter: force fastinter
12470 - fail-check: simulate a failed check, also forces fastinter (default)
12471 - sudden-death: simulate a pre-fatal failed health check, one more failed
12472 check will mark a server down, forces fastinter
12473 - mark-down: mark the server immediately down and force fastinter
12474
12475 See also the "check", "observe" and "error-limit".
12476
Simon Hormane0d1bfb2011-06-21 14:34:58 +090012477on-marked-down <action>
12478 Modify what occurs when a server is marked down.
12479 Currently one action is available:
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070012480 - shutdown-sessions: Shutdown peer sessions. When this setting is enabled,
12481 all connections to the server are immediately terminated when the server
12482 goes down. It might be used if the health check detects more complex cases
12483 than a simple connection status, and long timeouts would cause the service
12484 to remain unresponsive for too long a time. For instance, a health check
12485 might detect that a database is stuck and that there's no chance to reuse
12486 existing connections anymore. Connections killed this way are logged with
12487 a 'D' termination code (for "Down").
Simon Hormane0d1bfb2011-06-21 14:34:58 +090012488
12489 Actions are disabled by default
12490
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070012491on-marked-up <action>
12492 Modify what occurs when a server is marked up.
12493 Currently one action is available:
12494 - shutdown-backup-sessions: Shutdown sessions on all backup servers. This is
12495 done only if the server is not in backup state and if it is not disabled
12496 (it must have an effective weight > 0). This can be used sometimes to force
12497 an active server to take all the traffic back after recovery when dealing
Davor Ocelice9ed2812017-12-25 17:49:28 +010012498 with long sessions (e.g. LDAP, SQL, ...). Doing this can cause more trouble
12499 than it tries to solve (e.g. incomplete transactions), so use this feature
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070012500 with extreme care. Sessions killed because a server comes up are logged
12501 with an 'U' termination code (for "Up").
12502
12503 Actions are disabled by default
12504
Olivier Houchard006e3102018-12-10 18:30:32 +010012505pool-max-conn <max>
12506 Set the maximum number of idling connections for a server. -1 means unlimited
12507 connections, 0 means no idle connections. The default is -1. When idle
12508 connections are enabled, orphaned idle connections which do not belong to any
12509 client session anymore are moved to a dedicated pool so that they remain
12510 usable by future clients. This only applies to connections that can be shared
12511 according to the same principles as those applying to "http-reuse".
12512
Olivier Houchardb7b3faa2018-12-14 18:15:36 +010012513pool-purge-delay <delay>
12514 Sets the delay to start purging idle connections. Each <delay> interval, half
Olivier Houcharda56eebf2019-03-19 16:44:02 +010012515 of the idle connections are closed. 0 means we don't keep any idle connection.
Willy Tarreaufb553652019-06-04 14:06:31 +020012516 The default is 5s.
Olivier Houchardb7b3faa2018-12-14 18:15:36 +010012517
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012518port <port>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012519 Using the "port" parameter, it becomes possible to use a different port to
12520 send health-checks. On some servers, it may be desirable to dedicate a port
12521 to a specific component able to perform complex tests which are more suitable
12522 to health-checks than the application. It is common to run a simple script in
12523 inetd for instance. This parameter is ignored if the "check" parameter is not
12524 set. See also the "addr" parameter.
12525
Christopher Faulet8ed0a3e2018-04-10 14:45:45 +020012526proto <name>
12527
12528 Forces the multiplexer's protocol to use for the outgoing connections to this
12529 server. It must be compatible with the mode of the backend (TCP or HTTP). It
12530 must also be usable on the backend side. The list of available protocols is
12531 reported in haproxy -vv.
12532 Idea behind this optipon is to bypass the selection of the best multiplexer's
12533 protocol for all connections established to this server.
12534
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012535redir <prefix>
12536 The "redir" parameter enables the redirection mode for all GET and HEAD
12537 requests addressing this server. This means that instead of having HAProxy
12538 forward the request to the server, it will send an "HTTP 302" response with
12539 the "Location" header composed of this prefix immediately followed by the
12540 requested URI beginning at the leading '/' of the path component. That means
12541 that no trailing slash should be used after <prefix>. All invalid requests
12542 will be rejected, and all non-GET or HEAD requests will be normally served by
12543 the server. Note that since the response is completely forged, no header
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012544 mangling nor cookie insertion is possible in the response. However, cookies in
Davor Ocelice9ed2812017-12-25 17:49:28 +010012545 requests are still analyzed, making this solution completely usable to direct
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012546 users to a remote location in case of local disaster. Main use consists in
12547 increasing bandwidth for static servers by having the clients directly
12548 connect to them. Note: never use a relative location here, it would cause a
12549 loop between the client and HAProxy!
12550
12551 Example : server srv1 192.168.1.1:80 redir http://image1.mydomain.com check
12552
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012553rise <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012554 The "rise" parameter states that a server will be considered as operational
12555 after <count> consecutive successful health checks. This value defaults to 2
12556 if unspecified. See also the "check", "inter" and "fall" parameters.
12557
Baptiste Assmann8e2d9432018-06-22 15:04:43 +020012558resolve-opts <option>,<option>,...
12559 Comma separated list of options to apply to DNS resolution linked to this
12560 server.
12561
12562 Available options:
12563
12564 * allow-dup-ip
12565 By default, HAProxy prevents IP address duplication in a backend when DNS
12566 resolution at runtime is in operation.
12567 That said, for some cases, it makes sense that two servers (in the same
12568 backend, being resolved by the same FQDN) have the same IP address.
12569 For such case, simply enable this option.
12570 This is the opposite of prevent-dup-ip.
12571
Daniel Corbettf8716912019-11-17 09:48:56 -050012572 * ignore-weight
12573 Ignore any weight that is set within an SRV record. This is useful when
12574 you would like to control the weights using an alternate method, such as
12575 using an "agent-check" or through the runtime api.
12576
Baptiste Assmann8e2d9432018-06-22 15:04:43 +020012577 * prevent-dup-ip
12578 Ensure HAProxy's default behavior is enforced on a server: prevent re-using
12579 an IP address already set to a server in the same backend and sharing the
12580 same fqdn.
12581 This is the opposite of allow-dup-ip.
12582
12583 Example:
12584 backend b_myapp
12585 default-server init-addr none resolvers dns
12586 server s1 myapp.example.com:80 check resolve-opts allow-dup-ip
12587 server s2 myapp.example.com:81 check resolve-opts allow-dup-ip
12588
12589 With the option allow-dup-ip set:
12590 * if the nameserver returns a single IP address, then both servers will use
12591 it
12592 * If the nameserver returns 2 IP addresses, then each server will pick up a
12593 different address
12594
12595 Default value: not set
12596
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012597resolve-prefer <family>
12598 When DNS resolution is enabled for a server and multiple IP addresses from
12599 different families are returned, HAProxy will prefer using an IP address
12600 from the family mentioned in the "resolve-prefer" parameter.
12601 Available families: "ipv4" and "ipv6"
12602
Baptiste Assmannc4aabae2015-08-04 22:43:06 +020012603 Default value: ipv6
12604
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012605 Example:
12606
12607 server s1 app1.domain.com:80 resolvers mydns resolve-prefer ipv6
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012608
Thierry Fournierac88cfe2016-02-17 22:05:30 +010012609resolve-net <network>[,<network[,...]]
John Roeslerfb2fce12019-07-10 15:45:51 -050012610 This option prioritizes the choice of an ip address matching a network. This is
Thierry Fournierac88cfe2016-02-17 22:05:30 +010012611 useful with clouds to prefer a local ip. In some cases, a cloud high
Tim Düsterhus4896c442016-11-29 02:15:19 +010012612 availability service can be announced with many ip addresses on many
Davor Ocelice9ed2812017-12-25 17:49:28 +010012613 different datacenters. The latency between datacenter is not negligible, so
12614 this patch permits to prefer a local datacenter. If no address matches the
Thierry Fournierac88cfe2016-02-17 22:05:30 +010012615 configured network, another address is selected.
12616
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012617 Example:
12618
12619 server s1 app1.domain.com:80 resolvers mydns resolve-net 10.0.0.0/8
Thierry Fournierac88cfe2016-02-17 22:05:30 +010012620
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012621resolvers <id>
12622 Points to an existing "resolvers" section to resolve current server's
12623 hostname.
12624
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012625 Example:
12626
12627 server s1 app1.domain.com:80 check resolvers mydns
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012628
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012629 See also section 5.3
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012630
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010012631send-proxy
12632 The "send-proxy" parameter enforces use of the PROXY protocol over any
12633 connection established to this server. The PROXY protocol informs the other
12634 end about the layer 3/4 addresses of the incoming connection, so that it can
12635 know the client's address or the public address it accessed to, whatever the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010012636 upper layer protocol. For connections accepted by an "accept-proxy" or
12637 "accept-netscaler-cip" listener, the advertised address will be used. Only
12638 TCPv4 and TCPv6 address families are supported. Other families such as
12639 Unix sockets, will report an UNKNOWN family. Servers using this option can
12640 fully be chained to another instance of haproxy listening with an
12641 "accept-proxy" setting. This setting must not be used if the server isn't
12642 aware of the protocol. When health checks are sent to the server, the PROXY
12643 protocol is automatically used when this option is set, unless there is an
12644 explicit "port" or "addr" directive, in which case an explicit
12645 "check-send-proxy" directive would also be needed to use the PROXY protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010012646 See also the "no-send-proxy" option of this section and "accept-proxy" and
12647 "accept-netscaler-cip" option of the "bind" keyword.
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010012648
David Safb76832014-05-08 23:42:08 -040012649send-proxy-v2
12650 The "send-proxy-v2" parameter enforces use of the PROXY protocol version 2
12651 over any connection established to this server. The PROXY protocol informs
12652 the other end about the layer 3/4 addresses of the incoming connection, so
12653 that it can know the client's address or the public address it accessed to,
Emmanuel Hocdet404d9782017-10-24 10:55:14 +020012654 whatever the upper layer protocol. It also send ALPN information if an alpn
12655 have been negotiated. This setting must not be used if the server isn't aware
12656 of this version of the protocol. See also the "no-send-proxy-v2" option of
12657 this section and send-proxy" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040012658
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010012659proxy-v2-options <option>[,<option>]*
Tim Duesterhuscf6e0c82020-03-13 12:34:24 +010012660 The "proxy-v2-options" parameter add options to send in PROXY protocol
12661 version 2 when "send-proxy-v2" is used. Options available are:
12662
12663 - ssl : See also "send-proxy-v2-ssl".
12664 - cert-cn : See also "send-proxy-v2-ssl-cn".
12665 - ssl-cipher: Name of the used cipher.
12666 - cert-sig : Signature algorithm of the used certificate.
12667 - cert-key : Key algorithm of the used certificate
12668 - authority : Host name value passed by the client (only SNI from a TLS
12669 connection is supported).
12670 - crc32c : Checksum of the PROXYv2 header.
12671 - unique-id : Send a unique ID generated using the frontend's
12672 "unique-id-format" within the PROXYv2 header.
12673 This unique-id is primarily meant for "mode tcp". It can
12674 lead to unexpected results in "mode http", because the
12675 generated unique ID is also used for the first HTTP request
12676 within a Keep-Alive connection.
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010012677
David Safb76832014-05-08 23:42:08 -040012678send-proxy-v2-ssl
12679 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
12680 2 over any connection established to this server. The PROXY protocol informs
12681 the other end about the layer 3/4 addresses of the incoming connection, so
12682 that it can know the client's address or the public address it accessed to,
12683 whatever the upper layer protocol. In addition, the SSL information extension
12684 of the PROXY protocol is added to the PROXY protocol header. This setting
12685 must not be used if the server isn't aware of this version of the protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010012686 See also the "no-send-proxy-v2-ssl" option of this section and the
12687 "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040012688
12689send-proxy-v2-ssl-cn
12690 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
12691 2 over any connection established to this server. The PROXY protocol informs
12692 the other end about the layer 3/4 addresses of the incoming connection, so
12693 that it can know the client's address or the public address it accessed to,
12694 whatever the upper layer protocol. In addition, the SSL information extension
12695 of the PROXY protocol, along along with the Common Name from the subject of
12696 the client certificate (if any), is added to the PROXY protocol header. This
12697 setting must not be used if the server isn't aware of this version of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012698 protocol. See also the "no-send-proxy-v2-ssl-cn" option of this section and
12699 the "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040012700
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012701slowstart <start_time_in_ms>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012702 The "slowstart" parameter for a server accepts a value in milliseconds which
12703 indicates after how long a server which has just come back up will run at
12704 full speed. Just as with every other time-based parameter, it can be entered
12705 in any other explicit unit among { us, ms, s, m, h, d }. The speed grows
12706 linearly from 0 to 100% during this time. The limitation applies to two
12707 parameters :
12708
12709 - maxconn: the number of connections accepted by the server will grow from 1
12710 to 100% of the usual dynamic limit defined by (minconn,maxconn,fullconn).
12711
12712 - weight: when the backend uses a dynamic weighted algorithm, the weight
12713 grows linearly from 1 to 100%. In this case, the weight is updated at every
12714 health-check. For this reason, it is important that the "inter" parameter
12715 is smaller than the "slowstart", in order to maximize the number of steps.
12716
12717 The slowstart never applies when haproxy starts, otherwise it would cause
12718 trouble to running servers. It only applies when a server has been previously
12719 seen as failed.
12720
Willy Tarreau732eac42015-07-09 11:40:25 +020012721sni <expression>
12722 The "sni" parameter evaluates the sample fetch expression, converts it to a
12723 string and uses the result as the host name sent in the SNI TLS extension to
12724 the server. A typical use case is to send the SNI received from the client in
12725 a bridged HTTPS scenario, using the "ssl_fc_sni" sample fetch for the
Willy Tarreau2ab88672017-07-05 18:23:03 +020012726 expression, though alternatives such as req.hdr(host) can also make sense. If
12727 "verify required" is set (which is the recommended setting), the resulting
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012728 name will also be matched against the server certificate's names. See the
Jérôme Magninb36a6d22018-12-09 16:03:40 +010012729 "verify" directive for more details. If you want to set a SNI for health
12730 checks, see the "check-sni" directive for more details.
Willy Tarreau732eac42015-07-09 11:40:25 +020012731
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020012732source <addr>[:<pl>[-<ph>]] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +020012733source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020012734source <addr>[:<pl>[-<ph>]] [interface <name>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012735 The "source" parameter sets the source address which will be used when
12736 connecting to the server. It follows the exact same parameters and principle
12737 as the backend "source" keyword, except that it only applies to the server
12738 referencing it. Please consult the "source" keyword for details.
12739
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020012740 Additionally, the "source" statement on a server line allows one to specify a
12741 source port range by indicating the lower and higher bounds delimited by a
12742 dash ('-'). Some operating systems might require a valid IP address when a
12743 source port range is specified. It is permitted to have the same IP/range for
12744 several servers. Doing so makes it possible to bypass the maximum of 64k
12745 total concurrent connections. The limit will then reach 64k connections per
12746 server.
12747
Lukas Tribus7d56c6d2016-09-13 09:51:15 +000012748 Since Linux 4.2/libc 2.23 IP_BIND_ADDRESS_NO_PORT is set for connections
12749 specifying the source address without port(s).
12750
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012751ssl
Willy Tarreau44f65392013-06-25 07:56:20 +020012752 This option enables SSL ciphering on outgoing connections to the server. It
12753 is critical to verify server certificates using "verify" when using SSL to
12754 connect to servers, otherwise the communication is prone to trivial man in
12755 the-middle attacks rendering SSL useless. When this option is used, health
12756 checks are automatically sent in SSL too unless there is a "port" or an
12757 "addr" directive indicating the check should be sent to a different location.
Frédéric Lécailled2376272017-03-21 18:52:12 +010012758 See the "no-ssl" to disable "ssl" option and "check-ssl" option to force
12759 SSL health checks.
Willy Tarreau763a95b2012-10-04 23:15:39 +020012760
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012761ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
12762 This option enforces use of <version> or lower when SSL is used to communicate
12763 with the server. This option is also available on global statement
12764 "ssl-default-server-options". See also "ssl-min-ver".
12765
12766ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
12767 This option enforces use of <version> or upper when SSL is used to communicate
12768 with the server. This option is also available on global statement
12769 "ssl-default-server-options". See also "ssl-max-ver".
12770
Frédéric Lécailled2376272017-03-21 18:52:12 +010012771ssl-reuse
12772 This option may be used as "server" setting to reset any "no-ssl-reuse"
12773 setting which would have been inherited from "default-server" directive as
12774 default value.
12775 It may also be used as "default-server" setting to reset any previous
12776 "default-server" "no-ssl-reuse" setting.
12777
12778stick
12779 This option may be used as "server" setting to reset any "non-stick"
12780 setting which would have been inherited from "default-server" directive as
12781 default value.
12782 It may also be used as "default-server" setting to reset any previous
12783 "default-server" "non-stick" setting.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012784
Alexander Liu2a54bb72019-05-22 19:44:48 +080012785socks4 <addr>:<port>
John Roeslerfb2fce12019-07-10 15:45:51 -050012786 This option enables upstream socks4 tunnel for outgoing connections to the
Alexander Liu2a54bb72019-05-22 19:44:48 +080012787 server. Using this option won't force the health check to go via socks4 by
12788 default. You will have to use the keyword "check-via-socks4" to enable it.
12789
Willy Tarreau163d4622015-10-13 16:16:41 +020012790tcp-ut <delay>
12791 Sets the TCP User Timeout for all outgoing connections to this server. This
12792 option is available on Linux since version 2.6.37. It allows haproxy to
12793 configure a timeout for sockets which contain data not receiving an
Davor Ocelice9ed2812017-12-25 17:49:28 +010012794 acknowledgment for the configured delay. This is especially useful on
Willy Tarreau163d4622015-10-13 16:16:41 +020012795 long-lived connections experiencing long idle periods such as remote
12796 terminals or database connection pools, where the client and server timeouts
12797 must remain high to allow a long period of idle, but where it is important to
12798 detect that the server has disappeared in order to release all resources
12799 associated with its connection (and the client's session). One typical use
12800 case is also to force dead server connections to die when health checks are
12801 too slow or during a soft reload since health checks are then disabled. The
12802 argument is a delay expressed in milliseconds by default. This only works for
12803 regular TCP connections, and is ignored for other protocols.
12804
Willy Tarreau034c88c2017-01-23 23:36:45 +010012805tfo
12806 This option enables using TCP fast open when connecting to servers, on
12807 systems that support it (currently only the Linux kernel >= 4.11).
12808 See the "tfo" bind option for more information about TCP fast open.
12809 Please note that when using tfo, you should also use the "conn-failure",
12810 "empty-response" and "response-timeout" keywords for "retry-on", or haproxy
Frédéric Lécaille1b9423d2019-07-04 14:19:06 +020012811 won't be able to retry the connection on failure. See also "no-tfo".
Willy Tarreau034c88c2017-01-23 23:36:45 +010012812
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012813track [<proxy>/]<server>
Willy Tarreau32091232014-05-16 13:52:00 +020012814 This option enables ability to set the current state of the server by tracking
12815 another one. It is possible to track a server which itself tracks another
12816 server, provided that at the end of the chain, a server has health checks
12817 enabled. If <proxy> is omitted the current one is used. If disable-on-404 is
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012818 used, it has to be enabled on both proxies.
12819
Frédéric Lécailled2376272017-03-21 18:52:12 +010012820tls-tickets
12821 This option may be used as "server" setting to reset any "no-tls-tickets"
12822 setting which would have been inherited from "default-server" directive as
12823 default value.
Lukas Tribusbdb386d2020-03-10 00:56:09 +010012824 The TLS ticket mechanism is only used up to TLS 1.2.
12825 Forward Secrecy is compromised with TLS tickets, unless ticket keys
12826 are periodically rotated (via reload or by using "tls-ticket-keys").
Frédéric Lécailled2376272017-03-21 18:52:12 +010012827 It may also be used as "default-server" setting to reset any previous
Bjoern Jacke5ab7eb62020-02-13 14:16:16 +010012828 "default-server" "no-tls-tickets" setting.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012829
Emeric Brunef42d922012-10-11 16:11:36 +020012830verify [none|required]
12831 This setting is only available when support for OpenSSL was built in. If set
Emeric Brun850efd52014-01-29 12:24:34 +010012832 to 'none', server certificate is not verified. In the other case, The
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012833 certificate provided by the server is verified using CAs from 'ca-file' and
12834 optional CRLs from 'crl-file' after having checked that the names provided in
Davor Ocelice9ed2812017-12-25 17:49:28 +010012835 the certificate's subject and subjectAlternateNames attributes match either
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012836 the name passed using the "sni" directive, or if not provided, the static
12837 host name passed using the "verifyhost" directive. When no name is found, the
12838 certificate's names are ignored. For this reason, without SNI it's important
12839 to use "verifyhost". On verification failure the handshake is aborted. It is
12840 critically important to verify server certificates when using SSL to connect
12841 to servers, otherwise the communication is prone to trivial man-in-the-middle
12842 attacks rendering SSL totally useless. Unless "ssl_server_verify" appears in
12843 the global section, "verify" is set to "required" by default.
Emeric Brunef42d922012-10-11 16:11:36 +020012844
Evan Broderbe554312013-06-27 00:05:25 -070012845verifyhost <hostname>
12846 This setting is only available when support for OpenSSL was built in, and
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012847 only takes effect if 'verify required' is also specified. This directive sets
12848 a default static hostname to check the server's certificate against when no
12849 SNI was used to connect to the server. If SNI is not used, this is the only
12850 way to enable hostname verification. This static hostname, when set, will
12851 also be used for health checks (which cannot provide an SNI value). If none
12852 of the hostnames in the certificate match the specified hostname, the
12853 handshake is aborted. The hostnames in the server-provided certificate may
12854 include wildcards. See also "verify", "sni" and "no-verifyhost" options.
Evan Broderbe554312013-06-27 00:05:25 -070012855
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012856weight <weight>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012857 The "weight" parameter is used to adjust the server's weight relative to
12858 other servers. All servers will receive a load proportional to their weight
12859 relative to the sum of all weights, so the higher the weight, the higher the
Willy Tarreau6704d672009-06-15 10:56:05 +020012860 load. The default weight is 1, and the maximal value is 256. A value of 0
12861 means the server will not participate in load-balancing but will still accept
12862 persistent connections. If this parameter is used to distribute the load
12863 according to server's capacity, it is recommended to start with values which
12864 can both grow and shrink, for instance between 10 and 100 to leave enough
12865 room above and below for later adjustments.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012866
12867
Cyril Bonté46175dd2015-07-02 22:45:32 +0200128685.3. Server IP address resolution using DNS
12869-------------------------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012870
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012871HAProxy allows using a host name on the server line to retrieve its IP address
12872using name servers. By default, HAProxy resolves the name when parsing the
12873configuration file, at startup and cache the result for the process' life.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012874This is not sufficient in some cases, such as in Amazon where a server's IP
12875can change after a reboot or an ELB Virtual IP can change based on current
12876workload.
12877This chapter describes how HAProxy can be configured to process server's name
12878resolution at run time.
12879Whether run time server name resolution has been enable or not, HAProxy will
12880carry on doing the first resolution when parsing the configuration.
12881
12882
Cyril Bonté46175dd2015-07-02 22:45:32 +0200128835.3.1. Global overview
12884----------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012885
12886As we've seen in introduction, name resolution in HAProxy occurs at two
12887different steps of the process life:
12888
12889 1. when starting up, HAProxy parses the server line definition and matches a
12890 host name. It uses libc functions to get the host name resolved. This
12891 resolution relies on /etc/resolv.conf file.
12892
Christopher Faulet67957bd2017-09-27 11:00:59 +020012893 2. at run time, HAProxy performs periodically name resolutions for servers
12894 requiring DNS resolutions.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012895
12896A few other events can trigger a name resolution at run time:
12897 - when a server's health check ends up in a connection timeout: this may be
12898 because the server has a new IP address. So we need to trigger a name
12899 resolution to know this new IP.
12900
Christopher Faulet67957bd2017-09-27 11:00:59 +020012901When using resolvers, the server name can either be a hostname, or a SRV label.
Davor Ocelice9ed2812017-12-25 17:49:28 +010012902HAProxy considers anything that starts with an underscore as a SRV label. If a
Christopher Faulet67957bd2017-09-27 11:00:59 +020012903SRV label is specified, then the corresponding SRV records will be retrieved
12904from the DNS server, and the provided hostnames will be used. The SRV label
12905will be checked periodically, and if any server are added or removed, haproxy
12906will automatically do the same.
Olivier Houchardecfa18d2017-08-07 17:30:03 +020012907
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012908A few things important to notice:
John Roeslerfb2fce12019-07-10 15:45:51 -050012909 - all the name servers are queried in the meantime. HAProxy will process the
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012910 first valid response.
12911
12912 - a resolution is considered as invalid (NX, timeout, refused), when all the
12913 servers return an error.
12914
12915
Cyril Bonté46175dd2015-07-02 22:45:32 +0200129165.3.2. The resolvers section
12917----------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012918
12919This section is dedicated to host information related to name resolution in
Christopher Faulet67957bd2017-09-27 11:00:59 +020012920HAProxy. There can be as many as resolvers section as needed. Each section can
12921contain many name servers.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012922
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012923When multiple name servers are configured in a resolvers section, then HAProxy
12924uses the first valid response. In case of invalid responses, only the last one
12925is treated. Purpose is to give the chance to a slow server to deliver a valid
12926answer after a fast faulty or outdated server.
12927
12928When each server returns a different error type, then only the last error is
Christopher Faulet67957bd2017-09-27 11:00:59 +020012929used by HAProxy. The following processing is applied on this error:
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012930
Christopher Faulet67957bd2017-09-27 11:00:59 +020012931 1. HAProxy retries the same DNS query with a new query type. The A queries are
12932 switch to AAAA or the opposite. SRV queries are not concerned here. Timeout
12933 errors are also excluded.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012934
Christopher Faulet67957bd2017-09-27 11:00:59 +020012935 2. When the fallback on the query type was done (or not applicable), HAProxy
12936 retries the original DNS query, with the preferred query type.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012937
Christopher Faulet67957bd2017-09-27 11:00:59 +020012938 3. HAProxy retries previous steps <resolve_retires> times. If no valid
12939 response is received after that, it stops the DNS resolution and reports
12940 the error.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012941
Christopher Faulet67957bd2017-09-27 11:00:59 +020012942For example, with 2 name servers configured in a resolvers section, the
12943following scenarios are possible:
12944
12945 - First response is valid and is applied directly, second response is
12946 ignored
12947
12948 - First response is invalid and second one is valid, then second response is
12949 applied
12950
12951 - First response is a NX domain and second one a truncated response, then
12952 HAProxy retries the query with a new type
12953
12954 - First response is a NX domain and second one is a timeout, then HAProxy
12955 retries the query with a new type
12956
12957 - Query timed out for both name servers, then HAProxy retries it with the
12958 same query type
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012959
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012960As a DNS server may not answer all the IPs in one DNS request, haproxy keeps
12961a cache of previous answers, an answer will be considered obsolete after
Christopher Faulet67957bd2017-09-27 11:00:59 +020012962<hold obsolete> seconds without the IP returned.
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012963
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012964
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012965resolvers <resolvers id>
Davor Ocelice9ed2812017-12-25 17:49:28 +010012966 Creates a new name server list labeled <resolvers id>
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012967
12968A resolvers section accept the following parameters:
12969
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020012970accepted_payload_size <nb>
Davor Ocelice9ed2812017-12-25 17:49:28 +010012971 Defines the maximum payload size accepted by HAProxy and announced to all the
Christopher Faulet67957bd2017-09-27 11:00:59 +020012972 name servers configured in this resolvers section.
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020012973 <nb> is in bytes. If not set, HAProxy announces 512. (minimal value defined
12974 by RFC 6891)
12975
Baptiste Assmann9d8dbbc2017-08-18 23:35:08 +020012976 Note: the maximum allowed value is 8192.
12977
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012978nameserver <id> <ip>:<port>
12979 DNS server description:
12980 <id> : label of the server, should be unique
12981 <ip> : IP address of the server
12982 <port> : port where the DNS service actually runs
12983
Ben Draut44e609b2018-05-29 15:40:08 -060012984parse-resolv-conf
12985 Adds all nameservers found in /etc/resolv.conf to this resolvers nameservers
12986 list. Ordered as if each nameserver in /etc/resolv.conf was individually
12987 placed in the resolvers section in place of this directive.
12988
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012989hold <status> <period>
12990 Defines <period> during which the last name resolution should be kept based
12991 on last resolution <status>
Baptiste Assmann987e16d2016-11-02 22:23:31 +010012992 <status> : last name resolution status. Acceptable values are "nx",
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012993 "other", "refused", "timeout", "valid", "obsolete".
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012994 <period> : interval between two successive name resolution when the last
12995 answer was in <status>. It follows the HAProxy time format.
12996 <period> is in milliseconds by default.
12997
Baptiste Assmann686408b2017-08-18 10:15:42 +020012998 Default value is 10s for "valid", 0s for "obsolete" and 30s for others.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012999
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013000resolve_retries <nb>
13001 Defines the number <nb> of queries to send to resolve a server name before
13002 giving up.
13003 Default value: 3
13004
Baptiste Assmann62b75b42015-09-09 01:11:36 +020013005 A retry occurs on name server timeout or when the full sequence of DNS query
13006 type failover is over and we need to start up from the default ANY query
13007 type.
13008
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013009timeout <event> <time>
13010 Defines timeouts related to name resolution
13011 <event> : the event on which the <time> timeout period applies to.
13012 events available are:
Frédéric Lécaille93d33162019-03-06 09:35:59 +010013013 - resolve : default time to trigger name resolutions when no
13014 other time applied.
Christopher Faulet67957bd2017-09-27 11:00:59 +020013015 Default value: 1s
13016 - retry : time between two DNS queries, when no valid response
Frédéric Lécaille93d33162019-03-06 09:35:59 +010013017 have been received.
Christopher Faulet67957bd2017-09-27 11:00:59 +020013018 Default value: 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013019 <time> : time related to the event. It follows the HAProxy time format.
13020 <time> is expressed in milliseconds.
13021
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020013022 Example:
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013023
13024 resolvers mydns
13025 nameserver dns1 10.0.0.1:53
13026 nameserver dns2 10.0.0.2:53
Ben Draut44e609b2018-05-29 15:40:08 -060013027 parse-resolv-conf
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013028 resolve_retries 3
Christopher Faulet67957bd2017-09-27 11:00:59 +020013029 timeout resolve 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013030 timeout retry 1s
Baptiste Assmann987e16d2016-11-02 22:23:31 +010013031 hold other 30s
13032 hold refused 30s
13033 hold nx 30s
13034 hold timeout 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013035 hold valid 10s
Olivier Houcharda8c6db82017-07-06 18:46:47 +020013036 hold obsolete 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013037
13038
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200130396. Cache
13040---------
13041
13042HAProxy provides a cache, which was designed to perform cache on small objects
13043(favicon, css...). This is a minimalist low-maintenance cache which runs in
13044RAM.
13045
13046The cache is based on a memory which is shared between processes and threads,
13047this memory is split in blocks of 1k.
13048
13049If an object is not used anymore, it can be deleted to store a new object
13050independently of its expiration date. The oldest objects are deleted first
13051when we try to allocate a new one.
13052
13053The cache uses a hash of the host header and the URI as the key.
13054
13055It's possible to view the status of a cache using the Unix socket command
13056"show cache" consult section 9.3 "Unix Socket commands" of Management Guide
13057for more details.
13058
13059When an object is delivered from the cache, the server name in the log is
13060replaced by "<CACHE>".
13061
13062
130636.1. Limitation
13064----------------
13065
13066The cache won't store and won't deliver objects in these cases:
13067
13068- If the response is not a 200
13069- If the response contains a Vary header
13070- If the Content-Length + the headers size is greater than "max-object-size"
13071- If the response is not cacheable
13072
13073- If the request is not a GET
13074- If the HTTP version of the request is smaller than 1.1
13075- If the request contains an Authorization header
13076
13077
130786.2. Setup
13079-----------
13080
13081To setup a cache, you must define a cache section and use it in a proxy with
13082the corresponding http-request and response actions.
13083
13084
130856.2.1. Cache section
13086---------------------
13087
13088cache <name>
13089 Declare a cache section, allocate a shared cache memory named <name>, the
13090 size of cache is mandatory.
13091
13092total-max-size <megabytes>
13093 Define the size in RAM of the cache in megabytes. This size is split in
13094 blocks of 1kB which are used by the cache entries. Its maximum value is 4095.
13095
13096max-object-size <bytes>
13097 Define the maximum size of the objects to be cached. Must not be greater than
13098 an half of "total-max-size". If not set, it equals to a 256th of the cache size.
13099 All objects with sizes larger than "max-object-size" will not be cached.
13100
13101max-age <seconds>
13102 Define the maximum expiration duration. The expiration is set has the lowest
13103 value between the s-maxage or max-age (in this order) directive in the
13104 Cache-Control response header and this value. The default value is 60
13105 seconds, which means that you can't cache an object more than 60 seconds by
13106 default.
13107
13108
131096.2.2. Proxy section
13110---------------------
13111
13112http-request cache-use <name> [ { if | unless } <condition> ]
13113 Try to deliver a cached object from the cache <name>. This directive is also
13114 mandatory to store the cache as it calculates the cache hash. If you want to
13115 use a condition for both storage and delivering that's a good idea to put it
13116 after this one.
13117
13118http-response cache-store <name> [ { if | unless } <condition> ]
13119 Store an http-response within the cache. The storage of the response headers
13120 is done at this step, which means you can use others http-response actions
13121 to modify headers before or after the storage of the response. This action
13122 is responsible for the setup of the cache storage filter.
13123
13124
13125Example:
13126
13127 backend bck1
13128 mode http
13129
13130 http-request cache-use foobar
13131 http-response cache-store foobar
13132 server srv1 127.0.0.1:80
13133
13134 cache foobar
13135 total-max-size 4
13136 max-age 240
13137
13138
Willy Tarreau74ca5042013-06-11 23:12:07 +0200131397. Using ACLs and fetching samples
13140----------------------------------
13141
Davor Ocelice9ed2812017-12-25 17:49:28 +010013142HAProxy is capable of extracting data from request or response streams, from
Willy Tarreau74ca5042013-06-11 23:12:07 +020013143client or server information, from tables, environmental information etc...
13144The action of extracting such data is called fetching a sample. Once retrieved,
13145these samples may be used for various purposes such as a key to a stick-table,
13146but most common usages consist in matching them against predefined constant
13147data called patterns.
13148
13149
131507.1. ACL basics
13151---------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013152
13153The use of Access Control Lists (ACL) provides a flexible solution to perform
13154content switching and generally to take decisions based on content extracted
13155from the request, the response or any environmental status. The principle is
13156simple :
13157
Willy Tarreau74ca5042013-06-11 23:12:07 +020013158 - extract a data sample from a stream, table or the environment
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013159 - optionally apply some format conversion to the extracted sample
Willy Tarreau74ca5042013-06-11 23:12:07 +020013160 - apply one or multiple pattern matching methods on this sample
13161 - perform actions only when a pattern matches the sample
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013162
Willy Tarreau74ca5042013-06-11 23:12:07 +020013163The actions generally consist in blocking a request, selecting a backend, or
13164adding a header.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013165
13166In order to define a test, the "acl" keyword is used. The syntax is :
13167
Willy Tarreau74ca5042013-06-11 23:12:07 +020013168 acl <aclname> <criterion> [flags] [operator] [<value>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013169
13170This creates a new ACL <aclname> or completes an existing one with new tests.
13171Those tests apply to the portion of request/response specified in <criterion>
13172and may be adjusted with optional flags [flags]. Some criteria also support
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013173an operator which may be specified before the set of values. Optionally some
13174conversion operators may be applied to the sample, and they will be specified
13175as a comma-delimited list of keywords just after the first keyword. The values
13176are of the type supported by the criterion, and are separated by spaces.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013177
13178ACL names must be formed from upper and lower case letters, digits, '-' (dash),
13179'_' (underscore) , '.' (dot) and ':' (colon). ACL names are case-sensitive,
13180which means that "my_acl" and "My_Acl" are two different ACLs.
13181
13182There is no enforced limit to the number of ACLs. The unused ones do not affect
13183performance, they just consume a small amount of memory.
13184
Willy Tarreau74ca5042013-06-11 23:12:07 +020013185The criterion generally is the name of a sample fetch method, or one of its ACL
13186specific declinations. The default test method is implied by the output type of
13187this sample fetch method. The ACL declinations can describe alternate matching
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013188methods of a same sample fetch method. The sample fetch methods are the only
13189ones supporting a conversion.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013190
13191Sample fetch methods return data which can be of the following types :
13192 - boolean
13193 - integer (signed or unsigned)
13194 - IPv4 or IPv6 address
13195 - string
13196 - data block
13197
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013198Converters transform any of these data into any of these. For example, some
13199converters might convert a string to a lower-case string while other ones
13200would turn a string to an IPv4 address, or apply a netmask to an IP address.
13201The resulting sample is of the type of the last converter applied to the list,
13202which defaults to the type of the sample fetch method.
13203
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020013204Each sample or converter returns data of a specific type, specified with its
13205keyword in this documentation. When an ACL is declared using a standard sample
13206fetch method, certain types automatically involved a default matching method
13207which are summarized in the table below :
13208
13209 +---------------------+-----------------+
13210 | Sample or converter | Default |
13211 | output type | matching method |
13212 +---------------------+-----------------+
13213 | boolean | bool |
13214 +---------------------+-----------------+
13215 | integer | int |
13216 +---------------------+-----------------+
13217 | ip | ip |
13218 +---------------------+-----------------+
13219 | string | str |
13220 +---------------------+-----------------+
13221 | binary | none, use "-m" |
13222 +---------------------+-----------------+
13223
13224Note that in order to match a binary samples, it is mandatory to specify a
13225matching method, see below.
13226
Willy Tarreau74ca5042013-06-11 23:12:07 +020013227The ACL engine can match these types against patterns of the following types :
13228 - boolean
13229 - integer or integer range
13230 - IP address / network
13231 - string (exact, substring, suffix, prefix, subdir, domain)
13232 - regular expression
13233 - hex block
13234
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013235The following ACL flags are currently supported :
13236
Willy Tarreau2b5285d2010-05-09 23:45:24 +020013237 -i : ignore case during matching of all subsequent patterns.
13238 -f : load patterns from a file.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013239 -m : use a specific pattern matching method
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010013240 -n : forbid the DNS resolutions
Thierry FOURNIER9860c412014-01-29 14:23:29 +010013241 -M : load the file pointed by -f like a map file.
Thierry FOURNIER3534d882014-01-20 17:01:44 +010013242 -u : force the unique id of the ACL
Willy Tarreau6a06a402007-07-15 20:15:28 +020013243 -- : force end of flags. Useful when a string looks like one of the flags.
13244
Willy Tarreau74ca5042013-06-11 23:12:07 +020013245The "-f" flag is followed by the name of a file from which all lines will be
13246read as individual values. It is even possible to pass multiple "-f" arguments
13247if the patterns are to be loaded from multiple files. Empty lines as well as
13248lines beginning with a sharp ('#') will be ignored. All leading spaces and tabs
13249will be stripped. If it is absolutely necessary to insert a valid pattern
13250beginning with a sharp, just prefix it with a space so that it is not taken for
13251a comment. Depending on the data type and match method, haproxy may load the
13252lines into a binary tree, allowing very fast lookups. This is true for IPv4 and
13253exact string matching. In this case, duplicates will automatically be removed.
13254
Thierry FOURNIER9860c412014-01-29 14:23:29 +010013255The "-M" flag allows an ACL to use a map file. If this flag is set, the file is
13256parsed as two column file. The first column contains the patterns used by the
13257ACL, and the second column contain the samples. The sample can be used later by
13258a map. This can be useful in some rare cases where an ACL would just be used to
13259check for the existence of a pattern in a map before a mapping is applied.
13260
Thierry FOURNIER3534d882014-01-20 17:01:44 +010013261The "-u" flag forces the unique id of the ACL. This unique id is used with the
13262socket interface to identify ACL and dynamically change its values. Note that a
13263file is always identified by its name even if an id is set.
13264
Willy Tarreau74ca5042013-06-11 23:12:07 +020013265Also, note that the "-i" flag applies to subsequent entries and not to entries
13266loaded from files preceding it. For instance :
13267
13268 acl valid-ua hdr(user-agent) -f exact-ua.lst -i -f generic-ua.lst test
13269
13270In this example, each line of "exact-ua.lst" will be exactly matched against
13271the "user-agent" header of the request. Then each line of "generic-ua" will be
13272case-insensitively matched. Then the word "test" will be insensitively matched
13273as well.
13274
13275The "-m" flag is used to select a specific pattern matching method on the input
13276sample. All ACL-specific criteria imply a pattern matching method and generally
13277do not need this flag. However, this flag is useful with generic sample fetch
13278methods to describe how they're going to be matched against the patterns. This
13279is required for sample fetches which return data type for which there is no
Davor Ocelice9ed2812017-12-25 17:49:28 +010013280obvious matching method (e.g. string or binary). When "-m" is specified and
Willy Tarreau74ca5042013-06-11 23:12:07 +020013281followed by a pattern matching method name, this method is used instead of the
13282default one for the criterion. This makes it possible to match contents in ways
13283that were not initially planned, or with sample fetch methods which return a
13284string. The matching method also affects the way the patterns are parsed.
13285
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010013286The "-n" flag forbids the dns resolutions. It is used with the load of ip files.
13287By default, if the parser cannot parse ip address it considers that the parsed
13288string is maybe a domain name and try dns resolution. The flag "-n" disable this
13289resolution. It is useful for detecting malformed ip lists. Note that if the DNS
13290server is not reachable, the haproxy configuration parsing may last many minutes
John Roeslerfb2fce12019-07-10 15:45:51 -050013291waiting for the timeout. During this time no error messages are displayed. The
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010013292flag "-n" disable this behavior. Note also that during the runtime, this
13293function is disabled for the dynamic acl modifications.
13294
Willy Tarreau74ca5042013-06-11 23:12:07 +020013295There are some restrictions however. Not all methods can be used with all
13296sample fetch methods. Also, if "-m" is used in conjunction with "-f", it must
13297be placed first. The pattern matching method must be one of the following :
Willy Tarreau5adeda12013-03-31 22:13:34 +020013298
13299 - "found" : only check if the requested sample could be found in the stream,
13300 but do not compare it against any pattern. It is recommended not
Willy Tarreau74ca5042013-06-11 23:12:07 +020013301 to pass any pattern to avoid confusion. This matching method is
13302 particularly useful to detect presence of certain contents such
13303 as headers, cookies, etc... even if they are empty and without
13304 comparing them to anything nor counting them.
Willy Tarreau5adeda12013-03-31 22:13:34 +020013305
13306 - "bool" : check the value as a boolean. It can only be applied to fetches
13307 which return a boolean or integer value, and takes no pattern.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013308 Value zero or false does not match, all other values do match.
Willy Tarreau5adeda12013-03-31 22:13:34 +020013309
13310 - "int" : match the value as an integer. It can be used with integer and
Willy Tarreau74ca5042013-06-11 23:12:07 +020013311 boolean samples. Boolean false is integer 0, true is integer 1.
Willy Tarreau5adeda12013-03-31 22:13:34 +020013312
13313 - "ip" : match the value as an IPv4 or IPv6 address. It is compatible
Willy Tarreau74ca5042013-06-11 23:12:07 +020013314 with IP address samples only, so it is implied and never needed.
Willy Tarreau5adeda12013-03-31 22:13:34 +020013315
Davor Ocelice9ed2812017-12-25 17:49:28 +010013316 - "bin" : match the contents against a hexadecimal string representing a
Willy Tarreau5adeda12013-03-31 22:13:34 +020013317 binary sequence. This may be used with binary or string samples.
