blob: fbc914d8fbb7c3bf7d51e5cd689b6a5d2f7e909c [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 Tarreau71f95fa2020-01-22 10:34:58 +01007 2020/01/22
Willy Tarreau6a06a402007-07-15 20:15:28 +02008
9
10This document covers the configuration language as implemented in the version
Davor Ocelice9ed2812017-12-25 17:49:28 +010011specified above. It does not provide any hints, examples, or advice. For such
Willy Tarreau0ba27502007-12-24 16:55:16 +010012documentation, please refer to the Reference Manual or the Architecture Manual.
Davor Ocelice9ed2812017-12-25 17:49:28 +010013The summary below is meant to help you find sections by name and navigate
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014through the document.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016Note to documentation contributors :
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017 This document is formatted with 80 columns per line, with even number of
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018 spaces for indentation and without tabs. Please follow these rules strictly
19 so that it remains easily printable everywhere. If a line needs to be
20 printed verbatim and does not fit, please end each line with a backslash
Willy Tarreau62a36c42010-08-17 15:53:10 +020021 ('\') and continue on next line, indented by two characters. It is also
Davor Ocelice9ed2812017-12-25 17:49:28 +010022 sometimes useful to prefix all output lines (logs, console outputs) with 3
23 closing angle brackets ('>>>') in order to emphasize the difference between
24 inputs and outputs when they may be ambiguous. If you add sections,
Willy Tarreau62a36c42010-08-17 15:53:10 +020025 please update the summary below for easier searching.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020026
27
28Summary
29-------
30
311. Quick reminder about HTTP
321.1. The HTTP transaction model
331.2. HTTP request
Davor Ocelice9ed2812017-12-25 17:49:28 +0100341.2.1. The request line
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200351.2.2. The request headers
361.3. HTTP response
Davor Ocelice9ed2812017-12-25 17:49:28 +0100371.3.1. The response line
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200381.3.2. The response headers
39
402. Configuring HAProxy
412.1. Configuration file format
William Lallemandf9873ba2015-05-05 17:37:14 +0200422.2. Quoting and escaping
William Lallemandb2f07452015-05-12 14:27:13 +0200432.3. Environment variables
442.4. Time format
452.5. Examples
Willy Tarreauc57f0e22009-05-10 13:12:33 +020046
473. Global parameters
483.1. Process management and security
493.2. Performance tuning
503.3. Debugging
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +0100513.4. Userlists
Cyril Bontédc4d9032012-04-08 21:57:39 +0200523.5. Peers
Cyril Bonté307ee1e2015-09-28 23:16:06 +0200533.6. Mailers
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
Willy Tarreau6a06a402007-07-15 20:15:28 +0200604 - log
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200605 - log-tag
Joe Williamsdf5b38f2010-12-29 17:05:48 +0100606 - log-send-hostname
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200607 - lua-load
William Lallemand27edc4b2019-05-07 17:49:33 +0200608 - mworker-max-reloads
Willy Tarreau6a06a402007-07-15 20:15:28 +0200609 - nbproc
Christopher Fauletbe0faa22017-08-29 15:37:10 +0200610 - nbthread
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200611 - node
Willy Tarreau6a06a402007-07-15 20:15:28 +0200612 - pidfile
Willy Tarreau1d549722016-02-16 12:41:57 +0100613 - presetenv
614 - resetenv
Willy Tarreau6a06a402007-07-15 20:15:28 +0200615 - uid
616 - ulimit-n
617 - user
Willy Tarreau636848a2019-04-15 19:38:50 +0200618 - set-dumpable
Willy Tarreau1d549722016-02-16 12:41:57 +0100619 - setenv