13318
13319 - "len" : match the sample's length as an integer. This may be used with
13320 binary or string samples.
13321
Willy Tarreau74ca5042013-06-11 23:12:07 +020013322 - "str" : exact match : match the contents against a string. This may be
13323 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020013324
Willy Tarreau74ca5042013-06-11 23:12:07 +020013325 - "sub" : substring match : check that the contents contain at least one of
13326 the provided string patterns. This may be used with binary or
13327 string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020013328
Willy Tarreau74ca5042013-06-11 23:12:07 +020013329 - "reg" : regex match : match the contents against a list of regular
13330 expressions. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020013331
Willy Tarreau74ca5042013-06-11 23:12:07 +020013332 - "beg" : prefix match : check that the contents begin like the provided
13333 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020013334
Willy Tarreau74ca5042013-06-11 23:12:07 +020013335 - "end" : suffix match : check that the contents end like the provided
13336 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020013337
Willy Tarreau74ca5042013-06-11 23:12:07 +020013338 - "dir" : subdir match : check that a slash-delimited portion of the
13339 contents exactly matches one of the provided string patterns.
Willy Tarreau5adeda12013-03-31 22:13:34 +020013340 This may be used with binary or string samples.
13341
Willy Tarreau74ca5042013-06-11 23:12:07 +020013342 - "dom" : domain match : check that a dot-delimited portion of the contents
13343 exactly match one of the provided string patterns. This may be
13344 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020013345
13346For example, to quickly detect the presence of cookie "JSESSIONID" in an HTTP
13347request, it is possible to do :
13348
13349 acl jsess_present cook(JSESSIONID) -m found
13350
13351In order to apply a regular expression on the 500 first bytes of data in the
13352buffer, one would use the following acl :
13353
13354 acl script_tag payload(0,500) -m reg -i <script>
13355
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013356On systems where the regex library is much slower when using "-i", it is
13357possible to convert the sample to lowercase before matching, like this :
13358
13359 acl script_tag payload(0,500),lower -m reg <script>
13360
Willy Tarreau74ca5042013-06-11 23:12:07 +020013361All ACL-specific criteria imply a default matching method. Most often, these
13362criteria are composed by concatenating the name of the original sample fetch
13363method and the matching method. For example, "hdr_beg" applies the "beg" match
13364to samples retrieved using the "hdr" fetch method. Since all ACL-specific
13365criteria rely on a sample fetch method, it is always possible instead to use
13366the original sample fetch method and the explicit matching method using "-m".
Willy Tarreau2b5285d2010-05-09 23:45:24 +020013367
Willy Tarreau74ca5042013-06-11 23:12:07 +020013368If an alternate match is specified using "-m" on an ACL-specific criterion,
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013369the matching method is simply applied to the underlying sample fetch method.
13370For example, all ACLs below are exact equivalent :
Willy Tarreau2b5285d2010-05-09 23:45:24 +020013371
Willy Tarreau74ca5042013-06-11 23:12:07 +020013372 acl short_form hdr_beg(host) www.
13373 acl alternate1 hdr_beg(host) -m beg www.
13374 acl alternate2 hdr_dom(host) -m beg www.
13375 acl alternate3 hdr(host) -m beg www.
Willy Tarreau2b5285d2010-05-09 23:45:24 +020013376
Willy Tarreau2b5285d2010-05-09 23:45:24 +020013377
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020013378The table below summarizes the compatibility matrix between sample or converter
13379types and the pattern types to fetch against. It indicates for each compatible
13380combination the name of the matching method to be used, surrounded with angle
13381brackets ">" and "<" when the method is the default one and will work by
13382default without "-m".
Willy Tarreau0ba27502007-12-24 16:55:16 +010013383
Willy Tarreau74ca5042013-06-11 23:12:07 +020013384 +-------------------------------------------------+
13385 | Input sample type |
13386 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020013387 | pattern type | boolean | integer | ip | string | binary |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013388 +----------------------+---------+---------+---------+---------+---------+
13389 | none (presence only) | found | found | found | found | found |
13390 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020013391 | none (boolean value) |> bool <| bool | | bool | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013392 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020013393 | integer (value) | int |> int <| int | int | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013394 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010013395 | integer (length) | len | len | len | len | len |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013396 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020013397 | IP address | | |> ip <| ip | ip |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013398 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020013399 | exact string | str | str | str |> str <| str |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013400 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010013401 | prefix | beg | beg | beg | beg | beg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013402 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010013403 | suffix | end | end | end | end | end |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013404 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010013405 | substring | sub | sub | sub | sub | sub |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013406 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010013407 | subdir | dir | dir | dir | dir | dir |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013408 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010013409 | domain | dom | dom | dom | dom | dom |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013410 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010013411 | regex | reg | reg | reg | reg | reg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013412 +----------------------+---------+---------+---------+---------+---------+
13413 | hex block | | | | bin | bin |
13414 +----------------------+---------+---------+---------+---------+---------+
Willy Tarreau6a06a402007-07-15 20:15:28 +020013415
13416
Willy Tarreau74ca5042013-06-11 23:12:07 +0200134177.1.1. Matching booleans
13418------------------------
13419
13420In order to match a boolean, no value is needed and all values are ignored.
13421Boolean matching is used by default for all fetch methods of type "boolean".
13422When boolean matching is used, the fetched value is returned as-is, which means
13423that a boolean "true" will always match and a boolean "false" will never match.
13424
13425Boolean matching may also be enforced using "-m bool" on fetch methods which
13426return an integer value. Then, integer value 0 is converted to the boolean
13427"false" and all other values are converted to "true".
13428
Willy Tarreau6a06a402007-07-15 20:15:28 +020013429
Willy Tarreau74ca5042013-06-11 23:12:07 +0200134307.1.2. Matching integers
13431------------------------
13432
13433Integer matching applies by default to integer fetch methods. It can also be
13434enforced on boolean fetches using "-m int". In this case, "false" is converted
13435to the integer 0, and "true" is converted to the integer 1.
13436
13437Integer matching also supports integer ranges and operators. Note that integer
13438matching only applies to positive values. A range is a value expressed with a
13439lower and an upper bound separated with a colon, both of which may be omitted.
Willy Tarreau6a06a402007-07-15 20:15:28 +020013440
13441For instance, "1024:65535" is a valid range to represent a range of
13442unprivileged ports, and "1024:" would also work. "0:1023" is a valid
13443representation of privileged ports, and ":1023" would also work.
13444
Willy Tarreau62644772008-07-16 18:36:06 +020013445As a special case, some ACL functions support decimal numbers which are in fact
13446two integers separated by a dot. This is used with some version checks for
13447instance. All integer properties apply to those decimal numbers, including
13448ranges and operators.
13449
Willy Tarreau6a06a402007-07-15 20:15:28 +020013450For an easier usage, comparison operators are also supported. Note that using
Willy Tarreau0ba27502007-12-24 16:55:16 +010013451operators with ranges does not make much sense and is strongly discouraged.
13452Similarly, it does not make much sense to perform order comparisons with a set
13453of values.
Willy Tarreau6a06a402007-07-15 20:15:28 +020013454
Willy Tarreau0ba27502007-12-24 16:55:16 +010013455Available operators for integer matching are :
Willy Tarreau6a06a402007-07-15 20:15:28 +020013456
13457 eq : true if the tested value equals at least one value
13458 ge : true if the tested value is greater than or equal to at least one value
13459 gt : true if the tested value is greater than at least one value
13460 le : true if the tested value is less than or equal to at least one value
13461 lt : true if the tested value is less than at least one value
13462
Willy Tarreau0ba27502007-12-24 16:55:16 +010013463For instance, the following ACL matches any negative Content-Length header :
Willy Tarreau6a06a402007-07-15 20:15:28 +020013464
13465 acl negative-length hdr_val(content-length) lt 0
13466
Willy Tarreau62644772008-07-16 18:36:06 +020013467This one matches SSL versions between 3.0 and 3.1 (inclusive) :
13468
13469 acl sslv3 req_ssl_ver 3:3.1
13470
Willy Tarreau6a06a402007-07-15 20:15:28 +020013471
Willy Tarreau74ca5042013-06-11 23:12:07 +0200134727.1.3. Matching strings
13473-----------------------
13474
13475String matching applies to string or binary fetch methods, and exists in 6
13476different forms :
13477
13478 - exact match (-m str) : the extracted string must exactly match the
Davor Ocelice9ed2812017-12-25 17:49:28 +010013479 patterns;
Willy Tarreau74ca5042013-06-11 23:12:07 +020013480
13481 - substring match (-m sub) : the patterns are looked up inside the
Davor Ocelice9ed2812017-12-25 17:49:28 +010013482 extracted string, and the ACL matches if any of them is found inside;
Willy Tarreau74ca5042013-06-11 23:12:07 +020013483
13484 - prefix match (-m beg) : the patterns are compared with the beginning of
13485 the extracted string, and the ACL matches if any of them matches.
13486
13487 - suffix match (-m end) : the patterns are compared with the end of the
13488 extracted string, and the ACL matches if any of them matches.
13489
Baptiste Assmann33db6002016-03-06 23:32:10 +010013490 - subdir match (-m dir) : the patterns are looked up inside the extracted
Willy Tarreau74ca5042013-06-11 23:12:07 +020013491 string, delimited with slashes ("/"), and the ACL matches if any of them
13492 matches.
13493
13494 - domain match (-m dom) : the patterns are looked up inside the extracted
13495 string, delimited with dots ("."), and the ACL matches if any of them
13496 matches.
Willy Tarreau6a06a402007-07-15 20:15:28 +020013497
13498String matching applies to verbatim strings as they are passed, with the
13499exception of the backslash ("\") which makes it possible to escape some
13500characters such as the space. If the "-i" flag is passed before the first
13501string, then the matching will be performed ignoring the case. In order
13502to match the string "-i", either set it second, or pass the "--" flag
Willy Tarreau0ba27502007-12-24 16:55:16 +010013503before the first string. Same applies of course to match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020013504
Mathias Weiersmuellercb250fc2019-12-02 09:43:40 +010013505Do not use string matches for binary fetches which might contain null bytes
13506(0x00), as the comparison stops at the occurrence of the first null byte.
13507Instead, convert the binary fetch to a hex string with the hex converter first.
13508
13509Example:
13510 # matches if the string <tag> is present in the binary sample
13511 acl tag_found req.payload(0,0),hex -m sub 3C7461673E
13512
Willy Tarreau6a06a402007-07-15 20:15:28 +020013513
Willy Tarreau74ca5042013-06-11 23:12:07 +0200135147.1.4. Matching regular expressions (regexes)
13515---------------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020013516
13517Just like with string matching, regex matching applies to verbatim strings as
13518they are passed, with the exception of the backslash ("\") which makes it
13519possible to escape some characters such as the space. If the "-i" flag is
13520passed before the first regex, then the matching will be performed ignoring
13521the case. In order to match the string "-i", either set it second, or pass
Willy Tarreau0ba27502007-12-24 16:55:16 +010013522the "--" flag before the first string. Same principle applies of course to
13523match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020013524
13525
Willy Tarreau74ca5042013-06-11 23:12:07 +0200135267.1.5. Matching arbitrary data blocks
13527-------------------------------------
13528
13529It is possible to match some extracted samples against a binary block which may
13530not safely be represented as a string. For this, the patterns must be passed as
13531a series of hexadecimal digits in an even number, when the match method is set
13532to binary. Each sequence of two digits will represent a byte. The hexadecimal
13533digits may be used upper or lower case.
13534
13535Example :
13536 # match "Hello\n" in the input stream (\x48 \x65 \x6c \x6c \x6f \x0a)
13537 acl hello payload(0,6) -m bin 48656c6c6f0a
13538
13539
135407.1.6. Matching IPv4 and IPv6 addresses
13541---------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020013542
13543IPv4 addresses values can be specified either as plain addresses or with a
13544netmask appended, in which case the IPv4 address matches whenever it is
13545within the network. Plain addresses may also be replaced with a resolvable
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010013546host name, but this practice is generally discouraged as it makes it more
Willy Tarreau0ba27502007-12-24 16:55:16 +010013547difficult to read and debug configurations. If hostnames are used, you should
13548at least ensure that they are present in /etc/hosts so that the configuration
13549does not depend on any random DNS match at the moment the configuration is
13550parsed.
Willy Tarreau6a06a402007-07-15 20:15:28 +020013551
Daniel Schnellereba56342016-04-13 00:26:52 +020013552The dotted IPv4 address notation is supported in both regular as well as the
13553abbreviated form with all-0-octets omitted:
13554
13555 +------------------+------------------+------------------+
13556 | Example 1 | Example 2 | Example 3 |
13557 +------------------+------------------+------------------+
13558 | 192.168.0.1 | 10.0.0.12 | 127.0.0.1 |
13559 | 192.168.1 | 10.12 | 127.1 |
13560 | 192.168.0.1/22 | 10.0.0.12/8 | 127.0.0.1/8 |
13561 | 192.168.1/22 | 10.12/8 | 127.1/8 |
13562 +------------------+------------------+------------------+
13563
13564Notice that this is different from RFC 4632 CIDR address notation in which
13565192.168.42/24 would be equivalent to 192.168.42.0/24.
13566
Willy Tarreauceb4ac92012-04-28 00:41:46 +020013567IPv6 may be entered in their usual form, with or without a netmask appended.
13568Only bit counts are accepted for IPv6 netmasks. In order to avoid any risk of
13569trouble with randomly resolved IP addresses, host names are never allowed in
13570IPv6 patterns.
13571
13572HAProxy is also able to match IPv4 addresses with IPv6 addresses in the
13573following situations :
13574 - tested address is IPv4, pattern address is IPv4, the match applies
13575 in IPv4 using the supplied mask if any.
13576 - tested address is IPv6, pattern address is IPv6, the match applies
13577 in IPv6 using the supplied mask if any.
13578 - tested address is IPv6, pattern address is IPv4, the match applies in IPv4
13579 using the pattern's mask if the IPv6 address matches with 2002:IPV4::,
13580 ::IPV4 or ::ffff:IPV4, otherwise it fails.
13581 - tested address is IPv4, pattern address is IPv6, the IPv4 address is first
13582 converted to IPv6 by prefixing ::ffff: in front of it, then the match is
13583 applied in IPv6 using the supplied IPv6 mask.
13584
Willy Tarreau74ca5042013-06-11 23:12:07 +020013585
135867.2. Using ACLs to form conditions
13587----------------------------------
13588
13589Some actions are only performed upon a valid condition. A condition is a
13590combination of ACLs with operators. 3 operators are supported :
13591
13592 - AND (implicit)
13593 - OR (explicit with the "or" keyword or the "||" operator)
13594 - Negation with the exclamation mark ("!")
Willy Tarreau6a06a402007-07-15 20:15:28 +020013595
Willy Tarreau74ca5042013-06-11 23:12:07 +020013596A condition is formed as a disjunctive form:
Willy Tarreau6a06a402007-07-15 20:15:28 +020013597
Willy Tarreau74ca5042013-06-11 23:12:07 +020013598 [!]acl1 [!]acl2 ... [!]acln { or [!]acl1 [!]acl2 ... [!]acln } ...
Willy Tarreaubef91e72013-03-31 23:14:46 +020013599
Willy Tarreau74ca5042013-06-11 23:12:07 +020013600Such conditions are generally used after an "if" or "unless" statement,
13601indicating when the condition will trigger the action.
Willy Tarreaubef91e72013-03-31 23:14:46 +020013602
Willy Tarreau74ca5042013-06-11 23:12:07 +020013603For instance, to block HTTP requests to the "*" URL with methods other than
13604"OPTIONS", as well as POST requests without content-length, and GET or HEAD
13605requests with a content-length greater than 0, and finally every request which
13606is not either GET/HEAD/POST/OPTIONS !
13607
13608 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013609 http-request deny if HTTP_URL_STAR !METH_OPTIONS || METH_POST missing_cl
13610 http-request deny if METH_GET HTTP_CONTENT
13611 http-request deny unless METH_GET or METH_POST or METH_OPTIONS
Willy Tarreau74ca5042013-06-11 23:12:07 +020013612
13613To select a different backend for requests to static contents on the "www" site
13614and to every request on the "img", "video", "download" and "ftp" hosts :
13615
13616 acl url_static path_beg /static /images /img /css
13617 acl url_static path_end .gif .png .jpg .css .js
13618 acl host_www hdr_beg(host) -i www
13619 acl host_static hdr_beg(host) -i img. video. download. ftp.
13620
Davor Ocelice9ed2812017-12-25 17:49:28 +010013621 # now use backend "static" for all static-only hosts, and for static URLs
Willy Tarreau74ca5042013-06-11 23:12:07 +020013622 # of host "www". Use backend "www" for the rest.
13623 use_backend static if host_static or host_www url_static
13624 use_backend www if host_www
13625
13626It is also possible to form rules using "anonymous ACLs". Those are unnamed ACL
13627expressions that are built on the fly without needing to be declared. They must
13628be enclosed between braces, with a space before and after each brace (because
13629the braces must be seen as independent words). Example :
13630
13631 The following rule :
13632
13633 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013634 http-request deny if METH_POST missing_cl
Willy Tarreau74ca5042013-06-11 23:12:07 +020013635
13636 Can also be written that way :
13637
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013638 http-request deny if METH_POST { hdr_cnt(Content-length) eq 0 }
Willy Tarreau74ca5042013-06-11 23:12:07 +020013639
13640It is generally not recommended to use this construct because it's a lot easier
13641to leave errors in the configuration when written that way. However, for very
13642simple rules matching only one source IP address for instance, it can make more
13643sense to use them than to declare ACLs with random names. Another example of
13644good use is the following :
13645
13646 With named ACLs :
13647
13648 acl site_dead nbsrv(dynamic) lt 2
13649 acl site_dead nbsrv(static) lt 2
13650 monitor fail if site_dead
13651
13652 With anonymous ACLs :
13653
13654 monitor fail if { nbsrv(dynamic) lt 2 } || { nbsrv(static) lt 2 }
13655
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013656See section 4.2 for detailed help on the "http-request deny" and "use_backend"
13657keywords.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013658
13659
136607.3. Fetching samples
13661---------------------
13662
13663Historically, sample fetch methods were only used to retrieve data to match
13664against patterns using ACLs. With the arrival of stick-tables, a new class of
13665sample fetch methods was created, most often sharing the same syntax as their
13666ACL counterpart. These sample fetch methods are also known as "fetches". As
13667of now, ACLs and fetches have converged. All ACL fetch methods have been made
13668available as fetch methods, and ACLs may use any sample fetch method as well.
13669
13670This section details all available sample fetch methods and their output type.
13671Some sample fetch methods have deprecated aliases that are used to maintain
13672compatibility with existing configurations. They are then explicitly marked as
13673deprecated and should not be used in new setups.
13674
13675The ACL derivatives are also indicated when available, with their respective
13676matching methods. These ones all have a well defined default pattern matching
13677method, so it is never necessary (though allowed) to pass the "-m" option to
13678indicate how the sample will be matched using ACLs.
13679
13680As indicated in the sample type versus matching compatibility matrix above,
13681when using a generic sample fetch method in an ACL, the "-m" option is
13682mandatory unless the sample type is one of boolean, integer, IPv4 or IPv6. When
13683the same keyword exists as an ACL keyword and as a standard fetch method, the
13684ACL engine will automatically pick the ACL-only one by default.
13685
13686Some of these keywords support one or multiple mandatory arguments, and one or
13687multiple optional arguments. These arguments are strongly typed and are checked
13688when the configuration is parsed so that there is no risk of running with an
Davor Ocelice9ed2812017-12-25 17:49:28 +010013689incorrect argument (e.g. an unresolved backend name). Fetch function arguments
13690are passed between parenthesis and are delimited by commas. When an argument
Willy Tarreau74ca5042013-06-11 23:12:07 +020013691is optional, it will be indicated below between square brackets ('[ ]'). When
13692all arguments are optional, the parenthesis may be omitted.
13693
13694Thus, the syntax of a standard sample fetch method is one of the following :
13695 - name
13696 - name(arg1)
13697 - name(arg1,arg2)
13698
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013699
137007.3.1. Converters
13701-----------------
13702
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013703Sample fetch methods may be combined with transformations to be applied on top
13704of the fetched sample (also called "converters"). These combinations form what
13705is called "sample expressions" and the result is a "sample". Initially this
13706was only supported by "stick on" and "stick store-request" directives but this
Davor Ocelice9ed2812017-12-25 17:49:28 +010013707has now be extended to all places where samples may be used (ACLs, log-format,
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013708unique-id-format, add-header, ...).
13709
13710These transformations are enumerated as a series of specific keywords after the
13711sample fetch method. These keywords may equally be appended immediately after
13712the fetch keyword's argument, delimited by a comma. These keywords can also
Davor Ocelice9ed2812017-12-25 17:49:28 +010013713support some arguments (e.g. a netmask) which must be passed in parenthesis.
Willy Tarreau0ba27502007-12-24 16:55:16 +010013714
Willy Tarreau97707872015-01-27 15:12:13 +010013715A certain category of converters are bitwise and arithmetic operators which
13716support performing basic operations on integers. Some bitwise operations are
13717supported (and, or, xor, cpl) and some arithmetic operations are supported
13718(add, sub, mul, div, mod, neg). Some comparators are provided (odd, even, not,
13719bool) which make it possible to report a match without having to write an ACL.
13720
Willy Tarreau74ca5042013-06-11 23:12:07 +020013721The currently available list of transformation keywords include :
Willy Tarreau0ba27502007-12-24 16:55:16 +010013722
Ben Shillitof25e8e52016-12-02 14:25:37 +00001372351d.single(<prop>[,<prop>*])
13724 Returns values for the properties requested as a string, where values are
13725 separated by the delimiter specified with "51degrees-property-separator".
13726 The device is identified using the User-Agent header passed to the
13727 converter. The function can be passed up to five property names, and if a
13728 property name can't be found, the value "NoData" is returned.
13729
13730 Example :
Davor Ocelice9ed2812017-12-25 17:49:28 +010013731 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request,
13732 # containing values for the three properties requested by using the
Ben Shillitof25e8e52016-12-02 14:25:37 +000013733 # User-Agent passed to the converter.
13734 frontend http-in
13735 bind *:8081
13736 default_backend servers
13737 http-request set-header X-51D-DeviceTypeMobileTablet \
13738 %[req.fhdr(User-Agent),51d.single(DeviceType,IsMobile,IsTablet)]
13739
Willy Tarreau97707872015-01-27 15:12:13 +010013740add(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013741 Adds <value> to the input value of type signed integer, and returns the
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013742 result as a signed integer. <value> can be a numeric value or a variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013743 name. The name of the variable starts with an indication about its scope. The
13744 scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013745 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013746 "sess" : the variable is shared with the whole session
13747 "txn" : the variable is shared with the transaction (request and response)
13748 "req" : the variable is shared only during request processing
13749 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013750 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013751 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013752
Nenad Merdanovicc31499d2019-03-23 11:00:32 +010013753aes_gcm_dec(<bits>,<nonce>,<key>,<aead_tag>)
13754 Decrypts the raw byte input using the AES128-GCM, AES192-GCM or
13755 AES256-GCM algorithm, depending on the <bits> parameter. All other parameters
13756 need to be base64 encoded and the returned result is in raw byte format.
13757 If the <aead_tag> validation fails, the converter doesn't return any data.
13758 The <nonce>, <key> and <aead_tag> can either be strings or variables. This
13759 converter requires at least OpenSSL 1.0.1.
13760
13761 Example:
13762 http-response set-header X-Decrypted-Text %[var(txn.enc),\
13763 aes_gcm_dec(128,txn.nonce,Zm9vb2Zvb29mb29wZm9vbw==,txn.aead_tag)]
13764
Willy Tarreau97707872015-01-27 15:12:13 +010013765and(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013766 Performs a bitwise "AND" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013767 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010013768 numeric value or a variable name. The name of the variable starts with an
13769 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013770 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013771 "sess" : the variable is shared with the whole session
13772 "txn" : the variable is shared with the transaction (request and response)
13773 "req" : the variable is shared only during request processing
13774 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013775 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013776 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013777
Holger Just1bfc24b2017-05-06 00:56:53 +020013778b64dec
13779 Converts (decodes) a base64 encoded input string to its binary
13780 representation. It performs the inverse operation of base64().
13781
Emeric Brun53d1a982014-04-30 18:21:37 +020013782base64
13783 Converts a binary input sample to a base64 string. It is used to log or
Davor Ocelice9ed2812017-12-25 17:49:28 +010013784 transfer binary content in a way that can be reliably transferred (e.g.
Emeric Brun53d1a982014-04-30 18:21:37 +020013785 an SSL ID can be copied in a header).
13786
Willy Tarreau97707872015-01-27 15:12:13 +010013787bool
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013788 Returns a boolean TRUE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010013789 non-null, otherwise returns FALSE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010013790 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010013791 presence of a flag).
13792
Emeric Brun54c4ac82014-11-03 15:32:43 +010013793bytes(<offset>[,<length>])
13794 Extracts some bytes from an input binary sample. The result is a binary
13795 sample starting at an offset (in bytes) of the original sample and
Tim Düsterhus4896c442016-11-29 02:15:19 +010013796 optionally truncated at the given length.
Emeric Brun54c4ac82014-11-03 15:32:43 +010013797
Willy Tarreau280f42b2018-02-19 15:34:12 +010013798concat([<start>],[<var>],[<end>])
13799 Concatenates up to 3 fields after the current sample which is then turned to
13800 a string. The first one, <start>, is a constant string, that will be appended
13801 immediately after the existing sample. It may be omitted if not used. The
13802 second one, <var>, is a variable name. The variable will be looked up, its
13803 contents converted to a string, and it will be appended immediately after the
13804 <first> part. If the variable is not found, nothing is appended. It may be
13805 omitted as well. The third field, <end> is a constant string that will be
13806 appended after the variable. It may also be omitted. Together, these elements
13807 allow to concatenate variables with delimiters to an existing set of
13808 variables. This can be used to build new variables made of a succession of
Willy Tarreauef21fac2020-02-14 13:37:20 +010013809 other variables, such as colon-delimited values. If commas or closing
13810 parethesis are needed as delimiters, they must be protected by quotes or
13811 backslashes, themselves protected so that they are not stripped by the first
13812 level parser. See examples below.
Willy Tarreau280f42b2018-02-19 15:34:12 +010013813
13814 Example:
13815 tcp-request session set-var(sess.src) src
13816 tcp-request session set-var(sess.dn) ssl_c_s_dn
13817 tcp-request session set-var(txn.sig) str(),concat(<ip=,sess.ip,>),concat(<dn=,sess.dn,>)
Willy Tarreauef21fac2020-02-14 13:37:20 +010013818 tcp-request session set-var(txn.ipport) "str(),concat('addr=(',sess.ip),concat(',',sess.port,')')"
Willy Tarreau280f42b2018-02-19 15:34:12 +010013819 http-request set-header x-hap-sig %[var(txn.sig)]
13820
Willy Tarreau97707872015-01-27 15:12:13 +010013821cpl
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013822 Takes the input value of type signed integer, applies a ones-complement
13823 (flips all bits) and returns the result as an signed integer.
Willy Tarreau97707872015-01-27 15:12:13 +010013824
Willy Tarreau80599772015-01-20 19:35:24 +010013825crc32([<avalanche>])
13826 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32
13827 hash function. Optionally, it is possible to apply a full avalanche hash
13828 function to the output if the optional <avalanche> argument equals 1. This
13829 converter uses the same functions as used by the various hash-based load
13830 balancing algorithms, so it will provide exactly the same results. It is
13831 provided for compatibility with other software which want a CRC32 to be
13832 computed on some input keys, so it follows the most common implementation as
13833 found in Ethernet, Gzip, PNG, etc... It is slower than the other algorithms
13834 but may provide a better or at least less predictable distribution. It must
13835 not be used for security purposes as a 32-bit hash is trivial to break. See
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013836 also "djb2", "sdbm", "wt6", "crc32c" and the "hash-type" directive.
13837
13838crc32c([<avalanche>])
13839 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32C
13840 hash function. Optionally, it is possible to apply a full avalanche hash
13841 function to the output if the optional <avalanche> argument equals 1. This
13842 converter uses the same functions as described in RFC4960, Appendix B [8].
13843 It is provided for compatibility with other software which want a CRC32C to be
13844 computed on some input keys. It is slower than the other algorithms and it must
13845 not be used for security purposes as a 32-bit hash is trivial to break. See
13846 also "djb2", "sdbm", "wt6", "crc32" and the "hash-type" directive.
Willy Tarreau80599772015-01-20 19:35:24 +010013847
David Carlier29b3ca32015-09-25 14:09:21 +010013848da-csv-conv(<prop>[,<prop>*])
David Carlier4542b102015-06-01 13:54:29 +020013849 Asks the DeviceAtlas converter to identify the User Agent string passed on
13850 input, and to emit a string made of the concatenation of the properties
13851 enumerated in argument, delimited by the separator defined by the global
13852 keyword "deviceatlas-property-separator", or by default the pipe character
David Carlier840b0242016-03-16 10:09:55 +000013853 ('|'). There's a limit of 12 different properties imposed by the haproxy
David Carlier4542b102015-06-01 13:54:29 +020013854 configuration language.
13855
13856 Example:
13857 frontend www
Cyril Bonté307ee1e2015-09-28 23:16:06 +020013858 bind *:8881
13859 default_backend servers
David Carlier840b0242016-03-16 10:09:55 +000013860 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 +020013861
Willy Tarreau0851fd52019-12-17 10:07:25 +010013862debug([<prefix][,<destination>])
13863 This converter is used as debug tool. It takes a capture of the input sample
13864 and sends it to event sink <destination>, which may designate a ring buffer
13865 such as "buf0", as well as "stdout", or "stderr". Available sinks may be
13866 checked at run time by issuing "show events" on the CLI. When not specified,
13867 the output will be "buf0", which may be consulted via the CLI's "show events"
13868 command. An optional prefix <prefix> may be passed to help distinguish
13869 outputs from multiple expressions. It will then appear before the colon in
13870 the output message. The input sample is passed as-is on the output, so that
13871 it is safe to insert the debug converter anywhere in a chain, even with non-
13872 printable sample types.
13873
13874 Example:
13875 tcp-request connection track-sc0 src,debug(track-sc)
Thierry FOURNIER9687c772015-05-07 15:46:29 +020013876
Willy Tarreau97707872015-01-27 15:12:13 +010013877div(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013878 Divides the input value of type signed integer by <value>, and returns the
13879 result as an signed integer. If <value> is null, the largest unsigned
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013880 integer is returned (typically 2^63-1). <value> can be a numeric value or a
Daniel Schneller0b547052016-03-21 20:46:57 +010013881 variable name. The name of the variable starts with an indication about its
13882 scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013883 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013884 "sess" : the variable is shared with the whole session
13885 "txn" : the variable is shared with the transaction (request and response)
13886 "req" : the variable is shared only during request processing
13887 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013888 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013889 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013890
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013891djb2([<avalanche>])
13892 Hashes a binary input sample into an unsigned 32-bit quantity using the DJB2
13893 hash function. Optionally, it is possible to apply a full avalanche hash
13894 function to the output if the optional <avalanche> argument equals 1. This
13895 converter uses the same functions as used by the various hash-based load
13896 balancing algorithms, so it will provide exactly the same results. It is
13897 mostly intended for debugging, but can be used as a stick-table entry to
13898 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013899 32-bit hash is trivial to break. See also "crc32", "sdbm", "wt6", "crc32c",
13900 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013901
Willy Tarreau97707872015-01-27 15:12:13 +010013902even
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013903 Returns a boolean TRUE if the input value of type signed integer is even
Willy Tarreau97707872015-01-27 15:12:13 +010013904 otherwise returns FALSE. It is functionally equivalent to "not,and(1),bool".
13905
Marcin Deranek9631a282018-04-16 14:30:46 +020013906field(<index>,<delimiters>[,<count>])
13907 Extracts the substring at the given index counting from the beginning
13908 (positive index) or from the end (negative index) considering given delimiters
13909 from an input string. Indexes start at 1 or -1 and delimiters are a string
13910 formatted list of chars. Optionally you can specify <count> of fields to
13911 extract (default: 1). Value of 0 indicates extraction of all remaining
13912 fields.
13913
13914 Example :
13915 str(f1_f2_f3__f5),field(5,_) # f5
13916 str(f1_f2_f3__f5),field(2,_,0) # f2_f3__f5
13917 str(f1_f2_f3__f5),field(2,_,2) # f2_f3
13918 str(f1_f2_f3__f5),field(-2,_,3) # f2_f3_
13919 str(f1_f2_f3__f5),field(-3,_,0) # f1_f2_f3
Emeric Brunf399b0d2014-11-03 17:07:03 +010013920
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013921hex
Davor Ocelice9ed2812017-12-25 17:49:28 +010013922 Converts a binary input sample to a hex string containing two hex digits per
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013923 input byte. It is used to log or transfer hex dumps of some binary input data
Davor Ocelice9ed2812017-12-25 17:49:28 +010013924 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 +020013925 header).
Thierry FOURNIER2f49d6d2014-03-12 15:01:52 +010013926
Dragan Dosen3f957b22017-10-24 09:27:34 +020013927hex2i
13928 Converts a hex string containing two hex digits per input byte to an
John Roeslerfb2fce12019-07-10 15:45:51 -050013929 integer. If the input value cannot be converted, then zero is returned.
Dragan Dosen3f957b22017-10-24 09:27:34 +020013930
Cyril Bonté6bcd1822019-11-05 23:13:59 +010013931http_date([<offset],[<unit>])
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013932 Converts an integer supposed to contain a date since epoch to a string
13933 representing this date in a format suitable for use in HTTP header fields. If
Damien Claisseae6f1252019-10-30 15:57:28 +000013934 an offset value is specified, then it is added to the date before the
13935 conversion is operated. This is particularly useful to emit Date header fields,
13936 Expires values in responses when combined with a positive offset, or
13937 Last-Modified values when the offset is negative.
13938 If a unit value is specified, then consider the timestamp as either
13939 "s" for seconds (default behavior), "ms" for milliseconds, or "us" for
13940 microseconds since epoch. Offset is assumed to have the same unit as
13941 input timestamp.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013942
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013943in_table(<table>)
13944 Uses the string representation of the input sample to perform a look up in
13945 the specified table. If the key is not found in the table, a boolean false
13946 is returned. Otherwise a boolean true is returned. This can be used to verify
Davor Ocelice9ed2812017-12-25 17:49:28 +010013947 the presence of a certain key in a table tracking some elements (e.g. whether
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013948 or not a source IP address or an Authorization header was already seen).
13949
Tim Duesterhus1478aa72018-01-25 16:24:51 +010013950ipmask(<mask4>, [<mask6>])
13951 Apply a mask to an IP address, and use the result for lookups and storage.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013952 This can be used to make all hosts within a certain mask to share the same
Tim Duesterhus1478aa72018-01-25 16:24:51 +010013953 table entries and as such use the same server. The mask4 can be passed in
13954 dotted form (e.g. 255.255.255.0) or in CIDR form (e.g. 24). The mask6 can
13955 be passed in quadruplet form (e.g. ffff:ffff::) or in CIDR form (e.g. 64).
13956 If no mask6 is given IPv6 addresses will fail to convert for backwards
13957 compatibility reasons.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013958
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013959json([<input-code>])
Davor Ocelice9ed2812017-12-25 17:49:28 +010013960 Escapes the input string and produces an ASCII output string ready to use as a
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013961 JSON string. The converter tries to decode the input string according to the
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020013962 <input-code> parameter. It can be "ascii", "utf8", "utf8s", "utf8p" or
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013963 "utf8ps". The "ascii" decoder never fails. The "utf8" decoder detects 3 types
13964 of errors:
13965 - bad UTF-8 sequence (lone continuation byte, bad number of continuation
13966 bytes, ...)
13967 - invalid range (the decoded value is within a UTF-8 prohibited range),
13968 - code overlong (the value is encoded with more bytes than necessary).
13969
13970 The UTF-8 JSON encoding can produce a "too long value" error when the UTF-8
13971 character is greater than 0xffff because the JSON string escape specification
13972 only authorizes 4 hex digits for the value encoding. The UTF-8 decoder exists
13973 in 4 variants designated by a combination of two suffix letters : "p" for
13974 "permissive" and "s" for "silently ignore". The behaviors of the decoders
13975 are :
Davor Ocelice9ed2812017-12-25 17:49:28 +010013976 - "ascii" : never fails;
13977 - "utf8" : fails on any detected errors;
13978 - "utf8s" : never fails, but removes characters corresponding to errors;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013979 - "utf8p" : accepts and fixes the overlong errors, but fails on any other
Davor Ocelice9ed2812017-12-25 17:49:28 +010013980 error;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013981 - "utf8ps" : never fails, accepts and fixes the overlong errors, but removes
13982 characters corresponding to the other errors.
13983
13984 This converter is particularly useful for building properly escaped JSON for
Davor Ocelice9ed2812017-12-25 17:49:28 +010013985 logging to servers which consume JSON-formatted traffic logs.
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013986
13987 Example:
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013988 capture request header Host len 15
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020013989 capture request header user-agent len 150
13990 log-format '{"ip":"%[src]","user-agent":"%[capture.req.hdr(1),json(utf8s)]"}'
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013991
13992 Input request from client 127.0.0.1:
13993 GET / HTTP/1.0
13994 User-Agent: Very "Ugly" UA 1/2
13995
13996 Output log:
13997 {"ip":"127.0.0.1","user-agent":"Very \"Ugly\" UA 1\/2"}
13998
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013999language(<value>[,<default>])
14000 Returns the value with the highest q-factor from a list as extracted from the
14001 "accept-language" header using "req.fhdr". Values with no q-factor have a
14002 q-factor of 1. Values with a q-factor of 0 are dropped. Only values which
14003 belong to the list of semi-colon delimited <values> will be considered. The
14004 argument <value> syntax is "lang[;lang[;lang[;...]]]". If no value matches the
14005 given list and a default value is provided, it is returned. Note that language
14006 names may have a variant after a dash ('-'). If this variant is present in the
14007 list, it will be matched, but if it is not, only the base language is checked.
14008 The match is case-sensitive, and the output string is always one of those
Davor Ocelice9ed2812017-12-25 17:49:28 +010014009 provided in arguments. The ordering of arguments is meaningless, only the
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014010 ordering of the values in the request counts, as the first value among
14011 multiple sharing the same q-factor is used.
Thierry FOURNIERad903512014-04-11 17:51:01 +020014012
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014013 Example :
Thierry FOURNIERad903512014-04-11 17:51:01 +020014014
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014015 # this configuration switches to the backend matching a
14016 # given language based on the request :
Thierry FOURNIERad903512014-04-11 17:51:01 +020014017
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014018 acl es req.fhdr(accept-language),language(es;fr;en) -m str es
14019 acl fr req.fhdr(accept-language),language(es;fr;en) -m str fr
14020 acl en req.fhdr(accept-language),language(es;fr;en) -m str en
14021 use_backend spanish if es
14022 use_backend french if fr
14023 use_backend english if en
14024 default_backend choose_your_language
Thierry FOURNIERad903512014-04-11 17:51:01 +020014025
Willy Tarreau60a2ee72017-12-15 07:13:48 +010014026length
Etienne Carriereed0d24e2017-12-13 13:41:34 +010014027 Get the length of the string. This can only be placed after a string
14028 sample fetch function or after a transformation keyword returning a string
14029 type. The result is of type integer.
14030
Willy Tarreauffcb2e42014-07-10 16:29:08 +020014031lower
14032 Convert a string sample to lower case. This can only be placed after a string
14033 sample fetch function or after a transformation keyword returning a string
14034 type. The result is of type string.
14035
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020014036ltime(<format>[,<offset>])
14037 Converts an integer supposed to contain a date since epoch to a string
14038 representing this date in local time using a format defined by the <format>
14039 string using strftime(3). The purpose is to allow any date format to be used
14040 in logs. An optional <offset> in seconds may be applied to the input date
14041 (positive or negative). See the strftime() man page for the format supported
14042 by your operating system. See also the utime converter.
14043
14044 Example :
14045
14046 # Emit two colons, one with the local time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010014047 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020014048 log-format %[date,ltime(%Y%m%d%H%M%S)]\ %ci:%cp
14049
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014050map(<map_file>[,<default_value>])
14051map_<match_type>(<map_file>[,<default_value>])
14052map_<match_type>_<output_type>(<map_file>[,<default_value>])
14053 Search the input value from <map_file> using the <match_type> matching method,
14054 and return the associated value converted to the type <output_type>. If the
14055 input value cannot be found in the <map_file>, the converter returns the
14056 <default_value>. If the <default_value> is not set, the converter fails and
14057 acts as if no input value could be fetched. If the <match_type> is not set, it
14058 defaults to "str". Likewise, if the <output_type> is not set, it defaults to
14059 "str". For convenience, the "map" keyword is an alias for "map_str" and maps a
14060 string to another string.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010014061
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014062 It is important to avoid overlapping between the keys : IP addresses and
14063 strings are stored in trees, so the first of the finest match will be used.
14064 Other keys are stored in lists, so the first matching occurrence will be used.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010014065
Tim Düsterhus4896c442016-11-29 02:15:19 +010014066 The following array contains the list of all map functions available sorted by
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014067 input type, match type and output type.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010014068
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014069 input type | match method | output type str | output type int | output type ip
14070 -----------+--------------+-----------------+-----------------+---------------
14071 str | str | map_str | map_str_int | map_str_ip
14072 -----------+--------------+-----------------+-----------------+---------------
Willy Tarreau787a4c02014-05-10 07:55:30 +020014073 str | beg | map_beg | map_beg_int | map_end_ip
14074 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014075 str | sub | map_sub | map_sub_int | map_sub_ip
14076 -----------+--------------+-----------------+-----------------+---------------
14077 str | dir | map_dir | map_dir_int | map_dir_ip
14078 -----------+--------------+-----------------+-----------------+---------------
14079 str | dom | map_dom | map_dom_int | map_dom_ip
14080 -----------+--------------+-----------------+-----------------+---------------
14081 str | end | map_end | map_end_int | map_end_ip
14082 -----------+--------------+-----------------+-----------------+---------------
Ruoshan Huang3c5e3742016-12-02 16:25:31 +080014083 str | reg | map_reg | map_reg_int | map_reg_ip
14084 -----------+--------------+-----------------+-----------------+---------------
14085 str | reg | map_regm | map_reg_int | map_reg_ip
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014086 -----------+--------------+-----------------+-----------------+---------------
14087 int | int | map_int | map_int_int | map_int_ip
14088 -----------+--------------+-----------------+-----------------+---------------
14089 ip | ip | map_ip | map_ip_int | map_ip_ip
14090 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010014091
Thierry Fournier8feaa662016-02-10 22:55:20 +010014092 The special map called "map_regm" expect matching zone in the regular
14093 expression and modify the output replacing back reference (like "\1") by
14094 the corresponding match text.
14095
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014096 The file contains one key + value per line. Lines which start with '#' are
14097 ignored, just like empty lines. Leading tabs and spaces are stripped. The key
14098 is then the first "word" (series of non-space/tabs characters), and the value
14099 is what follows this series of space/tab till the end of the line excluding
14100 trailing spaces/tabs.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010014101
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014102 Example :
14103
14104 # this is a comment and is ignored
14105 2.22.246.0/23 United Kingdom \n
14106 <-><-----------><--><------------><---->
14107 | | | | `- trailing spaces ignored
14108 | | | `---------- value
14109 | | `-------------------- middle spaces ignored
14110 | `---------------------------- key
14111 `------------------------------------ leading spaces ignored
14112
Willy Tarreau97707872015-01-27 15:12:13 +010014113mod(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014114 Divides the input value of type signed integer by <value>, and returns the
14115 remainder as an signed integer. If <value> is null, then zero is returned.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014116 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010014117 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010014118 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010014119 "sess" : the variable is shared with the whole session
14120 "txn" : the variable is shared with the transaction (request and response)
14121 "req" : the variable is shared only during request processing
14122 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010014123 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014124 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010014125
14126mul(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014127 Multiplies the input value of type signed integer by <value>, and returns
Thierry FOURNIER00c005c2015-07-08 01:10:21 +020014128 the product as an signed integer. In case of overflow, the largest possible
14129 value for the sign is returned so that the operation doesn't wrap around.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014130 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010014131 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010014132 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010014133 "sess" : the variable is shared with the whole session
14134 "txn" : the variable is shared with the transaction (request and response)
14135 "req" : the variable is shared only during request processing
14136 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010014137 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014138 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010014139
Nenad Merdanovicb7e7c472017-03-12 21:56:55 +010014140nbsrv
14141 Takes an input value of type string, interprets it as a backend name and
14142 returns the number of usable servers in that backend. Can be used in places
14143 where we want to look up a backend from a dynamic name, like a result of a
14144 map lookup.
14145
Willy Tarreau97707872015-01-27 15:12:13 +010014146neg
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014147 Takes the input value of type signed integer, computes the opposite value,
14148 and returns the remainder as an signed integer. 0 is identity. This operator
14149 is provided for reversed subtracts : in order to subtract the input from a
14150 constant, simply perform a "neg,add(value)".
Willy Tarreau97707872015-01-27 15:12:13 +010014151
14152not
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014153 Returns a boolean FALSE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010014154 non-null, otherwise returns TRUE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010014155 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010014156 absence of a flag).
14157
14158odd
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014159 Returns a boolean TRUE if the input value of type signed integer is odd
Willy Tarreau97707872015-01-27 15:12:13 +010014160 otherwise returns FALSE. It is functionally equivalent to "and(1),bool".
14161
14162or(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014163 Performs a bitwise "OR" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014164 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010014165 numeric value or a variable name. The name of the variable starts with an
14166 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010014167 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010014168 "sess" : the variable is shared with the whole session
14169 "txn" : the variable is shared with the transaction (request and response)
14170 "req" : the variable is shared only during request processing
14171 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010014172 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014173 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010014174
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010014175protobuf(<field_number>,[<field_type>])
14176 This extracts the protocol buffers message field in raw mode of an input binary
14177 sample representation of a protocol buffer message with <field_number> as field
14178 number (dotted notation) if <field_type> is not present, or as an integer sample
14179 if this field is present (see also "ungrpc" below).
14180 The list of the authorized types is the following one: "int32", "int64", "uint32",
14181 "uint64", "sint32", "sint64", "bool", "enum" for the "varint" wire type 0
14182 "fixed64", "sfixed64", "double" for the 64bit wire type 1, "fixed32", "sfixed32",
14183 "float" for the wire type 5. Note that "string" is considered as a length-delimited
14184 type, so it does not require any <field_type> argument to be extracted.
14185 More information may be found here about the protocol buffers message field types:
14186 https://developers.google.com/protocol-buffers/docs/encoding
14187
Willy Tarreauc4dc3502015-01-23 20:39:28 +010014188regsub(<regex>,<subst>[,<flags>])
Willy Tarreau7eda8492015-01-20 19:47:06 +010014189 Applies a regex-based substitution to the input string. It does the same
14190 operation as the well-known "sed" utility with "s/<regex>/<subst>/". By
14191 default it will replace in the input string the first occurrence of the
14192 largest part matching the regular expression <regex> with the substitution
14193 string <subst>. It is possible to replace all occurrences instead by adding
14194 the flag "g" in the third argument <flags>. It is also possible to make the
14195 regex case insensitive by adding the flag "i" in <flags>. Since <flags> is a
14196 string, it is made up from the concatenation of all desired flags. Thus if
14197 both "i" and "g" are desired, using "gi" or "ig" will have the same effect.
Willy Tarreauef21fac2020-02-14 13:37:20 +010014198 The first use of this converter is to replace certain characters or sequence
14199 of characters with other ones.
14200
14201 It is highly recommended to enclose the regex part using protected quotes to
14202 improve clarity and never have a closing parenthesis from the regex mixed up
14203 with the parenthesis from the function. Just like in Bourne shell, the first
14204 level of quotes is processed when delimiting word groups on the line, a
14205 second level is usable for argument. It is recommended to use single quotes
14206 outside since these ones do not try to resolve backslashes nor dollar signs.
Willy Tarreau7eda8492015-01-20 19:47:06 +010014207
Willy Tarreaucd0d2ed2020-02-14 17:33:06 +010014208 Examples:
Willy Tarreau7eda8492015-01-20 19:47:06 +010014209
14210 # de-duplicate "/" in header "x-path".
14211 # input: x-path: /////a///b/c/xzxyz/
14212 # output: x-path: /a/b/c/xzxyz/
Willy Tarreauef21fac2020-02-14 13:37:20 +010014213 http-request set-header x-path "%[hdr(x-path),regsub('/+','/','g')]"
Willy Tarreau7eda8492015-01-20 19:47:06 +010014214
Willy Tarreaucd0d2ed2020-02-14 17:33:06 +010014215 # copy query string to x-query and drop all leading '?', ';' and '&'
14216 http-request set-header x-query "%[query,regsub([?;&]*,'')]"
14217
Jerome Magnin07e1e3c2020-02-16 19:20:19 +010014218 # capture groups and backreferences
14219 # both lines do the same.
14220 http-request redirect location %[url,'regsub("(foo|bar)([0-9]+)?","\2\1",i)]'
14221 http-request redirect location %[url,regsub(\"(foo|bar)([0-9]+)?\",\"\2\1\",i)]
14222
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020014223capture-req(<id>)
14224 Capture the string entry in the request slot <id> and returns the entry as
14225 is. If the slot doesn't exist, the capture fails silently.
14226
14227 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020014228 "http-response capture", "capture.req.hdr" and
14229 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020014230
14231capture-res(<id>)
14232 Capture the string entry in the response slot <id> and returns the entry as
14233 is. If the slot doesn't exist, the capture fails silently.
14234
14235 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020014236 "http-response capture", "capture.req.hdr" and
14237 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020014238
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020014239sdbm([<avalanche>])
14240 Hashes a binary input sample into an unsigned 32-bit quantity using the SDBM
14241 hash function. Optionally, it is possible to apply a full avalanche hash
14242 function to the output if the optional <avalanche> argument equals 1. This
14243 converter uses the same functions as used by the various hash-based load
14244 balancing algorithms, so it will provide exactly the same results. It is
14245 mostly intended for debugging, but can be used as a stick-table entry to
14246 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010014247 32-bit hash is trivial to break. See also "crc32", "djb2", "wt6", "crc32c",
14248 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020014249
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014250set-var(<var name>)
Davor Ocelice9ed2812017-12-25 17:49:28 +010014251 Sets a variable with the input content and returns the content on the output
14252 as-is. The variable keeps the value and the associated input type. The name of
14253 the variable starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010014254 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010014255 "sess" : the variable is shared with the whole session
14256 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014257 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010014258 "req" : the variable is shared only during request processing,
14259 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014260 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014261 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014262
Dragan Dosen6e5a9ca2017-10-24 09:18:23 +020014263sha1
Tim Duesterhusd4376302019-06-17 12:41:44 +020014264 Converts a binary input sample to a SHA-1 digest. The result is a binary
Dragan Dosen6e5a9ca2017-10-24 09:18:23 +020014265 sample with length of 20 bytes.
14266
Tim Duesterhusd4376302019-06-17 12:41:44 +020014267sha2([<bits>])
14268 Converts a binary input sample to a digest in the SHA-2 family. The result
14269 is a binary sample with length of <bits>/8 bytes.
14270
14271 Valid values for <bits> are 224, 256, 384, 512, each corresponding to
14272 SHA-<bits>. The default value is 256.
14273
14274 Please note that this converter is only available when haproxy has been
14275 compiled with USE_OPENSSL.
14276
Nenad Merdanovic177adc92019-08-27 01:58:13 +020014277srv_queue
14278 Takes an input value of type string, either a server name or <backend>/<server>
14279 format and returns the number of queued sessions on that server. Can be used
14280 in places where we want to look up queued sessions from a dynamic name, like a
14281 cookie value (e.g. req.cook(SRVID),srv_queue) and then make a decision to break
14282 persistence or direct a request elsewhere.
14283
Tim Duesterhusca097c12018-04-27 21:18:45 +020014284strcmp(<var>)
14285 Compares the contents of <var> with the input value of type string. Returns
14286 the result as a signed integer compatible with strcmp(3): 0 if both strings
14287 are identical. A value less than 0 if the left string is lexicographically
14288 smaller than the right string or if the left string is shorter. A value greater
14289 than 0 otherwise (right string greater than left string or the right string is
14290 shorter).
14291
14292 Example :
14293
14294 http-request set-var(txn.host) hdr(host)
14295 # Check whether the client is attempting domain fronting.
14296 acl ssl_sni_http_host_match ssl_fc_sni,strcmp(txn.host) eq 0
14297
14298
Willy Tarreau97707872015-01-27 15:12:13 +010014299sub(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014300 Subtracts <value> from the input value of type signed integer, and returns
14301 the result as an signed integer. Note: in order to subtract the input from
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014302 a constant, simply perform a "neg,add(value)". <value> can be a numeric value
Daniel Schneller0b547052016-03-21 20:46:57 +010014303 or a variable name. The name of the variable starts with an indication about
14304 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010014305 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010014306 "sess" : the variable is shared with the whole session
14307 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014308 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010014309 "req" : the variable is shared only during request processing,
14310 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014311 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014312 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010014313
Willy Tarreaud9f316a2014-07-10 14:03:38 +020014314table_bytes_in_rate(<table>)
14315 Uses the string representation of the input sample to perform a look up in
14316 the specified table. If the key is not found in the table, integer value zero
14317 is returned. Otherwise the converter returns the average client-to-server
14318 bytes rate associated with the input sample in the designated table, measured
14319 in amount of bytes over the period configured in the table. See also the
14320 sc_bytes_in_rate sample fetch keyword.
14321
14322
14323table_bytes_out_rate(<table>)
14324 Uses the string representation of the input sample to perform a look up in
14325 the specified table. If the key is not found in the table, integer value zero
14326 is returned. Otherwise the converter returns the average server-to-client
14327 bytes rate associated with the input sample in the designated table, measured
14328 in amount of bytes over the period configured in the table. See also the
14329 sc_bytes_out_rate sample fetch keyword.
14330
14331table_conn_cnt(<table>)
14332 Uses the string representation of the input sample to perform a look up in
14333 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010014334 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020014335 connections associated with the input sample in the designated table. See
14336 also the sc_conn_cnt sample fetch keyword.
14337
14338table_conn_cur(<table>)
14339 Uses the string representation of the input sample to perform a look up in
14340 the specified table. If the key is not found in the table, integer value zero
14341 is returned. Otherwise the converter returns the current amount of concurrent
14342 tracked connections associated with the input sample in the designated table.
14343 See also the sc_conn_cur sample fetch keyword.
14344
14345table_conn_rate(<table>)
14346 Uses the string representation of the input sample to perform a look up in
14347 the specified table. If the key is not found in the table, integer value zero
14348 is returned. Otherwise the converter returns the average incoming connection
14349 rate associated with the input sample in the designated table. See also the
14350 sc_conn_rate sample fetch keyword.
14351
Thierry FOURNIER236657b2015-08-19 08:25:14 +020014352table_gpt0(<table>)
14353 Uses the string representation of the input sample to perform a look up in
14354 the specified table. If the key is not found in the table, boolean value zero
14355 is returned. Otherwise the converter returns the current value of the first
14356 general purpose tag associated with the input sample in the designated table.
14357 See also the sc_get_gpt0 sample fetch keyword.
14358
Willy Tarreaud9f316a2014-07-10 14:03:38 +020014359table_gpc0(<table>)
14360 Uses the string representation of the input sample to perform a look up in
14361 the specified table. If the key is not found in the table, integer value zero
14362 is returned. Otherwise the converter returns the current value of the first
14363 general purpose counter associated with the input sample in the designated
14364 table. See also the sc_get_gpc0 sample fetch keyword.
14365
14366table_gpc0_rate(<table>)
14367 Uses the string representation of the input sample to perform a look up in
14368 the specified table. If the key is not found in the table, integer value zero
14369 is returned. Otherwise the converter returns the frequency which the gpc0
14370 counter was incremented over the configured period in the table, associated
14371 with the input sample in the designated table. See also the sc_get_gpc0_rate
14372 sample fetch keyword.
14373
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014374table_gpc1(<table>)
14375 Uses the string representation of the input sample to perform a look up in
14376 the specified table. If the key is not found in the table, integer value zero
14377 is returned. Otherwise the converter returns the current value of the second
14378 general purpose counter associated with the input sample in the designated
14379 table. See also the sc_get_gpc1 sample fetch keyword.
14380
14381table_gpc1_rate(<table>)
14382 Uses the string representation of the input sample to perform a look up in
14383 the specified table. If the key is not found in the table, integer value zero
14384 is returned. Otherwise the converter returns the frequency which the gpc1
14385 counter was incremented over the configured period in the table, associated
14386 with the input sample in the designated table. See also the sc_get_gpc1_rate
14387 sample fetch keyword.
14388
Willy Tarreaud9f316a2014-07-10 14:03:38 +020014389table_http_err_cnt(<table>)
14390 Uses the string representation of the input sample to perform a look up in
14391 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010014392 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020014393 errors associated with the input sample in the designated table. See also the
14394 sc_http_err_cnt sample fetch keyword.
14395
14396table_http_err_rate(<table>)
14397 Uses the string representation of the input sample to perform a look up in
14398 the specified table. If the key is not found in the table, integer value zero
14399 is returned. Otherwise the average rate of HTTP errors associated with the
14400 input sample in the designated table, measured in amount of errors over the
14401 period configured in the table. See also the sc_http_err_rate sample fetch
14402 keyword.
14403
14404table_http_req_cnt(<table>)
14405 Uses the string representation of the input sample to perform a look up in
14406 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010014407 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020014408 requests associated with the input sample in the designated table. See also
14409 the sc_http_req_cnt sample fetch keyword.
14410
14411table_http_req_rate(<table>)
14412 Uses the string representation of the input sample to perform a look up in
14413 the specified table. If the key is not found in the table, integer value zero
14414 is returned. Otherwise the average rate of HTTP requests associated with the
14415 input sample in the designated table, measured in amount of requests over the
14416 period configured in the table. See also the sc_http_req_rate sample fetch
14417 keyword.
14418
14419table_kbytes_in(<table>)
14420 Uses the string representation of the input sample to perform a look up in
14421 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010014422 is returned. Otherwise the converter returns the cumulative number of client-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020014423 to-server data associated with the input sample in the designated table,
14424 measured in kilobytes. The test is currently performed on 32-bit integers,
14425 which limits values to 4 terabytes. See also the sc_kbytes_in sample fetch
14426 keyword.
14427
14428table_kbytes_out(<table>)
14429 Uses the string representation of the input sample to perform a look up in
14430 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010014431 is returned. Otherwise the converter returns the cumulative number of server-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020014432 to-client data associated with the input sample in the designated table,
14433 measured in kilobytes. The test is currently performed on 32-bit integers,
14434 which limits values to 4 terabytes. See also the sc_kbytes_out sample fetch
14435 keyword.
14436
14437table_server_id(<table>)
14438 Uses the string representation of the input sample to perform a look up in
14439 the specified table. If the key is not found in the table, integer value zero
14440 is returned. Otherwise the converter returns the server ID associated with
14441 the input sample in the designated table. A server ID is associated to a
14442 sample by a "stick" rule when a connection to a server succeeds. A server ID
14443 zero means that no server is associated with this key.
14444
14445table_sess_cnt(<table>)
14446 Uses the string representation of the input sample to perform a look up in
14447 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010014448 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020014449 sessions associated with the input sample in the designated table. Note that
14450 a session here refers to an incoming connection being accepted by the
14451 "tcp-request connection" rulesets. See also the sc_sess_cnt sample fetch
14452 keyword.
14453
14454table_sess_rate(<table>)
14455 Uses the string representation of the input sample to perform a look up in
14456 the specified table. If the key is not found in the table, integer value zero
14457 is returned. Otherwise the converter returns the average incoming session
14458 rate associated with the input sample in the designated table. Note that a
14459 session here refers to an incoming connection being accepted by the
14460 "tcp-request connection" rulesets. See also the sc_sess_rate sample fetch
14461 keyword.
14462
14463table_trackers(<table>)
14464 Uses the string representation of the input sample to perform a look up in
14465 the specified table. If the key is not found in the table, integer value zero
14466 is returned. Otherwise the converter returns the current amount of concurrent
14467 connections tracking the same key as the input sample in the designated
14468 table. It differs from table_conn_cur in that it does not rely on any stored
14469 information but on the table's reference count (the "use" value which is
14470 returned by "show table" on the CLI). This may sometimes be more suited for
14471 layer7 tracking. It can be used to tell a server how many concurrent
14472 connections there are from a given address for example. See also the
14473 sc_trackers sample fetch keyword.
14474
Willy Tarreauffcb2e42014-07-10 16:29:08 +020014475upper
14476 Convert a string sample to upper case. This can only be placed after a string
14477 sample fetch function or after a transformation keyword returning a string
14478 type. The result is of type string.
14479
Thierry FOURNIER82ff3c92015-05-07 15:46:20 +020014480url_dec
14481 Takes an url-encoded string provided as input and returns the decoded
14482 version as output. The input and the output are of type string.
14483
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010014484ungrpc(<field_number>,[<field_type>])
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010014485 This extracts the protocol buffers message field in raw mode of an input binary
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010014486 sample representation of a gRPC message with <field_number> as field number
14487 (dotted notation) if <field_type> is not present, or as an integer sample if this
14488 field is present.
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010014489 The list of the authorized types is the following one: "int32", "int64", "uint32",
14490 "uint64", "sint32", "sint64", "bool", "enum" for the "varint" wire type 0
14491 "fixed64", "sfixed64", "double" for the 64bit wire type 1, "fixed32", "sfixed32",
14492 "float" for the wire type 5. Note that "string" is considered as a length-delimited
Frédéric Lécaille93d33162019-03-06 09:35:59 +010014493 type, so it does not require any <field_type> argument to be extracted.
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010014494 More information may be found here about the protocol buffers message field types:
14495 https://developers.google.com/protocol-buffers/docs/encoding
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010014496
14497 Example:
14498 // with such a protocol buffer .proto file content adapted from
14499 // https://github.com/grpc/grpc/blob/master/examples/protos/route_guide.proto
14500
14501 message Point {
14502 int32 latitude = 1;
14503 int32 longitude = 2;
14504 }
14505
14506 message PPoint {
14507 Point point = 59;
14508 }
14509
14510 message Rectangle {
14511 // One corner of the rectangle.
14512 PPoint lo = 48;
14513 // The other corner of the rectangle.
14514 PPoint hi = 49;
14515 }
14516
14517 let's say a body request is made of a "Rectangle" object value (two PPoint
14518 protocol buffers messages), the four protocol buffers fields could be
14519 extracted with these "ungrpc" directives:
14520
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010014521 req.body,ungrpc(48.59.1,int32) # "latitude" of "lo" first PPoint
14522 req.body,ungrpc(48.59.2,int32) # "longitude" of "lo" first PPoint
John Roeslerfb2fce12019-07-10 15:45:51 -050014523 req.body,ungrpc(49.59.1,int32) # "latitude" of "hi" second PPoint
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010014524 req.body,ungrpc(49.59.2,int32) # "longitude" of "hi" second PPoint
14525
Frédéric Lécaille93d33162019-03-06 09:35:59 +010014526 We could also extract the intermediary 48.59 field as a binary sample as follows:
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010014527
Frédéric Lécaille93d33162019-03-06 09:35:59 +010014528 req.body,ungrpc(48.59)
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010014529
John Roeslerfb2fce12019-07-10 15:45:51 -050014530 As a gRPC message is always made of a gRPC header followed by protocol buffers
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010014531 messages, in the previous example the "latitude" of "lo" first PPoint
14532 could be extracted with these equivalent directives:
14533
14534 req.body,ungrpc(48.59),protobuf(1,int32)
14535 req.body,ungrpc(48),protobuf(59.1,int32)
14536 req.body,ungrpc(48),protobuf(59),protobuf(1,int32)
14537
14538 Note that the first convert must be "ungrpc", the remaining ones must be
14539 "protobuf" and only the last one may have or not a second argument to
14540 interpret the previous binary sample.
14541
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010014542
Christopher Faulet85d79c92016-11-09 16:54:56 +010014543unset-var(<var name>)
14544 Unsets a variable if the input content is defined. The name of the variable
14545 starts with an indication about its scope. The scopes allowed are:
14546 "proc" : the variable is shared with the whole process
14547 "sess" : the variable is shared with the whole session
14548 "txn" : the variable is shared with the transaction (request and
14549 response),
14550 "req" : the variable is shared only during request processing,
14551 "res" : the variable is shared only during response processing.
14552 This prefix is followed by a name. The separator is a '.'. The name may only
14553 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
14554
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020014555utime(<format>[,<offset>])
14556 Converts an integer supposed to contain a date since epoch to a string
14557 representing this date in UTC time using a format defined by the <format>
14558 string using strftime(3). The purpose is to allow any date format to be used
14559 in logs. An optional <offset> in seconds may be applied to the input date
14560 (positive or negative). See the strftime() man page for the format supported
14561 by your operating system. See also the ltime converter.
14562
14563 Example :
14564
14565 # Emit two colons, one with the UTC time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010014566 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020014567 log-format %[date,utime(%Y%m%d%H%M%S)]\ %ci:%cp
14568
Marcin Deranek9631a282018-04-16 14:30:46 +020014569word(<index>,<delimiters>[,<count>])
14570 Extracts the nth word counting from the beginning (positive index) or from
14571 the end (negative index) considering given delimiters from an input string.
14572 Indexes start at 1 or -1 and delimiters are a string formatted list of chars.
Jerome Magnin88209322020-01-28 13:33:44 +010014573 Delimiters at the beginning or end of the input string are ignored.
Marcin Deranek9631a282018-04-16 14:30:46 +020014574 Optionally you can specify <count> of words to extract (default: 1).
14575 Value of 0 indicates extraction of all remaining words.
14576
14577 Example :
14578 str(f1_f2_f3__f5),word(4,_) # f5
14579 str(f1_f2_f3__f5),word(2,_,0) # f2_f3__f5
14580 str(f1_f2_f3__f5),word(3,_,2) # f3__f5
14581 str(f1_f2_f3__f5),word(-2,_,3) # f1_f2_f3
14582 str(f1_f2_f3__f5),word(-3,_,0) # f1_f2
Jerome Magnin88209322020-01-28 13:33:44 +010014583 str(/f1/f2/f3/f4),word(1,/) # f1
Emeric Brunc9a0f6d2014-11-25 14:09:01 +010014584
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020014585wt6([<avalanche>])
14586 Hashes a binary input sample into an unsigned 32-bit quantity using the WT6
14587 hash function. Optionally, it is possible to apply a full avalanche hash
14588 function to the output if the optional <avalanche> argument equals 1. This
14589 converter uses the same functions as used by the various hash-based load
14590 balancing algorithms, so it will provide exactly the same results. It is
14591 mostly intended for debugging, but can be used as a stick-table entry to
14592 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010014593 32-bit hash is trivial to break. See also "crc32", "djb2", "sdbm", "crc32c",
14594 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020014595
Willy Tarreau97707872015-01-27 15:12:13 +010014596xor(<value>)
14597 Performs a bitwise "XOR" (exclusive OR) between <value> and the input value
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014598 of type signed integer, and returns the result as an signed integer.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014599 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010014600 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010014601 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010014602 "sess" : the variable is shared with the whole session
14603 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014604 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010014605 "req" : the variable is shared only during request processing,
14606 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014607 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014608 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010014609
Thierry FOURNIER01e09742016-12-26 11:46:11 +010014610xxh32([<seed>])
14611 Hashes a binary input sample into an unsigned 32-bit quantity using the 32-bit
14612 variant of the XXHash hash function. This hash supports a seed which defaults
14613 to zero but a different value maybe passed as the <seed> argument. This hash
14614 is known to be very good and very fast so it can be used to hash URLs and/or
14615 URL parameters for use as stick-table keys to collect statistics with a low
14616 collision rate, though care must be taken as the algorithm is not considered
14617 as cryptographically secure.
14618
14619xxh64([<seed>])
14620 Hashes a binary input sample into a signed 64-bit quantity using the 64-bit
14621 variant of the XXHash hash function. This hash supports a seed which defaults
14622 to zero but a different value maybe passed as the <seed> argument. This hash
14623 is known to be very good and very fast so it can be used to hash URLs and/or
14624 URL parameters for use as stick-table keys to collect statistics with a low
14625 collision rate, though care must be taken as the algorithm is not considered
14626 as cryptographically secure.
14627
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010014628
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200146297.3.2. Fetching samples from internal states
Willy Tarreau74ca5042013-06-11 23:12:07 +020014630--------------------------------------------
14631
14632A first set of sample fetch methods applies to internal information which does
14633not even relate to any client information. These ones are sometimes used with
14634"monitor-fail" directives to report an internal status to external watchers.
14635The sample fetch methods described in this section are usable anywhere.
14636
14637always_false : boolean
14638 Always returns the boolean "false" value. It may be used with ACLs as a
14639 temporary replacement for another one when adjusting configurations.
14640
14641always_true : boolean
14642 Always returns the boolean "true" value. It may be used with ACLs as a
14643 temporary replacement for another one when adjusting configurations.
14644
14645avg_queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010014646 Returns the total number of queued connections of the designated backend
Willy Tarreau74ca5042013-06-11 23:12:07 +020014647 divided by the number of active servers. The current backend is used if no
14648 backend is specified. This is very similar to "queue" except that the size of
14649 the farm is considered, in order to give a more accurate measurement of the
14650 time it may take for a new connection to be processed. The main usage is with
14651 ACL to return a sorry page to new users when it becomes certain they will get
14652 a degraded service, or to pass to the backend servers in a header so that
14653 they decide to work in degraded mode or to disable some functions to speed up
14654 the processing a bit. Note that in the event there would not be any active
14655 server anymore, twice the number of queued connections would be considered as
14656 the measured value. This is a fair estimate, as we expect one server to get
14657 back soon anyway, but we still prefer to send new traffic to another backend
14658 if in better shape. See also the "queue", "be_conn", and "be_sess_rate"
14659 sample fetches.
Krzysztof Piotr Oledzki346f76d2010-01-12 21:59:30 +010014660
Willy Tarreau74ca5042013-06-11 23:12:07 +020014661be_conn([<backend>]) : integer
Willy Tarreaua36af912009-10-10 12:02:45 +020014662 Applies to the number of currently established connections on the backend,
14663 possibly including the connection being evaluated. If no backend name is
14664 specified, the current one is used. But it is also possible to check another
14665 backend. It can be used to use a specific farm when the nominal one is full.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040014666 See also the "fe_conn", "queue", "be_conn_free", and "be_sess_rate" criteria.
14667
14668be_conn_free([<backend>]) : integer
14669 Returns an integer value corresponding to the number of available connections
14670 across available servers in the backend. Queue slots are not included. Backup
14671 servers are also not included, unless all other servers are down. If no
14672 backend name is specified, the current one is used. But it is also possible
14673 to check another backend. It can be used to use a specific farm when the
Patrick Hemmer155e93e2018-06-14 18:01:35 -040014674 nominal one is full. See also the "be_conn", "connslots", and "srv_conn_free"
14675 criteria.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040014676
14677 OTHER CAVEATS AND NOTES: if any of the server maxconn, or maxqueue is 0
14678 (meaning unlimited), then this fetch clearly does not make sense, in which
14679 case the value returned will be -1.
Willy Tarreau6a06a402007-07-15 20:15:28 +020014680
Willy Tarreau74ca5042013-06-11 23:12:07 +020014681be_sess_rate([<backend>]) : integer
14682 Returns an integer value corresponding to the sessions creation rate on the
14683 backend, in number of new sessions per second. This is used with ACLs to
14684 switch to an alternate backend when an expensive or fragile one reaches too
Davor Ocelice9ed2812017-12-25 17:49:28 +010014685 high a session rate, or to limit abuse of service (e.g. prevent sucking of an
Willy Tarreau74ca5042013-06-11 23:12:07 +020014686 online dictionary). It can also be useful to add this element to logs using a
14687 log-format directive.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014688
14689 Example :
14690 # Redirect to an error page if the dictionary is requested too often
14691 backend dynamic
14692 mode http
14693 acl being_scanned be_sess_rate gt 100
14694 redirect location /denied.html if being_scanned
Willy Tarreau0ba27502007-12-24 16:55:16 +010014695
Davor Ocelice9ed2812017-12-25 17:49:28 +010014696bin(<hex>) : bin
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014697 Returns a binary chain. The input is the hexadecimal representation
14698 of the string.
14699
14700bool(<bool>) : bool
14701 Returns a boolean value. <bool> can be 'true', 'false', '1' or '0'.
14702 'false' and '0' are the same. 'true' and '1' are the same.
14703
Willy Tarreau74ca5042013-06-11 23:12:07 +020014704connslots([<backend>]) : integer
14705 Returns an integer value corresponding to the number of connection slots
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014706 still available in the backend, by totaling the maximum amount of
Willy Tarreau74ca5042013-06-11 23:12:07 +020014707 connections on all servers and the maximum queue size. This is probably only
14708 used with ACLs.
Tait Clarridge7896d522012-12-05 21:39:31 -050014709
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014710 The basic idea here is to be able to measure the number of connection "slots"
Willy Tarreau55165fe2009-05-10 12:02:55 +020014711 still available (connection + queue), so that anything beyond that (intended
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014712 usage; see "use_backend" keyword) can be redirected to a different backend.
14713
Willy Tarreau55165fe2009-05-10 12:02:55 +020014714 'connslots' = number of available server connection slots, + number of
14715 available server queue slots.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014716
Willy Tarreaua36af912009-10-10 12:02:45 +020014717 Note that while "fe_conn" may be used, "connslots" comes in especially
Willy Tarreau55165fe2009-05-10 12:02:55 +020014718 useful when you have a case of traffic going to one single ip, splitting into
Willy Tarreau74ca5042013-06-11 23:12:07 +020014719 multiple backends (perhaps using ACLs to do name-based load balancing) and
Willy Tarreau55165fe2009-05-10 12:02:55 +020014720 you want to be able to differentiate between different backends, and their
Davor Ocelice9ed2812017-12-25 17:49:28 +010014721 available "connslots". Also, whereas "nbsrv" only measures servers that are
Willy Tarreau74ca5042013-06-11 23:12:07 +020014722 actually *down*, this fetch is more fine-grained and looks into the number of
Willy Tarreaua36af912009-10-10 12:02:45 +020014723 available connection slots as well. See also "queue" and "avg_queue".
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014724
Willy Tarreau55165fe2009-05-10 12:02:55 +020014725 OTHER CAVEATS AND NOTES: at this point in time, the code does not take care
14726 of dynamic connections. Also, if any of the server maxconn, or maxqueue is 0,
Willy Tarreau74ca5042013-06-11 23:12:07 +020014727 then this fetch clearly does not make sense, in which case the value returned
Willy Tarreau55165fe2009-05-10 12:02:55 +020014728 will be -1.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014729
Willy Tarreau70fe9442018-11-22 16:07:39 +010014730cpu_calls : integer
14731 Returns the number of calls to the task processing the stream or current
14732 request since it was allocated. This number is reset for each new request on
14733 the same connections in case of HTTP keep-alive. This value should usually be
14734 low and stable (around 2 calls for a typically simple request) but may become
14735 high if some processing (compression, caching or analysis) is performed. This
14736 is purely for performance monitoring purposes.
14737
14738cpu_ns_avg : integer
14739 Returns the average number of nanoseconds spent in each call to the task
14740 processing the stream or current request. This number is reset for each new
14741 request on the same connections in case of HTTP keep-alive. This value
14742 indicates the overall cost of processing the request or the connection for
14743 each call. There is no good nor bad value but the time spent in a call
14744 automatically causes latency for other processing (see lat_ns_avg below),
14745 and may affect other connection's apparent response time. Certain operations
14746 like compression, complex regex matching or heavy Lua operations may directly
14747 affect this value, and having it in the logs will make it easier to spot the
14748 faulty processing that needs to be fixed to recover decent performance.
14749 Note: this value is exactly cpu_ns_tot divided by cpu_calls.
14750
14751cpu_ns_tot : integer
14752 Returns the total number of nanoseconds spent in each call to the task
14753 processing the stream or current request. This number is reset for each new
14754 request on the same connections in case of HTTP keep-alive. This value
14755 indicates the overall cost of processing the request or the connection for
14756 each call. There is no good nor bad value but the time spent in a call
14757 automatically causes latency for other processing (see lat_ns_avg below),
14758 induces CPU costs on the machine, and may affect other connection's apparent
14759 response time. Certain operations like compression, complex regex matching or
14760 heavy Lua operations may directly affect this value, and having it in the
14761 logs will make it easier to spot the faulty processing that needs to be fixed
14762 to recover decent performance. The value may be artificially high due to a
14763 high cpu_calls count, for example when processing many HTTP chunks, and for
14764 this reason it is often preferred to log cpu_ns_avg instead.
14765
Cyril Bonté6bcd1822019-11-05 23:13:59 +010014766date([<offset>],[<unit>]) : integer
Willy Tarreau6236d3a2013-07-25 14:28:25 +020014767 Returns the current date as the epoch (number of seconds since 01/01/1970).
Damien Claisseae6f1252019-10-30 15:57:28 +000014768
14769 If an offset value is specified, then it is added to the current date before
14770 returning the value. This is particularly useful to compute relative dates,
14771 as both positive and negative offsets are allowed.
Willy Tarreau276fae92013-07-25 14:36:01 +020014772 It is useful combined with the http_date converter.
14773
Damien Claisseae6f1252019-10-30 15:57:28 +000014774 <unit> is facultative, and can be set to "s" for seconds (default behavior),
14775 "ms" for milliseconds or "us" for microseconds.
14776 If unit is set, return value is an integer reflecting either seconds,
14777 milliseconds or microseconds since epoch, plus offset.
14778 It is useful when a time resolution of less than a second is needed.
14779
Willy Tarreau276fae92013-07-25 14:36:01 +020014780 Example :
14781
14782 # set an expires header to now+1 hour in every response
14783 http-response set-header Expires %[date(3600),http_date]
Willy Tarreau6236d3a2013-07-25 14:28:25 +020014784
Damien Claisseae6f1252019-10-30 15:57:28 +000014785 # set an expires header to now+1 hour in every response, with
14786 # millisecond granularity
14787 http-response set-header Expires %[date(3600000,ms),http_date(0,ms)]
14788
Etienne Carrierea792a0a2018-01-17 13:43:24 +010014789date_us : integer
14790 Return the microseconds part of the date (the "second" part is returned by
14791 date sample). This sample is coherent with the date sample as it is comes
14792 from the same timeval structure.
14793
Willy Tarreaud716f9b2017-10-13 11:03:15 +020014794distcc_body(<token>[,<occ>]) : binary
14795 Parses a distcc message and returns the body associated to occurrence #<occ>
14796 of the token <token>. Occurrences start at 1, and when unspecified, any may
14797 match though in practice only the first one is checked for now. This can be
14798 used to extract file names or arguments in files built using distcc through
14799 haproxy. Please refer to distcc's protocol documentation for the complete
14800 list of supported tokens.
14801
14802distcc_param(<token>[,<occ>]) : integer
14803 Parses a distcc message and returns the parameter associated to occurrence
14804 #<occ> of the token <token>. Occurrences start at 1, and when unspecified,
14805 any may match though in practice only the first one is checked for now. This
14806 can be used to extract certain information such as the protocol version, the
14807 file size or the argument in files built using distcc through haproxy.
14808 Another use case consists in waiting for the start of the preprocessed file
14809 contents before connecting to the server to avoid keeping idle connections.
14810 Please refer to distcc's protocol documentation for the complete list of
14811 supported tokens.
14812
14813 Example :
14814 # wait up to 20s for the pre-processed file to be uploaded
14815 tcp-request inspect-delay 20s
14816 tcp-request content accept if { distcc_param(DOTI) -m found }
14817 # send large files to the big farm
14818 use_backend big_farm if { distcc_param(DOTI) gt 1000000 }
14819
Willy Tarreau595ec542013-06-12 21:34:28 +020014820env(<name>) : string
14821 Returns a string containing the value of environment variable <name>. As a
14822 reminder, environment variables are per-process and are sampled when the
14823 process starts. This can be useful to pass some information to a next hop
14824 server, or with ACLs to take specific action when the process is started a
14825 certain way.
14826
14827 Examples :
14828 # Pass the Via header to next hop with the local hostname in it
14829 http-request add-header Via 1.1\ %[env(HOSTNAME)]
14830
14831 # reject cookie-less requests when the STOP environment variable is set
14832 http-request deny if !{ cook(SESSIONID) -m found } { env(STOP) -m found }
14833
Willy Tarreau74ca5042013-06-11 23:12:07 +020014834fe_conn([<frontend>]) : integer
14835 Returns the number of currently established connections on the frontend,
Willy Tarreaud63335a2010-02-26 12:56:52 +010014836 possibly including the connection being evaluated. If no frontend name is
14837 specified, the current one is used. But it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020014838 frontend. It can be used to return a sorry page before hard-blocking, or to
14839 use a specific backend to drain new requests when the farm is considered
Davor Ocelice9ed2812017-12-25 17:49:28 +010014840 full. This is mostly used with ACLs but can also be used to pass some
Willy Tarreau74ca5042013-06-11 23:12:07 +020014841 statistics to servers in HTTP headers. See also the "dst_conn", "be_conn",
14842 "fe_sess_rate" fetches.
Willy Tarreaua36af912009-10-10 12:02:45 +020014843
Nenad Merdanovicad9a7e92016-10-03 04:57:37 +020014844fe_req_rate([<frontend>]) : integer
14845 Returns an integer value corresponding to the number of HTTP requests per
14846 second sent to a frontend. This number can differ from "fe_sess_rate" in
14847 situations where client-side keep-alive is enabled.
14848
Willy Tarreau74ca5042013-06-11 23:12:07 +020014849fe_sess_rate([<frontend>]) : integer
14850 Returns an integer value corresponding to the sessions creation rate on the
14851 frontend, in number of new sessions per second. This is used with ACLs to
14852 limit the incoming session rate to an acceptable range in order to prevent
14853 abuse of service at the earliest moment, for example when combined with other
14854 layer 4 ACLs in order to force the clients to wait a bit for the rate to go
14855 down below the limit. It can also be useful to add this element to logs using
14856 a log-format directive. See also the "rate-limit sessions" directive for use
14857 in frontends.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010014858
14859 Example :
14860 # This frontend limits incoming mails to 10/s with a max of 100
14861 # concurrent connections. We accept any connection below 10/s, and
14862 # force excess clients to wait for 100 ms. Since clients are limited to
14863 # 100 max, there cannot be more than 10 incoming mails per second.
14864 frontend mail
14865 bind :25
14866 mode tcp
14867 maxconn 100
14868 acl too_fast fe_sess_rate ge 10
14869 tcp-request inspect-delay 100ms
14870 tcp-request content accept if ! too_fast
14871 tcp-request content accept if WAIT_END
Willy Tarreaud72758d2010-01-12 10:42:19 +010014872
Nenad Merdanovic807a6e72017-03-12 22:00:00 +010014873hostname : string
14874 Returns the system hostname.
14875
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014876int(<integer>) : signed integer
14877 Returns a signed integer.
14878
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014879ipv4(<ipv4>) : ipv4
14880 Returns an ipv4.
14881
14882ipv6(<ipv6>) : ipv6
14883 Returns an ipv6.
14884
Willy Tarreau70fe9442018-11-22 16:07:39 +010014885lat_ns_avg : integer
14886 Returns the average number of nanoseconds spent between the moment the task
14887 handling the stream is woken up and the moment it is effectively called. This
14888 number is reset for each new request on the same connections in case of HTTP
14889 keep-alive. This value indicates the overall latency inflicted to the current
14890 request by all other requests being processed in parallel, and is a direct
14891 indicator of perceived performance due to noisy neighbours. In order to keep
14892 the value low, it is possible to reduce the scheduler's run queue depth using
14893 "tune.runqueue-depth", to reduce the number of concurrent events processed at
14894 once using "tune.maxpollevents", to decrease the stream's nice value using
14895 the "nice" option on the "bind" lines or in the frontend, or to look for
14896 other heavy requests in logs (those exhibiting large values of "cpu_ns_avg"),
14897 whose processing needs to be adjusted or fixed. Compression of large buffers
14898 could be a culprit, like heavy regex or long lists of regex.
14899 Note: this value is exactly lat_ns_tot divided by cpu_calls.
14900
14901lat_ns_tot : integer
14902 Returns the total number of nanoseconds spent between the moment the task
14903 handling the stream is woken up and the moment it is effectively called. This
14904 number is reset for each new request on the same connections in case of HTTP
14905 keep-alive. This value indicates the overall latency inflicted to the current
14906 request by all other requests being processed in parallel, and is a direct
14907 indicator of perceived performance due to noisy neighbours. In order to keep
14908 the value low, it is possible to reduce the scheduler's run queue depth using
14909 "tune.runqueue-depth", to reduce the number of concurrent events processed at
14910 once using "tune.maxpollevents", to decrease the stream's nice value using
14911 the "nice" option on the "bind" lines or in the frontend, or to look for
14912 other heavy requests in logs (those exhibiting large values of "cpu_ns_avg"),
14913 whose processing needs to be adjusted or fixed. Compression of large buffers
14914 could be a culprit, like heavy regex or long lists of regex. Note: while it
14915 may intuitively seem that the total latency adds to a transfer time, it is
14916 almost never true because while a task waits for the CPU, network buffers
14917 continue to fill up and the next call will process more at once. The value
14918 may be artificially high due to a high cpu_calls count, for example when
14919 processing many HTTP chunks, and for this reason it is often preferred to log
14920 lat_ns_avg instead, which is a more relevant performance indicator.
14921
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014922meth(<method>) : method
14923 Returns a method.
14924
Willy Tarreau0f30d262014-11-24 16:02:05 +010014925nbproc : integer
14926 Returns an integer value corresponding to the number of processes that were
14927 started (it equals the global "nbproc" setting). This is useful for logging
14928 and debugging purposes.
14929
Willy Tarreau74ca5042013-06-11 23:12:07 +020014930nbsrv([<backend>]) : integer
14931 Returns an integer value corresponding to the number of usable servers of
14932 either the current backend or the named backend. This is mostly used with
14933 ACLs but can also be useful when added to logs. This is normally used to
Willy Tarreaud63335a2010-02-26 12:56:52 +010014934 switch to an alternate backend when the number of servers is too low to
14935 to handle some load. It is useful to report a failure when combined with
14936 "monitor fail".
Willy Tarreau079ff0a2009-03-05 21:34:28 +010014937
Patrick Hemmerfabb24f2018-08-13 14:07:57 -040014938prio_class : integer
14939 Returns the priority class of the current session for http mode or connection
14940 for tcp mode. The value will be that set by the last call to "http-request
14941 set-priority-class" or "tcp-request content set-priority-class".
14942
14943prio_offset : integer
14944 Returns the priority offset of the current session for http mode or
14945 connection for tcp mode. The value will be that set by the last call to
14946 "http-request set-priority-offset" or "tcp-request content
14947 set-priority-offset".
14948
Willy Tarreau0f30d262014-11-24 16:02:05 +010014949proc : integer
14950 Returns an integer value corresponding to the position of the process calling
14951 the function, between 1 and global.nbproc. This is useful for logging and
14952 debugging purposes.