Willy Tarreaufbee7132007-10-18 13:53:22 +0200620 - stats
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200621 - ssl-default-bind-ciphers
Dirkjan Bussink415150f2018-09-14 11:14:21 +0200622 - ssl-default-bind-ciphersuites
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200623 - ssl-default-bind-options
624 - ssl-default-server-ciphers
Dirkjan Bussink415150f2018-09-14 11:14:21 +0200625 - ssl-default-server-ciphersuites
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200626 - ssl-default-server-options
627 - ssl-dh-param-file
Emeric Brun850efd52014-01-29 12:24:34 +0100628 - ssl-server-verify
Willy Tarreauceb24bc2010-11-09 12:46:41 +0100629 - unix-bind
Willy Tarreau1d549722016-02-16 12:41:57 +0100630 - unsetenv
Thomas Holmesdb04f192015-05-18 13:21:39 +0100631 - 51degrees-data-file
632 - 51degrees-property-name-list
Dragan Dosen93b38d92015-06-29 16:43:25 +0200633 - 51degrees-property-separator
Dragan Dosenae6d39a2015-06-29 16:43:27 +0200634 - 51degrees-cache-size
Willy Tarreaub3cc9f22019-04-19 16:03:32 +0200635 - wurfl-data-file
636 - wurfl-information-list
637 - wurfl-information-list-separator
Willy Tarreaub3cc9f22019-04-19 16:03:32 +0200638 - wurfl-cache-size
William Dauchy0fec3ab2019-10-27 20:08:11 +0100639 - strict-limits
Willy Tarreaud72758d2010-01-12 10:42:19 +0100640
Willy Tarreau6a06a402007-07-15 20:15:28 +0200641 * Performance tuning
William Dauchy0a8824f2019-10-27 20:08:09 +0100642 - busy-polling
Willy Tarreau1746eec2014-04-25 10:46:47 +0200643 - max-spread-checks
Willy Tarreau6a06a402007-07-15 20:15:28 +0200644 - maxconn
Willy Tarreau81c25d02011-09-07 15:17:21 +0200645 - maxconnrate
William Lallemandd85f9172012-11-09 17:05:39 +0100646 - maxcomprate
William Lallemand072a2bf2012-11-20 17:01:01 +0100647 - maxcompcpuusage
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100648 - maxpipes
Willy Tarreau93e7c002013-10-07 18:51:07 +0200649 - maxsessrate
Willy Tarreau403edff2012-09-06 11:58:37 +0200650 - maxsslconn
Willy Tarreaue43d5322013-10-07 20:01:52 +0200651 - maxsslrate
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200652 - maxzlibmem
Willy Tarreau6a06a402007-07-15 20:15:28 +0200653 - noepoll
654 - nokqueue
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +0000655 - noevports
Willy Tarreau6a06a402007-07-15 20:15:28 +0200656 - nopoll
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100657 - nosplice
Jarno Huuskonen0e82b922014-04-12 18:22:19 +0300658 - nogetaddrinfo
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +0000659 - noreuseport
Willy Tarreau75c62c22018-11-22 11:02:09 +0100660 - profiling.tasks
Willy Tarreaufe255b72007-10-14 23:09:26 +0200661 - spread-checks
Baptiste Assmann5626f482015-08-23 10:00:10 +0200662 - server-state-base
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +0200663 - server-state-file
Grant Zhang872f9c22017-01-21 01:10:18 +0000664 - ssl-engine
Grant Zhangfa6c7ee2017-01-14 01:42:15 +0000665 - ssl-mode-async
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200666 - tune.buffers.limit
667 - tune.buffers.reserve
Willy Tarreau27a674e2009-08-17 07:23:33 +0200668 - tune.bufsize
Willy Tarreau43961d52010-10-04 20:39:20 +0200669 - tune.chksize
William Lallemandf3747832012-11-09 12:33:10 +0100670 - tune.comp.maxlevel
Willy Tarreaufe20e5b2017-07-27 11:42:14 +0200671 - tune.h2.header-table-size
Willy Tarreaue6baec02017-07-27 11:45:11 +0200672 - tune.h2.initial-window-size
Willy Tarreau5242ef82017-07-27 11:47:28 +0200673 - tune.h2.max-concurrent-streams
Willy Tarreau193b8c62012-11-22 00:17:38 +0100674 - tune.http.cookielen
Stéphane Cottin23e9e932017-05-18 08:58:41 +0200675 - tune.http.logurilen
Willy Tarreauac1932d2011-10-24 19:14:41 +0200676 - tune.http.maxhdr
Willy Tarreau7e312732014-02-12 16:35:14 +0100677 - tune.idletimer