14953
Willy Tarreau74ca5042013-06-11 23:12:07 +020014954queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010014955 Returns the total number of queued connections of the designated backend,
14956 including all the connections in server queues. If no backend name is
14957 specified, the current one is used, but it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020014958 one. This is useful with ACLs or to pass statistics to backend servers. This
14959 can be used to take actions when queuing goes above a known level, generally
14960 indicating a surge of traffic or a massive slowdown on the servers. One
14961 possible action could be to reject new users but still accept old ones. See
14962 also the "avg_queue", "be_conn", and "be_sess_rate" fetches.
14963
Willy Tarreau84310e22014-02-14 11:59:04 +010014964rand([<range>]) : integer
14965 Returns a random integer value within a range of <range> possible values,
14966 starting at zero. If the range is not specified, it defaults to 2^32, which
14967 gives numbers between 0 and 4294967295. It can be useful to pass some values
14968 needed to take some routing decisions for example, or just for debugging
14969 purposes. This random must not be used for security purposes.
14970
Luca Schimweg8a694b82019-09-10 15:42:52 +020014971uuid([<version>]) : string
14972 Returns a UUID following the RFC4122 standard. If the version is not
14973 specified, a UUID version 4 (fully random) is returned.
14974 Currently, only version 4 is supported.
14975
Willy Tarreau74ca5042013-06-11 23:12:07 +020014976srv_conn([<backend>/]<server>) : integer
14977 Returns an integer value corresponding to the number of currently established
14978 connections on the designated server, possibly including the connection being
14979 evaluated. If <backend> is omitted, then the server is looked up in the
14980 current backend. It can be used to use a specific farm when one server is
14981 full, or to inform the server about our view of the number of active
Patrick Hemmer155e93e2018-06-14 18:01:35 -040014982 connections with it. See also the "fe_conn", "be_conn", "queue", and
14983 "srv_conn_free" fetch methods.
14984
14985srv_conn_free([<backend>/]<server>) : integer
14986 Returns an integer value corresponding to the number of available connections
14987 on the designated server, possibly including the connection being evaluated.
14988 The value does not include queue slots. If <backend> is omitted, then the
14989 server is looked up in the current backend. It can be used to use a specific
14990 farm when one server is full, or to inform the server about our view of the
14991 number of active connections with it. See also the "be_conn_free" and
14992 "srv_conn" fetch methods.
14993
14994 OTHER CAVEATS AND NOTES: If the server maxconn is 0, then this fetch clearly
14995 does not make sense, in which case the value returned will be -1.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014996
14997srv_is_up([<backend>/]<server>) : boolean
14998 Returns true when the designated server is UP, and false when it is either
14999 DOWN or in maintenance mode. If <backend> is omitted, then the server is
15000 looked up in the current backend. It is mainly used to take action based on
Davor Ocelice9ed2812017-12-25 17:49:28 +010015001 an external status reported via a health check (e.g. a geographical site's
Willy Tarreau74ca5042013-06-11 23:12:07 +020015002 availability). Another possible use which is more of a hack consists in
15003 using dummy servers as boolean variables that can be enabled or disabled from
15004 the CLI, so that rules depending on those ACLs can be tweaked in realtime.
15005
Willy Tarreauff2b7af2017-10-13 11:46:26 +020015006srv_queue([<backend>/]<server>) : integer
15007 Returns an integer value corresponding to the number of connections currently
15008 pending in the designated server's queue. If <backend> is omitted, then the
15009 server is looked up in the current backend. It can sometimes be used together
15010 with the "use-server" directive to force to use a known faster server when it
15011 is not much loaded. See also the "srv_conn", "avg_queue" and "queue" sample
15012 fetch methods.
15013
Willy Tarreau74ca5042013-06-11 23:12:07 +020015014srv_sess_rate([<backend>/]<server>) : integer
15015 Returns an integer corresponding to the sessions creation rate on the
15016 designated server, in number of new sessions per second. If <backend> is
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015017 omitted, then the server is looked up in the current backend. This is mostly
Willy Tarreau74ca5042013-06-11 23:12:07 +020015018 used with ACLs but can make sense with logs too. This is used to switch to an
15019 alternate backend when an expensive or fragile one reaches too high a session
Davor Ocelice9ed2812017-12-25 17:49:28 +010015020 rate, or to limit abuse of service (e.g. prevent latent requests from
Willy Tarreau74ca5042013-06-11 23:12:07 +020015021 overloading servers).
15022
15023 Example :
15024 # Redirect to a separate back
15025 acl srv1_full srv_sess_rate(be1/srv1) gt 50
15026 acl srv2_full srv_sess_rate(be1/srv2) gt 50
15027 use_backend be2 if srv1_full or srv2_full
15028
Willy Tarreau0f30d262014-11-24 16:02:05 +010015029stopping : boolean
15030 Returns TRUE if the process calling the function is currently stopping. This
15031 can be useful for logging, or for relaxing certain checks or helping close
15032 certain connections upon graceful shutdown.
15033
Thierry FOURNIERcc103292015-06-06 19:30:17 +020015034str(<string>) : string
15035 Returns a string.
15036
Willy Tarreau74ca5042013-06-11 23:12:07 +020015037table_avl([<table>]) : integer
15038 Returns the total number of available entries in the current proxy's
15039 stick-table or in the designated stick-table. See also table_cnt.
15040
15041table_cnt([<table>]) : integer
15042 Returns the total number of entries currently in use in the current proxy's
15043 stick-table or in the designated stick-table. See also src_conn_cnt and
15044 table_avl for other entry counting methods.
15045
Christopher Faulet34adb2a2017-11-21 21:45:38 +010015046thread : integer
15047 Returns an integer value corresponding to the position of the thread calling
15048 the function, between 0 and (global.nbthread-1). This is useful for logging
15049 and debugging purposes.
15050
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020015051var(<var-name>) : undefined
15052 Returns a variable with the stored type. If the variable is not set, the
Daniel Schneller0b547052016-03-21 20:46:57 +010015053 sample fetch fails. The name of the variable starts with an indication
15054 about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015055 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015056 "sess" : the variable is shared with the whole session
15057 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020015058 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010015059 "req" : the variable is shared only during request processing,
15060 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020015061 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015062 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020015063
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200150647.3.3. Fetching samples at Layer 4
Willy Tarreau74ca5042013-06-11 23:12:07 +020015065----------------------------------
15066
15067The layer 4 usually describes just the transport layer which in haproxy is
15068closest to the connection, where no content is yet made available. The fetch
15069methods described here are usable as low as the "tcp-request connection" rule
15070sets unless they require some future information. Those generally include
15071TCP/IP addresses and ports, as well as elements from stick-tables related to
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015072the incoming connection. For retrieving a value from a sticky counters, the
15073counter number can be explicitly set as 0, 1, or 2 using the pre-defined
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020015074"sc0_", "sc1_", or "sc2_" prefix. These three pre-defined prefixes can only be
15075used if MAX_SESS_STKCTR value does not exceed 3, otherwise the counter number
15076can be specified as the first integer argument when using the "sc_" prefix.
15077Starting from "sc_0" to "sc_N" where N is (MAX_SESS_STKCTR-1). An optional
15078table may be specified with the "sc*" form, in which case the currently
15079tracked key will be looked up into this alternate table instead of the table
15080currently being tracked.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015081
Jérôme Magnin35e53a62019-01-16 14:38:37 +010015082bc_http_major : integer
Jérôme Magnin86577422018-12-07 09:03:11 +010015083 Returns the backend connection's HTTP major version encoding, which may be 1
15084 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
15085 encoding and not the version present in the request header.
15086
Willy Tarreau74ca5042013-06-11 23:12:07 +020015087be_id : integer
15088 Returns an integer containing the current backend's id. It can be used in
15089 frontends with responses to check which backend processed the request.
15090
Marcin Deranekd2471c22016-12-12 14:08:05 +010015091be_name : string
15092 Returns a string containing the current backend's name. It can be used in
15093 frontends with responses to check which backend processed the request.
15094
Willy Tarreau74ca5042013-06-11 23:12:07 +020015095dst : ip
15096 This is the destination IPv4 address of the connection on the client side,
15097 which is the address the client connected to. It can be useful when running
15098 in transparent mode. It is of type IP and works on both IPv4 and IPv6 tables.
15099 On IPv6 tables, IPv4 address is mapped to its IPv6 equivalent, according to
Willy Tarreau64ded3d2019-01-23 10:02:15 +010015100 RFC 4291. When the incoming connection passed through address translation or
15101 redirection involving connection tracking, the original destination address
15102 before the redirection will be reported. On Linux systems, the source and
15103 destination may seldom appear reversed if the nf_conntrack_tcp_loose sysctl
15104 is set, because a late response may reopen a timed out connection and switch
15105 what is believed to be the source and the destination.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015106
15107dst_conn : integer
15108 Returns an integer value corresponding to the number of currently established
15109 connections on the same socket including the one being evaluated. It is
15110 normally used with ACLs but can as well be used to pass the information to
15111 servers in an HTTP header or in logs. It can be used to either return a sorry
15112 page before hard-blocking, or to use a specific backend to drain new requests
15113 when the socket is considered saturated. This offers the ability to assign
15114 different limits to different listening ports or addresses. See also the
15115 "fe_conn" and "be_conn" fetches.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015116
Willy Tarreau16e01562016-08-09 16:46:18 +020015117dst_is_local : boolean
15118 Returns true if the destination address of the incoming connection is local
15119 to the system, or false if the address doesn't exist on the system, meaning
15120 that it was intercepted in transparent mode. It can be useful to apply
15121 certain rules by default to forwarded traffic and other rules to the traffic
Davor Ocelice9ed2812017-12-25 17:49:28 +010015122 targeting the real address of the machine. For example the stats page could
Willy Tarreau16e01562016-08-09 16:46:18 +020015123 be delivered only on this address, or SSH access could be locally redirected.
15124 Please note that the check involves a few system calls, so it's better to do
15125 it only once per connection.
15126
Willy Tarreau74ca5042013-06-11 23:12:07 +020015127dst_port : integer
15128 Returns an integer value corresponding to the destination TCP port of the
15129 connection on the client side, which is the port the client connected to.
15130 This might be used when running in transparent mode, when assigning dynamic
15131 ports to some clients for a whole application session, to stick all users to
15132 a same server, or to pass the destination port information to a server using
15133 an HTTP header.
15134
Willy Tarreau60ca10a2017-08-18 15:26:54 +020015135fc_http_major : integer
15136 Reports the front connection's HTTP major version encoding, which may be 1
15137 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
15138 encoding and not on the version present in the request header.
15139
Geoff Simmons7185b782019-08-27 18:31:16 +020015140fc_pp_authority : string
15141 Returns the authority TLV sent by the client in the PROXY protocol header,
15142 if any.
15143
Tim Duesterhusd1b15b62020-03-13 12:34:23 +010015144fc_pp_unique_id : string
15145 Returns the unique ID TLV sent by the client in the PROXY protocol header,
15146 if any.
15147
Emeric Brun4f603012017-01-05 15:11:44 +010015148fc_rcvd_proxy : boolean
15149 Returns true if the client initiated the connection with a PROXY protocol
15150 header.
15151
Thierry Fournier / OZON.IO6310bef2016-07-24 20:16:50 +020015152fc_rtt(<unit>) : integer
15153 Returns the Round Trip Time (RTT) measured by the kernel for the client
15154 connection. <unit> is facultative, by default the unit is milliseconds. <unit>
15155 can be set to "ms" for milliseconds or "us" for microseconds. If the server
15156 connection is not established, if the connection is not TCP or if the
15157 operating system does not support TCP_INFO, for example Linux kernels before
15158 2.4, the sample fetch fails.
15159
15160fc_rttvar(<unit>) : integer
15161 Returns the Round Trip Time (RTT) variance measured by the kernel for the
15162 client connection. <unit> is facultative, by default the unit is milliseconds.
15163 <unit> can be set to "ms" for milliseconds or "us" for microseconds. If the
15164 server connection is not established, if the connection is not TCP or if the
15165 operating system does not support TCP_INFO, for example Linux kernels before
15166 2.4, the sample fetch fails.
15167
Christopher Fauletba0c53e2019-10-17 14:40:48 +020015168fc_unacked : integer
Joe Williams30fcd392016-08-10 07:06:44 -070015169 Returns the unacked counter measured by the kernel for the client connection.
15170 If the server connection is not established, if the connection is not TCP or
15171 if the operating system does not support TCP_INFO, for example Linux kernels
15172 before 2.4, the sample fetch fails.
15173
Christopher Fauletba0c53e2019-10-17 14:40:48 +020015174fc_sacked : integer
Joe Williams30fcd392016-08-10 07:06:44 -070015175 Returns the sacked counter measured by the kernel for the client connection.
15176 If the server connection is not established, if the connection is not TCP or
15177 if the operating system does not support TCP_INFO, for example Linux kernels
15178 before 2.4, the sample fetch fails.
15179
Christopher Fauletba0c53e2019-10-17 14:40:48 +020015180fc_retrans : integer
Joe Williams30fcd392016-08-10 07:06:44 -070015181 Returns the retransmits counter measured by the kernel for the client
15182 connection. If the server connection is not established, if the connection is
15183 not TCP or if the operating system does not support TCP_INFO, for example
15184 Linux kernels before 2.4, the sample fetch fails.
15185
Christopher Fauletba0c53e2019-10-17 14:40:48 +020015186fc_fackets : integer
Joe Williams30fcd392016-08-10 07:06:44 -070015187 Returns the fack counter measured by the kernel for the client
15188 connection. If the server connection is not established, if the connection is
15189 not TCP or if the operating system does not support TCP_INFO, for example
15190 Linux kernels before 2.4, the sample fetch fails.
15191
Christopher Fauletba0c53e2019-10-17 14:40:48 +020015192fc_lost : integer
Joe Williams30fcd392016-08-10 07:06:44 -070015193 Returns the lost counter measured by the kernel for the client
15194 connection. If the server connection is not established, if the connection is
15195 not TCP or if the operating system does not support TCP_INFO, for example
15196 Linux kernels before 2.4, the sample fetch fails.
15197
Christopher Fauletba0c53e2019-10-17 14:40:48 +020015198fc_reordering : integer
Joe Williams30fcd392016-08-10 07:06:44 -070015199 Returns the reordering counter measured by the kernel for the client
15200 connection. If the server connection is not established, if the connection is
15201 not TCP or if the operating system does not support TCP_INFO, for example
15202 Linux kernels before 2.4, the sample fetch fails.
15203
Marcin Deranek9a66dfb2018-04-13 14:37:50 +020015204fe_defbe : string
15205 Returns a string containing the frontend's default backend name. It can be
15206 used in frontends to check which backend will handle requests by default.
15207
Willy Tarreau74ca5042013-06-11 23:12:07 +020015208fe_id : integer
15209 Returns an integer containing the current frontend's id. It can be used in
Marcin Deranek6e413ed2016-12-13 12:40:01 +010015210 backends to check from which frontend it was called, or to stick all users
Willy Tarreau74ca5042013-06-11 23:12:07 +020015211 coming via a same frontend to the same server.
15212
Marcin Deranekd2471c22016-12-12 14:08:05 +010015213fe_name : string
15214 Returns a string containing the current frontend's name. It can be used in
15215 backends to check from which frontend it was called, or to stick all users
15216 coming via a same frontend to the same server.
15217
Cyril Bonté62ba8702014-04-22 23:52:25 +020015218sc_bytes_in_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015219sc0_bytes_in_rate([<table>]) : integer
15220sc1_bytes_in_rate([<table>]) : integer
15221sc2_bytes_in_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020015222 Returns the average client-to-server bytes rate from the currently tracked
15223 counters, measured in amount of bytes over the period configured in the
15224 table. See also src_bytes_in_rate.
15225
Cyril Bonté62ba8702014-04-22 23:52:25 +020015226sc_bytes_out_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015227sc0_bytes_out_rate([<table>]) : integer
15228sc1_bytes_out_rate([<table>]) : integer
15229sc2_bytes_out_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020015230 Returns the average server-to-client bytes rate from the currently tracked
15231 counters, measured in amount of bytes over the period configured in the
15232 table. See also src_bytes_out_rate.
15233
Cyril Bonté62ba8702014-04-22 23:52:25 +020015234sc_clr_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015235sc0_clr_gpc0([<table>]) : integer
15236sc1_clr_gpc0([<table>]) : integer
15237sc2_clr_gpc0([<table>]) : integer
Willy Tarreauf73cd112011-08-13 01:45:16 +020015238 Clears the first General Purpose Counter associated to the currently tracked
15239 counters, and returns its previous value. Before the first invocation, the
Willy Tarreau869948b2013-01-04 14:14:57 +010015240 stored value is zero, so first invocation will always return zero. This is
15241 typically used as a second ACL in an expression in order to mark a connection
15242 when a first ACL was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020015243
Jarno Huuskonen676f6222017-03-30 09:19:45 +030015244 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020015245 # block if 5 consecutive requests continue to come faster than 10 sess
15246 # per second, and reset the counter as soon as the traffic slows down.
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020015247 acl abuse sc0_http_req_rate gt 10
15248 acl kill sc0_inc_gpc0 gt 5
15249 acl save sc0_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020015250 tcp-request connection accept if !abuse save
15251 tcp-request connection reject if abuse kill
15252
Frédéric Lécaille6778b272018-01-29 15:22:53 +010015253sc_clr_gpc1(<ctr>[,<table>]) : integer
15254sc0_clr_gpc1([<table>]) : integer
15255sc1_clr_gpc1([<table>]) : integer
15256sc2_clr_gpc1([<table>]) : integer
15257 Clears the second General Purpose Counter associated to the currently tracked
15258 counters, and returns its previous value. Before the first invocation, the
15259 stored value is zero, so first invocation will always return zero. This is
15260 typically used as a second ACL in an expression in order to mark a connection
15261 when a first ACL was verified.
15262
Cyril Bonté62ba8702014-04-22 23:52:25 +020015263sc_conn_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015264sc0_conn_cnt([<table>]) : integer
15265sc1_conn_cnt([<table>]) : integer
15266sc2_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010015267 Returns the cumulative number of incoming connections from currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020015268 counters. See also src_conn_cnt.
15269
Cyril Bonté62ba8702014-04-22 23:52:25 +020015270sc_conn_cur(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015271sc0_conn_cur([<table>]) : integer
15272sc1_conn_cur([<table>]) : integer
15273sc2_conn_cur([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020015274 Returns the current amount of concurrent connections tracking the same
15275 tracked counters. This number is automatically incremented when tracking
15276 begins and decremented when tracking stops. See also src_conn_cur.
15277
Cyril Bonté62ba8702014-04-22 23:52:25 +020015278sc_conn_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015279sc0_conn_rate([<table>]) : integer
15280sc1_conn_rate([<table>]) : integer
15281sc2_conn_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020015282 Returns the average connection rate from the currently tracked counters,
15283 measured in amount of connections over the period configured in the table.
15284 See also src_conn_rate.
15285
Cyril Bonté62ba8702014-04-22 23:52:25 +020015286sc_get_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015287sc0_get_gpc0([<table>]) : integer
15288sc1_get_gpc0([<table>]) : integer
15289sc2_get_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020015290 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015291 currently tracked counters. See also src_get_gpc0 and sc/sc0/sc1/sc2_inc_gpc0.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020015292
Frédéric Lécaille6778b272018-01-29 15:22:53 +010015293sc_get_gpc1(<ctr>[,<table>]) : integer
15294sc0_get_gpc1([<table>]) : integer
15295sc1_get_gpc1([<table>]) : integer
15296sc2_get_gpc1([<table>]) : integer
15297 Returns the value of the second General Purpose Counter associated to the
15298 currently tracked counters. See also src_get_gpc1 and sc/sc0/sc1/sc2_inc_gpc1.
15299
Thierry FOURNIER236657b2015-08-19 08:25:14 +020015300sc_get_gpt0(<ctr>[,<table>]) : integer
15301sc0_get_gpt0([<table>]) : integer
15302sc1_get_gpt0([<table>]) : integer
15303sc2_get_gpt0([<table>]) : integer
15304 Returns the value of the first General Purpose Tag associated to the
15305 currently tracked counters. See also src_get_gpt0.
15306
Cyril Bonté62ba8702014-04-22 23:52:25 +020015307sc_gpc0_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015308sc0_gpc0_rate([<table>]) : integer
15309sc1_gpc0_rate([<table>]) : integer
15310sc2_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020015311 Returns the average increment rate of the first General Purpose Counter
15312 associated to the currently tracked counters. It reports the frequency
15313 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015314 src_gpc0_rate, sc/sc0/sc1/sc2_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
15315 that the "gpc0_rate" counter must be stored in the stick-table for a value to
15316 be returned, as "gpc0" only holds the event count.
Willy Tarreaue9656522010-08-17 15:40:09 +020015317
Frédéric Lécaille6778b272018-01-29 15:22:53 +010015318sc_gpc1_rate(<ctr>[,<table>]) : integer
15319sc0_gpc1_rate([<table>]) : integer
15320sc1_gpc1_rate([<table>]) : integer
15321sc2_gpc1_rate([<table>]) : integer
15322 Returns the average increment rate of the second General Purpose Counter
15323 associated to the currently tracked counters. It reports the frequency
15324 which the gpc1 counter was incremented over the configured period. See also
15325 src_gpcA_rate, sc/sc0/sc1/sc2_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
15326 that the "gpc1_rate" counter must be stored in the stick-table for a value to
15327 be returned, as "gpc1" only holds the event count.
15328
Cyril Bonté62ba8702014-04-22 23:52:25 +020015329sc_http_err_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015330sc0_http_err_cnt([<table>]) : integer
15331sc1_http_err_cnt([<table>]) : integer
15332sc2_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010015333 Returns the cumulative number of HTTP errors from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020015334 counters. This includes the both request errors and 4xx error responses.
15335 See also src_http_err_cnt.
15336
Cyril Bonté62ba8702014-04-22 23:52:25 +020015337sc_http_err_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015338sc0_http_err_rate([<table>]) : integer
15339sc1_http_err_rate([<table>]) : integer
15340sc2_http_err_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020015341 Returns the average rate of HTTP errors from the currently tracked counters,
15342 measured in amount of errors over the period configured in the table. This
15343 includes the both request errors and 4xx error responses. See also
15344 src_http_err_rate.
15345
Cyril Bonté62ba8702014-04-22 23:52:25 +020015346sc_http_req_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015347sc0_http_req_cnt([<table>]) : integer
15348sc1_http_req_cnt([<table>]) : integer
15349sc2_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010015350 Returns the cumulative number of HTTP requests from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020015351 counters. This includes every started request, valid or not. See also
15352 src_http_req_cnt.
15353
Cyril Bonté62ba8702014-04-22 23:52:25 +020015354sc_http_req_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015355sc0_http_req_rate([<table>]) : integer
15356sc1_http_req_rate([<table>]) : integer
15357sc2_http_req_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020015358 Returns the average rate of HTTP requests from the currently tracked
15359 counters, measured in amount of requests over the period configured in
15360 the table. This includes every started request, valid or not. See also
15361 src_http_req_rate.
15362
Cyril Bonté62ba8702014-04-22 23:52:25 +020015363sc_inc_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015364sc0_inc_gpc0([<table>]) : integer
15365sc1_inc_gpc0([<table>]) : integer
15366sc2_inc_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020015367 Increments the first General Purpose Counter associated to the currently
Willy Tarreau869948b2013-01-04 14:14:57 +010015368 tracked counters, and returns its new value. Before the first invocation,
15369 the stored value is zero, so first invocation will increase it to 1 and will
15370 return 1. This is typically used as a second ACL in an expression in order
15371 to mark a connection when a first ACL was verified :
Willy Tarreaue9656522010-08-17 15:40:09 +020015372
Jarno Huuskonen676f6222017-03-30 09:19:45 +030015373 Example:
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020015374 acl abuse sc0_http_req_rate gt 10
15375 acl kill sc0_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020015376 tcp-request connection reject if abuse kill
15377
Frédéric Lécaille6778b272018-01-29 15:22:53 +010015378sc_inc_gpc1(<ctr>[,<table>]) : integer
15379sc0_inc_gpc1([<table>]) : integer
15380sc1_inc_gpc1([<table>]) : integer
15381sc2_inc_gpc1([<table>]) : integer
15382 Increments the second General Purpose Counter associated to the currently
15383 tracked counters, and returns its new value. Before the first invocation,
15384 the stored value is zero, so first invocation will increase it to 1 and will
15385 return 1. This is typically used as a second ACL in an expression in order
15386 to mark a connection when a first ACL was verified.
15387
Cyril Bonté62ba8702014-04-22 23:52:25 +020015388sc_kbytes_in(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015389sc0_kbytes_in([<table>]) : integer
15390sc1_kbytes_in([<table>]) : integer
15391sc2_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020015392 Returns the total amount of client-to-server data from the currently tracked
15393 counters, measured in kilobytes. The test is currently performed on 32-bit
15394 integers, which limits values to 4 terabytes. See also src_kbytes_in.
Willy Tarreaue9656522010-08-17 15:40:09 +020015395
Cyril Bonté62ba8702014-04-22 23:52:25 +020015396sc_kbytes_out(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015397sc0_kbytes_out([<table>]) : integer
15398sc1_kbytes_out([<table>]) : integer
15399sc2_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020015400 Returns the total amount of server-to-client data from the currently tracked
15401 counters, measured in kilobytes. The test is currently performed on 32-bit
15402 integers, which limits values to 4 terabytes. See also src_kbytes_out.
Willy Tarreaue9656522010-08-17 15:40:09 +020015403
Cyril Bonté62ba8702014-04-22 23:52:25 +020015404sc_sess_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015405sc0_sess_cnt([<table>]) : integer
15406sc1_sess_cnt([<table>]) : integer
15407sc2_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010015408 Returns the cumulative number of incoming connections that were transformed
Willy Tarreaue9656522010-08-17 15:40:09 +020015409 into sessions, which means that they were accepted by a "tcp-request
15410 connection" rule, from the currently tracked counters. A backend may count
15411 more sessions than connections because each connection could result in many
Jamie Gloudonaaa21002012-08-25 00:18:33 -040015412 backend sessions if some HTTP keep-alive is performed over the connection
Willy Tarreaue9656522010-08-17 15:40:09 +020015413 with the client. See also src_sess_cnt.
15414
Cyril Bonté62ba8702014-04-22 23:52:25 +020015415sc_sess_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015416sc0_sess_rate([<table>]) : integer
15417sc1_sess_rate([<table>]) : integer
15418sc2_sess_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020015419 Returns the average session rate from the currently tracked counters,
15420 measured in amount of sessions over the period configured in the table. A
15421 session is a connection that got past the early "tcp-request connection"
15422 rules. A backend may count more sessions than connections because each
15423 connection could result in many backend sessions if some HTTP keep-alive is
Jamie Gloudonaaa21002012-08-25 00:18:33 -040015424 performed over the connection with the client. See also src_sess_rate.
Willy Tarreaue9656522010-08-17 15:40:09 +020015425
Cyril Bonté62ba8702014-04-22 23:52:25 +020015426sc_tracked(<ctr>[,<table>]) : boolean
Willy Tarreau0f791d42013-07-23 19:56:43 +020015427sc0_tracked([<table>]) : boolean
15428sc1_tracked([<table>]) : boolean
15429sc2_tracked([<table>]) : boolean
Willy Tarreau6f1615f2013-06-03 15:15:22 +020015430 Returns true if the designated session counter is currently being tracked by
15431 the current session. This can be useful when deciding whether or not we want
15432 to set some values in a header passed to the server.
15433
Cyril Bonté62ba8702014-04-22 23:52:25 +020015434sc_trackers(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015435sc0_trackers([<table>]) : integer
15436sc1_trackers([<table>]) : integer
15437sc2_trackers([<table>]) : integer
Willy Tarreau2406db42012-12-09 12:16:43 +010015438 Returns the current amount of concurrent connections tracking the same
15439 tracked counters. This number is automatically incremented when tracking
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020015440 begins and decremented when tracking stops. It differs from sc0_conn_cur in
Willy Tarreau2406db42012-12-09 12:16:43 +010015441 that it does not rely on any stored information but on the table's reference
15442 count (the "use" value which is returned by "show table" on the CLI). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020015443 may sometimes be more suited for layer7 tracking. It can be used to tell a
15444 server how many concurrent connections there are from a given address for
15445 example.
Willy Tarreau2406db42012-12-09 12:16:43 +010015446
Willy Tarreau74ca5042013-06-11 23:12:07 +020015447so_id : integer
15448 Returns an integer containing the current listening socket's id. It is useful
15449 in frontends involving many "bind" lines, or to stick all users coming via a
15450 same socket to the same server.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015451
Willy Tarreau74ca5042013-06-11 23:12:07 +020015452src : ip
Davor Ocelice9ed2812017-12-25 17:49:28 +010015453 This is the source IPv4 address of the client of the session. It is of type
Willy Tarreau74ca5042013-06-11 23:12:07 +020015454 IP and works on both IPv4 and IPv6 tables. On IPv6 tables, IPv4 addresses are
15455 mapped to their IPv6 equivalent, according to RFC 4291. Note that it is the
15456 TCP-level source address which is used, and not the address of a client
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010015457 behind a proxy. However if the "accept-proxy" or "accept-netscaler-cip" bind
15458 directive is used, it can be the address of a client behind another
15459 PROXY-protocol compatible component for all rule sets except
Willy Tarreau64ded3d2019-01-23 10:02:15 +010015460 "tcp-request connection" which sees the real address. When the incoming
15461 connection passed through address translation or redirection involving
15462 connection tracking, the original destination address before the redirection
15463 will be reported. On Linux systems, the source and destination may seldom
15464 appear reversed if the nf_conntrack_tcp_loose sysctl is set, because a late
15465 response may reopen a timed out connection and switch what is believed to be
15466 the source and the destination.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015467
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010015468 Example:
15469 # add an HTTP header in requests with the originating address' country
15470 http-request set-header X-Country %[src,map_ip(geoip.lst)]
15471
Willy Tarreau74ca5042013-06-11 23:12:07 +020015472src_bytes_in_rate([<table>]) : integer
15473 Returns the average bytes rate from the incoming connection's source address
15474 in the current proxy's stick-table or in the designated stick-table, measured
15475 in amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015476 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_in_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015477
Willy Tarreau74ca5042013-06-11 23:12:07 +020015478src_bytes_out_rate([<table>]) : integer
15479 Returns the average bytes rate to the incoming connection's source address in
15480 the current proxy's stick-table or in the designated stick-table, measured in
Willy Tarreauc9705a12010-07-27 20:05:50 +020015481 amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015482 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_out_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015483
Willy Tarreau74ca5042013-06-11 23:12:07 +020015484src_clr_gpc0([<table>]) : integer
15485 Clears the first General Purpose Counter associated to the incoming
15486 connection's source address in the current proxy's stick-table or in the
15487 designated stick-table, and returns its previous value. If the address is not
15488 found, an entry is created and 0 is returned. This is typically used as a
15489 second ACL in an expression in order to mark a connection when a first ACL
15490 was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020015491
Jarno Huuskonen676f6222017-03-30 09:19:45 +030015492 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020015493 # block if 5 consecutive requests continue to come faster than 10 sess
15494 # per second, and reset the counter as soon as the traffic slows down.
15495 acl abuse src_http_req_rate gt 10
15496 acl kill src_inc_gpc0 gt 5
Willy Tarreau869948b2013-01-04 14:14:57 +010015497 acl save src_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020015498 tcp-request connection accept if !abuse save
15499 tcp-request connection reject if abuse kill
15500
Frédéric Lécaille6778b272018-01-29 15:22:53 +010015501src_clr_gpc1([<table>]) : integer
15502 Clears the second General Purpose Counter associated to the incoming
15503 connection's source address in the current proxy's stick-table or in the
15504 designated stick-table, and returns its previous value. If the address is not
15505 found, an entry is created and 0 is returned. This is typically used as a
15506 second ACL in an expression in order to mark a connection when a first ACL
15507 was verified.
15508
Willy Tarreau74ca5042013-06-11 23:12:07 +020015509src_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010015510 Returns the cumulative number of connections initiated from the current
Willy Tarreau74ca5042013-06-11 23:12:07 +020015511 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020015512 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015513 See also sc/sc0/sc1/sc2_conn_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015514
Willy Tarreau74ca5042013-06-11 23:12:07 +020015515src_conn_cur([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020015516 Returns the current amount of concurrent connections initiated from the
Willy Tarreau74ca5042013-06-11 23:12:07 +020015517 current incoming connection's source address in the current proxy's
15518 stick-table or in the designated stick-table. If the address is not found,
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015519 zero is returned. See also sc/sc0/sc1/sc2_conn_cur.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015520
Willy Tarreau74ca5042013-06-11 23:12:07 +020015521src_conn_rate([<table>]) : integer
15522 Returns the average connection rate from the incoming connection's source
15523 address in the current proxy's stick-table or in the designated stick-table,
15524 measured in amount of connections over the period configured in the table. If
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015525 the address is not found, zero is returned. See also sc/sc0/sc1/sc2_conn_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015526
Willy Tarreau74ca5042013-06-11 23:12:07 +020015527src_get_gpc0([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020015528 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau74ca5042013-06-11 23:12:07 +020015529 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020015530 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015531 See also sc/sc0/sc1/sc2_get_gpc0 and src_inc_gpc0.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015532
Frédéric Lécaille6778b272018-01-29 15:22:53 +010015533src_get_gpc1([<table>]) : integer
15534 Returns the value of the second General Purpose Counter associated to the
15535 incoming connection's source address in the current proxy's stick-table or in
15536 the designated stick-table. If the address is not found, zero is returned.
15537 See also sc/sc0/sc1/sc2_get_gpc1 and src_inc_gpc1.
15538
Thierry FOURNIER236657b2015-08-19 08:25:14 +020015539src_get_gpt0([<table>]) : integer
15540 Returns the value of the first General Purpose Tag associated to the
15541 incoming connection's source address in the current proxy's stick-table or in
15542 the designated stick-table. If the address is not found, zero is returned.
15543 See also sc/sc0/sc1/sc2_get_gpt0.
15544
Willy Tarreau74ca5042013-06-11 23:12:07 +020015545src_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020015546 Returns the average increment rate of the first General Purpose Counter
Willy Tarreau74ca5042013-06-11 23:12:07 +020015547 associated to the incoming connection's source address in the current proxy's
Willy Tarreauba2ffd12013-05-29 15:54:14 +020015548 stick-table or in the designated stick-table. It reports the frequency
15549 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015550 sc/sc0/sc1/sc2_gpc0_rate, src_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
15551 that the "gpc0_rate" counter must be stored in the stick-table for a value to
15552 be returned, as "gpc0" only holds the event count.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020015553
Frédéric Lécaille6778b272018-01-29 15:22:53 +010015554src_gpc1_rate([<table>]) : integer
15555 Returns the average increment rate of the second General Purpose Counter
15556 associated to the incoming connection's source address in the current proxy's
15557 stick-table or in the designated stick-table. It reports the frequency
15558 which the gpc1 counter was incremented over the configured period. See also
15559 sc/sc0/sc1/sc2_gpc1_rate, src_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
15560 that the "gpc1_rate" counter must be stored in the stick-table for a value to
15561 be returned, as "gpc1" only holds the event count.
15562
Willy Tarreau74ca5042013-06-11 23:12:07 +020015563src_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010015564 Returns the cumulative number of HTTP errors from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020015565 source address in the current proxy's stick-table or in the designated
Willy Tarreauc9705a12010-07-27 20:05:50 +020015566 stick-table. This includes the both request errors and 4xx error responses.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015567 See also sc/sc0/sc1/sc2_http_err_cnt. If the address is not found, zero is
Willy Tarreau74ca5042013-06-11 23:12:07 +020015568 returned.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015569
Willy Tarreau74ca5042013-06-11 23:12:07 +020015570src_http_err_rate([<table>]) : integer
15571 Returns the average rate of HTTP errors from the incoming connection's source
15572 address in the current proxy's stick-table or in the designated stick-table,
15573 measured in amount of errors over the period configured in the table. This
15574 includes the both request errors and 4xx error responses. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015575 not found, zero is returned. See also sc/sc0/sc1/sc2_http_err_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015576
Willy Tarreau74ca5042013-06-11 23:12:07 +020015577src_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010015578 Returns the cumulative number of HTTP requests from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020015579 source address in the current proxy's stick-table or in the designated stick-
15580 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015581 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015582
Willy Tarreau74ca5042013-06-11 23:12:07 +020015583src_http_req_rate([<table>]) : integer
15584 Returns the average rate of HTTP requests from the incoming connection's
15585 source address in the current proxy's stick-table or in the designated stick-
15586 table, measured in amount of requests over the period configured in the
Willy Tarreauc9705a12010-07-27 20:05:50 +020015587 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015588 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015589
Willy Tarreau74ca5042013-06-11 23:12:07 +020015590src_inc_gpc0([<table>]) : integer
15591 Increments the first General Purpose Counter associated to the incoming
15592 connection's source address in the current proxy's stick-table or in the
15593 designated stick-table, and returns its new value. If the address is not
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020015594 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc0.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015595 This is typically used as a second ACL in an expression in order to mark a
15596 connection when a first ACL was verified :
Willy Tarreauc9705a12010-07-27 20:05:50 +020015597
Jarno Huuskonen676f6222017-03-30 09:19:45 +030015598 Example:
Willy Tarreauc9705a12010-07-27 20:05:50 +020015599 acl abuse src_http_req_rate gt 10
Willy Tarreau869948b2013-01-04 14:14:57 +010015600 acl kill src_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020015601 tcp-request connection reject if abuse kill
Willy Tarreauc9705a12010-07-27 20:05:50 +020015602
Frédéric Lécaille6778b272018-01-29 15:22:53 +010015603src_inc_gpc1([<table>]) : integer
15604 Increments the second General Purpose Counter associated to the incoming
15605 connection's source address in the current proxy's stick-table or in the
15606 designated stick-table, and returns its new value. If the address is not
15607 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc1.
15608 This is typically used as a second ACL in an expression in order to mark a
15609 connection when a first ACL was verified.
15610
Willy Tarreau16e01562016-08-09 16:46:18 +020015611src_is_local : boolean
15612 Returns true if the source address of the incoming connection is local to the
15613 system, or false if the address doesn't exist on the system, meaning that it
15614 comes from a remote machine. Note that UNIX addresses are considered local.
15615 It can be useful to apply certain access restrictions based on where the
Davor Ocelice9ed2812017-12-25 17:49:28 +010015616 client comes from (e.g. require auth or https for remote machines). Please
Willy Tarreau16e01562016-08-09 16:46:18 +020015617 note that the check involves a few system calls, so it's better to do it only
15618 once per connection.
15619
Willy Tarreau74ca5042013-06-11 23:12:07 +020015620src_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020015621 Returns the total amount of data received from the incoming connection's
15622 source address in the current proxy's stick-table or in the designated
15623 stick-table, measured in kilobytes. If the address is not found, zero is
15624 returned. The test is currently performed on 32-bit integers, which limits
15625 values to 4 terabytes. See also sc/sc0/sc1/sc2_kbytes_in.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015626
Willy Tarreau74ca5042013-06-11 23:12:07 +020015627src_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020015628 Returns the total amount of data sent to the incoming connection's source
15629 address in the current proxy's stick-table or in the designated stick-table,
15630 measured in kilobytes. If the address is not found, zero is returned. The
15631 test is currently performed on 32-bit integers, which limits values to 4
15632 terabytes. See also sc/sc0/sc1/sc2_kbytes_out.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020015633
Willy Tarreau74ca5042013-06-11 23:12:07 +020015634src_port : integer
15635 Returns an integer value corresponding to the TCP source port of the
15636 connection on the client side, which is the port the client connected from.
15637 Usage of this function is very limited as modern protocols do not care much
15638 about source ports nowadays.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010015639
Willy Tarreau74ca5042013-06-11 23:12:07 +020015640src_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010015641 Returns the cumulative number of connections initiated from the incoming
Willy Tarreauc9705a12010-07-27 20:05:50 +020015642 connection's source IPv4 address in the current proxy's stick-table or in the
15643 designated stick-table, that were transformed into sessions, which means that
15644 they were accepted by "tcp-request" rules. If the address is not found, zero
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015645 is returned. See also sc/sc0/sc1/sc2_sess_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015646
Willy Tarreau74ca5042013-06-11 23:12:07 +020015647src_sess_rate([<table>]) : integer
15648 Returns the average session rate from the incoming connection's source
15649 address in the current proxy's stick-table or in the designated stick-table,
15650 measured in amount of sessions over the period configured in the table. A
15651 session is a connection that went past the early "tcp-request" rules. If the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015652 address is not found, zero is returned. See also sc/sc0/sc1/sc2_sess_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015653
Willy Tarreau74ca5042013-06-11 23:12:07 +020015654src_updt_conn_cnt([<table>]) : integer
15655 Creates or updates the entry associated to the incoming connection's source
15656 address in the current proxy's stick-table or in the designated stick-table.
15657 This table must be configured to store the "conn_cnt" data type, otherwise
15658 the match will be ignored. The current count is incremented by one, and the
15659 expiration timer refreshed. The updated count is returned, so this match
15660 can't return zero. This was used to reject service abusers based on their
15661 source address. Note: it is recommended to use the more complete "track-sc*"
15662 actions in "tcp-request" rules instead.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020015663
15664 Example :
15665 # This frontend limits incoming SSH connections to 3 per 10 second for
15666 # each source address, and rejects excess connections until a 10 second
15667 # silence is observed. At most 20 addresses are tracked.
15668 listen ssh
15669 bind :22
15670 mode tcp
15671 maxconn 100
Willy Tarreauc9705a12010-07-27 20:05:50 +020015672 stick-table type ip size 20 expire 10s store conn_cnt
Willy Tarreau74ca5042013-06-11 23:12:07 +020015673 tcp-request content reject if { src_updt_conn_cnt gt 3 }
Willy Tarreaua975b8f2010-06-05 19:13:27 +020015674 server local 127.0.0.1:22
15675
Willy Tarreau74ca5042013-06-11 23:12:07 +020015676srv_id : integer
15677 Returns an integer containing the server's id when processing the response.
15678 While it's almost only used with ACLs, it may be used for logging or
15679 debugging.
Hervé COMMOWICKdaa824e2011-08-05 12:09:44 +020015680
vkill1dfd1652019-10-30 16:58:14 +080015681srv_name : string
15682 Returns a string containing the server's name when processing the response.
15683 While it's almost only used with ACLs, it may be used for logging or
15684 debugging.
15685
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200156867.3.4. Fetching samples at Layer 5
Willy Tarreau74ca5042013-06-11 23:12:07 +020015687----------------------------------
Willy Tarreau0b1cd942010-05-16 22:18:27 +020015688
Willy Tarreau74ca5042013-06-11 23:12:07 +020015689The layer 5 usually describes just the session layer which in haproxy is
15690closest to the session once all the connection handshakes are finished, but
15691when no content is yet made available. The fetch methods described here are
15692usable as low as the "tcp-request content" rule sets unless they require some
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015693future information. Those generally include the results of SSL negotiations.
Willy Tarreauc735a072011-03-29 00:57:02 +020015694
Ben Shillitof25e8e52016-12-02 14:25:37 +00001569551d.all(<prop>[,<prop>*]) : string
15696 Returns values for the properties requested as a string, where values are
15697 separated by the delimiter specified with "51degrees-property-separator".
15698 The device is identified using all the important HTTP headers from the
15699 request. The function can be passed up to five property names, and if a
15700 property name can't be found, the value "NoData" is returned.
15701
15702 Example :
15703 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request
15704 # containing the three properties requested using all relevant headers from
15705 # the request.
15706 frontend http-in
15707 bind *:8081
15708 default_backend servers
15709 http-request set-header X-51D-DeviceTypeMobileTablet \
15710 %[51d.all(DeviceType,IsMobile,IsTablet)]
15711
Emeric Brun645ae792014-04-30 14:21:06 +020015712ssl_bc : boolean
15713 Returns true when the back connection was made via an SSL/TLS transport
15714 layer and is locally deciphered. This means the outgoing connection was made
15715 other a server with the "ssl" option.
15716
15717ssl_bc_alg_keysize : integer
15718 Returns the symmetric cipher key size supported in bits when the outgoing
15719 connection was made over an SSL/TLS transport layer.
15720
Olivier Houchard6b77f492018-11-22 18:18:29 +010015721ssl_bc_alpn : string
15722 This extracts the Application Layer Protocol Negotiation field from an
15723 outgoing connection made via a TLS transport layer.
Michael Prokop4438c602019-05-24 10:25:45 +020015724 The result is a string containing the protocol name negotiated with the
Olivier Houchard6b77f492018-11-22 18:18:29 +010015725 server. The SSL library must have been built with support for TLS
15726 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
15727 not advertised unless the "alpn" keyword on the "server" line specifies a
15728 protocol list. Also, nothing forces the server to pick a protocol from this
15729 list, any other one may be requested. The TLS ALPN extension is meant to
15730 replace the TLS NPN extension. See also "ssl_bc_npn".
15731
Emeric Brun645ae792014-04-30 14:21:06 +020015732ssl_bc_cipher : string
15733 Returns the name of the used cipher when the outgoing connection was made
15734 over an SSL/TLS transport layer.
15735
Patrick Hemmer65674662019-06-04 08:13:03 -040015736ssl_bc_client_random : binary
15737 Returns the client random of the back connection when the incoming connection
15738 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
15739 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
15740
Emeric Brun74f7ffa2018-02-19 16:14:12 +010015741ssl_bc_is_resumed : boolean
15742 Returns true when the back connection was made over an SSL/TLS transport
15743 layer and the newly created SSL session was resumed using a cached
15744 session or a TLS ticket.
15745
Olivier Houchard6b77f492018-11-22 18:18:29 +010015746ssl_bc_npn : string
15747 This extracts the Next Protocol Negotiation field from an outgoing connection
15748 made via a TLS transport layer. The result is a string containing the
Michael Prokop4438c602019-05-24 10:25:45 +020015749 protocol name negotiated with the server . The SSL library must have been
Olivier Houchard6b77f492018-11-22 18:18:29 +010015750 built with support for TLS extensions enabled (check haproxy -vv). Note that
15751 the TLS NPN extension is not advertised unless the "npn" keyword on the
15752 "server" line specifies a protocol list. Also, nothing forces the server to
15753 pick a protocol from this list, any other one may be used. Please note that
15754 the TLS NPN extension was replaced with ALPN.
15755
Emeric Brun645ae792014-04-30 14:21:06 +020015756ssl_bc_protocol : string
15757 Returns the name of the used protocol when the outgoing connection was made
15758 over an SSL/TLS transport layer.
15759
Emeric Brunb73a9b02014-04-30 18:49:19 +020015760ssl_bc_unique_id : binary
Emeric Brun645ae792014-04-30 14:21:06 +020015761 When the outgoing connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020015762 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
15763 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
Emeric Brun645ae792014-04-30 14:21:06 +020015764
Patrick Hemmer65674662019-06-04 08:13:03 -040015765ssl_bc_server_random : binary
15766 Returns the server random of the back connection when the incoming connection
15767 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
15768 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
15769
Emeric Brun645ae792014-04-30 14:21:06 +020015770ssl_bc_session_id : binary
15771 Returns the SSL ID of the back connection when the outgoing connection was
15772 made over an SSL/TLS transport layer. It is useful to log if we want to know
15773 if session was reused or not.
15774
Patrick Hemmere0275472018-04-28 19:15:51 -040015775ssl_bc_session_key : binary
15776 Returns the SSL session master key of the back connection when the outgoing
15777 connection was made over an SSL/TLS transport layer. It is useful to decrypt
15778 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
15779 BoringSSL.
15780
Emeric Brun645ae792014-04-30 14:21:06 +020015781ssl_bc_use_keysize : integer
15782 Returns the symmetric cipher key size used in bits when the outgoing
15783 connection was made over an SSL/TLS transport layer.
15784
Willy Tarreau74ca5042013-06-11 23:12:07 +020015785ssl_c_ca_err : integer
15786 When the incoming connection was made over an SSL/TLS transport layer,
15787 returns the ID of the first error detected during verification of the client
15788 certificate at depth > 0, or 0 if no error was encountered during this
15789 verification process. Please refer to your SSL library's documentation to
15790 find the exhaustive list of error codes.
Willy Tarreauc735a072011-03-29 00:57:02 +020015791
Willy Tarreau74ca5042013-06-11 23:12:07 +020015792ssl_c_ca_err_depth : integer
15793 When the incoming connection was made over an SSL/TLS transport layer,
15794 returns the depth in the CA chain of the first error detected during the
15795 verification of the client certificate. If no error is encountered, 0 is
15796 returned.
Willy Tarreau0ba27502007-12-24 16:55:16 +010015797
Emeric Brun43e79582014-10-29 19:03:26 +010015798ssl_c_der : binary
15799 Returns the DER formatted certificate presented by the client when the
15800 incoming connection was made over an SSL/TLS transport layer. When used for
15801 an ACL, the value(s) to match against can be passed in hexadecimal form.
15802
Willy Tarreau74ca5042013-06-11 23:12:07 +020015803ssl_c_err : integer
15804 When the incoming connection was made over an SSL/TLS transport layer,
15805 returns the ID of the first error detected during verification at depth 0, or
15806 0 if no error was encountered during this verification process. Please refer
15807 to your SSL library's documentation to find the exhaustive list of error
15808 codes.
Willy Tarreau62644772008-07-16 18:36:06 +020015809
Elliot Otchet71f82972020-01-15 08:12:14 -050015810ssl_c_i_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020015811 When the incoming connection was made over an SSL/TLS transport layer,
15812 returns the full distinguished name of the issuer of the certificate
15813 presented by the client when no <entry> is specified, or the value of the
15814 first given entry found from the beginning of the DN. If a positive/negative
15815 occurrence number is specified as the optional second argument, it returns
15816 the value of the nth given entry value from the beginning/end of the DN.
15817 For instance, "ssl_c_i_dn(OU,2)" the second organization unit, and
15818 "ssl_c_i_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050015819 The <format> parameter allows you to receive the DN suitable for
15820 consumption by different protocols. Currently supported is rfc2253 for
15821 LDAP v3.
15822 If you'd like to modify the format only you can specify an empty string
15823 and zero for the first two parameters. Example: ssl_c_i_dn(,0,rfc2253)
Willy Tarreau62644772008-07-16 18:36:06 +020015824
Willy Tarreau74ca5042013-06-11 23:12:07 +020015825ssl_c_key_alg : string
15826 Returns the name of the algorithm used to generate the key of the certificate
15827 presented by the client when the incoming connection was made over an SSL/TLS
15828 transport layer.
Willy Tarreau62644772008-07-16 18:36:06 +020015829
Willy Tarreau74ca5042013-06-11 23:12:07 +020015830ssl_c_notafter : string
15831 Returns the end date presented by the client as a formatted string
15832 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15833 transport layer.
Emeric Brunbede3d02009-06-30 17:54:00 +020015834
Willy Tarreau74ca5042013-06-11 23:12:07 +020015835ssl_c_notbefore : string
15836 Returns the start date presented by the client as a formatted string
15837 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15838 transport layer.
Willy Tarreaub6672b52011-12-12 17:23:41 +010015839
Elliot Otchet71f82972020-01-15 08:12:14 -050015840ssl_c_s_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020015841 When the incoming connection was made over an SSL/TLS transport layer,
15842 returns the full distinguished name of the subject of the certificate
15843 presented by the client when no <entry> is specified, or the value of the
15844 first given entry found from the beginning of the DN. If a positive/negative
15845 occurrence number is specified as the optional second argument, it returns
15846 the value of the nth given entry value from the beginning/end of the DN.
15847 For instance, "ssl_c_s_dn(OU,2)" the second organization unit, and
15848 "ssl_c_s_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050015849 The <format> parameter allows you to receive the DN suitable for
15850 consumption by different protocols. Currently supported is rfc2253 for
15851 LDAP v3.
15852 If you'd like to modify the format only you can specify an empty string
15853 and zero for the first two parameters. Example: ssl_c_s_dn(,0,rfc2253)
Willy Tarreaub6672b52011-12-12 17:23:41 +010015854
Willy Tarreau74ca5042013-06-11 23:12:07 +020015855ssl_c_serial : binary
15856 Returns the serial of the certificate presented by the client when the
15857 incoming connection was made over an SSL/TLS transport layer. When used for
15858 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun2525b6b2012-10-18 15:59:43 +020015859
Willy Tarreau74ca5042013-06-11 23:12:07 +020015860ssl_c_sha1 : binary
15861 Returns the SHA-1 fingerprint of the certificate presented by the client when
15862 the incoming connection was made over an SSL/TLS transport layer. This can be
15863 used to stick a client to a server, or to pass this information to a server.
Willy Tarreau2d0caa32014-07-02 19:01:22 +020015864 Note that the output is binary, so if you want to pass that signature to the
15865 server, you need to encode it in hex or base64, such as in the example below:
15866
Jarno Huuskonen676f6222017-03-30 09:19:45 +030015867 Example:
Willy Tarreau2d0caa32014-07-02 19:01:22 +020015868 http-request set-header X-SSL-Client-SHA1 %[ssl_c_sha1,hex]
Emeric Brun2525b6b2012-10-18 15:59:43 +020015869
Willy Tarreau74ca5042013-06-11 23:12:07 +020015870ssl_c_sig_alg : string
15871 Returns the name of the algorithm used to sign the certificate presented by
15872 the client when the incoming connection was made over an SSL/TLS transport
15873 layer.
Emeric Brun87855892012-10-17 17:39:35 +020015874
Willy Tarreau74ca5042013-06-11 23:12:07 +020015875ssl_c_used : boolean
15876 Returns true if current SSL session uses a client certificate even if current
15877 connection uses SSL session resumption. See also "ssl_fc_has_crt".
Emeric Brun7f56e742012-10-19 18:15:40 +020015878
Willy Tarreau74ca5042013-06-11 23:12:07 +020015879ssl_c_verify : integer
15880 Returns the verify result error ID when the incoming connection was made over
15881 an SSL/TLS transport layer, otherwise zero if no error is encountered. Please
15882 refer to your SSL library's documentation for an exhaustive list of error
15883 codes.
Emeric Brunce5ad802012-10-22 14:11:22 +020015884
Willy Tarreau74ca5042013-06-11 23:12:07 +020015885ssl_c_version : integer
15886 Returns the version of the certificate presented by the client when the
15887 incoming connection was made over an SSL/TLS transport layer.
Emeric Brunce5ad802012-10-22 14:11:22 +020015888
Emeric Brun43e79582014-10-29 19:03:26 +010015889ssl_f_der : binary
15890 Returns the DER formatted certificate presented by the frontend when the
15891 incoming connection was made over an SSL/TLS transport layer. When used for
15892 an ACL, the value(s) to match against can be passed in hexadecimal form.
15893
Elliot Otchet71f82972020-01-15 08:12:14 -050015894ssl_f_i_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020015895 When the incoming connection was made over an SSL/TLS transport layer,
15896 returns the full distinguished name of the issuer of the certificate
15897 presented by the frontend when no <entry> is specified, or the value of the
15898 first given entry found from the beginning of the DN. If a positive/negative
Emeric Brun87855892012-10-17 17:39:35 +020015899 occurrence number is specified as the optional second argument, it returns
Willy Tarreau74ca5042013-06-11 23:12:07 +020015900 the value of the nth given entry value from the beginning/end of the DN.
15901 For instance, "ssl_f_i_dn(OU,2)" the second organization unit, and
15902 "ssl_f_i_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050015903 The <format> parameter allows you to receive the DN suitable for
15904 consumption by different protocols. Currently supported is rfc2253 for
15905 LDAP v3.
15906 If you'd like to modify the format only you can specify an empty string
15907 and zero for the first two parameters. Example: ssl_f_i_dn(,0,rfc2253)
Emeric Brun87855892012-10-17 17:39:35 +020015908
Willy Tarreau74ca5042013-06-11 23:12:07 +020015909ssl_f_key_alg : string
15910 Returns the name of the algorithm used to generate the key of the certificate
15911 presented by the frontend when the incoming connection was made over an
15912 SSL/TLS transport layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020015913
Willy Tarreau74ca5042013-06-11 23:12:07 +020015914ssl_f_notafter : string
15915 Returns the end date presented by the frontend as a formatted string
15916 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15917 transport layer.
Emeric Brun2525b6b2012-10-18 15:59:43 +020015918
Willy Tarreau74ca5042013-06-11 23:12:07 +020015919ssl_f_notbefore : string
15920 Returns the start date presented by the frontend as a formatted string
15921 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15922 transport layer.
Emeric Brun87855892012-10-17 17:39:35 +020015923
Elliot Otchet71f82972020-01-15 08:12:14 -050015924ssl_f_s_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020015925 When the incoming connection was made over an SSL/TLS transport layer,
15926 returns the full distinguished name of the subject of the certificate
15927 presented by the frontend when no <entry> is specified, or the value of the
15928 first given entry found from the beginning of the DN. If a positive/negative
15929 occurrence number is specified as the optional second argument, it returns
15930 the value of the nth given entry value from the beginning/end of the DN.
15931 For instance, "ssl_f_s_dn(OU,2)" the second organization unit, and
15932 "ssl_f_s_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050015933 The <format> parameter allows you to receive the DN suitable for
15934 consumption by different protocols. Currently supported is rfc2253 for
15935 LDAP v3.
15936 If you'd like to modify the format only you can specify an empty string
15937 and zero for the first two parameters. Example: ssl_f_s_dn(,0,rfc2253)
Emeric Brunce5ad802012-10-22 14:11:22 +020015938
Willy Tarreau74ca5042013-06-11 23:12:07 +020015939ssl_f_serial : binary
15940 Returns the serial of the certificate presented by the frontend when the
15941 incoming connection was made over an SSL/TLS transport layer. When used for
15942 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun87855892012-10-17 17:39:35 +020015943
Emeric Brun55f4fa82014-04-30 17:11:25 +020015944ssl_f_sha1 : binary
15945 Returns the SHA-1 fingerprint of the certificate presented by the frontend
15946 when the incoming connection was made over an SSL/TLS transport layer. This
15947 can be used to know which certificate was chosen using SNI.
15948
Willy Tarreau74ca5042013-06-11 23:12:07 +020015949ssl_f_sig_alg : string
15950 Returns the name of the algorithm used to sign the certificate presented by
15951 the frontend when the incoming connection was made over an SSL/TLS transport
15952 layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020015953
Willy Tarreau74ca5042013-06-11 23:12:07 +020015954ssl_f_version : integer
15955 Returns the version of the certificate presented by the frontend when the
15956 incoming connection was made over an SSL/TLS transport layer.
15957
15958ssl_fc : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020015959 Returns true when the front connection was made via an SSL/TLS transport
15960 layer and is locally deciphered. This means it has matched a socket declared
15961 with a "bind" line having the "ssl" option.
15962
Willy Tarreau74ca5042013-06-11 23:12:07 +020015963 Example :
15964 # This passes "X-Proto: https" to servers when client connects over SSL
15965 listen http-https
15966 bind :80
15967 bind :443 ssl crt /etc/haproxy.pem
15968 http-request add-header X-Proto https if { ssl_fc }
15969
15970ssl_fc_alg_keysize : integer
15971 Returns the symmetric cipher key size supported in bits when the incoming
15972 connection was made over an SSL/TLS transport layer.
15973
15974ssl_fc_alpn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015975 This extracts the Application Layer Protocol Negotiation field from an
Willy Tarreau74ca5042013-06-11 23:12:07 +020015976 incoming connection made via a TLS transport layer and locally deciphered by
15977 haproxy. The result is a string containing the protocol name advertised by
15978 the client. The SSL library must have been built with support for TLS
15979 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
15980 not advertised unless the "alpn" keyword on the "bind" line specifies a
15981 protocol list. Also, nothing forces the client to pick a protocol from this
15982 list, any other one may be requested. The TLS ALPN extension is meant to
15983 replace the TLS NPN extension. See also "ssl_fc_npn".
15984
Willy Tarreau74ca5042013-06-11 23:12:07 +020015985ssl_fc_cipher : string
15986 Returns the name of the used cipher when the incoming connection was made
15987 over an SSL/TLS transport layer.
Willy Tarreauab861d32013-04-02 02:30:41 +020015988
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015989ssl_fc_cipherlist_bin : binary
15990 Returns the binary form of the client hello cipher list. The maximum returned
15991 value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010015992 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015993
15994ssl_fc_cipherlist_hex : string
15995 Returns the binary form of the client hello cipher list encoded as
15996 hexadecimal. The maximum returned value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010015997 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015998
15999ssl_fc_cipherlist_str : string
16000 Returns the decoded text form of the client hello cipher list. The maximum
16001 number of ciphers returned is according with the value of
16002 "tune.ssl.capture-cipherlist-size". Note that this sample-fetch is only
Davor Ocelice9ed2812017-12-25 17:49:28 +010016003 available with OpenSSL >= 1.0.2. If the function is not enabled, this
Emmanuel Hocdetddcde192017-09-01 17:32:08 +020016004 sample-fetch returns the hash like "ssl_fc_cipherlist_xxh".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010016005
16006ssl_fc_cipherlist_xxh : integer
16007 Returns a xxh64 of the cipher list. This hash can be return only is the value
16008 "tune.ssl.capture-cipherlist-size" is set greater than 0, however the hash
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010016009 take in account all the data of the cipher list.
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010016010
Patrick Hemmer65674662019-06-04 08:13:03 -040016011ssl_fc_client_random : binary
16012 Returns the client random of the front connection when the incoming connection
16013 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
16014 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
16015
Willy Tarreau74ca5042013-06-11 23:12:07 +020016016ssl_fc_has_crt : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020016017 Returns true if a client certificate is present in an incoming connection over
16018 SSL/TLS transport layer. Useful if 'verify' statement is set to 'optional'.
Emeric Brun9143d372012-12-20 15:44:16 +010016019 Note: on SSL session resumption with Session ID or TLS ticket, client
16020 certificate is not present in the current connection but may be retrieved
16021 from the cache or the ticket. So prefer "ssl_c_used" if you want to check if
16022 current SSL session uses a client certificate.
Emeric Brun2525b6b2012-10-18 15:59:43 +020016023
Olivier Houchardccaa7de2017-10-02 11:51:03 +020016024ssl_fc_has_early : boolean
16025 Returns true if early data were sent, and the handshake didn't happen yet. As
16026 it has security implications, it is useful to be able to refuse those, or
16027 wait until the handshake happened.
16028
Willy Tarreau74ca5042013-06-11 23:12:07 +020016029ssl_fc_has_sni : boolean
16030 This checks for the presence of a Server Name Indication TLS extension (SNI)
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020016031 in an incoming connection was made over an SSL/TLS transport layer. Returns
16032 true when the incoming connection presents a TLS SNI field. This requires
John Roeslerfb2fce12019-07-10 15:45:51 -050016033 that the SSL library is built with support for TLS extensions enabled (check
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020016034 haproxy -vv).
Willy Tarreau7875d092012-09-10 08:20:03 +020016035
Nenad Merdanovic1516fe32016-05-17 03:31:21 +020016036ssl_fc_is_resumed : boolean
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020016037 Returns true if the SSL/TLS session has been resumed through the use of
Jérôme Magnin4a326cb2018-01-15 14:01:17 +010016038 SSL session cache or TLS tickets on an incoming connection over an SSL/TLS
16039 transport layer.
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020016040
Willy Tarreau74ca5042013-06-11 23:12:07 +020016041ssl_fc_npn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016042 This extracts the Next Protocol Negotiation field from an incoming connection
Willy Tarreau74ca5042013-06-11 23:12:07 +020016043 made via a TLS transport layer and locally deciphered by haproxy. The result
16044 is a string containing the protocol name advertised by the client. The SSL
16045 library must have been built with support for TLS extensions enabled (check
16046 haproxy -vv). Note that the TLS NPN extension is not advertised unless the
16047 "npn" keyword on the "bind" line specifies a protocol list. Also, nothing
16048 forces the client to pick a protocol from this list, any other one may be
16049 requested. Please note that the TLS NPN extension was replaced with ALPN.
Willy Tarreaua33c6542012-10-15 13:19:06 +020016050
Willy Tarreau74ca5042013-06-11 23:12:07 +020016051ssl_fc_protocol : string
16052 Returns the name of the used protocol when the incoming connection was made
16053 over an SSL/TLS transport layer.
Willy Tarreau7875d092012-09-10 08:20:03 +020016054
Emeric Brunb73a9b02014-04-30 18:49:19 +020016055ssl_fc_unique_id : binary
David Sc1ad52e2014-04-08 18:48:47 -040016056 When the incoming connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020016057 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
16058 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
David Sc1ad52e2014-04-08 18:48:47 -040016059
Patrick Hemmer65674662019-06-04 08:13:03 -040016060ssl_fc_server_random : binary
16061 Returns the server random of the front connection when the incoming connection
16062 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
16063 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
16064
Willy Tarreau74ca5042013-06-11 23:12:07 +020016065ssl_fc_session_id : binary
16066 Returns the SSL ID of the front connection when the incoming connection was
16067 made over an SSL/TLS transport layer. It is useful to stick a given client to
16068 a server. It is important to note that some browsers refresh their session ID
16069 every few minutes.
Willy Tarreau7875d092012-09-10 08:20:03 +020016070
Patrick Hemmere0275472018-04-28 19:15:51 -040016071ssl_fc_session_key : binary
16072 Returns the SSL session master key of the front connection when the incoming
16073 connection was made over an SSL/TLS transport layer. It is useful to decrypt
16074 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
16075 BoringSSL.
16076
16077
Willy Tarreau74ca5042013-06-11 23:12:07 +020016078ssl_fc_sni : string
16079 This extracts the Server Name Indication TLS extension (SNI) field from an
16080 incoming connection made via an SSL/TLS transport layer and locally
16081 deciphered by haproxy. The result (when present) typically is a string
16082 matching the HTTPS host name (253 chars or less). The SSL library must have
16083 been built with support for TLS extensions enabled (check haproxy -vv).
16084
16085 This fetch is different from "req_ssl_sni" above in that it applies to the
16086 connection being deciphered by haproxy and not to SSL contents being blindly
16087 forwarded. See also "ssl_fc_sni_end" and "ssl_fc_sni_reg" below. This
John Roeslerfb2fce12019-07-10 15:45:51 -050016088 requires that the SSL library is built with support for TLS extensions
Cyril Bonté9c1eb1e2012-10-09 22:45:34 +020016089 enabled (check haproxy -vv).
Willy Tarreau62644772008-07-16 18:36:06 +020016090
Willy Tarreau74ca5042013-06-11 23:12:07 +020016091 ACL derivatives :
Willy Tarreau74ca5042013-06-11 23:12:07 +020016092 ssl_fc_sni_end : suffix match
16093 ssl_fc_sni_reg : regex match
Emeric Brun589fcad2012-10-16 14:13:26 +020016094
Willy Tarreau74ca5042013-06-11 23:12:07 +020016095ssl_fc_use_keysize : integer
16096 Returns the symmetric cipher key size used in bits when the incoming
16097 connection was made over an SSL/TLS transport layer.
Willy Tarreaub6fb4202008-07-20 11:18:28 +020016098
Willy Tarreaub6fb4202008-07-20 11:18:28 +020016099
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200161007.3.5. Fetching samples from buffer contents (Layer 6)
Willy Tarreau74ca5042013-06-11 23:12:07 +020016101------------------------------------------------------
Willy Tarreaub6fb4202008-07-20 11:18:28 +020016102
Willy Tarreau74ca5042013-06-11 23:12:07 +020016103Fetching samples from buffer contents is a bit different from the previous
16104sample fetches above because the sampled data are ephemeral. These data can
16105only be used when they're available and will be lost when they're forwarded.
16106For this reason, samples fetched from buffer contents during a request cannot
16107be used in a response for example. Even while the data are being fetched, they
16108can change. Sometimes it is necessary to set some delays or combine multiple
16109sample fetch methods to ensure that the expected data are complete and usable,
16110for example through TCP request content inspection. Please see the "tcp-request
16111content" keyword for more detailed information on the subject.
Willy Tarreau62644772008-07-16 18:36:06 +020016112
Willy Tarreau74ca5042013-06-11 23:12:07 +020016113payload(<offset>,<length>) : binary (deprecated)
Davor Ocelice9ed2812017-12-25 17:49:28 +010016114 This is an alias for "req.payload" when used in the context of a request (e.g.
Willy Tarreau74ca5042013-06-11 23:12:07 +020016115 "stick on", "stick match"), and for "res.payload" when used in the context of
16116 a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010016117
Willy Tarreau74ca5042013-06-11 23:12:07 +020016118payload_lv(<offset1>,<length>[,<offset2>]) : binary (deprecated)
16119 This is an alias for "req.payload_lv" when used in the context of a request
Davor Ocelice9ed2812017-12-25 17:49:28 +010016120 (e.g. "stick on", "stick match"), and for "res.payload_lv" when used in the
Willy Tarreau74ca5042013-06-11 23:12:07 +020016121 context of a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010016122
Thierry FOURNIERd7d88812017-04-19 15:15:14 +020016123req.hdrs : string
16124 Returns the current request headers as string including the last empty line
16125 separating headers from the request body. The last empty line can be used to
16126 detect a truncated header block. This sample fetch is useful for some SPOE
16127 headers analyzers and for advanced logging.
16128
Thierry FOURNIER5617dce2017-04-09 05:38:19 +020016129req.hdrs_bin : binary
16130 Returns the current request headers contained in preparsed binary form. This
16131 is useful for offloading some processing with SPOE. Each string is described
16132 by a length followed by the number of bytes indicated in the length. The
16133 length is represented using the variable integer encoding detailed in the
16134 SPOE documentation. The end of the list is marked by a couple of empty header
16135 names and values (length of 0 for both).
16136
16137 *(<str:header-name><str:header-value>)<empty string><empty string>
16138
16139 int: refer to the SPOE documentation for the encoding
16140 str: <int:length><bytes>
16141
Willy Tarreau74ca5042013-06-11 23:12:07 +020016142req.len : integer
16143req_len : integer (deprecated)
16144 Returns an integer value corresponding to the number of bytes present in the
16145 request buffer. This is mostly used in ACL. It is important to understand
16146 that this test does not return false as long as the buffer is changing. This
16147 means that a check with equality to zero will almost always immediately match
16148 at the beginning of the session, while a test for more data will wait for
16149 that data to come in and return false only when haproxy is certain that no
16150 more data will come in. This test was designed to be used with TCP request
16151 content inspection.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020016152
Willy Tarreau74ca5042013-06-11 23:12:07 +020016153req.payload(<offset>,<length>) : binary
16154 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020016155 in the request buffer. As a special case, if the <length> argument is zero,
16156 the the whole buffer from <offset> to the end is extracted. This can be used
16157 with ACLs in order to check for the presence of some content in a buffer at
16158 any location.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020016159
Willy Tarreau74ca5042013-06-11 23:12:07 +020016160 ACL alternatives :
16161 payload(<offset>,<length>) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020016162
Willy Tarreau74ca5042013-06-11 23:12:07 +020016163req.payload_lv(<offset1>,<length>[,<offset2>]) : binary
16164 This extracts a binary block whose size is specified at <offset1> for <length>
16165 bytes, and which starts at <offset2> if specified or just after the length in
16166 the request buffer. The <offset2> parameter also supports relative offsets if
16167 prepended with a '+' or '-' sign.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020016168
Willy Tarreau74ca5042013-06-11 23:12:07 +020016169 ACL alternatives :
16170 payload_lv(<offset1>,<length>[,<offset2>]) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020016171
Willy Tarreau74ca5042013-06-11 23:12:07 +020016172 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020016173
Willy Tarreau74ca5042013-06-11 23:12:07 +020016174req.proto_http : boolean
16175req_proto_http : boolean (deprecated)
16176 Returns true when data in the request buffer look like HTTP and correctly
16177 parses as such. It is the same parser as the common HTTP request parser which
16178 is used so there should be no surprises. The test does not match until the
16179 request is complete, failed or timed out. This test may be used to report the
16180 protocol in TCP logs, but the biggest use is to block TCP request analysis
16181 until a complete HTTP request is present in the buffer, for example to track
16182 a header.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020016183
Willy Tarreau74ca5042013-06-11 23:12:07 +020016184 Example:
16185 # track request counts per "base" (concatenation of Host+URL)
16186 tcp-request inspect-delay 10s
16187 tcp-request content reject if !HTTP
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020016188 tcp-request content track-sc0 base table req-rate
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020016189
Willy Tarreau74ca5042013-06-11 23:12:07 +020016190req.rdp_cookie([<name>]) : string
16191rdp_cookie([<name>]) : string (deprecated)
16192 When the request buffer looks like the RDP protocol, extracts the RDP cookie
16193 <name>, or any cookie if unspecified. The parser only checks for the first
16194 cookie, as illustrated in the RDP protocol specification. The cookie name is
16195 case insensitive. Generally the "MSTS" cookie name will be used, as it can
16196 contain the user name of the client connecting to the server if properly
16197 configured on the client. The "MSTSHASH" cookie is often used as well for
16198 session stickiness to servers.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020016199
Willy Tarreau74ca5042013-06-11 23:12:07 +020016200 This differs from "balance rdp-cookie" in that any balancing algorithm may be
16201 used and thus the distribution of clients to backend servers is not linked to
16202 a hash of the RDP cookie. It is envisaged that using a balancing algorithm
16203 such as "balance roundrobin" or "balance leastconn" will lead to a more even
16204 distribution of clients to backend servers than the hash used by "balance
16205 rdp-cookie".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020016206
Willy Tarreau74ca5042013-06-11 23:12:07 +020016207 ACL derivatives :
16208 req_rdp_cookie([<name>]) : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020016209
Willy Tarreau74ca5042013-06-11 23:12:07 +020016210 Example :
16211 listen tse-farm
16212 bind 0.0.0.0:3389
16213 # wait up to 5s for an RDP cookie in the request
16214 tcp-request inspect-delay 5s
16215 tcp-request content accept if RDP_COOKIE
16216 # apply RDP cookie persistence
16217 persist rdp-cookie
16218 # Persist based on the mstshash cookie
16219 # This is only useful makes sense if
16220 # balance rdp-cookie is not used
16221 stick-table type string size 204800
16222 stick on req.rdp_cookie(mstshash)
16223 server srv1 1.1.1.1:3389
16224 server srv1 1.1.1.2:3389
Willy Tarreau04aa6a92012-04-06 18:57:55 +020016225
Willy Tarreau74ca5042013-06-11 23:12:07 +020016226 See also : "balance rdp-cookie", "persist rdp-cookie", "tcp-request" and the
16227 "req_rdp_cookie" ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020016228
Willy Tarreau74ca5042013-06-11 23:12:07 +020016229req.rdp_cookie_cnt([name]) : integer
16230rdp_cookie_cnt([name]) : integer (deprecated)
16231 Tries to parse the request buffer as RDP protocol, then returns an integer
16232 corresponding to the number of RDP cookies found. If an optional cookie name
16233 is passed, only cookies matching this name are considered. This is mostly
16234 used in ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020016235
Willy Tarreau74ca5042013-06-11 23:12:07 +020016236 ACL derivatives :
16237 req_rdp_cookie_cnt([<name>]) : integer match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020016238
Alex Zorin4afdd132018-12-30 13:56:28 +110016239req.ssl_alpn : string
16240 Returns a string containing the values of the Application-Layer Protocol
16241 Negotiation (ALPN) TLS extension (RFC7301), sent by the client within the SSL
16242 ClientHello message. Note that this only applies to raw contents found in the
16243 request buffer and not to the contents deciphered via an SSL data layer, so
16244 this will not work with "bind" lines having the "ssl" option. This is useful
16245 in ACL to make a routing decision based upon the ALPN preferences of a TLS
Jarno Huuskonene504f812019-01-03 07:56:49 +020016246 client, like in the example below. See also "ssl_fc_alpn".
Alex Zorin4afdd132018-12-30 13:56:28 +110016247
16248 Examples :
16249 # Wait for a client hello for at most 5 seconds
16250 tcp-request inspect-delay 5s
16251 tcp-request content accept if { req_ssl_hello_type 1 }
Jarno Huuskonene504f812019-01-03 07:56:49 +020016252 use_backend bk_acme if { req.ssl_alpn acme-tls/1 }
Alex Zorin4afdd132018-12-30 13:56:28 +110016253 default_backend bk_default
16254
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020016255req.ssl_ec_ext : boolean
16256 Returns a boolean identifying if client sent the Supported Elliptic Curves
16257 Extension as defined in RFC4492, section 5.1. within the SSL ClientHello
Cyril Bonté307ee1e2015-09-28 23:16:06 +020016258 message. This can be used to present ECC compatible clients with EC
16259 certificate and to use RSA for all others, on the same IP address. Note that
16260 this only applies to raw contents found in the request buffer and not to
16261 contents deciphered via an SSL data layer, so this will not work with "bind"
16262 lines having the "ssl" option.
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020016263
Willy Tarreau74ca5042013-06-11 23:12:07 +020016264req.ssl_hello_type : integer
16265req_ssl_hello_type : integer (deprecated)
16266 Returns an integer value containing the type of the SSL hello message found
16267 in the request buffer if the buffer contains data that parse as a complete
16268 SSL (v3 or superior) client hello message. Note that this only applies to raw
16269 contents found in the request buffer and not to contents deciphered via an
16270 SSL data layer, so this will not work with "bind" lines having the "ssl"
16271 option. This is mostly used in ACL to detect presence of an SSL hello message
16272 that is supposed to contain an SSL session ID usable for stickiness.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020016273
Willy Tarreau74ca5042013-06-11 23:12:07 +020016274req.ssl_sni : string
16275req_ssl_sni : string (deprecated)
16276 Returns a string containing the value of the Server Name TLS extension sent
16277 by a client in a TLS stream passing through the request buffer if the buffer
16278 contains data that parse as a complete SSL (v3 or superior) client hello
16279 message. Note that this only applies to raw contents found in the request
16280 buffer and not to contents deciphered via an SSL data layer, so this will not
16281 work with "bind" lines having the "ssl" option. SNI normally contains the
16282 name of the host the client tries to connect to (for recent browsers). SNI is
16283 useful for allowing or denying access to certain hosts when SSL/TLS is used
16284 by the client. This test was designed to be used with TCP request content
16285 inspection. If content switching is needed, it is recommended to first wait
16286 for a complete client hello (type 1), like in the example below. See also
16287 "ssl_fc_sni".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020016288
Willy Tarreau74ca5042013-06-11 23:12:07 +020016289 ACL derivatives :
16290 req_ssl_sni : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020016291
Willy Tarreau74ca5042013-06-11 23:12:07 +020016292 Examples :
16293 # Wait for a client hello for at most 5 seconds
16294 tcp-request inspect-delay 5s
16295 tcp-request content accept if { req_ssl_hello_type 1 }
16296 use_backend bk_allow if { req_ssl_sni -f allowed_sites }
16297 default_backend bk_sorry_page
Willy Tarreau04aa6a92012-04-06 18:57:55 +020016298
Pradeep Jindalbb2acf52015-09-29 10:12:57 +053016299req.ssl_st_ext : integer
16300 Returns 0 if the client didn't send a SessionTicket TLS Extension (RFC5077)
16301 Returns 1 if the client sent SessionTicket TLS Extension
16302 Returns 2 if the client also sent non-zero length TLS SessionTicket
16303 Note that this only applies to raw contents found in the request buffer and
16304 not to contents deciphered via an SSL data layer, so this will not work with
16305 "bind" lines having the "ssl" option. This can for example be used to detect
16306 whether the client sent a SessionTicket or not and stick it accordingly, if
16307 no SessionTicket then stick on SessionID or don't stick as there's no server
16308 side state is there when SessionTickets are in use.
16309
Willy Tarreau74ca5042013-06-11 23:12:07 +020016310req.ssl_ver : integer
16311req_ssl_ver : integer (deprecated)
16312 Returns an integer value containing the version of the SSL/TLS protocol of a
16313 stream present in the request buffer. Both SSLv2 hello messages and SSLv3
16314 messages are supported. TLSv1 is announced as SSL version 3.1. The value is
16315 composed of the major version multiplied by 65536, added to the minor
16316 version. Note that this only applies to raw contents found in the request
16317 buffer and not to contents deciphered via an SSL data layer, so this will not
16318 work with "bind" lines having the "ssl" option. The ACL version of the test
Davor Ocelice9ed2812017-12-25 17:49:28 +010016319 matches against a decimal notation in the form MAJOR.MINOR (e.g. 3.1). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020016320 fetch is mostly used in ACL.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016321
Willy Tarreau74ca5042013-06-11 23:12:07 +020016322 ACL derivatives :
16323 req_ssl_ver : decimal match
Willy Tarreaud63335a2010-02-26 12:56:52 +010016324
Willy Tarreau47e8eba2013-09-11 23:28:46 +020016325res.len : integer
16326 Returns an integer value corresponding to the number of bytes present in the
16327 response buffer. This is mostly used in ACL. It is important to understand
16328 that this test does not return false as long as the buffer is changing. This
16329 means that a check with equality to zero will almost always immediately match
16330 at the beginning of the session, while a test for more data will wait for
16331 that data to come in and return false only when haproxy is certain that no
16332 more data will come in. This test was designed to be used with TCP response
16333 content inspection.
16334
Willy Tarreau74ca5042013-06-11 23:12:07 +020016335res.payload(<offset>,<length>) : binary
16336 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020016337 in the response buffer. As a special case, if the <length> argument is zero,
16338 the the whole buffer from <offset> to the end is extracted. This can be used
16339 with ACLs in order to check for the presence of some content in a buffer at
16340 any location.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016341
Willy Tarreau74ca5042013-06-11 23:12:07 +020016342res.payload_lv(<offset1>,<length>[,<offset2>]) : binary
16343 This extracts a binary block whose size is specified at <offset1> for <length>
16344 bytes, and which starts at <offset2> if specified or just after the length in
16345 the response buffer. The <offset2> parameter also supports relative offsets
16346 if prepended with a '+' or '-' sign.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016347
Willy Tarreau74ca5042013-06-11 23:12:07 +020016348 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016349
Willy Tarreau971f7b62015-09-29 14:06:59 +020016350res.ssl_hello_type : integer
16351rep_ssl_hello_type : integer (deprecated)
16352 Returns an integer value containing the type of the SSL hello message found
16353 in the response buffer if the buffer contains data that parses as a complete
16354 SSL (v3 or superior) hello message. Note that this only applies to raw
16355 contents found in the response buffer and not to contents deciphered via an
16356 SSL data layer, so this will not work with "server" lines having the "ssl"
16357 option. This is mostly used in ACL to detect presence of an SSL hello message
16358 that is supposed to contain an SSL session ID usable for stickiness.
16359
Willy Tarreau74ca5042013-06-11 23:12:07 +020016360wait_end : boolean
16361 This fetch either returns true when the inspection period is over, or does
16362 not fetch. It is only used in ACLs, in conjunction with content analysis to
Davor Ocelice9ed2812017-12-25 17:49:28 +010016363 avoid returning a wrong verdict early. It may also be used to delay some
Willy Tarreau74ca5042013-06-11 23:12:07 +020016364 actions, such as a delayed reject for some special addresses. Since it either
16365 stops the rules evaluation or immediately returns true, it is recommended to
Davor Ocelice9ed2812017-12-25 17:49:28 +010016366 use this acl as the last one in a rule. Please note that the default ACL
Willy Tarreau74ca5042013-06-11 23:12:07 +020016367 "WAIT_END" is always usable without prior declaration. This test was designed
16368 to be used with TCP request content inspection.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016369
Willy Tarreau74ca5042013-06-11 23:12:07 +020016370 Examples :
16371 # delay every incoming request by 2 seconds
16372 tcp-request inspect-delay 2s
16373 tcp-request content accept if WAIT_END
Willy Tarreaud63335a2010-02-26 12:56:52 +010016374
Willy Tarreau74ca5042013-06-11 23:12:07 +020016375 # don't immediately tell bad guys they are rejected
16376 tcp-request inspect-delay 10s
16377 acl goodguys src 10.0.0.0/24
16378 acl badguys src 10.0.1.0/24
16379 tcp-request content accept if goodguys
16380 tcp-request content reject if badguys WAIT_END
16381 tcp-request content reject
16382
16383
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200163847.3.6. Fetching HTTP samples (Layer 7)
Willy Tarreau74ca5042013-06-11 23:12:07 +020016385--------------------------------------
16386
16387It is possible to fetch samples from HTTP contents, requests and responses.
16388This application layer is also called layer 7. It is only possible to fetch the
16389data in this section when a full HTTP request or response has been parsed from
16390its respective request or response buffer. This is always the case with all
16391HTTP specific rules and for sections running with "mode http". When using TCP
16392content inspection, it may be necessary to support an inspection delay in order
16393to let the request or response come in first. These fetches may require a bit
16394more CPU resources than the layer 4 ones, but not much since the request and
16395response are indexed.
16396
16397base : string
16398 This returns the concatenation of the first Host header and the path part of
16399 the request, which starts at the first slash and ends before the question
16400 mark. It can be useful in virtual hosted environments to detect URL abuses as
16401 well as to improve shared caches efficiency. Using this with a limited size
16402 stick table also allows one to collect statistics about most commonly
16403 requested objects by host/path. With ACLs it can allow simple content
16404 switching rules involving the host and the path at the same time, such as
16405 "www.example.com/favicon.ico". See also "path" and "uri".
16406
16407 ACL derivatives :
16408 base : exact string match
16409 base_beg : prefix match
16410 base_dir : subdir match
16411 base_dom : domain match
16412 base_end : suffix match
16413 base_len : length match
16414 base_reg : regex match
16415 base_sub : substring match
16416
16417base32 : integer
16418 This returns a 32-bit hash of the value returned by the "base" fetch method
16419 above. This is useful to track per-URL activity on high traffic sites without
16420 having to store all URLs. Instead a shorter hash is stored, saving a lot of
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020016421 memory. The output type is an unsigned integer. The hash function used is
16422 SDBM with full avalanche on the output. Technically, base32 is exactly equal
16423 to "base,sdbm(1)".
Willy Tarreau74ca5042013-06-11 23:12:07 +020016424
16425base32+src : binary
16426 This returns the concatenation of the base32 fetch above and the src fetch
16427 below. The resulting type is of type binary, with a size of 8 or 20 bytes
16428 depending on the source address family. This can be used to track per-IP,
16429 per-URL counters.
16430
William Lallemand65ad6e12014-01-31 15:08:02 +010016431capture.req.hdr(<idx>) : string
16432 This extracts the content of the header captured by the "capture request
16433 header", idx is the position of the capture keyword in the configuration.
16434 The first entry is an index of 0. See also: "capture request header".
16435
16436capture.req.method : string
16437 This extracts the METHOD of an HTTP request. It can be used in both request
16438 and response. Unlike "method", it can be used in both request and response
16439 because it's allocated.
16440
16441capture.req.uri : string
16442 This extracts the request's URI, which starts at the first slash and ends
16443 before the first space in the request (without the host part). Unlike "path"
16444 and "url", it can be used in both request and response because it's
16445 allocated.
16446
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020016447capture.req.ver : string
16448 This extracts the request's HTTP version and returns either "HTTP/1.0" or
16449 "HTTP/1.1". Unlike "req.ver", it can be used in both request, response, and
16450 logs because it relies on a persistent flag.
16451
William Lallemand65ad6e12014-01-31 15:08:02 +010016452capture.res.hdr(<idx>) : string
16453 This extracts the content of the header captured by the "capture response
16454 header", idx is the position of the capture keyword in the configuration.
16455 The first entry is an index of 0.
16456 See also: "capture response header"
16457
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020016458capture.res.ver : string
16459 This extracts the response's HTTP version and returns either "HTTP/1.0" or
16460 "HTTP/1.1". Unlike "res.ver", it can be used in logs because it relies on a
16461 persistent flag.
16462
Willy Tarreaua5910cc2015-05-02 00:46:08 +020016463req.body : binary
16464 This returns the HTTP request's available body as a block of data. It
16465 requires that the request body has been buffered made available using
16466 "option http-buffer-request". In case of chunked-encoded body, currently only
16467 the first chunk is analyzed.
16468
Thierry FOURNIER9826c772015-05-20 15:50:54 +020016469req.body_param([<name>) : string
16470 This fetch assumes that the body of the POST request is url-encoded. The user
16471 can check if the "content-type" contains the value
16472 "application/x-www-form-urlencoded". This extracts the first occurrence of the
16473 parameter <name> in the body, which ends before '&'. The parameter name is
16474 case-sensitive. If no name is given, any parameter will match, and the first
16475 one will be returned. The result is a string corresponding to the value of the
16476 parameter <name> as presented in the request body (no URL decoding is
16477 performed). Note that the ACL version of this fetch iterates over multiple
16478 parameters and will iteratively report all parameters values if no name is
16479 given.
16480
Willy Tarreaua5910cc2015-05-02 00:46:08 +020016481req.body_len : integer
16482 This returns the length of the HTTP request's available body in bytes. It may
16483 be lower than the advertised length if the body is larger than the buffer. It
16484 requires that the request body has been buffered made available using
16485 "option http-buffer-request".
16486
16487req.body_size : integer
16488 This returns the advertised length of the HTTP request's body in bytes. It
16489 will represent the advertised Content-Length header, or the size of the first
16490 chunk in case of chunked encoding. In order to parse the chunks, it requires
16491 that the request body has been buffered made available using
16492 "option http-buffer-request".
16493
Willy Tarreau74ca5042013-06-11 23:12:07 +020016494req.cook([<name>]) : string
16495cook([<name>]) : string (deprecated)
16496 This extracts the last occurrence of the cookie name <name> on a "Cookie"
16497 header line from the request, and returns its value as string. If no name is
16498 specified, the first cookie value is returned. When used with ACLs, all
16499 matching cookies are evaluated. Spaces around the name and the value are
16500 ignored as requested by the Cookie header specification (RFC6265). The cookie
16501 name is case-sensitive. Empty cookies are valid, so an empty cookie may very
16502 well return an empty value if it is present. Use the "found" match to detect
16503 presence. Use the res.cook() variant for response cookies sent by the server.
16504
16505 ACL derivatives :
16506 cook([<name>]) : exact string match
16507 cook_beg([<name>]) : prefix match
16508 cook_dir([<name>]) : subdir match
16509 cook_dom([<name>]) : domain match
16510 cook_end([<name>]) : suffix match
16511 cook_len([<name>]) : length match
16512 cook_reg([<name>]) : regex match
16513 cook_sub([<name>]) : substring match
Willy Tarreaud63335a2010-02-26 12:56:52 +010016514
Willy Tarreau74ca5042013-06-11 23:12:07 +020016515req.cook_cnt([<name>]) : integer
16516cook_cnt([<name>]) : integer (deprecated)
16517 Returns an integer value representing the number of occurrences of the cookie
16518 <name> in the request, or all cookies if <name> is not specified.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016519
Willy Tarreau74ca5042013-06-11 23:12:07 +020016520req.cook_val([<name>]) : integer
16521cook_val([<name>]) : integer (deprecated)
16522 This extracts the last occurrence of the cookie name <name> on a "Cookie"
16523 header line from the request, and converts its value to an integer which is
16524 returned. If no name is specified, the first cookie value is returned. When
16525 used in ACLs, all matching names are iterated over until a value matches.
Willy Tarreau0e698542011-09-16 08:32:32 +020016526
Willy Tarreau74ca5042013-06-11 23:12:07 +020016527cookie([<name>]) : string (deprecated)
16528 This extracts the last occurrence of the cookie name <name> on a "Cookie"
16529 header line from the request, or a "Set-Cookie" header from the response, and
16530 returns its value as a string. A typical use is to get multiple clients
16531 sharing a same profile use the same server. This can be similar to what
Willy Tarreau294d0f02015-08-10 19:40:12 +020016532 "appsession" did with the "request-learn" statement, but with support for
Willy Tarreau74ca5042013-06-11 23:12:07 +020016533 multi-peer synchronization and state keeping across restarts. If no name is
16534 specified, the first cookie value is returned. This fetch should not be used
16535 anymore and should be replaced by req.cook() or res.cook() instead as it
16536 ambiguously uses the direction based on the context where it is used.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016537
Willy Tarreau74ca5042013-06-11 23:12:07 +020016538hdr([<name>[,<occ>]]) : string
16539 This is equivalent to req.hdr() when used on requests, and to res.hdr() when
16540 used on responses. Please refer to these respective fetches for more details.
16541 In case of doubt about the fetch direction, please use the explicit ones.
16542 Note that contrary to the hdr() sample fetch method, the hdr_* ACL keywords
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016543 unambiguously apply to the request headers.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016544
Willy Tarreau74ca5042013-06-11 23:12:07 +020016545req.fhdr(<name>[,<occ>]) : string
16546 This extracts the last occurrence of header <name> in an HTTP request. When
16547 used from an ACL, all occurrences are iterated over until a match is found.
16548 Optionally, a specific occurrence might be specified as a position number.
16549 Positive values indicate a position from the first occurrence, with 1 being
16550 the first one. Negative values indicate positions relative to the last one,
16551 with -1 being the last one. It differs from req.hdr() in that any commas
16552 present in the value are returned and are not used as delimiters. This is
16553 sometimes useful with headers such as User-Agent.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016554
Willy Tarreau74ca5042013-06-11 23:12:07 +020016555req.fhdr_cnt([<name>]) : integer
16556 Returns an integer value representing the number of occurrences of request
16557 header field name <name>, or the total number of header fields if <name> is
16558 not specified. Contrary to its req.hdr_cnt() cousin, this function returns
16559 the number of full line headers and does not stop on commas.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016560
Willy Tarreau74ca5042013-06-11 23:12:07 +020016561req.hdr([<name>[,<occ>]]) : string
16562 This extracts the last occurrence of header <name> in an HTTP request. When
16563 used from an ACL, all occurrences are iterated over until a match is found.
16564 Optionally, a specific occurrence might be specified as a position number.
16565 Positive values indicate a position from the first occurrence, with 1 being
16566 the first one. Negative values indicate positions relative to the last one,
16567 with -1 being the last one. A typical use is with the X-Forwarded-For header
16568 once converted to IP, associated with an IP stick-table. The function
16569 considers any comma as a delimiter for distinct values. If full-line headers
Lukas Tribus23953682017-04-28 13:24:30 +000016570 are desired instead, use req.fhdr(). Please carefully check RFC7231 to know
Willy Tarreau74ca5042013-06-11 23:12:07 +020016571 how certain headers are supposed to be parsed. Also, some of them are case
Davor Ocelice9ed2812017-12-25 17:49:28 +010016572 insensitive (e.g. Connection).
Willy Tarreaud63335a2010-02-26 12:56:52 +010016573
Willy Tarreau74ca5042013-06-11 23:12:07 +020016574 ACL derivatives :
16575 hdr([<name>[,<occ>]]) : exact string match
16576 hdr_beg([<name>[,<occ>]]) : prefix match
16577 hdr_dir([<name>[,<occ>]]) : subdir match
16578 hdr_dom([<name>[,<occ>]]) : domain match
16579 hdr_end([<name>[,<occ>]]) : suffix match
16580 hdr_len([<name>[,<occ>]]) : length match
16581 hdr_reg([<name>[,<occ>]]) : regex match
16582 hdr_sub([<name>[,<occ>]]) : substring match
16583
16584req.hdr_cnt([<name>]) : integer
16585hdr_cnt([<header>]) : integer (deprecated)
16586 Returns an integer value representing the number of occurrences of request
16587 header field name <name>, or the total number of header field values if
16588 <name> is not specified. It is important to remember that one header line may
16589 count as several headers if it has several values. The function considers any
16590 comma as a delimiter for distinct values. If full-line headers are desired
16591 instead, req.fhdr_cnt() should be used instead. With ACLs, it can be used to
16592 detect presence, absence or abuse of a specific header, as well as to block
16593 request smuggling attacks by rejecting requests which contain more than one
16594 of certain headers. See "req.hdr" for more information on header matching.
16595
16596req.hdr_ip([<name>[,<occ>]]) : ip
16597hdr_ip([<name>[,<occ>]]) : ip (deprecated)
16598 This extracts the last occurrence of header <name> in an HTTP request,
16599 converts it to an IPv4 or IPv6 address and returns this address. When used
16600 with ACLs, all occurrences are checked, and if <name> is omitted, every value
16601 of every header is checked. Optionally, a specific occurrence might be
16602 specified as a position number. Positive values indicate a position from the
Davor Ocelice9ed2812017-12-25 17:49:28 +010016603 first occurrence, with 1 being the first one. Negative values indicate
Willy Tarreau74ca5042013-06-11 23:12:07 +020016604 positions relative to the last one, with -1 being the last one. A typical use
16605 is with the X-Forwarded-For and X-Client-IP headers.
16606
16607req.hdr_val([<name>[,<occ>]]) : integer
16608hdr_val([<name>[,<occ>]]) : integer (deprecated)
16609 This extracts the last occurrence of header <name> in an HTTP request, and
16610 converts it to an integer value. When used with ACLs, all occurrences are
16611 checked, and if <name> is omitted, every value of every header is checked.
16612 Optionally, a specific occurrence might be specified as a position number.
16613 Positive values indicate a position from the first occurrence, with 1 being
16614 the first one. Negative values indicate positions relative to the last one,
16615 with -1 being the last one. A typical use is with the X-Forwarded-For header.
16616
Frédéric Lécailleec891192019-02-26 15:02:35 +010016617
16618
Willy Tarreau74ca5042013-06-11 23:12:07 +020016619http_auth(<userlist>) : boolean
16620 Returns a boolean indicating whether the authentication data received from
16621 the client match a username & password stored in the specified userlist. This
16622 fetch function is not really useful outside of ACLs. Currently only http
16623 basic auth is supported.
16624
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010016625http_auth_group(<userlist>) : string
16626 Returns a string corresponding to the user name found in the authentication
16627 data received from the client if both the user name and password are valid
16628 according to the specified userlist. The main purpose is to use it in ACLs
16629 where it is then checked whether the user belongs to any group within a list.
Willy Tarreau74ca5042013-06-11 23:12:07 +020016630 This fetch function is not really useful outside of ACLs. Currently only http
16631 basic auth is supported.
16632
16633 ACL derivatives :
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010016634 http_auth_group(<userlist>) : group ...
16635 Returns true when the user extracted from the request and whose password is
16636 valid according to the specified userlist belongs to at least one of the
16637 groups.
Willy Tarreau74ca5042013-06-11 23:12:07 +020016638
Christopher Fauleta4063562019-08-02 11:51:37 +020016639http_auth_pass : string
Willy Tarreauc9c6cdb2020-03-05 16:03:58 +010016640 Returns the user's password found in the authentication data received from
16641 the client, as supplied in the Authorization header. Not checks are
16642 performed by this sample fetch. Only Basic authentication is supported.
Christopher Fauleta4063562019-08-02 11:51:37 +020016643
16644http_auth_type : string
Willy Tarreauc9c6cdb2020-03-05 16:03:58 +010016645 Returns the authentication method found in the authentication data received from
16646 the client, as supplied in the Authorization header. Not checks are
16647 performed by this sample fetch. Only Basic authentication is supported.
Christopher Fauleta4063562019-08-02 11:51:37 +020016648
16649http_auth_user : string
Willy Tarreauc9c6cdb2020-03-05 16:03:58 +010016650 Returns the user name found in the authentication data received from the
16651 client, as supplied in the Authorization header. Not checks are performed by
16652 this sample fetch. Only Basic authentication is supported.
Christopher Fauleta4063562019-08-02 11:51:37 +020016653
Willy Tarreau74ca5042013-06-11 23:12:07 +020016654http_first_req : boolean
Willy Tarreau7f18e522010-10-22 20:04:13 +020016655 Returns true when the request being processed is the first one of the
16656 connection. This can be used to add or remove headers that may be missing
Willy Tarreau74ca5042013-06-11 23:12:07 +020016657 from some requests when a request is not the first one, or to help grouping
16658 requests in the logs.
Willy Tarreau7f18e522010-10-22 20:04:13 +020016659
Willy Tarreau74ca5042013-06-11 23:12:07 +020016660method : integer + string
16661 Returns an integer value corresponding to the method in the HTTP request. For
16662 example, "GET" equals 1 (check sources to establish the matching). Value 9
16663 means "other method" and may be converted to a string extracted from the
16664 stream. This should not be used directly as a sample, this is only meant to
16665 be used from ACLs, which transparently convert methods from patterns to these
16666 integer + string values. Some predefined ACL already check for most common
16667 methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016668
Willy Tarreau74ca5042013-06-11 23:12:07 +020016669 ACL derivatives :
16670 method : case insensitive method match
Willy Tarreau6a06a402007-07-15 20:15:28 +020016671
Willy Tarreau74ca5042013-06-11 23:12:07 +020016672 Example :
16673 # only accept GET and HEAD requests
16674 acl valid_method method GET HEAD
16675 http-request deny if ! valid_method
Willy Tarreau6a06a402007-07-15 20:15:28 +020016676
Willy Tarreau74ca5042013-06-11 23:12:07 +020016677path : string
16678 This extracts the request's URL path, which starts at the first slash and
16679 ends before the question mark (without the host part). A typical use is with
16680 prefetch-capable caches, and with portals which need to aggregate multiple
16681 information from databases and keep them in caches. Note that with outgoing
16682 caches, it would be wiser to use "url" instead. With ACLs, it's typically
Davor Ocelice9ed2812017-12-25 17:49:28 +010016683 used to match exact file names (e.g. "/login.php"), or directory parts using
Willy Tarreau74ca5042013-06-11 23:12:07 +020016684 the derivative forms. See also the "url" and "base" fetch methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016685
Willy Tarreau74ca5042013-06-11 23:12:07 +020016686 ACL derivatives :
16687 path : exact string match
16688 path_beg : prefix match
16689 path_dir : subdir match
16690 path_dom : domain match
16691 path_end : suffix match
16692 path_len : length match
16693 path_reg : regex match
16694 path_sub : substring match
Willy Tarreau6a06a402007-07-15 20:15:28 +020016695
Willy Tarreau49ad95c2015-01-19 15:06:26 +010016696query : string
16697 This extracts the request's query string, which starts after the first
16698 question mark. If no question mark is present, this fetch returns nothing. If
16699 a question mark is present but nothing follows, it returns an empty string.
16700 This means it's possible to easily know whether a query string is present
Tim Düsterhus4896c442016-11-29 02:15:19 +010016701 using the "found" matching method. This fetch is the complement of "path"
Willy Tarreau49ad95c2015-01-19 15:06:26 +010016702 which stops before the question mark.
16703
Willy Tarreaueb27ec72015-02-20 13:55:29 +010016704req.hdr_names([<delim>]) : string
16705 This builds a string made from the concatenation of all header names as they
16706 appear in the request when the rule is evaluated. The default delimiter is
16707 the comma (',') but it may be overridden as an optional argument <delim>. In
16708 this case, only the first character of <delim> is considered.
16709
Willy Tarreau74ca5042013-06-11 23:12:07 +020016710req.ver : string
16711req_ver : string (deprecated)
16712 Returns the version string from the HTTP request, for example "1.1". This can
16713 be useful for logs, but is mostly there for ACL. Some predefined ACL already
16714 check for versions 1.0 and 1.1.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016715
Willy Tarreau74ca5042013-06-11 23:12:07 +020016716 ACL derivatives :
16717 req_ver : exact string match
Willy Tarreau0e698542011-09-16 08:32:32 +020016718
Willy Tarreau74ca5042013-06-11 23:12:07 +020016719res.comp : boolean
16720 Returns the boolean "true" value if the response has been compressed by
16721 HAProxy, otherwise returns boolean "false". This may be used to add
16722 information in the logs.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016723
Willy Tarreau74ca5042013-06-11 23:12:07 +020016724res.comp_algo : string
16725 Returns a string containing the name of the algorithm used if the response
16726 was compressed by HAProxy, for example : "deflate". This may be used to add
16727 some information in the logs.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016728
Willy Tarreau74ca5042013-06-11 23:12:07 +020016729res.cook([<name>]) : string
16730scook([<name>]) : string (deprecated)
16731 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
16732 header line from the response, and returns its value as string. If no name is
16733 specified, the first cookie value is returned.
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020016734
Willy Tarreau74ca5042013-06-11 23:12:07 +020016735 ACL derivatives :
16736 scook([<name>] : exact string match
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020016737
Willy Tarreau74ca5042013-06-11 23:12:07 +020016738res.cook_cnt([<name>]) : integer
16739scook_cnt([<name>]) : integer (deprecated)
16740 Returns an integer value representing the number of occurrences of the cookie
16741 <name> in the response, or all cookies if <name> is not specified. This is
16742 mostly useful when combined with ACLs to detect suspicious responses.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016743
Willy Tarreau74ca5042013-06-11 23:12:07 +020016744res.cook_val([<name>]) : integer
16745scook_val([<name>]) : integer (deprecated)
16746 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
16747 header line from the response, and converts its value to an integer which is
16748 returned. If no name is specified, the first cookie value is returned.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016749
Willy Tarreau74ca5042013-06-11 23:12:07 +020016750res.fhdr([<name>[,<occ>]]) : string
16751 This extracts the last occurrence of header <name> in an HTTP response, or of
16752 the last header if no <name> is specified. Optionally, a specific occurrence
16753 might be specified as a position number. Positive values indicate a position
16754 from the first occurrence, with 1 being the first one. Negative values
16755 indicate positions relative to the last one, with -1 being the last one. It
16756 differs from res.hdr() in that any commas present in the value are returned
16757 and are not used as delimiters. If this is not desired, the res.hdr() fetch
16758 should be used instead. This is sometimes useful with headers such as Date or
16759 Expires.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016760
Willy Tarreau74ca5042013-06-11 23:12:07 +020016761res.fhdr_cnt([<name>]) : integer
16762 Returns an integer value representing the number of occurrences of response
16763 header field name <name>, or the total number of header fields if <name> is
16764 not specified. Contrary to its res.hdr_cnt() cousin, this function returns
16765 the number of full line headers and does not stop on commas. If this is not
16766 desired, the res.hdr_cnt() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016767
Willy Tarreau74ca5042013-06-11 23:12:07 +020016768res.hdr([<name>[,<occ>]]) : string
16769shdr([<name>[,<occ>]]) : string (deprecated)
16770 This extracts the last occurrence of header <name> in an HTTP response, or of
16771 the last header if no <name> is specified. Optionally, a specific occurrence
16772 might be specified as a position number. Positive values indicate a position
16773 from the first occurrence, with 1 being the first one. Negative values
16774 indicate positions relative to the last one, with -1 being the last one. This
16775 can be useful to learn some data into a stick-table. The function considers
16776 any comma as a delimiter for distinct values. If this is not desired, the
16777 res.fhdr() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016778
Willy Tarreau74ca5042013-06-11 23:12:07 +020016779 ACL derivatives :
16780 shdr([<name>[,<occ>]]) : exact string match
16781 shdr_beg([<name>[,<occ>]]) : prefix match
16782 shdr_dir([<name>[,<occ>]]) : subdir match
16783 shdr_dom([<name>[,<occ>]]) : domain match
16784 shdr_end([<name>[,<occ>]]) : suffix match
16785 shdr_len([<name>[,<occ>]]) : length match
16786 shdr_reg([<name>[,<occ>]]) : regex match
16787 shdr_sub([<name>[,<occ>]]) : substring match
16788
16789res.hdr_cnt([<name>]) : integer
16790shdr_cnt([<name>]) : integer (deprecated)
16791 Returns an integer value representing the number of occurrences of response
16792 header field name <name>, or the total number of header fields if <name> is
16793 not specified. The function considers any comma as a delimiter for distinct
16794 values. If this is not desired, the res.fhdr_cnt() fetch should be used
16795 instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016796
Willy Tarreau74ca5042013-06-11 23:12:07 +020016797res.hdr_ip([<name>[,<occ>]]) : ip
16798shdr_ip([<name>[,<occ>]]) : ip (deprecated)
16799 This extracts the last occurrence of header <name> in an HTTP response,
16800 convert it to an IPv4 or IPv6 address and returns this address. Optionally, a
16801 specific occurrence might be specified as a position number. Positive values
16802 indicate a position from the first occurrence, with 1 being the first one.
16803 Negative values indicate positions relative to the last one, with -1 being
16804 the last one. This can be useful to learn some data into a stick table.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016805
Willy Tarreaueb27ec72015-02-20 13:55:29 +010016806res.hdr_names([<delim>]) : string
16807 This builds a string made from the concatenation of all header names as they
16808 appear in the response when the rule is evaluated. The default delimiter is
16809 the comma (',') but it may be overridden as an optional argument <delim>. In
16810 this case, only the first character of <delim> is considered.
16811
Willy Tarreau74ca5042013-06-11 23:12:07 +020016812res.hdr_val([<name>[,<occ>]]) : integer
16813shdr_val([<name>[,<occ>]]) : integer (deprecated)
16814 This extracts the last occurrence of header <name> in an HTTP response, and
16815 converts it to an integer value. Optionally, a specific occurrence might be
16816 specified as a position number. Positive values indicate a position from the
16817 first occurrence, with 1 being the first one. Negative values indicate
16818 positions relative to the last one, with -1 being the last one. This can be
16819 useful to learn some data into a stick table.
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010016820
Willy Tarreau74ca5042013-06-11 23:12:07 +020016821res.ver : string
16822resp_ver : string (deprecated)
16823 Returns the version string from the HTTP response, for example "1.1". This
16824 can be useful for logs, but is mostly there for ACL.
Willy Tarreau0e698542011-09-16 08:32:32 +020016825
Willy Tarreau74ca5042013-06-11 23:12:07 +020016826 ACL derivatives :
16827 resp_ver : exact string match
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010016828
Willy Tarreau74ca5042013-06-11 23:12:07 +020016829set-cookie([<name>]) : string (deprecated)
16830 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
16831 header line from the response and uses the corresponding value to match. This
Willy Tarreau294d0f02015-08-10 19:40:12 +020016832 can be comparable to what "appsession" did with default options, but with
Willy Tarreau74ca5042013-06-11 23:12:07 +020016833 support for multi-peer synchronization and state keeping across restarts.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010016834
Willy Tarreau74ca5042013-06-11 23:12:07 +020016835 This fetch function is deprecated and has been superseded by the "res.cook"
16836 fetch. This keyword will disappear soon.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010016837
Willy Tarreau74ca5042013-06-11 23:12:07 +020016838status : integer
16839 Returns an integer containing the HTTP status code in the HTTP response, for
16840 example, 302. It is mostly used within ACLs and integer ranges, for example,
16841 to remove any Location header if the response is not a 3xx.
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016842
Thierry Fournier0e00dca2016-04-07 15:47:40 +020016843unique-id : string
16844 Returns the unique-id attached to the request. The directive
16845 "unique-id-format" must be set. If it is not set, the unique-id sample fetch
16846 fails. Note that the unique-id is usually used with HTTP requests, however this
16847 sample fetch can be used with other protocols. Obviously, if it is used with
16848 other protocols than HTTP, the unique-id-format directive must not contain
16849 HTTP parts. See: unique-id-format and unique-id-header
16850
Willy Tarreau74ca5042013-06-11 23:12:07 +020016851url : string
16852 This extracts the request's URL as presented in the request. A typical use is
16853 with prefetch-capable caches, and with portals which need to aggregate
16854 multiple information from databases and keep them in caches. With ACLs, using
16855 "path" is preferred over using "url", because clients may send a full URL as
16856 is normally done with proxies. The only real use is to match "*" which does
16857 not match in "path", and for which there is already a predefined ACL. See
16858 also "path" and "base".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016859
Willy Tarreau74ca5042013-06-11 23:12:07 +020016860 ACL derivatives :
16861 url : exact string match
16862 url_beg : prefix match
16863 url_dir : subdir match
16864 url_dom : domain match
16865 url_end : suffix match
16866 url_len : length match
16867 url_reg : regex match
16868 url_sub : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016869
Willy Tarreau74ca5042013-06-11 23:12:07 +020016870url_ip : ip
16871 This extracts the IP address from the request's URL when the host part is
16872 presented as an IP address. Its use is very limited. For instance, a
16873 monitoring system might use this field as an alternative for the source IP in
16874 order to test what path a given source address would follow, or to force an
16875 entry in a table for a given source address. With ACLs it can be used to
16876 restrict access to certain systems through a proxy, for example when combined
16877 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016878
Willy Tarreau74ca5042013-06-11 23:12:07 +020016879url_port : integer
16880 This extracts the port part from the request's URL. Note that if the port is
16881 not specified in the request, port 80 is assumed. With ACLs it can be used to
16882 restrict access to certain systems through a proxy, for example when combined
16883 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016884
Willy Tarreau1ede1da2015-05-07 16:06:18 +020016885urlp([<name>[,<delim>]]) : string
16886url_param([<name>[,<delim>]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020016887 This extracts the first occurrence of the parameter <name> in the query
16888 string, which begins after either '?' or <delim>, and which ends before '&',
Willy Tarreau1ede1da2015-05-07 16:06:18 +020016889 ';' or <delim>. The parameter name is case-sensitive. If no name is given,
16890 any parameter will match, and the first one will be returned. The result is
16891 a string corresponding to the value of the parameter <name> as presented in
16892 the request (no URL decoding is performed). This can be used for session
Willy Tarreau74ca5042013-06-11 23:12:07 +020016893 stickiness based on a client ID, to extract an application cookie passed as a
16894 URL parameter, or in ACLs to apply some checks. Note that the ACL version of
Willy Tarreau1ede1da2015-05-07 16:06:18 +020016895 this fetch iterates over multiple parameters and will iteratively report all
16896 parameters values if no name is given
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016897
Willy Tarreau74ca5042013-06-11 23:12:07 +020016898 ACL derivatives :
16899 urlp(<name>[,<delim>]) : exact string match
16900 urlp_beg(<name>[,<delim>]) : prefix match
16901 urlp_dir(<name>[,<delim>]) : subdir match
16902 urlp_dom(<name>[,<delim>]) : domain match
16903 urlp_end(<name>[,<delim>]) : suffix match
16904 urlp_len(<name>[,<delim>]) : length match
16905 urlp_reg(<name>[,<delim>]) : regex match
16906 urlp_sub(<name>[,<delim>]) : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016907
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016908
Willy Tarreau74ca5042013-06-11 23:12:07 +020016909 Example :
16910 # match http://example.com/foo?PHPSESSIONID=some_id
16911 stick on urlp(PHPSESSIONID)
16912 # match http://example.com/foo;JSESSIONID=some_id
16913 stick on urlp(JSESSIONID,;)
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016914
Jarno Huuskonen676f6222017-03-30 09:19:45 +030016915urlp_val([<name>[,<delim>]]) : integer
Willy Tarreau74ca5042013-06-11 23:12:07 +020016916 See "urlp" above. This one extracts the URL parameter <name> in the request
16917 and converts it to an integer value. This can be used for session stickiness
16918 based on a user ID for example, or with ACLs to match a page number or price.
Willy Tarreaua9fddca2012-07-31 07:51:48 +020016919
Dragan Dosen0070cd52016-06-16 12:19:49 +020016920url32 : integer
16921 This returns a 32-bit hash of the value obtained by concatenating the first
16922 Host header and the whole URL including parameters (not only the path part of
16923 the request, as in the "base32" fetch above). This is useful to track per-URL
16924 activity. A shorter hash is stored, saving a lot of memory. The output type
16925 is an unsigned integer.
16926
16927url32+src : binary
16928 This returns the concatenation of the "url32" fetch and the "src" fetch. The
16929 resulting type is of type binary, with a size of 8 or 20 bytes depending on
16930 the source address family. This can be used to track per-IP, per-URL counters.
16931
Willy Tarreau198a7442008-01-17 12:05:32 +010016932
Christopher Fauletd47941d2020-01-08 14:40:19 +0100169337.3.7. Fetching samples for developers
16934---------------------------------------
16935
16936This set of sample fetch methods is reserved to developers and must never be
16937used on a production environment, except on developer demand, for debugging
16938purposes. Moreover, no special care will be taken on backwards compatibility.
16939There is no warranty the following sample fetches will never change, be renamed
16940or simply removed. So be really careful if you should use one of them. To avoid
16941any ambiguity, these sample fetches are placed in the dedicated scope "internal",
16942for instance "internal.strm.is_htx".
16943
16944internal.htx.data : integer
16945 Returns the size in bytes used by data in the HTX message associated to a
16946 channel. The channel is chosen depending on the sample direction.
16947
16948internal.htx.free : integer
16949 Returns the free space (size - used) in bytes in the HTX message associated
16950 to a channel. The channel is chosen depending on the sample direction.
16951
16952internal.htx.free_data : integer
16953 Returns the free space for the data in bytes in the HTX message associated to
16954 a channel. The channel is chosen depending on the sample direction.
16955
16956internal.htx.has_eom : boolean
16957 Returns true if the HTX message associated to a channel contains an
16958 end-of-message block (EOM). Otherwise, it returns false. The channel is
16959 chosen depending on the sample direction.
16960
16961internal.htx.nbblks : integer
16962 Returns the number of blocks present in the HTX message associated to a
16963 channel. The channel is chosen depending on the sample direction.
16964
16965internal.htx.size : integer
16966 Returns the total size in bytes of the HTX message associated to a
16967 channel. The channel is chosen depending on the sample direction.
16968
16969internal.htx.used : integer
16970 Returns the total size used in bytes (data + metadata) in the HTX message
16971 associated to a channel. The channel is chosen depending on the sample
16972 direction.
16973
16974internal.htx_blk.size(<idx>) : integer
16975 Returns the size of the block at the position <idx> in the HTX message
16976 associated to a channel or 0 if it does not exist. The channel is chosen
16977 depending on the sample direction. <idx> may be any positive integer or one
16978 of the special value :
16979 * head : The oldest inserted block
16980 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050016981 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010016982
16983internal.htx_blk.type(<idx>) : string
16984 Returns the type of the block at the position <idx> in the HTX message
16985 associated to a channel or "HTX_BLK_UNUSED" if it does not exist. The channel
16986 is chosen depending on the sample direction. <idx> may be any positive
16987 integer or one of the special value :
16988 * head : The oldest inserted block
16989 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050016990 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010016991
16992internal.htx_blk.data(<idx>) : binary
16993 Returns the value of the DATA block at the position <idx> in the HTX message
16994 associated to a channel or an empty string if it does not exist or if it is
16995 not a DATA block. The channel is chosen depending on the sample direction.
16996 <idx> may be any positive integer or one of the special value :
16997
16998 * head : The oldest inserted block
16999 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050017000 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010017001
17002internal.htx_blk.hdrname(<idx>) : string
17003 Returns the header name of the HEADER block at the position <idx> in the HTX
17004 message associated to a channel or an empty string if it does not exist or if
17005 it is not an HEADER block. The channel is chosen depending on the sample
17006 direction. <idx> may be any positive integer or one of the special value :
17007
17008 * head : The oldest inserted block
17009 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050017010 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010017011
17012internal.htx_blk.hdrval(<idx>) : string
17013 Returns the header value of the HEADER block at the position <idx> in the HTX
17014 message associated to a channel or an empty string if it does not exist or if
17015 it is not an HEADER block. The channel is chosen depending on the sample
17016 direction. <idx> may be any positive integer or one of the special value :
17017
17018 * head : The oldest inserted block
17019 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050017020 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010017021
17022internal.htx_blk.start_line(<idx>) : string
17023 Returns the value of the REQ_SL or RES_SL block at the position <idx> in the
17024 HTX message associated to a channel or an empty string if it does not exist
17025 or if it is not a SL block. The channel is chosen depending on the sample
17026 direction. <idx> may be any positive integer or one of the special value :
17027
17028 * head : The oldest inserted block
17029 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050017030 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010017031
17032internal.strm.is_htx : boolean
17033 Returns true if the current stream is an HTX stream. It means the data in the
17034 channels buffers are stored using the internal HTX representation. Otherwise,
17035 it returns false.
17036
17037
Willy Tarreau74ca5042013-06-11 23:12:07 +0200170387.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017039---------------------
Willy Tarreauced27012008-01-17 20:35:34 +010017040
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017041Some predefined ACLs are hard-coded so that they do not have to be declared in
17042every frontend which needs them. They all have their names in upper case in
Patrick Mézard2382ad62010-05-09 10:43:32 +020017043order to avoid confusion. Their equivalence is provided below.
Willy Tarreauced27012008-01-17 20:35:34 +010017044
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017045ACL name Equivalent to Usage
17046---------------+-----------------------------+---------------------------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017047FALSE always_false never match
Willy Tarreau2492d5b2009-07-11 00:06:00 +020017048HTTP req_proto_http match if protocol is valid HTTP
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017049HTTP_1.0 req_ver 1.0 match HTTP version 1.0
17050HTTP_1.1 req_ver 1.1 match HTTP version 1.1
Willy Tarreaud63335a2010-02-26 12:56:52 +010017051HTTP_CONTENT hdr_val(content-length) gt 0 match an existing content-length
17052HTTP_URL_ABS url_reg ^[^/:]*:// match absolute URL with scheme
17053HTTP_URL_SLASH url_beg / match URL beginning with "/"
17054HTTP_URL_STAR url * match URL equal to "*"
17055LOCALHOST src 127.0.0.1/8 match connection from local host
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017056METH_CONNECT method CONNECT match HTTP CONNECT method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020017057METH_DELETE method DELETE match HTTP DELETE method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017058METH_GET method GET HEAD match HTTP GET or HEAD method
17059METH_HEAD method HEAD match HTTP HEAD method
17060METH_OPTIONS method OPTIONS match HTTP OPTIONS method
17061METH_POST method POST match HTTP POST method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020017062METH_PUT method PUT match HTTP PUT method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017063METH_TRACE method TRACE match HTTP TRACE method
Emeric Brunbede3d02009-06-30 17:54:00 +020017064RDP_COOKIE req_rdp_cookie_cnt gt 0 match presence of an RDP cookie
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017065REQ_CONTENT req_len gt 0 match data in the request buffer
Willy Tarreaud63335a2010-02-26 12:56:52 +010017066TRUE always_true always match
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017067WAIT_END wait_end wait for end of content analysis
17068---------------+-----------------------------+---------------------------------
Willy Tarreauced27012008-01-17 20:35:34 +010017069
Willy Tarreaub937b7e2010-01-12 15:27:54 +010017070
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200170718. Logging
17072----------
Willy Tarreau844e3c52008-01-11 16:28:18 +010017073
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017074One of HAProxy's strong points certainly lies is its precise logs. It probably
17075provides the finest level of information available for such a product, which is
17076very important for troubleshooting complex environments. Standard information
17077provided in logs include client ports, TCP/HTTP state timers, precise session
17078state at termination and precise termination cause, information about decisions
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010017079to direct traffic to a server, and of course the ability to capture arbitrary
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017080headers.
17081
17082In order to improve administrators reactivity, it offers a great transparency
17083about encountered problems, both internal and external, and it is possible to
17084send logs to different sources at the same time with different level filters :
17085
17086 - global process-level logs (system errors, start/stop, etc..)
17087 - per-instance system and internal errors (lack of resource, bugs, ...)
17088 - per-instance external troubles (servers up/down, max connections)
17089 - per-instance activity (client connections), either at the establishment or
17090 at the termination.
Davor Ocelice9ed2812017-12-25 17:49:28 +010017091 - per-request control of log-level, e.g.
Jim Freeman9e8714b2015-05-26 09:16:34 -060017092 http-request set-log-level silent if sensitive_request
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017093
17094The ability to distribute different levels of logs to different log servers
17095allow several production teams to interact and to fix their problems as soon
17096as possible. For example, the system team might monitor system-wide errors,
17097while the application team might be monitoring the up/down for their servers in
17098real time, and the security team might analyze the activity logs with one hour
17099delay.
17100
17101
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200171028.1. Log levels
17103---------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017104
Simon Hormandf791f52011-05-29 15:01:10 +090017105TCP and HTTP connections can be logged with information such as the date, time,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017106source IP address, destination address, connection duration, response times,
Simon Hormandf791f52011-05-29 15:01:10 +090017107HTTP request, HTTP return code, number of bytes transmitted, conditions
17108in which the session ended, and even exchanged cookies values. For example
17109track a particular user's problems. All messages may be sent to up to two
17110syslog servers. Check the "log" keyword in section 4.2 for more information
17111about log facilities.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017112
17113
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200171148.2. Log formats
17115----------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017116
William Lallemand48940402012-01-30 16:47:22 +010017117HAProxy supports 5 log formats. Several fields are common between these formats
Simon Hormandf791f52011-05-29 15:01:10 +090017118and will be detailed in the following sections. A few of them may vary
17119slightly with the configuration, due to indicators specific to certain
17120options. The supported formats are as follows :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017121
17122 - the default format, which is very basic and very rarely used. It only
17123 provides very basic information about the incoming connection at the moment
17124 it is accepted : source IP:port, destination IP:port, and frontend-name.
17125 This mode will eventually disappear so it will not be described to great
17126 extents.
17127
17128 - the TCP format, which is more advanced. This format is enabled when "option
17129 tcplog" is set on the frontend. HAProxy will then usually wait for the
17130 connection to terminate before logging. This format provides much richer
17131 information, such as timers, connection counts, queue size, etc... This
17132 format is recommended for pure TCP proxies.
17133
17134 - the HTTP format, which is the most advanced for HTTP proxying. This format
17135 is enabled when "option httplog" is set on the frontend. It provides the
17136 same information as the TCP format with some HTTP-specific fields such as
17137 the request, the status code, and captures of headers and cookies. This
17138 format is recommended for HTTP proxies.
17139
Emeric Brun3a058f32009-06-30 18:26:00 +020017140 - the CLF HTTP format, which is equivalent to the HTTP format, but with the
17141 fields arranged in the same order as the CLF format. In this mode, all
17142 timers, captures, flags, etc... appear one per field after the end of the
17143 common fields, in the same order they appear in the standard HTTP format.
17144
William Lallemand48940402012-01-30 16:47:22 +010017145 - the custom log format, allows you to make your own log line.
17146
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017147Next sections will go deeper into details for each of these formats. Format
17148specification will be performed on a "field" basis. Unless stated otherwise, a
17149field is a portion of text delimited by any number of spaces. Since syslog
17150servers are susceptible of inserting fields at the beginning of a line, it is
17151always assumed that the first field is the one containing the process name and
17152identifier.
17153
17154Note : Since log lines may be quite long, the log examples in sections below
17155 might be broken into multiple lines. The example log lines will be
17156 prefixed with 3 closing angle brackets ('>>>') and each time a log is
17157 broken into multiple lines, each non-final line will end with a
17158 backslash ('\') and the next line will start indented by two characters.
17159
17160
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200171618.2.1. Default log format
17162-------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017163
17164This format is used when no specific option is set. The log is emitted as soon
17165as the connection is accepted. One should note that this currently is the only
17166format which logs the request's destination IP and ports.
17167
17168 Example :
17169 listen www
17170 mode http
17171 log global
17172 server srv1 127.0.0.1:8000
17173
17174 >>> Feb 6 12:12:09 localhost \
17175 haproxy[14385]: Connect from 10.0.1.2:33312 to 10.0.3.31:8012 \
17176 (www/HTTP)
17177
17178 Field Format Extract from the example above
17179 1 process_name '[' pid ']:' haproxy[14385]:
17180 2 'Connect from' Connect from
17181 3 source_ip ':' source_port 10.0.1.2:33312
17182 4 'to' to
17183 5 destination_ip ':' destination_port 10.0.3.31:8012
17184 6 '(' frontend_name '/' mode ')' (www/HTTP)
17185
17186Detailed fields description :
17187 - "source_ip" is the IP address of the client which initiated the connection.
17188 - "source_port" is the TCP port of the client which initiated the connection.
17189 - "destination_ip" is the IP address the client connected to.
17190 - "destination_port" is the TCP port the client connected to.
17191 - "frontend_name" is the name of the frontend (or listener) which received
17192 and processed the connection.
17193 - "mode is the mode the frontend is operating (TCP or HTTP).
17194
Willy Tarreauceb24bc2010-11-09 12:46:41 +010017195In case of a UNIX socket, the source and destination addresses are marked as
17196"unix:" and the ports reflect the internal ID of the socket which accepted the
17197connection (the same ID as reported in the stats).
17198
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017199It is advised not to use this deprecated format for newer installations as it
17200will eventually disappear.
17201
17202
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200172038.2.2. TCP log format
17204---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017205
17206The TCP format is used when "option tcplog" is specified in the frontend, and
17207is the recommended format for pure TCP proxies. It provides a lot of precious
17208information for troubleshooting. Since this format includes timers and byte
17209counts, the log is normally emitted at the end of the session. It can be
17210emitted earlier if "option logasap" is specified, which makes sense in most
17211environments with long sessions such as remote terminals. Sessions which match
17212the "monitor" rules are never logged. It is also possible not to emit logs for
17213sessions for which no data were exchanged between the client and the server, by
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020017214specifying "option dontlognull" in the frontend. Successful connections will
17215not be logged if "option dontlog-normal" is specified in the frontend. A few
17216fields may slightly vary depending on some configuration options, those are
17217marked with a star ('*') after the field name below.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017218
17219 Example :
17220 frontend fnt
17221 mode tcp
17222 option tcplog
17223 log global
17224 default_backend bck
17225
17226 backend bck
17227 server srv1 127.0.0.1:8000
17228
17229 >>> Feb 6 12:12:56 localhost \
17230 haproxy[14387]: 10.0.1.2:33313 [06/Feb/2009:12:12:51.443] fnt \
17231 bck/srv1 0/0/5007 212 -- 0/0/0/0/3 0/0
17232
17233 Field Format Extract from the example above
17234 1 process_name '[' pid ']:' haproxy[14387]:
17235 2 client_ip ':' client_port 10.0.1.2:33313
17236 3 '[' accept_date ']' [06/Feb/2009:12:12:51.443]
17237 4 frontend_name fnt
17238 5 backend_name '/' server_name bck/srv1
17239 6 Tw '/' Tc '/' Tt* 0/0/5007
17240 7 bytes_read* 212
17241 8 termination_state --
17242 9 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 0/0/0/0/3
17243 10 srv_queue '/' backend_queue 0/0
17244
17245Detailed fields description :
17246 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010017247 connection to haproxy. If the connection was accepted on a UNIX socket
17248 instead, the IP address would be replaced with the word "unix". Note that
17249 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010017250 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010017251 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010017252 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017253
17254 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010017255 If the connection was accepted on a UNIX socket instead, the port would be
17256 replaced with the ID of the accepting socket, which is also reported in the
17257 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017258
17259 - "accept_date" is the exact date when the connection was received by haproxy
17260 (which might be very slightly different from the date observed on the
17261 network if there was some queuing in the system's backlog). This is usually
Willy Tarreau590a0512018-09-05 11:56:48 +020017262 the same date which may appear in any upstream firewall's log. When used in
17263 HTTP mode, the accept_date field will be reset to the first moment the
17264 connection is ready to receive a new request (end of previous response for
17265 HTTP/1, immediately after previous request for HTTP/2).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017266
17267 - "frontend_name" is the name of the frontend (or listener) which received
17268 and processed the connection.
17269
17270 - "backend_name" is the name of the backend (or listener) which was selected
17271 to manage the connection to the server. This will be the same as the
17272 frontend if no switching rule has been applied, which is common for TCP
17273 applications.
17274
17275 - "server_name" is the name of the last server to which the connection was
17276 sent, which might differ from the first one if there were connection errors
17277 and a redispatch occurred. Note that this server belongs to the backend
17278 which processed the request. If the connection was aborted before reaching
17279 a server, "<NOSRV>" is indicated instead of a server name.
17280
17281 - "Tw" is the total time in milliseconds spent waiting in the various queues.
17282 It can be "-1" if the connection was aborted before reaching the queue.
17283 See "Timers" below for more details.
17284
17285 - "Tc" is the total time in milliseconds spent waiting for the connection to
17286 establish to the final server, including retries. It can be "-1" if the
17287 connection was aborted before a connection could be established. See
17288 "Timers" below for more details.
17289
17290 - "Tt" is the total time in milliseconds elapsed between the accept and the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017291 last close. It covers all possible processing. There is one exception, if
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017292 "option logasap" was specified, then the time counting stops at the moment
17293 the log is emitted. In this case, a '+' sign is prepended before the value,
17294 indicating that the final one will be larger. See "Timers" below for more
17295 details.
17296
17297 - "bytes_read" is the total number of bytes transmitted from the server to
17298 the client when the log is emitted. If "option logasap" is specified, the
17299 this value will be prefixed with a '+' sign indicating that the final one
17300 may be larger. Please note that this value is a 64-bit counter, so log
17301 analysis tools must be able to handle it without overflowing.
17302
17303 - "termination_state" is the condition the session was in when the session
17304 ended. This indicates the session state, which side caused the end of
17305 session to happen, and for what reason (timeout, error, ...). The normal
17306 flags should be "--", indicating the session was closed by either end with
17307 no data remaining in buffers. See below "Session state at disconnection"
17308 for more details.
17309
17310 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017311 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017312 limits have been reached. For instance, if actconn is close to 512 when
17313 multiple connection errors occur, chances are high that the system limits
17314 the process to use a maximum of 1024 file descriptors and that all of them
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017315 are used. See section 3 "Global parameters" to find how to tune the system.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017316
17317 - "feconn" is the total number of concurrent connections on the frontend when
17318 the session was logged. It is useful to estimate the amount of resource
17319 required to sustain high loads, and to detect when the frontend's "maxconn"
17320 has been reached. Most often when this value increases by huge jumps, it is
17321 because there is congestion on the backend servers, but sometimes it can be
17322 caused by a denial of service attack.
17323
17324 - "beconn" is the total number of concurrent connections handled by the
17325 backend when the session was logged. It includes the total number of
17326 concurrent connections active on servers as well as the number of
17327 connections pending in queues. It is useful to estimate the amount of
17328 additional servers needed to support high loads for a given application.
17329 Most often when this value increases by huge jumps, it is because there is
17330 congestion on the backend servers, but sometimes it can be caused by a
17331 denial of service attack.
17332
17333 - "srv_conn" is the total number of concurrent connections still active on
17334 the server when the session was logged. It can never exceed the server's
17335 configured "maxconn" parameter. If this value is very often close or equal
17336 to the server's "maxconn", it means that traffic regulation is involved a
17337 lot, meaning that either the server's maxconn value is too low, or that
17338 there aren't enough servers to process the load with an optimal response
17339 time. When only one of the server's "srv_conn" is high, it usually means
17340 that this server has some trouble causing the connections to take longer to
17341 be processed than on other servers.
17342
17343 - "retries" is the number of connection retries experienced by this session
17344 when trying to connect to the server. It must normally be zero, unless a
17345 server is being stopped at the same moment the connection was attempted.
17346 Frequent retries generally indicate either a network problem between
17347 haproxy and the server, or a misconfigured system backlog on the server
17348 preventing new connections from being queued. This field may optionally be
17349 prefixed with a '+' sign, indicating that the session has experienced a
17350 redispatch after the maximal retry count has been reached on the initial
17351 server. In this case, the server name appearing in the log is the one the
17352 connection was redispatched to, and not the first one, though both may
17353 sometimes be the same in case of hashing for instance. So as a general rule
17354 of thumb, when a '+' is present in front of the retry count, this count
17355 should not be attributed to the logged server.
17356
17357 - "srv_queue" is the total number of requests which were processed before
17358 this one in the server queue. It is zero when the request has not gone
17359 through the server queue. It makes it possible to estimate the approximate
17360 server's response time by dividing the time spent in queue by the number of
17361 requests in the queue. It is worth noting that if a session experiences a
17362 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010017363 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017364 backend queue unless a redispatch occurs.
17365
17366 - "backend_queue" is the total number of requests which were processed before
17367 this one in the backend's global queue. It is zero when the request has not
17368 gone through the global queue. It makes it possible to estimate the average
17369 queue length, which easily translates into a number of missing servers when
17370 divided by a server's "maxconn" parameter. It is worth noting that if a
17371 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010017372 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017373 through both the server queue and the backend queue unless a redispatch
17374 occurs.
17375
17376
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200173778.2.3. HTTP log format
17378----------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017379
17380The HTTP format is the most complete and the best suited for HTTP proxies. It
17381is enabled by when "option httplog" is specified in the frontend. It provides
17382the same level of information as the TCP format with additional features which
17383are specific to the HTTP protocol. Just like the TCP format, the log is usually
17384emitted at the end of the session, unless "option logasap" is specified, which
17385generally only makes sense for download sites. A session which matches the
17386"monitor" rules will never logged. It is also possible not to log sessions for
17387which no data were sent by the client by specifying "option dontlognull" in the
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020017388frontend. Successful connections will not be logged if "option dontlog-normal"
17389is specified in the frontend.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017390
17391Most fields are shared with the TCP log, some being different. A few fields may
17392slightly vary depending on some configuration options. Those ones are marked
17393with a star ('*') after the field name below.
17394
17395 Example :
17396 frontend http-in
17397 mode http
17398 option httplog
17399 log global
17400 default_backend bck
17401
17402 backend static
17403 server srv1 127.0.0.1:8000
17404
17405 >>> Feb 6 12:14:14 localhost \
17406 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
17407 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 +010017408 {} "GET /index.html HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017409
17410 Field Format Extract from the example above
17411 1 process_name '[' pid ']:' haproxy[14389]:
17412 2 client_ip ':' client_port 10.0.1.2:33317
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017413 3 '[' request_date ']' [06/Feb/2009:12:14:14.655]
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017414 4 frontend_name http-in
17415 5 backend_name '/' server_name static/srv1
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017416 6 TR '/' Tw '/' Tc '/' Tr '/' Ta* 10/0/30/69/109
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017417 7 status_code 200
17418 8 bytes_read* 2750
17419 9 captured_request_cookie -
17420 10 captured_response_cookie -
17421 11 termination_state ----
17422 12 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 1/1/1/1/0
17423 13 srv_queue '/' backend_queue 0/0
17424 14 '{' captured_request_headers* '}' {haproxy.1wt.eu}
17425 15 '{' captured_response_headers* '}' {}
17426 16 '"' http_request '"' "GET /index.html HTTP/1.1"
Willy Tarreaud72758d2010-01-12 10:42:19 +010017427
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017428Detailed fields description :
17429 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010017430 connection to haproxy. If the connection was accepted on a UNIX socket
17431 instead, the IP address would be replaced with the word "unix". Note that
17432 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010017433 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010017434 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010017435 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017436
17437 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010017438 If the connection was accepted on a UNIX socket instead, the port would be
17439 replaced with the ID of the accepting socket, which is also reported in the
17440 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017441
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017442 - "request_date" is the exact date when the first byte of the HTTP request
17443 was received by haproxy (log field %tr).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017444
17445 - "frontend_name" is the name of the frontend (or listener) which received
17446 and processed the connection.
17447
17448 - "backend_name" is the name of the backend (or listener) which was selected
17449 to manage the connection to the server. This will be the same as the
17450 frontend if no switching rule has been applied.
17451
17452 - "server_name" is the name of the last server to which the connection was
17453 sent, which might differ from the first one if there were connection errors
17454 and a redispatch occurred. Note that this server belongs to the backend
17455 which processed the request. If the request was aborted before reaching a
17456 server, "<NOSRV>" is indicated instead of a server name. If the request was
17457 intercepted by the stats subsystem, "<STATS>" is indicated instead.
17458
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017459 - "TR" is the total time in milliseconds spent waiting for a full HTTP
17460 request from the client (not counting body) after the first byte was
17461 received. It can be "-1" if the connection was aborted before a complete
John Roeslerfb2fce12019-07-10 15:45:51 -050017462 request could be received or a bad request was received. It should
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017463 always be very small because a request generally fits in one single packet.
17464 Large times here generally indicate network issues between the client and
Willy Tarreau590a0512018-09-05 11:56:48 +020017465 haproxy or requests being typed by hand. See section 8.4 "Timing Events"
17466 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017467
17468 - "Tw" is the total time in milliseconds spent waiting in the various queues.
17469 It can be "-1" if the connection was aborted before reaching the queue.
Willy Tarreau590a0512018-09-05 11:56:48 +020017470 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017471
17472 - "Tc" is the total time in milliseconds spent waiting for the connection to
17473 establish to the final server, including retries. It can be "-1" if the
Willy Tarreau590a0512018-09-05 11:56:48 +020017474 request was aborted before a connection could be established. See section
17475 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017476
17477 - "Tr" is the total time in milliseconds spent waiting for the server to send
17478 a full HTTP response, not counting data. It can be "-1" if the request was
17479 aborted before a complete response could be received. It generally matches
17480 the server's processing time for the request, though it may be altered by
17481 the amount of data sent by the client to the server. Large times here on
Willy Tarreau590a0512018-09-05 11:56:48 +020017482 "GET" requests generally indicate an overloaded server. See section 8.4
17483 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017484
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017485 - "Ta" is the time the request remained active in haproxy, which is the total
17486 time in milliseconds elapsed between the first byte of the request was
17487 received and the last byte of response was sent. It covers all possible
17488 processing except the handshake (see Th) and idle time (see Ti). There is
17489 one exception, if "option logasap" was specified, then the time counting
17490 stops at the moment the log is emitted. In this case, a '+' sign is
17491 prepended before the value, indicating that the final one will be larger.
Willy Tarreau590a0512018-09-05 11:56:48 +020017492 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017493
17494 - "status_code" is the HTTP status code returned to the client. This status
17495 is generally set by the server, but it might also be set by haproxy when
17496 the server cannot be reached or when its response is blocked by haproxy.
17497
17498 - "bytes_read" is the total number of bytes transmitted to the client when
17499 the log is emitted. This does include HTTP headers. If "option logasap" is
John Roeslerfb2fce12019-07-10 15:45:51 -050017500 specified, this value will be prefixed with a '+' sign indicating that
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017501 the final one may be larger. Please note that this value is a 64-bit
17502 counter, so log analysis tools must be able to handle it without
17503 overflowing.
17504
17505 - "captured_request_cookie" is an optional "name=value" entry indicating that
17506 the client had this cookie in the request. The cookie name and its maximum
17507 length are defined by the "capture cookie" statement in the frontend
17508 configuration. The field is a single dash ('-') when the option is not
17509 set. Only one cookie may be captured, it is generally used to track session
17510 ID exchanges between a client and a server to detect session crossing
17511 between clients due to application bugs. For more details, please consult
17512 the section "Capturing HTTP headers and cookies" below.
17513
17514 - "captured_response_cookie" is an optional "name=value" entry indicating
17515 that the server has returned a cookie with its response. The cookie name
17516 and its maximum length are defined by the "capture cookie" statement in the
17517 frontend configuration. The field is a single dash ('-') when the option is
17518 not set. Only one cookie may be captured, it is generally used to track
17519 session ID exchanges between a client and a server to detect session
17520 crossing between clients due to application bugs. For more details, please
17521 consult the section "Capturing HTTP headers and cookies" below.
17522
17523 - "termination_state" is the condition the session was in when the session
17524 ended. This indicates the session state, which side caused the end of
17525 session to happen, for what reason (timeout, error, ...), just like in TCP
17526 logs, and information about persistence operations on cookies in the last
17527 two characters. The normal flags should begin with "--", indicating the
17528 session was closed by either end with no data remaining in buffers. See
17529 below "Session state at disconnection" for more details.
17530
17531 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017532 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017533 limits have been reached. For instance, if actconn is close to 512 or 1024
17534 when multiple connection errors occur, chances are high that the system
17535 limits the process to use a maximum of 1024 file descriptors and that all
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017536 of them are used. See section 3 "Global parameters" to find how to tune the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017537 system.
17538
17539 - "feconn" is the total number of concurrent connections on the frontend when
17540 the session was logged. It is useful to estimate the amount of resource
17541 required to sustain high loads, and to detect when the frontend's "maxconn"
17542 has been reached. Most often when this value increases by huge jumps, it is
17543 because there is congestion on the backend servers, but sometimes it can be
17544 caused by a denial of service attack.
17545
17546 - "beconn" is the total number of concurrent connections handled by the
17547 backend when the session was logged. It includes the total number of
17548 concurrent connections active on servers as well as the number of
17549 connections pending in queues. It is useful to estimate the amount of
17550 additional servers needed to support high loads for a given application.
17551 Most often when this value increases by huge jumps, it is because there is
17552 congestion on the backend servers, but sometimes it can be caused by a
17553 denial of service attack.
17554
17555 - "srv_conn" is the total number of concurrent connections still active on
17556 the server when the session was logged. It can never exceed the server's
17557 configured "maxconn" parameter. If this value is very often close or equal
17558 to the server's "maxconn", it means that traffic regulation is involved a
17559 lot, meaning that either the server's maxconn value is too low, or that
17560 there aren't enough servers to process the load with an optimal response
17561 time. When only one of the server's "srv_conn" is high, it usually means
17562 that this server has some trouble causing the requests to take longer to be
17563 processed than on other servers.
17564
17565 - "retries" is the number of connection retries experienced by this session
17566 when trying to connect to the server. It must normally be zero, unless a
17567 server is being stopped at the same moment the connection was attempted.
17568 Frequent retries generally indicate either a network problem between
17569 haproxy and the server, or a misconfigured system backlog on the server
17570 preventing new connections from being queued. This field may optionally be
17571 prefixed with a '+' sign, indicating that the session has experienced a
17572 redispatch after the maximal retry count has been reached on the initial
17573 server. In this case, the server name appearing in the log is the one the
17574 connection was redispatched to, and not the first one, though both may
17575 sometimes be the same in case of hashing for instance. So as a general rule
17576 of thumb, when a '+' is present in front of the retry count, this count
17577 should not be attributed to the logged server.
17578
17579 - "srv_queue" is the total number of requests which were processed before
17580 this one in the server queue. It is zero when the request has not gone
17581 through the server queue. It makes it possible to estimate the approximate
17582 server's response time by dividing the time spent in queue by the number of
17583 requests in the queue. It is worth noting that if a session experiences a
17584 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010017585 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017586 backend queue unless a redispatch occurs.
17587
17588 - "backend_queue" is the total number of requests which were processed before
17589 this one in the backend's global queue. It is zero when the request has not
17590 gone through the global queue. It makes it possible to estimate the average
17591 queue length, which easily translates into a number of missing servers when
17592 divided by a server's "maxconn" parameter. It is worth noting that if a
17593 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010017594 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017595 through both the server queue and the backend queue unless a redispatch
17596 occurs.
17597
17598 - "captured_request_headers" is a list of headers captured in the request due
17599 to the presence of the "capture request header" statement in the frontend.
17600 Multiple headers can be captured, they will be delimited by a vertical bar
17601 ('|'). When no capture is enabled, the braces do not appear, causing a
17602 shift of remaining fields. It is important to note that this field may
17603 contain spaces, and that using it requires a smarter log parser than when
17604 it's not used. Please consult the section "Capturing HTTP headers and
17605 cookies" below for more details.
17606
17607 - "captured_response_headers" is a list of headers captured in the response
17608 due to the presence of the "capture response header" statement in the
17609 frontend. Multiple headers can be captured, they will be delimited by a
17610 vertical bar ('|'). When no capture is enabled, the braces do not appear,
17611 causing a shift of remaining fields. It is important to note that this
17612 field may contain spaces, and that using it requires a smarter log parser
17613 than when it's not used. Please consult the section "Capturing HTTP headers
17614 and cookies" below for more details.
17615
17616 - "http_request" is the complete HTTP request line, including the method,
17617 request and HTTP version string. Non-printable characters are encoded (see
17618 below the section "Non-printable characters"). This is always the last
17619 field, and it is always delimited by quotes and is the only one which can
17620 contain quotes. If new fields are added to the log format, they will be
17621 added before this field. This field might be truncated if the request is
17622 huge and does not fit in the standard syslog buffer (1024 characters). This
17623 is the reason why this field must always remain the last one.
17624
17625
Cyril Bontédc4d9032012-04-08 21:57:39 +0200176268.2.4. Custom log format
17627------------------------
William Lallemand48940402012-01-30 16:47:22 +010017628
Willy Tarreau2beef582012-12-20 17:22:52 +010017629The directive log-format allows you to customize the logs in http mode and tcp
William Lallemandbddd4fd2012-02-27 11:23:10 +010017630mode. It takes a string as argument.
William Lallemand48940402012-01-30 16:47:22 +010017631
Davor Ocelice9ed2812017-12-25 17:49:28 +010017632HAProxy understands some log format variables. % precedes log format variables.
William Lallemand48940402012-01-30 16:47:22 +010017633Variables can take arguments using braces ('{}'), and multiple arguments are
17634separated by commas within the braces. Flags may be added or removed by
17635prefixing them with a '+' or '-' sign.
17636
17637Special variable "%o" may be used to propagate its flags to all other
17638variables on the same format string. This is particularly handy with quoted
Dragan Dosen835b9212016-02-12 13:23:03 +010017639("Q") and escaped ("E") string formats.
William Lallemand48940402012-01-30 16:47:22 +010017640
Willy Tarreauc8368452012-12-21 00:09:23 +010017641If a variable is named between square brackets ('[' .. ']') then it is used
Willy Tarreaube722a22014-06-13 16:31:59 +020017642as a sample expression rule (see section 7.3). This it useful to add some
Willy Tarreauc8368452012-12-21 00:09:23 +010017643less common information such as the client's SSL certificate's DN, or to log
17644the key that would be used to store an entry into a stick table.
17645
William Lallemand48940402012-01-30 16:47:22 +010017646Note: spaces must be escaped. A space character is considered as a separator.
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017647In order to emit a verbatim '%', it must be preceded by another '%' resulting
Willy Tarreau06d97f92013-12-02 17:45:48 +010017648in '%%'. HAProxy will automatically merge consecutive separators.
William Lallemand48940402012-01-30 16:47:22 +010017649
Dragan Dosen835b9212016-02-12 13:23:03 +010017650Note: when using the RFC5424 syslog message format, the characters '"',
17651'\' and ']' inside PARAM-VALUE should be escaped with '\' as prefix (see
17652https://tools.ietf.org/html/rfc5424#section-6.3.3 for more details). In
17653such cases, the use of the flag "E" should be considered.
17654
William Lallemand48940402012-01-30 16:47:22 +010017655Flags are :
17656 * Q: quote a string
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017657 * X: hexadecimal representation (IPs, Ports, %Ts, %rt, %pid)
Dragan Dosen835b9212016-02-12 13:23:03 +010017658 * E: escape characters '"', '\' and ']' in a string with '\' as prefix
17659 (intended purpose is for the RFC5424 structured-data log formats)
William Lallemand48940402012-01-30 16:47:22 +010017660
17661 Example:
17662
17663 log-format %T\ %t\ Some\ Text
17664 log-format %{+Q}o\ %t\ %s\ %{-Q}r
17665
Dragan Dosen835b9212016-02-12 13:23:03 +010017666 log-format-sd %{+Q,+E}o\ [exampleSDID@1234\ header=%[capture.req.hdr(0)]]
17667
William Lallemand48940402012-01-30 16:47:22 +010017668At the moment, the default HTTP format is defined this way :
17669
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017670 log-format "%ci:%cp [%tr] %ft %b/%s %TR/%Tw/%Tc/%Tr/%Ta %ST %B %CC \
17671 %CS %tsc %ac/%fc/%bc/%sc/%rc %sq/%bq %hr %hs %{+Q}r"
William Lallemand48940402012-01-30 16:47:22 +010017672
William Lallemandbddd4fd2012-02-27 11:23:10 +010017673the default CLF format is defined this way :
William Lallemand48940402012-01-30 16:47:22 +010017674
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017675 log-format "%{+Q}o %{-Q}ci - - [%trg] %r %ST %B \"\" \"\" %cp \
17676 %ms %ft %b %s %TR %Tw %Tc %Tr %Ta %tsc %ac %fc \
17677 %bc %sc %rc %sq %bq %CC %CS %hrl %hsl"
William Lallemand48940402012-01-30 16:47:22 +010017678
William Lallemandbddd4fd2012-02-27 11:23:10 +010017679and the default TCP format is defined this way :
17680
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017681 log-format "%ci:%cp [%t] %ft %b/%s %Tw/%Tc/%Tt %B %ts \
17682 %ac/%fc/%bc/%sc/%rc %sq/%bq"
William Lallemandbddd4fd2012-02-27 11:23:10 +010017683
William Lallemand48940402012-01-30 16:47:22 +010017684Please refer to the table below for currently defined variables :
17685
William Lallemandbddd4fd2012-02-27 11:23:10 +010017686 +---+------+-----------------------------------------------+-------------+
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020017687 | R | var | field name (8.2.2 and 8.2.3 for description) | type |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017688 +---+------+-----------------------------------------------+-------------+
17689 | | %o | special variable, apply flags on all next var | |
17690 +---+------+-----------------------------------------------+-------------+
Willy Tarreau2beef582012-12-20 17:22:52 +010017691 | | %B | bytes_read (from server to client) | numeric |
17692 | H | %CC | captured_request_cookie | string |
17693 | H | %CS | captured_response_cookie | string |
William Lallemand5f232402012-04-05 18:02:55 +020017694 | | %H | hostname | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000017695 | H | %HM | HTTP method (ex: POST) | string |
17696 | H | %HP | HTTP request URI without query string (path) | string |
Andrew Hayworthe63ac872015-07-31 16:14:16 +000017697 | H | %HQ | HTTP request URI query string (ex: ?bar=baz) | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000017698 | H | %HU | HTTP request URI (ex: /foo?bar=baz) | string |
17699 | H | %HV | HTTP version (ex: HTTP/1.0) | string |
William Lallemanda73203e2012-03-12 12:48:57 +010017700 | | %ID | unique-id | string |
Willy Tarreau4bf99632014-06-13 12:21:40 +020017701 | | %ST | status_code | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020017702 | | %T | gmt_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017703 | | %Ta | Active time of the request (from TR to end) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017704 | | %Tc | Tc | numeric |
Willy Tarreau27b639d2016-05-17 17:55:27 +020017705 | | %Td | Td = Tt - (Tq + Tw + Tc + Tr) | numeric |
Yuxans Yao4e25b012012-10-19 10:36:09 +080017706 | | %Tl | local_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017707 | | %Th | connection handshake time (SSL, PROXY proto) | numeric |
17708 | H | %Ti | idle time before the HTTP request | numeric |
17709 | H | %Tq | Th + Ti + TR | numeric |
17710 | H | %TR | time to receive the full request from 1st byte| numeric |
17711 | H | %Tr | Tr (response time) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020017712 | | %Ts | timestamp | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017713 | | %Tt | Tt | numeric |
17714 | | %Tw | Tw | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010017715 | | %U | bytes_uploaded (from client to server) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017716 | | %ac | actconn | numeric |
17717 | | %b | backend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010017718 | | %bc | beconn (backend concurrent connections) | numeric |
17719 | | %bi | backend_source_ip (connecting address) | IP |
17720 | | %bp | backend_source_port (connecting address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017721 | | %bq | backend_queue | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010017722 | | %ci | client_ip (accepted address) | IP |
17723 | | %cp | client_port (accepted address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017724 | | %f | frontend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010017725 | | %fc | feconn (frontend concurrent connections) | numeric |
17726 | | %fi | frontend_ip (accepting address) | IP |
17727 | | %fp | frontend_port (accepting address) | numeric |
Willy Tarreau773d65f2012-10-12 14:56:11 +020017728 | | %ft | frontend_name_transport ('~' suffix for SSL) | string |
Willy Tarreau7346acb2014-08-28 15:03:15 +020017729 | | %lc | frontend_log_counter | numeric |
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020017730 | | %hr | captured_request_headers default style | string |
17731 | | %hrl | captured_request_headers CLF style | string list |
17732 | | %hs | captured_response_headers default style | string |
17733 | | %hsl | captured_response_headers CLF style | string list |
Willy Tarreau812c88e2015-08-09 10:56:35 +020017734 | | %ms | accept date milliseconds (left-padded with 0) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020017735 | | %pid | PID | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020017736 | H | %r | http_request | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017737 | | %rc | retries | numeric |
Willy Tarreau1f0da242014-01-25 11:01:50 +010017738 | | %rt | request_counter (HTTP req or TCP session) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017739 | | %s | server_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010017740 | | %sc | srv_conn (server concurrent connections) | numeric |
17741 | | %si | server_IP (target address) | IP |
17742 | | %sp | server_port (target address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017743 | | %sq | srv_queue | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020017744 | S | %sslc| ssl_ciphers (ex: AES-SHA) | string |
17745 | S | %sslv| ssl_version (ex: TLSv1) | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010017746 | | %t | date_time (with millisecond resolution) | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017747 | H | %tr | date_time of HTTP request | date |
17748 | H | %trg | gmt_date_time of start of HTTP request | date |
Jens Bissinger15c64ff2018-08-23 14:11:27 +020017749 | H | %trl | local_date_time of start of HTTP request | date |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017750 | | %ts | termination_state | string |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020017751 | H | %tsc | termination_state with cookie status | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017752 +---+------+-----------------------------------------------+-------------+
William Lallemand48940402012-01-30 16:47:22 +010017753
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020017754 R = Restrictions : H = mode http only ; S = SSL only
William Lallemand48940402012-01-30 16:47:22 +010017755
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010017756
177578.2.5. Error log format
17758-----------------------
17759
17760When an incoming connection fails due to an SSL handshake or an invalid PROXY
17761protocol header, haproxy will log the event using a shorter, fixed line format.
17762By default, logs are emitted at the LOG_INFO level, unless the option
17763"log-separate-errors" is set in the backend, in which case the LOG_ERR level
Davor Ocelice9ed2812017-12-25 17:49:28 +010017764will be used. Connections on which no data are exchanged (e.g. probes) are not
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010017765logged if the "dontlognull" option is set.
17766
17767The format looks like this :
17768
17769 >>> Dec 3 18:27:14 localhost \
17770 haproxy[6103]: 127.0.0.1:56059 [03/Dec/2012:17:35:10.380] frt/f1: \
17771 Connection error during SSL handshake
17772
17773 Field Format Extract from the example above
17774 1 process_name '[' pid ']:' haproxy[6103]:
17775 2 client_ip ':' client_port 127.0.0.1:56059
17776 3 '[' accept_date ']' [03/Dec/2012:17:35:10.380]
17777 4 frontend_name "/" bind_name ":" frt/f1:
17778 5 message Connection error during SSL handshake
17779
17780These fields just provide minimal information to help debugging connection
17781failures.
17782
17783
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200177848.3. Advanced logging options
17785-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017786
17787Some advanced logging options are often looked for but are not easy to find out
17788just by looking at the various options. Here is an entry point for the few
17789options which can enable better logging. Please refer to the keywords reference
17790for more information about their usage.
17791
17792
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200177938.3.1. Disabling logging of external tests
17794------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017795
17796It is quite common to have some monitoring tools perform health checks on
17797haproxy. Sometimes it will be a layer 3 load-balancer such as LVS or any
17798commercial load-balancer, and sometimes it will simply be a more complete
17799monitoring system such as Nagios. When the tests are very frequent, users often
17800ask how to disable logging for those checks. There are three possibilities :
17801
17802 - if connections come from everywhere and are just TCP probes, it is often
17803 desired to simply disable logging of connections without data exchange, by
17804 setting "option dontlognull" in the frontend. It also disables logging of
17805 port scans, which may or may not be desired.
17806
17807 - if the connection come from a known source network, use "monitor-net" to
17808 declare this network as monitoring only. Any host in this network will then
17809 only be able to perform health checks, and their requests will not be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017810 logged. This is generally appropriate to designate a list of equipment
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017811 such as other load-balancers.
17812
17813 - if the tests are performed on a known URI, use "monitor-uri" to declare
17814 this URI as dedicated to monitoring. Any host sending this request will
17815 only get the result of a health-check, and the request will not be logged.
17816
17817
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200178188.3.2. Logging before waiting for the session to terminate
17819----------------------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017820
17821The problem with logging at end of connection is that you have no clue about
17822what is happening during very long sessions, such as remote terminal sessions
17823or large file downloads. This problem can be worked around by specifying
Davor Ocelice9ed2812017-12-25 17:49:28 +010017824"option logasap" in the frontend. HAProxy will then log as soon as possible,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017825just before data transfer begins. This means that in case of TCP, it will still
17826log the connection status to the server, and in case of HTTP, it will log just
17827after processing the server headers. In this case, the number of bytes reported
17828is the number of header bytes sent to the client. In order to avoid confusion
17829with normal logs, the total time field and the number of bytes are prefixed
17830with a '+' sign which means that real numbers are certainly larger.
17831
17832
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200178338.3.3. Raising log level upon errors
17834------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020017835
17836Sometimes it is more convenient to separate normal traffic from errors logs,
17837for instance in order to ease error monitoring from log files. When the option
17838"log-separate-errors" is used, connections which experience errors, timeouts,
17839retries, redispatches or HTTP status codes 5xx will see their syslog level
17840raised from "info" to "err". This will help a syslog daemon store the log in
17841a separate file. It is very important to keep the errors in the normal traffic
17842file too, so that log ordering is not altered. You should also be careful if
17843you already have configured your syslog daemon to store all logs higher than
17844"notice" in an "admin" file, because the "err" level is higher than "notice".
17845
17846
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200178478.3.4. Disabling logging of successful connections
17848--------------------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020017849
17850Although this may sound strange at first, some large sites have to deal with
17851multiple thousands of logs per second and are experiencing difficulties keeping
17852them intact for a long time or detecting errors within them. If the option
17853"dontlog-normal" is set on the frontend, all normal connections will not be
17854logged. In this regard, a normal connection is defined as one without any
17855error, timeout, retry nor redispatch. In HTTP, the status code is checked too,
17856and a response with a status 5xx is not considered normal and will be logged
17857too. Of course, doing is is really discouraged as it will remove most of the
17858useful information from the logs. Do this only if you have no other
17859alternative.
17860
17861
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200178628.4. Timing events
17863------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017864
17865Timers provide a great help in troubleshooting network problems. All values are
17866reported in milliseconds (ms). These timers should be used in conjunction with
17867the session termination flags. In TCP mode with "option tcplog" set on the
17868frontend, 3 control points are reported under the form "Tw/Tc/Tt", and in HTTP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017869mode, 5 control points are reported under the form "TR/Tw/Tc/Tr/Ta". In
17870addition, three other measures are provided, "Th", "Ti", and "Tq".
17871
Guillaume de Lafondf27cddc2016-12-23 17:32:43 +010017872Timings events in HTTP mode:
17873
17874 first request 2nd request
17875 |<-------------------------------->|<-------------- ...
17876 t tr t tr ...
17877 ---|----|----|----|----|----|----|----|----|--
17878 : Th Ti TR Tw Tc Tr Td : Ti ...
17879 :<---- Tq ---->: :
17880 :<-------------- Tt -------------->:
17881 :<--------- Ta --------->:
17882
17883Timings events in TCP mode:
17884
17885 TCP session
17886 |<----------------->|
17887 t t
17888 ---|----|----|----|----|---
17889 | Th Tw Tc Td |
17890 |<------ Tt ------->|
17891
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017892 - Th: total time to accept tcp connection and execute handshakes for low level
Davor Ocelice9ed2812017-12-25 17:49:28 +010017893 protocols. Currently, these protocols are proxy-protocol and SSL. This may
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017894 only happen once during the whole connection's lifetime. A large time here
17895 may indicate that the client only pre-established the connection without
17896 speaking, that it is experiencing network issues preventing it from
Davor Ocelice9ed2812017-12-25 17:49:28 +010017897 completing a handshake in a reasonable time (e.g. MTU issues), or that an
Willy Tarreau590a0512018-09-05 11:56:48 +020017898 SSL handshake was very expensive to compute. Please note that this time is
17899 reported only before the first request, so it is safe to average it over
17900 all request to calculate the amortized value. The second and subsequent
17901 request will always report zero here.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017902
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017903 - Ti: is the idle time before the HTTP request (HTTP mode only). This timer
17904 counts between the end of the handshakes and the first byte of the HTTP
17905 request. When dealing with a second request in keep-alive mode, it starts
Willy Tarreau590a0512018-09-05 11:56:48 +020017906 to count after the end of the transmission the previous response. When a
17907 multiplexed protocol such as HTTP/2 is used, it starts to count immediately
17908 after the previous request. Some browsers pre-establish connections to a
17909 server in order to reduce the latency of a future request, and keep them
17910 pending until they need it. This delay will be reported as the idle time. A
17911 value of -1 indicates that nothing was received on the connection.
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017912
17913 - TR: total time to get the client request (HTTP mode only). It's the time
17914 elapsed between the first bytes received and the moment the proxy received
17915 the empty line marking the end of the HTTP headers. The value "-1"
17916 indicates that the end of headers has never been seen. This happens when
17917 the client closes prematurely or times out. This time is usually very short
17918 since most requests fit in a single packet. A large time may indicate a
17919 request typed by hand during a test.
17920
17921 - Tq: total time to get the client request from the accept date or since the
17922 emission of the last byte of the previous response (HTTP mode only). It's
Davor Ocelice9ed2812017-12-25 17:49:28 +010017923 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 +020017924 returns -1 as well. This timer used to be very useful before the arrival of
17925 HTTP keep-alive and browsers' pre-connect feature. It's recommended to drop
17926 it in favor of TR nowadays, as the idle time adds a lot of noise to the
17927 reports.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017928
17929 - Tw: total time spent in the queues waiting for a connection slot. It
17930 accounts for backend queue as well as the server queues, and depends on the
17931 queue size, and the time needed for the server to complete previous
17932 requests. The value "-1" means that the request was killed before reaching
17933 the queue, which is generally what happens with invalid or denied requests.
17934
17935 - Tc: total time to establish the TCP connection to the server. It's the time
17936 elapsed between the moment the proxy sent the connection request, and the
17937 moment it was acknowledged by the server, or between the TCP SYN packet and
17938 the matching SYN/ACK packet in return. The value "-1" means that the
17939 connection never established.
17940
17941 - Tr: server response time (HTTP mode only). It's the time elapsed between
17942 the moment the TCP connection was established to the server and the moment
17943 the server sent its complete response headers. It purely shows its request
17944 processing time, without the network overhead due to the data transmission.
17945 It is worth noting that when the client has data to send to the server, for
17946 instance during a POST request, the time already runs, and this can distort
17947 apparent response time. For this reason, it's generally wise not to trust
17948 too much this field for POST requests initiated from clients behind an
17949 untrusted network. A value of "-1" here means that the last the response
17950 header (empty line) was never seen, most likely because the server timeout
17951 stroke before the server managed to process the request.
17952
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017953 - Ta: total active time for the HTTP request, between the moment the proxy
17954 received the first byte of the request header and the emission of the last
17955 byte of the response body. The exception is when the "logasap" option is
17956 specified. In this case, it only equals (TR+Tw+Tc+Tr), and is prefixed with
17957 a '+' sign. From this field, we can deduce "Td", the data transmission time,
17958 by subtracting other timers when valid :
17959
17960 Td = Ta - (TR + Tw + Tc + Tr)
17961
17962 Timers with "-1" values have to be excluded from this equation. Note that
17963 "Ta" can never be negative.
17964
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017965 - Tt: total session duration time, between the moment the proxy accepted it
17966 and the moment both ends were closed. The exception is when the "logasap"
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017967 option is specified. In this case, it only equals (Th+Ti+TR+Tw+Tc+Tr), and
17968 is prefixed with a '+' sign. From this field, we can deduce "Td", the data
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017969 transmission time, by subtracting other timers when valid :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017970
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017971 Td = Tt - (Th + Ti + TR + Tw + Tc + Tr)
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017972
17973 Timers with "-1" values have to be excluded from this equation. In TCP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017974 mode, "Ti", "Tq" and "Tr" have to be excluded too. Note that "Tt" can never
17975 be negative and that for HTTP, Tt is simply equal to (Th+Ti+Ta).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017976
17977These timers provide precious indications on trouble causes. Since the TCP
17978protocol defines retransmit delays of 3, 6, 12... seconds, we know for sure
17979that timers close to multiples of 3s are nearly always related to lost packets
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017980due to network problems (wires, negotiation, congestion). Moreover, if "Ta" or
17981"Tt" is close to a timeout value specified in the configuration, it often means
17982that a session has been aborted on timeout.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017983
17984Most common cases :
17985
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017986 - If "Th" or "Ti" are close to 3000, a packet has probably been lost between
17987 the client and the proxy. This is very rare on local networks but might
17988 happen when clients are on far remote networks and send large requests. It
17989 may happen that values larger than usual appear here without any network
17990 cause. Sometimes, during an attack or just after a resource starvation has
17991 ended, haproxy may accept thousands of connections in a few milliseconds.
17992 The time spent accepting these connections will inevitably slightly delay
17993 processing of other connections, and it can happen that request times in the
17994 order of a few tens of milliseconds are measured after a few thousands of
17995 new connections have been accepted at once. Using one of the keep-alive
17996 modes may display larger idle times since "Ti" measures the time spent
Patrick Mezard105faca2010-06-12 17:02:46 +020017997 waiting for additional requests.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017998
17999 - If "Tc" is close to 3000, a packet has probably been lost between the
18000 server and the proxy during the server connection phase. This value should
18001 always be very low, such as 1 ms on local networks and less than a few tens
18002 of ms on remote networks.
18003
Willy Tarreau55165fe2009-05-10 12:02:55 +020018004 - If "Tr" is nearly always lower than 3000 except some rare values which seem
18005 to be the average majored by 3000, there are probably some packets lost
18006 between the proxy and the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018007
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020018008 - If "Ta" is large even for small byte counts, it generally is because
18009 neither the client nor the server decides to close the connection while
18010 haproxy is running in tunnel mode and both have agreed on a keep-alive
18011 connection mode. In order to solve this issue, it will be needed to specify
18012 one of the HTTP options to manipulate keep-alive or close options on either
18013 the frontend or the backend. Having the smallest possible 'Ta' or 'Tt' is
18014 important when connection regulation is used with the "maxconn" option on
18015 the servers, since no new connection will be sent to the server until
18016 another one is released.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018017
18018Other noticeable HTTP log cases ('xx' means any value to be ignored) :
18019
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020018020 TR/Tw/Tc/Tr/+Ta The "option logasap" is present on the frontend and the log
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018021 was emitted before the data phase. All the timers are valid
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020018022 except "Ta" which is shorter than reality.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018023
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020018024 -1/xx/xx/xx/Ta The client was not able to send a complete request in time
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018025 or it aborted too early. Check the session termination flags
18026 then "timeout http-request" and "timeout client" settings.
18027
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020018028 TR/-1/xx/xx/Ta It was not possible to process the request, maybe because
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018029 servers were out of order, because the request was invalid
18030 or forbidden by ACL rules. Check the session termination
18031 flags.
18032
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020018033 TR/Tw/-1/xx/Ta The connection could not establish on the server. Either it
18034 actively refused it or it timed out after Ta-(TR+Tw) ms.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018035 Check the session termination flags, then check the
18036 "timeout connect" setting. Note that the tarpit action might
18037 return similar-looking patterns, with "Tw" equal to the time
18038 the client connection was maintained open.
18039
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020018040 TR/Tw/Tc/-1/Ta The server has accepted the connection but did not return
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030018041 a complete response in time, or it closed its connection
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020018042 unexpectedly after Ta-(TR+Tw+Tc) ms. Check the session
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018043 termination flags, then check the "timeout server" setting.
18044
18045
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200180468.5. Session state at disconnection
18047-----------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018048
18049TCP and HTTP logs provide a session termination indicator in the
18050"termination_state" field, just before the number of active connections. It is
180512-characters long in TCP mode, and is extended to 4 characters in HTTP mode,
18052each of which has a special meaning :
18053
18054 - On the first character, a code reporting the first event which caused the
18055 session to terminate :
18056
18057 C : the TCP session was unexpectedly aborted by the client.
18058
18059 S : the TCP session was unexpectedly aborted by the server, or the
18060 server explicitly refused it.
18061
18062 P : the session was prematurely aborted by the proxy, because of a
18063 connection limit enforcement, because a DENY filter was matched,
18064 because of a security check which detected and blocked a dangerous
18065 error in server response which might have caused information leak
Davor Ocelice9ed2812017-12-25 17:49:28 +010018066 (e.g. cacheable cookie).
Willy Tarreau570f2212013-06-10 16:42:09 +020018067
18068 L : the session was locally processed by haproxy and was not passed to
18069 a server. This is what happens for stats and redirects.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018070
18071 R : a resource on the proxy has been exhausted (memory, sockets, source
18072 ports, ...). Usually, this appears during the connection phase, and
18073 system logs should contain a copy of the precise error. If this
18074 happens, it must be considered as a very serious anomaly which
18075 should be fixed as soon as possible by any means.
18076
18077 I : an internal error was identified by the proxy during a self-check.
18078 This should NEVER happen, and you are encouraged to report any log
18079 containing this, because this would almost certainly be a bug. It
18080 would be wise to preventively restart the process after such an
18081 event too, in case it would be caused by memory corruption.
18082
Simon Horman752dc4a2011-06-21 14:34:59 +090018083 D : the session was killed by haproxy because the server was detected
18084 as down and was configured to kill all connections when going down.
18085
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070018086 U : the session was killed by haproxy on this backup server because an
18087 active server was detected as up and was configured to kill all
18088 backup connections when going up.
18089
Willy Tarreaua2a64e92011-09-07 23:01:56 +020018090 K : the session was actively killed by an admin operating on haproxy.
18091
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018092 c : the client-side timeout expired while waiting for the client to
18093 send or receive data.
18094
18095 s : the server-side timeout expired while waiting for the server to
18096 send or receive data.
18097
18098 - : normal session completion, both the client and the server closed
18099 with nothing left in the buffers.
18100
18101 - on the second character, the TCP or HTTP session state when it was closed :
18102
Willy Tarreauf7b30a92010-12-06 22:59:17 +010018103 R : the proxy was waiting for a complete, valid REQUEST from the client
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018104 (HTTP mode only). Nothing was sent to any server.
18105
18106 Q : the proxy was waiting in the QUEUE for a connection slot. This can
18107 only happen when servers have a 'maxconn' parameter set. It can
18108 also happen in the global queue after a redispatch consecutive to
18109 a failed attempt to connect to a dying server. If no redispatch is
18110 reported, then no connection attempt was made to any server.
18111
18112 C : the proxy was waiting for the CONNECTION to establish on the
18113 server. The server might at most have noticed a connection attempt.
18114
18115 H : the proxy was waiting for complete, valid response HEADERS from the
18116 server (HTTP only).
18117
18118 D : the session was in the DATA phase.
18119
18120 L : the proxy was still transmitting LAST data to the client while the
18121 server had already finished. This one is very rare as it can only
18122 happen when the client dies while receiving the last packets.
18123
18124 T : the request was tarpitted. It has been held open with the client
18125 during the whole "timeout tarpit" duration or until the client
18126 closed, both of which will be reported in the "Tw" timer.
18127
18128 - : normal session completion after end of data transfer.
18129
18130 - the third character tells whether the persistence cookie was provided by
18131 the client (only in HTTP mode) :
18132
18133 N : the client provided NO cookie. This is usually the case for new
18134 visitors, so counting the number of occurrences of this flag in the
18135 logs generally indicate a valid trend for the site frequentation.
18136
18137 I : the client provided an INVALID cookie matching no known server.
18138 This might be caused by a recent configuration change, mixed
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +020018139 cookies between HTTP/HTTPS sites, persistence conditionally
18140 ignored, or an attack.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018141
18142 D : the client provided a cookie designating a server which was DOWN,
18143 so either "option persist" was used and the client was sent to
18144 this server, or it was not set and the client was redispatched to
18145 another server.
18146
Willy Tarreau996a92c2010-10-13 19:30:47 +020018147 V : the client provided a VALID cookie, and was sent to the associated
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018148 server.
18149
Willy Tarreau996a92c2010-10-13 19:30:47 +020018150 E : the client provided a valid cookie, but with a last date which was
18151 older than what is allowed by the "maxidle" cookie parameter, so
18152 the cookie is consider EXPIRED and is ignored. The request will be
18153 redispatched just as if there was no cookie.
18154
18155 O : the client provided a valid cookie, but with a first date which was
18156 older than what is allowed by the "maxlife" cookie parameter, so
18157 the cookie is consider too OLD and is ignored. The request will be
18158 redispatched just as if there was no cookie.
18159
Willy Tarreauc89ccb62012-04-05 21:18:22 +020018160 U : a cookie was present but was not used to select the server because
18161 some other server selection mechanism was used instead (typically a
18162 "use-server" rule).
18163
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018164 - : does not apply (no cookie set in configuration).
18165
18166 - the last character reports what operations were performed on the persistence
18167 cookie returned by the server (only in HTTP mode) :
18168
18169 N : NO cookie was provided by the server, and none was inserted either.
18170
18171 I : no cookie was provided by the server, and the proxy INSERTED one.
18172 Note that in "cookie insert" mode, if the server provides a cookie,
18173 it will still be overwritten and reported as "I" here.
18174
Willy Tarreau996a92c2010-10-13 19:30:47 +020018175 U : the proxy UPDATED the last date in the cookie that was presented by
18176 the client. This can only happen in insert mode with "maxidle". It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030018177 happens every time there is activity at a different date than the
Willy Tarreau996a92c2010-10-13 19:30:47 +020018178 date indicated in the cookie. If any other change happens, such as
18179 a redispatch, then the cookie will be marked as inserted instead.
18180
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018181 P : a cookie was PROVIDED by the server and transmitted as-is.
18182
18183 R : the cookie provided by the server was REWRITTEN by the proxy, which
18184 happens in "cookie rewrite" or "cookie prefix" modes.
18185
18186 D : the cookie provided by the server was DELETED by the proxy.
18187
18188 - : does not apply (no cookie set in configuration).
18189
Willy Tarreau996a92c2010-10-13 19:30:47 +020018190The combination of the two first flags gives a lot of information about what
18191was happening when the session terminated, and why it did terminate. It can be
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018192helpful to detect server saturation, network troubles, local system resource
18193starvation, attacks, etc...
18194
18195The most common termination flags combinations are indicated below. They are
18196alphabetically sorted, with the lowercase set just after the upper case for
18197easier finding and understanding.
18198
18199 Flags Reason
18200
18201 -- Normal termination.
18202
18203 CC The client aborted before the connection could be established to the
18204 server. This can happen when haproxy tries to connect to a recently
18205 dead (or unchecked) server, and the client aborts while haproxy is
18206 waiting for the server to respond or for "timeout connect" to expire.
18207
18208 CD The client unexpectedly aborted during data transfer. This can be
18209 caused by a browser crash, by an intermediate equipment between the
18210 client and haproxy which decided to actively break the connection,
18211 by network routing issues between the client and haproxy, or by a
18212 keep-alive session between the server and the client terminated first
18213 by the client.
Willy Tarreaud72758d2010-01-12 10:42:19 +010018214
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018215 cD The client did not send nor acknowledge any data for as long as the
18216 "timeout client" delay. This is often caused by network failures on
Cyril Bontédc4d9032012-04-08 21:57:39 +020018217 the client side, or the client simply leaving the net uncleanly.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018218
18219 CH The client aborted while waiting for the server to start responding.
18220 It might be the server taking too long to respond or the client
18221 clicking the 'Stop' button too fast.
18222
18223 cH The "timeout client" stroke while waiting for client data during a
18224 POST request. This is sometimes caused by too large TCP MSS values
18225 for PPPoE networks which cannot transport full-sized packets. It can
18226 also happen when client timeout is smaller than server timeout and
18227 the server takes too long to respond.
18228
18229 CQ The client aborted while its session was queued, waiting for a server
18230 with enough empty slots to accept it. It might be that either all the
18231 servers were saturated or that the assigned server was taking too
18232 long a time to respond.
18233
18234 CR The client aborted before sending a full HTTP request. Most likely
18235 the request was typed by hand using a telnet client, and aborted
18236 too early. The HTTP status code is likely a 400 here. Sometimes this
18237 might also be caused by an IDS killing the connection between haproxy
Willy Tarreau0f228a02015-05-01 15:37:53 +020018238 and the client. "option http-ignore-probes" can be used to ignore
18239 connections without any data transfer.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018240
18241 cR The "timeout http-request" stroke before the client sent a full HTTP
18242 request. This is sometimes caused by too large TCP MSS values on the
18243 client side for PPPoE networks which cannot transport full-sized
18244 packets, or by clients sending requests by hand and not typing fast
18245 enough, or forgetting to enter the empty line at the end of the
Willy Tarreau2705a612014-05-23 17:38:34 +020018246 request. The HTTP status code is likely a 408 here. Note: recently,
Willy Tarreau0f228a02015-05-01 15:37:53 +020018247 some browsers started to implement a "pre-connect" feature consisting
18248 in speculatively connecting to some recently visited web sites just
18249 in case the user would like to visit them. This results in many
18250 connections being established to web sites, which end up in 408
18251 Request Timeout if the timeout strikes first, or 400 Bad Request when
18252 the browser decides to close them first. These ones pollute the log
18253 and feed the error counters. Some versions of some browsers have even
18254 been reported to display the error code. It is possible to work
Davor Ocelice9ed2812017-12-25 17:49:28 +010018255 around the undesirable effects of this behavior by adding "option
Willy Tarreau0f228a02015-05-01 15:37:53 +020018256 http-ignore-probes" in the frontend, resulting in connections with
18257 zero data transfer to be totally ignored. This will definitely hide
18258 the errors of people experiencing connectivity issues though.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018259
18260 CT The client aborted while its session was tarpitted. It is important to
18261 check if this happens on valid requests, in order to be sure that no
Willy Tarreau55165fe2009-05-10 12:02:55 +020018262 wrong tarpit rules have been written. If a lot of them happen, it
18263 might make sense to lower the "timeout tarpit" value to something
18264 closer to the average reported "Tw" timer, in order not to consume
18265 resources for just a few attackers.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018266
Willy Tarreau570f2212013-06-10 16:42:09 +020018267 LR The request was intercepted and locally handled by haproxy. Generally
18268 it means that this was a redirect or a stats request.
18269
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010018270 SC The server or an equipment between it and haproxy explicitly refused
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018271 the TCP connection (the proxy received a TCP RST or an ICMP message
18272 in return). Under some circumstances, it can also be the network
Davor Ocelice9ed2812017-12-25 17:49:28 +010018273 stack telling the proxy that the server is unreachable (e.g. no route,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018274 or no ARP response on local network). When this happens in HTTP mode,
18275 the status code is likely a 502 or 503 here.
18276
18277 sC The "timeout connect" stroke before a connection to the server could
18278 complete. When this happens in HTTP mode, the status code is likely a
18279 503 or 504 here.
18280
18281 SD The connection to the server died with an error during the data
18282 transfer. This usually means that haproxy has received an RST from
18283 the server or an ICMP message from an intermediate equipment while
18284 exchanging data with the server. This can be caused by a server crash
18285 or by a network issue on an intermediate equipment.
18286
18287 sD The server did not send nor acknowledge any data for as long as the
18288 "timeout server" setting during the data phase. This is often caused
Davor Ocelice9ed2812017-12-25 17:49:28 +010018289 by too short timeouts on L4 equipment before the server (firewalls,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018290 load-balancers, ...), as well as keep-alive sessions maintained
18291 between the client and the server expiring first on haproxy.
18292
18293 SH The server aborted before sending its full HTTP response headers, or
18294 it crashed while processing the request. Since a server aborting at
18295 this moment is very rare, it would be wise to inspect its logs to
18296 control whether it crashed and why. The logged request may indicate a
18297 small set of faulty requests, demonstrating bugs in the application.
18298 Sometimes this might also be caused by an IDS killing the connection
18299 between haproxy and the server.
18300
18301 sH The "timeout server" stroke before the server could return its
18302 response headers. This is the most common anomaly, indicating too
18303 long transactions, probably caused by server or database saturation.
18304 The immediate workaround consists in increasing the "timeout server"
18305 setting, but it is important to keep in mind that the user experience
18306 will suffer from these long response times. The only long term
18307 solution is to fix the application.
18308
18309 sQ The session spent too much time in queue and has been expired. See
18310 the "timeout queue" and "timeout connect" settings to find out how to
18311 fix this if it happens too often. If it often happens massively in
18312 short periods, it may indicate general problems on the affected
18313 servers due to I/O or database congestion, or saturation caused by
18314 external attacks.
18315
18316 PC The proxy refused to establish a connection to the server because the
18317 process' socket limit has been reached while attempting to connect.
Cyril Bontédc4d9032012-04-08 21:57:39 +020018318 The global "maxconn" parameter may be increased in the configuration
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018319 so that it does not happen anymore. This status is very rare and
18320 might happen when the global "ulimit-n" parameter is forced by hand.
18321
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010018322 PD The proxy blocked an incorrectly formatted chunked encoded message in
18323 a request or a response, after the server has emitted its headers. In
18324 most cases, this will indicate an invalid message from the server to
Davor Ocelice9ed2812017-12-25 17:49:28 +010018325 the client. HAProxy supports chunk sizes of up to 2GB - 1 (2147483647
Willy Tarreauf3a3e132013-08-31 08:16:26 +020018326 bytes). Any larger size will be considered as an error.
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010018327
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018328 PH The proxy blocked the server's response, because it was invalid,
18329 incomplete, dangerous (cache control), or matched a security filter.
18330 In any case, an HTTP 502 error is sent to the client. One possible
18331 cause for this error is an invalid syntax in an HTTP header name
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010018332 containing unauthorized characters. It is also possible but quite
18333 rare, that the proxy blocked a chunked-encoding request from the
18334 client due to an invalid syntax, before the server responded. In this
18335 case, an HTTP 400 error is sent to the client and reported in the
18336 logs.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018337
18338 PR The proxy blocked the client's HTTP request, either because of an
18339 invalid HTTP syntax, in which case it returned an HTTP 400 error to
18340 the client, or because a deny filter matched, in which case it
18341 returned an HTTP 403 error.
18342
18343 PT The proxy blocked the client's request and has tarpitted its
18344 connection before returning it a 500 server error. Nothing was sent
18345 to the server. The connection was maintained open for as long as
18346 reported by the "Tw" timer field.
18347
18348 RC A local resource has been exhausted (memory, sockets, source ports)
18349 preventing the connection to the server from establishing. The error
18350 logs will tell precisely what was missing. This is very rare and can
18351 only be solved by proper system tuning.
18352
Willy Tarreau996a92c2010-10-13 19:30:47 +020018353The combination of the two last flags gives a lot of information about how
18354persistence was handled by the client, the server and by haproxy. This is very
18355important to troubleshoot disconnections, when users complain they have to
18356re-authenticate. The commonly encountered flags are :
18357
18358 -- Persistence cookie is not enabled.
18359
18360 NN No cookie was provided by the client, none was inserted in the
18361 response. For instance, this can be in insert mode with "postonly"
18362 set on a GET request.
18363
18364 II A cookie designating an invalid server was provided by the client,
18365 a valid one was inserted in the response. This typically happens when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040018366 a "server" entry is removed from the configuration, since its cookie
Willy Tarreau996a92c2010-10-13 19:30:47 +020018367 value can be presented by a client when no other server knows it.
18368
18369 NI No cookie was provided by the client, one was inserted in the
18370 response. This typically happens for first requests from every user
18371 in "insert" mode, which makes it an easy way to count real users.
18372
18373 VN A cookie was provided by the client, none was inserted in the
18374 response. This happens for most responses for which the client has
18375 already got a cookie.
18376
18377 VU A cookie was provided by the client, with a last visit date which is
18378 not completely up-to-date, so an updated cookie was provided in
18379 response. This can also happen if there was no date at all, or if
18380 there was a date but the "maxidle" parameter was not set, so that the
18381 cookie can be switched to unlimited time.
18382
18383 EI A cookie was provided by the client, with a last visit date which is
18384 too old for the "maxidle" parameter, so the cookie was ignored and a
18385 new cookie was inserted in the response.
18386
18387 OI A cookie was provided by the client, with a first visit date which is
18388 too old for the "maxlife" parameter, so the cookie was ignored and a
18389 new cookie was inserted in the response.
18390
18391 DI The server designated by the cookie was down, a new server was
18392 selected and a new cookie was emitted in the response.
18393
18394 VI The server designated by the cookie was not marked dead but could not
18395 be reached. A redispatch happened and selected another one, which was
18396 then advertised in the response.
18397
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018398
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200183998.6. Non-printable characters
18400-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018401
18402In order not to cause trouble to log analysis tools or terminals during log
18403consulting, non-printable characters are not sent as-is into log files, but are
18404converted to the two-digits hexadecimal representation of their ASCII code,
18405prefixed by the character '#'. The only characters that can be logged without
18406being escaped are comprised between 32 and 126 (inclusive). Obviously, the
18407escape character '#' itself is also encoded to avoid any ambiguity ("#23"). It
18408is the same for the character '"' which becomes "#22", as well as '{', '|' and
18409'}' when logging headers.
18410
18411Note that the space character (' ') is not encoded in headers, which can cause
18412issues for tools relying on space count to locate fields. A typical header
18413containing spaces is "User-Agent".
18414
18415Last, it has been observed that some syslog daemons such as syslog-ng escape
18416the quote ('"') with a backslash ('\'). The reverse operation can safely be
18417performed since no quote may appear anywhere else in the logs.
18418
18419
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200184208.7. Capturing HTTP cookies
18421---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018422
18423Cookie capture simplifies the tracking a complete user session. This can be
18424achieved using the "capture cookie" statement in the frontend. Please refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018425section 4.2 for more details. Only one cookie can be captured, and the same
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018426cookie will simultaneously be checked in the request ("Cookie:" header) and in
18427the response ("Set-Cookie:" header). The respective values will be reported in
18428the HTTP logs at the "captured_request_cookie" and "captured_response_cookie"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018429locations (see section 8.2.3 about HTTP log format). When either cookie is
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018430not seen, a dash ('-') replaces the value. This way, it's easy to detect when a
18431user switches to a new session for example, because the server will reassign it
18432a new cookie. It is also possible to detect if a server unexpectedly sets a
18433wrong cookie to a client, leading to session crossing.
18434
18435 Examples :
18436 # capture the first cookie whose name starts with "ASPSESSION"
18437 capture cookie ASPSESSION len 32
18438
18439 # capture the first cookie whose name is exactly "vgnvisitor"
18440 capture cookie vgnvisitor= len 32
18441
18442
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200184438.8. Capturing HTTP headers
18444---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018445
18446Header captures are useful to track unique request identifiers set by an upper
18447proxy, virtual host names, user-agents, POST content-length, referrers, etc. In
18448the response, one can search for information about the response length, how the
18449server asked the cache to behave, or an object location during a redirection.
18450
18451Header captures are performed using the "capture request header" and "capture
18452response header" statements in the frontend. Please consult their definition in
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018453section 4.2 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018454
18455It is possible to include both request headers and response headers at the same
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010018456time. Non-existent headers are logged as empty strings, and if one header
18457appears more than once, only its last occurrence will be logged. Request headers
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018458are grouped within braces '{' and '}' in the same order as they were declared,
18459and delimited with a vertical bar '|' without any space. Response headers
18460follow the same representation, but are displayed after a space following the
18461request headers block. These blocks are displayed just before the HTTP request
18462in the logs.
18463
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020018464As a special case, it is possible to specify an HTTP header capture in a TCP
18465frontend. The purpose is to enable logging of headers which will be parsed in
18466an HTTP backend if the request is then switched to this HTTP backend.
18467
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018468 Example :
18469 # This instance chains to the outgoing proxy
18470 listen proxy-out
18471 mode http
18472 option httplog
18473 option logasap
18474 log global
18475 server cache1 192.168.1.1:3128
18476
18477 # log the name of the virtual server
18478 capture request header Host len 20
18479
18480 # log the amount of data uploaded during a POST
18481 capture request header Content-Length len 10
18482
18483 # log the beginning of the referrer
18484 capture request header Referer len 20
18485
18486 # server name (useful for outgoing proxies only)
18487 capture response header Server len 20
18488
18489 # logging the content-length is useful with "option logasap"
18490 capture response header Content-Length len 10
18491
Davor Ocelice9ed2812017-12-25 17:49:28 +010018492 # log the expected cache behavior on the response
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018493 capture response header Cache-Control len 8
18494
18495 # the Via header will report the next proxy's name
18496 capture response header Via len 20
18497
18498 # log the URL location during a redirection
18499 capture response header Location len 20
18500
18501 >>> Aug 9 20:26:09 localhost \
18502 haproxy[2022]: 127.0.0.1:34014 [09/Aug/2004:20:26:09] proxy-out \
18503 proxy-out/cache1 0/0/0/162/+162 200 +350 - - ---- 0/0/0/0/0 0/0 \
18504 {fr.adserver.yahoo.co||http://fr.f416.mail.} {|864|private||} \
18505 "GET http://fr.adserver.yahoo.com/"
18506
18507 >>> Aug 9 20:30:46 localhost \
18508 haproxy[2022]: 127.0.0.1:34020 [09/Aug/2004:20:30:46] proxy-out \
18509 proxy-out/cache1 0/0/0/182/+182 200 +279 - - ---- 0/0/0/0/0 0/0 \
18510 {w.ods.org||} {Formilux/0.1.8|3495|||} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010018511 "GET http://trafic.1wt.eu/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018512
18513 >>> Aug 9 20:30:46 localhost \
18514 haproxy[2022]: 127.0.0.1:34028 [09/Aug/2004:20:30:46] proxy-out \
18515 proxy-out/cache1 0/0/2/126/+128 301 +223 - - ---- 0/0/0/0/0 0/0 \
18516 {www.sytadin.equipement.gouv.fr||http://trafic.1wt.eu/} \
18517 {Apache|230|||http://www.sytadin.} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010018518 "GET http://www.sytadin.equipement.gouv.fr/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018519
18520
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200185218.9. Examples of logs
18522---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018523
18524These are real-world examples of logs accompanied with an explanation. Some of
18525them have been made up by hand. The syslog part has been removed for better
18526reading. Their sole purpose is to explain how to decipher them.
18527
18528 >>> haproxy[674]: 127.0.0.1:33318 [15/Oct/2003:08:31:57.130] px-http \
18529 px-http/srv1 6559/0/7/147/6723 200 243 - - ---- 5/3/3/1/0 0/0 \
18530 "HEAD / HTTP/1.0"
18531
18532 => long request (6.5s) entered by hand through 'telnet'. The server replied
18533 in 147 ms, and the session ended normally ('----')
18534
18535 >>> haproxy[674]: 127.0.0.1:33319 [15/Oct/2003:08:31:57.149] px-http \
18536 px-http/srv1 6559/1230/7/147/6870 200 243 - - ---- 324/239/239/99/0 \
18537 0/9 "HEAD / HTTP/1.0"
18538
18539 => Idem, but the request was queued in the global queue behind 9 other
18540 requests, and waited there for 1230 ms.
18541
18542 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.654] px-http \
18543 px-http/srv1 9/0/7/14/+30 200 +243 - - ---- 3/3/3/1/0 0/0 \
18544 "GET /image.iso HTTP/1.0"
18545
18546 => request for a long data transfer. The "logasap" option was specified, so
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010018547 the log was produced just before transferring data. The server replied in
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018548 14 ms, 243 bytes of headers were sent to the client, and total time from
18549 accept to first data byte is 30 ms.
18550
18551 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.925] px-http \
18552 px-http/srv1 9/0/7/14/30 502 243 - - PH-- 3/2/2/0/0 0/0 \
18553 "GET /cgi-bin/bug.cgi? HTTP/1.0"
18554
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020018555 => the proxy blocked a server response either because of an "http-response
18556 deny" rule, or because the response was improperly formatted and not
18557 HTTP-compliant, or because it blocked sensitive information which risked
18558 being cached. In this case, the response is replaced with a "502 bad
18559 gateway". The flags ("PH--") tell us that it was haproxy who decided to
18560 return the 502 and not the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018561
18562 >>> haproxy[18113]: 127.0.0.1:34548 [15/Oct/2003:15:18:55.798] px-http \
Willy Tarreaud72758d2010-01-12 10:42:19 +010018563 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 +010018564
18565 => the client never completed its request and aborted itself ("C---") after
18566 8.5s, while the proxy was waiting for the request headers ("-R--").
18567 Nothing was sent to any server.
18568
18569 >>> haproxy[18113]: 127.0.0.1:34549 [15/Oct/2003:15:19:06.103] px-http \
18570 px-http/<NOSRV> -1/-1/-1/-1/50001 408 0 - - cR-- 2/2/2/0/0 0/0 ""
18571
18572 => The client never completed its request, which was aborted by the
18573 time-out ("c---") after 50s, while the proxy was waiting for the request
Davor Ocelice9ed2812017-12-25 17:49:28 +010018574 headers ("-R--"). Nothing was sent to any server, but the proxy could
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018575 send a 408 return code to the client.
18576
18577 >>> haproxy[18989]: 127.0.0.1:34550 [15/Oct/2003:15:24:28.312] px-tcp \
18578 px-tcp/srv1 0/0/5007 0 cD 0/0/0/0/0 0/0
18579
18580 => This log was produced with "option tcplog". The client timed out after
18581 5 seconds ("c----").
18582
18583 >>> haproxy[18989]: 10.0.0.1:34552 [15/Oct/2003:15:26:31.462] px-http \
18584 px-http/srv1 3183/-1/-1/-1/11215 503 0 - - SC-- 205/202/202/115/3 \
Willy Tarreaud72758d2010-01-12 10:42:19 +010018585 0/0 "HEAD / HTTP/1.0"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018586
18587 => The request took 3s to complete (probably a network problem), and the
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018588 connection to the server failed ('SC--') after 4 attempts of 2 seconds
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018589 (config says 'retries 3'), and no redispatch (otherwise we would have
18590 seen "/+3"). Status code 503 was returned to the client. There were 115
18591 connections on this server, 202 connections on this proxy, and 205 on
18592 the global process. It is possible that the server refused the
18593 connection because of too many already established.
Willy Tarreau844e3c52008-01-11 16:28:18 +010018594
Willy Tarreau52b2d222011-09-07 23:48:48 +020018595
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200185969. Supported filters
18597--------------------
18598
18599Here are listed officially supported filters with the list of parameters they
18600accept. Depending on compile options, some of these filters might be
18601unavailable. The list of available filters is reported in haproxy -vv.
18602
18603See also : "filter"
18604
186059.1. Trace
18606----------
18607
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010018608filter trace [name <name>] [random-parsing] [random-forwarding] [hexdump]
Christopher Fauletc3fe5332016-04-07 15:30:10 +020018609
18610 Arguments:
18611 <name> is an arbitrary name that will be reported in
18612 messages. If no name is provided, "TRACE" is used.
18613
18614 <random-parsing> enables the random parsing of data exchanged between
18615 the client and the server. By default, this filter
18616 parses all available data. With this parameter, it
18617 only parses a random amount of the available data.
18618
Davor Ocelice9ed2812017-12-25 17:49:28 +010018619 <random-forwarding> enables the random forwarding of parsed data. By
Christopher Fauletc3fe5332016-04-07 15:30:10 +020018620 default, this filter forwards all previously parsed
18621 data. With this parameter, it only forwards a random
18622 amount of the parsed data.
18623
Davor Ocelice9ed2812017-12-25 17:49:28 +010018624 <hexdump> dumps all forwarded data to the server and the client.
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010018625
Christopher Fauletc3fe5332016-04-07 15:30:10 +020018626This filter can be used as a base to develop new filters. It defines all
18627callbacks and print a message on the standard error stream (stderr) with useful
18628information for all of them. It may be useful to debug the activity of other
18629filters or, quite simply, HAProxy's activity.
18630
18631Using <random-parsing> and/or <random-forwarding> parameters is a good way to
18632tests the behavior of a filter that parses data exchanged between a client and
18633a server by adding some latencies in the processing.
18634
18635
186369.2. HTTP compression
18637---------------------
18638
18639filter compression
18640
18641The HTTP compression has been moved in a filter in HAProxy 1.7. "compression"
18642keyword must still be used to enable and configure the HTTP compression. And
Christopher Fauletb30b3102019-09-12 23:03:09 +020018643when no other filter is used, it is enough. When used with the cache or the
18644fcgi-app enabled, it is also enough. In this case, the compression is always
18645done after the response is stored in the cache. But it is mandatory to
18646explicitly use a filter line to enable the HTTP compression when at least one
18647filter other than the cache or the fcgi-app is used for the same
18648listener/frontend/backend. This is important to know the filters evaluation
18649order.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020018650
Christopher Fauletb30b3102019-09-12 23:03:09 +020018651See also : "compression", section 9.4 about the cache filter and section 9.5
18652 about the fcgi-app filter.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020018653
18654
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +0200186559.3. Stream Processing Offload Engine (SPOE)
18656--------------------------------------------
18657
18658filter spoe [engine <name>] config <file>
18659
18660 Arguments :
18661
18662 <name> is the engine name that will be used to find the right scope in
18663 the configuration file. If not provided, all the file will be
18664 parsed.
18665
18666 <file> is the path of the engine configuration file. This file can
18667 contain configuration of several engines. In this case, each
18668 part must be placed in its own scope.
18669
18670The Stream Processing Offload Engine (SPOE) is a filter communicating with
18671external components. It allows the offload of some specifics processing on the
Davor Ocelice9ed2812017-12-25 17:49:28 +010018672streams in tiered applications. These external components and information
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020018673exchanged with them are configured in dedicated files, for the main part. It
18674also requires dedicated backends, defined in HAProxy configuration.
18675
18676SPOE communicates with external components using an in-house binary protocol,
18677the Stream Processing Offload Protocol (SPOP).
18678
Tim Düsterhus4896c442016-11-29 02:15:19 +010018679For all information about the SPOE configuration and the SPOP specification, see
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020018680"doc/SPOE.txt".
18681
18682Important note:
18683 The SPOE filter is highly experimental for now and was not heavily
18684 tested. It is really not production ready. So use it carefully.
18685
Christopher Faulet99a17a22018-12-11 09:18:27 +0100186869.4. Cache
18687----------
18688
18689filter cache <name>
18690
18691 Arguments :
18692
18693 <name> is name of the cache section this filter will use.
18694
18695The cache uses a filter to store cacheable responses. The HTTP rules
18696"cache-store" and "cache-use" must be used to define how and when to use a
John Roeslerfb2fce12019-07-10 15:45:51 -050018697cache. By default the corresponding filter is implicitly defined. And when no
Christopher Fauletb30b3102019-09-12 23:03:09 +020018698other filters than fcgi-app or compression are used, it is enough. In such
18699case, the compression filter is always evaluated after the cache filter. But it
18700is mandatory to explicitly use a filter line to use a cache when at least one
18701filter other than the compression or the fcgi-app is used for the same
Christopher Faulet27d93c32018-12-15 22:32:02 +010018702listener/frontend/backend. This is important to know the filters evaluation
18703order.
Christopher Faulet99a17a22018-12-11 09:18:27 +010018704
Christopher Fauletb30b3102019-09-12 23:03:09 +020018705See also : section 9.2 about the compression filter, section 9.5 about the
18706 fcgi-app filter and section 6 about cache.
18707
18708
187099.5. Fcgi-app
18710-------------
18711
18712filter fcg-app <name>
18713
18714 Arguments :
18715
18716 <name> is name of the fcgi-app section this filter will use.
18717
18718The FastCGI application uses a filter to evaluate all custom parameters on the
18719request path, and to process the headers on the response path. the <name> must
18720reference an existing fcgi-app section. The directive "use-fcgi-app" should be
18721used to define the application to use. By default the corresponding filter is
18722implicitly defined. And when no other filters than cache or compression are
18723used, it is enough. But it is mandatory to explicitly use a filter line to a
18724fcgi-app when at least one filter other than the compression or the cache is
18725used for the same backend. This is important to know the filters evaluation
18726order.
18727
18728See also: "use-fcgi-app", section 9.2 about the compression filter, section 9.4
18729 about the cache filter and section 10 about FastCGI application.
18730
18731
1873210. FastCGI applications
18733-------------------------
18734
18735HAProxy is able to send HTTP requests to Responder FastCGI applications. This
18736feature was added in HAProxy 2.1. To do so, servers must be configured to use
18737the FastCGI protocol (using the keyword "proto fcgi" on the server line) and a
18738FastCGI application must be configured and used by the backend managing these
18739servers (using the keyword "use-fcgi-app" into the proxy section). Several
18740FastCGI applications may be defined, but only one can be used at a time by a
18741backend.
18742
18743HAProxy implements all features of the FastCGI specification for Responder
18744application. Especially it is able to multiplex several requests on a simple
18745connection.
18746
1874710.1. Setup
18748-----------
18749
1875010.1.1. Fcgi-app section
18751--------------------------
18752
18753fcgi-app <name>
18754 Declare a FastCGI application named <name>. To be valid, at least the
18755 document root must be defined.
18756
18757acl <aclname> <criterion> [flags] [operator] <value> ...
18758 Declare or complete an access list.
18759
18760 See "acl" keyword in section 4.2 and section 7 about ACL usage for
18761 details. ACLs defined for a FastCGI application are private. They cannot be
18762 used by any other application or by any proxy. In the same way, ACLs defined
18763 in any other section are not usable by a FastCGI application. However,
18764 Pre-defined ACLs are available.
18765
18766docroot <path>
18767 Define the document root on the remote host. <path> will be used to build
18768 the default value of FastCGI parameters SCRIPT_FILENAME and
18769 PATH_TRANSLATED. It is a mandatory setting.
18770
18771index <script-name>
18772 Define the script name that will be appended after an URI that ends with a
18773 slash ("/") to set the default value of the FastCGI parameter SCRIPT_NAME. It
18774 is an optional setting.
18775
18776 Example :
18777 index index.php
18778
18779log-stderr global
18780log-stderr <address> [len <length>] [format <format>]
18781 [sample <ranges>:<smp_size>] <facility> [<level> [<minlevel>]]
18782 Enable logging of STDERR messages reported by the FastCGI application.
18783
18784 See "log" keyword in section 4.2 for details. It is an optional setting. By
18785 default STDERR messages are ignored.
18786
18787pass-header <name> [ { if | unless } <condition> ]
18788 Specify the name of a request header which will be passed to the FastCGI
18789 application. It may optionally be followed by an ACL-based condition, in
18790 which case it will only be evaluated if the condition is true.
18791
18792 Most request headers are already available to the FastCGI application,
18793 prefixed with "HTTP_". Thus, this directive is only required to pass headers
18794 that are purposefully omitted. Currently, the headers "Authorization",
18795 "Proxy-Authorization" and hop-by-hop headers are omitted.
18796
18797 Note that the headers "Content-type" and "Content-length" are never passed to
18798 the FastCGI application because they are already converted into parameters.
18799
18800path-info <regex>
Christopher Faulet28cb3662020-02-14 14:47:37 +010018801 Define a regular expression to extract the script-name and the path-info from
Christopher Faulet6c57f2d2020-02-14 16:55:52 +010018802 the URL-decoded path. Thus, <regex> may have two captures: the first one to
18803 capture the script name and the second one to capture the path-info. The
18804 first one is mandatory, the second one is optional. This way, it is possible
18805 to extract the script-name from the path ignoring the path-info. It is an
18806 optional setting. If it is not defined, no matching is performed on the
18807 path. and the FastCGI parameters PATH_INFO and PATH_TRANSLATED are not
18808 filled.
Christopher Faulet28cb3662020-02-14 14:47:37 +010018809
18810 For security reason, when this regular expression is defined, the newline and
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050018811 the null characters are forbidden from the path, once URL-decoded. The reason
Christopher Faulet28cb3662020-02-14 14:47:37 +010018812 to such limitation is because otherwise the matching always fails (due to a
18813 limitation one the way regular expression are executed in HAProxy). So if one
18814 of these two characters is found in the URL-decoded path, an error is
18815 returned to the client. The principle of least astonishment is applied here.
Christopher Fauletb30b3102019-09-12 23:03:09 +020018816
18817 Example :
Christopher Faulet6c57f2d2020-02-14 16:55:52 +010018818 path-info ^(/.+\.php)(/.*)?$ # both script-name and path-info may be set
18819 path-info ^(/.+\.php) # the path-info is ignored
Christopher Fauletb30b3102019-09-12 23:03:09 +020018820
18821option get-values
18822no option get-values
18823 Enable or disable the retrieve of variables about connection management.
18824
18825 HAproxy is able to send the record FCGI_GET_VALUES on connection
18826 establishment to retrieve the value for following variables:
18827
18828 * FCGI_MAX_REQS The maximum number of concurrent requests this
18829 application will accept.
18830
William Lallemand93e548e2019-09-30 13:54:02 +020018831 * FCGI_MPXS_CONNS "0" if this application does not multiplex connections,
18832 "1" otherwise.
Christopher Fauletb30b3102019-09-12 23:03:09 +020018833
18834 Some FastCGI applications does not support this feature. Some others close
18835 the connexion immediately after sending their response. So, by default, this
18836 option is disabled.
18837
18838 Note that the maximum number of concurrent requests accepted by a FastCGI
18839 application is a connection variable. It only limits the number of streams
18840 per connection. If the global load must be limited on the application, the
18841 server parameters "maxconn" and "pool-max-conn" must be set. In addition, if
18842 an application does not support connection multiplexing, the maximum number
18843 of concurrent requests is automatically set to 1.
18844
18845option keep-conn
18846no option keep-conn
18847 Instruct the FastCGI application to keep the connection open or not after
18848 sending a response.
18849
18850 If disabled, the FastCGI application closes the connection after responding
18851 to this request. By default, this option is enabled.
18852
18853option max-reqs <reqs>
18854 Define the maximum number of concurrent requests this application will
18855 accept.
18856
18857 This option may be overwritten if the variable FCGI_MAX_REQS is retrieved
18858 during connection establishment. Furthermore, if the application does not
18859 support connection multiplexing, this option will be ignored. By default set
18860 to 1.
18861
18862option mpxs-conns
18863no option mpxs-conns
18864 Enable or disable the support of connection multiplexing.
18865
18866 This option may be overwritten if the variable FCGI_MPXS_CONNS is retrieved
18867 during connection establishment. It is disabled by default.
18868
18869set-param <name> <fmt> [ { if | unless } <condition> ]
18870 Set a FastCGI parameter that should be passed to this application. Its
18871 value, defined by <fmt> must follows the log-format rules (see section 8.2.4
18872 "Custom Log format"). It may optionally be followed by an ACL-based
18873 condition, in which case it will only be evaluated if the condition is true.
18874
18875 With this directive, it is possible to overwrite the value of default FastCGI
18876 parameters. If the value is evaluated to an empty string, the rule is
18877 ignored. These directives are evaluated in their declaration order.
18878
18879 Example :
18880 # PHP only, required if PHP was built with --enable-force-cgi-redirect
18881 set-param REDIRECT_STATUS 200
18882
18883 set-param PHP_AUTH_DIGEST %[req.hdr(Authorization)]
18884
18885
1888610.1.2. Proxy section
18887---------------------
18888
18889use-fcgi-app <name>
18890 Define the FastCGI application to use for the backend.
18891
18892 Arguments :
18893 <name> is the name of the FastCGI application to use.
18894
18895 This keyword is only available for HTTP proxies with the backend capability
18896 and with at least one FastCGI server. However, FastCGI servers can be mixed
18897 with HTTP servers. But except there is a good reason to do so, it is not
18898 recommended (see section 10.3 about the limitations for details). Only one
18899 application may be defined at a time per backend.
18900
18901 Note that, once a FastCGI application is referenced for a backend, depending
18902 on the configuration some processing may be done even if the request is not
18903 sent to a FastCGI server. Rules to set parameters or pass headers to an
18904 application are evaluated.
18905
18906
1890710.1.3. Example
18908---------------
18909
18910 frontend front-http
18911 mode http
18912 bind *:80
18913 bind *:
18914
18915 use_backend back-dynamic if { path_reg ^/.+\.php(/.*)?$ }
18916 default_backend back-static
18917
18918 backend back-static
18919 mode http
18920 server www A.B.C.D:80
18921
18922 backend back-dynamic
18923 mode http
18924 use-fcgi-app php-fpm
18925 server php-fpm A.B.C.D:9000 proto fcgi
18926
18927 fcgi-app php-fpm
18928 log-stderr global
18929 option keep-conn
18930
18931 docroot /var/www/my-app
18932 index index.php
18933 path-info ^(/.+\.php)(/.*)?$
18934
18935
1893610.2. Default parameters
18937------------------------
18938
18939A Responder FastCGI application has the same purpose as a CGI/1.1 program. In
18940the CGI/1.1 specification (RFC3875), several variables must be passed to the
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050018941script. So HAProxy set them and some others commonly used by FastCGI
Christopher Fauletb30b3102019-09-12 23:03:09 +020018942applications. All these variables may be overwritten, with caution though.
18943
18944 +-------------------+-----------------------------------------------------+
18945 | AUTH_TYPE | Identifies the mechanism, if any, used by HAProxy |
18946 | | to authenticate the user. Concretely, only the |
18947 | | BASIC authentication mechanism is supported. |
18948 | | |
18949 +-------------------+-----------------------------------------------------+
18950 | CONTENT_LENGTH | Contains the size of the message-body attached to |
18951 | | the request. It means only requests with a known |
18952 | | size are considered as valid and sent to the |
18953 | | application. |
18954 | | |
18955 +-------------------+-----------------------------------------------------+
18956 | CONTENT_TYPE | Contains the type of the message-body attached to |
18957 | | the request. It may not be set. |
18958 | | |
18959 +-------------------+-----------------------------------------------------+
18960 | DOCUMENT_ROOT | Contains the document root on the remote host under |
18961 | | which the script should be executed, as defined in |
18962 | | the application's configuration. |
18963 | | |
18964 +-------------------+-----------------------------------------------------+
18965 | GATEWAY_INTERFACE | Contains the dialect of CGI being used by HAProxy |
18966 | | to communicate with the FastCGI application. |
18967 | | Concretely, it is set to "CGI/1.1". |
18968 | | |
18969 +-------------------+-----------------------------------------------------+
18970 | PATH_INFO | Contains the portion of the URI path hierarchy |
18971 | | following the part that identifies the script |
18972 | | itself. To be set, the directive "path-info" must |
18973 | | be defined. |
18974 | | |
18975 +-------------------+-----------------------------------------------------+
18976 | PATH_TRANSLATED | If PATH_INFO is set, it is its translated version. |
18977 | | It is the concatenation of DOCUMENT_ROOT and |
18978 | | PATH_INFO. If PATH_INFO is not set, this parameters |
18979 | | is not set too. |
18980 | | |
18981 +-------------------+-----------------------------------------------------+
18982 | QUERY_STRING | Contains the request's query string. It may not be |
18983 | | set. |
18984 | | |
18985 +-------------------+-----------------------------------------------------+
18986 | REMOTE_ADDR | Contains the network address of the client sending |
18987 | | the request. |
18988 | | |
18989 +-------------------+-----------------------------------------------------+
18990 | REMOTE_USER | Contains the user identification string supplied by |
18991 | | client as part of user authentication. |
18992 | | |
18993 +-------------------+-----------------------------------------------------+
18994 | REQUEST_METHOD | Contains the method which should be used by the |
18995 | | script to process the request. |
18996 | | |
18997 +-------------------+-----------------------------------------------------+
18998 | REQUEST_URI | Contains the request's URI. |
18999 | | |
19000 +-------------------+-----------------------------------------------------+
19001 | SCRIPT_FILENAME | Contains the absolute pathname of the script. it is |
19002 | | the concatenation of DOCUMENT_ROOT and SCRIPT_NAME. |
19003 | | |
19004 +-------------------+-----------------------------------------------------+
19005 | SCRIPT_NAME | Contains the name of the script. If the directive |
19006 | | "path-info" is defined, it is the first part of the |
19007 | | URI path hierarchy, ending with the script name. |
19008 | | Otherwise, it is the entire URI path. |
19009 | | |
19010 +-------------------+-----------------------------------------------------+
19011 | SERVER_NAME | Contains the name of the server host to which the |
19012 | | client request is directed. It is the value of the |
19013 | | header "Host", if defined. Otherwise, the |
19014 | | destination address of the connection on the client |
19015 | | side. |
19016 | | |
19017 +-------------------+-----------------------------------------------------+
19018 | SERVER_PORT | Contains the destination TCP port of the connection |
19019 | | on the client side, which is the port the client |
19020 | | connected to. |
19021 | | |
19022 +-------------------+-----------------------------------------------------+
19023 | SERVER_PROTOCOL | Contains the request's protocol. |
19024 | | |
19025 +-------------------+-----------------------------------------------------+
19026 | HTTPS | Set to a non-empty value ("on") if the script was |
19027 | | queried through the HTTPS protocol. |
19028 | | |
19029 +-------------------+-----------------------------------------------------+
19030
19031
1903210.3. Limitations
19033------------------
19034
19035The current implementation have some limitations. The first one is about the
19036way some request headers are hidden to the FastCGI applications. This happens
19037during the headers analysis, on the backend side, before the connection
19038establishment. At this stage, HAProxy know the backend is using a FastCGI
19039application but it don't know if the request will be routed to a FastCGI server
19040or not. But to hide request headers, it simply removes them from the HTX
19041message. So, if the request is finally routed to an HTTP server, it never see
19042these headers. For this reason, it is not recommended to mix FastCGI servers
19043and HTTP servers under the same backend.
19044
19045Similarly, the rules "set-param" and "pass-header" are evaluated during the
19046request headers analysis. So the evaluation is always performed, even if the
19047requests is finally forwarded to an HTTP server.
19048
19049About the rules "set-param", when a rule is applied, a pseudo header is added
19050into the HTX message. So, the same way than for HTTP header rewrites, it may
19051fail if the buffer is full. The rules "set-param" will compete with
19052"http-request" ones.
19053
19054Finally, all FastCGI params and HTTP headers are sent into a unique record
19055FCGI_PARAM. Encoding of this record must be done in one pass, otherwise a
19056processing error is returned. It means the record FCGI_PARAM, once encoded,
19057must not exceeds the size of a buffer. However, there is no reserve to respect
19058here.
William Lallemand86d0df02017-11-24 21:36:45 +010019059
Willy Tarreau0ba27502007-12-24 16:55:16 +010019060/*
19061 * Local variables:
19062 * fill-column: 79
19063 * End:
19064 */