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100678 - tune.lua.forced-yield
Willy Tarreau32f61e22015-03-18 17:54:59 +0100679 - tune.lua.maxmem
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100680 - tune.lua.session-timeout
681 - tune.lua.task-timeout
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +0200682 - tune.lua.service-timeout
Willy Tarreaua0250ba2008-01-06 11:22:57 +0100683 - tune.maxaccept
684 - tune.maxpollevents
Willy Tarreau27a674e2009-08-17 07:23:33 +0200685 - tune.maxrewrite
Willy Tarreauf3045d22015-04-29 16:24:50 +0200686 - tune.pattern.cache-size
Willy Tarreaubd9a0a72011-10-23 21:14:29 +0200687 - tune.pipesize
Willy Tarreaue803de22010-01-21 17:43:04 +0100688 - tune.rcvbuf.client
689 - tune.rcvbuf.server
Willy Tarreaub22fc302015-12-14 12:04:35 +0100690 - tune.recv_enough
Olivier Houchard1599b802018-05-24 18:59:04 +0200691 - tune.runqueue-depth
Willy Tarreaue803de22010-01-21 17:43:04 +0100692 - tune.sndbuf.client
693 - tune.sndbuf.server
Willy Tarreau6ec58db2012-11-16 16:32:15 +0100694 - tune.ssl.cachesize
Willy Tarreaubfd59462013-02-21 07:46:09 +0100695 - tune.ssl.lifetime
Emeric Brun8dc60392014-05-09 13:52:00 +0200696 - tune.ssl.force-private-cache
Willy Tarreaubfd59462013-02-21 07:46:09 +0100697 - tune.ssl.maxrecord
Remi Gacognef46cd6e2014-06-12 14:58:40 +0200698 - tune.ssl.default-dh-param
Christopher Faulet31af49d2015-06-09 17:29:50 +0200699 - tune.ssl.ssl-ctx-cache-size
Thierry FOURNIER5bf77322017-02-25 12:45:22 +0100700 - tune.ssl.capture-cipherlist-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200701 - tune.vars.global-max-size
Christopher Fauletff2613e2016-11-09 11:36:17 +0100702 - tune.vars.proc-max-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200703 - tune.vars.reqres-max-size
704 - tune.vars.sess-max-size
705 - tune.vars.txn-max-size
William Lallemanda509e4c2012-11-07 16:54:34 +0100706 - tune.zlib.memlevel
707 - tune.zlib.windowsize
Willy Tarreaud72758d2010-01-12 10:42:19 +0100708
Willy Tarreau6a06a402007-07-15 20:15:28 +0200709 * Debugging
710 - debug
711 - quiet
Willy Tarreau6a06a402007-07-15 20:15:28 +0200712
713
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007143.1. Process management and security
Willy Tarreau6a06a402007-07-15 20:15:28 +0200715------------------------------------
716
Emeric Brunc8e8d122012-10-02 18:42:10 +0200717ca-base <dir>
718 Assigns a default directory to fetch SSL CA certificates and CRLs from when a
Emeric Brunfd33a262012-10-11 16:28:27 +0200719 relative path is used with "ca-file" or "crl-file" directives. Absolute
720 locations specified in "ca-file" and "crl-file" prevail and ignore "ca-base".
Emeric Brunc8e8d122012-10-02 18:42:10 +0200721
Willy Tarreau6a06a402007-07-15 20:15:28 +0200722chroot <jail dir>
723 Changes current directory to <jail dir> and performs a chroot() there before
724 dropping privileges. This increases the security level in case an unknown
725 vulnerability would be exploited, since it would make it very hard for the
726 attacker to exploit the system. This only works when the process is started
727 with superuser privileges. It is important to ensure that <jail_dir> is both
Davor Ocelice9ed2812017-12-25 17:49:28 +0100728 empty and non-writable to anyone.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100729
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100730cpu-map [auto:]<process-set>[/<thread-set>] <cpu-set>...
731 On Linux 2.6 and above, it is possible to bind a process or a thread to a
732 specific CPU set. This means that the process or the thread will never run on
733 other CPUs. The "cpu-map" directive specifies CPU sets for process or thread
734 sets. The first argument is a process set, eventually followed by a thread
735 set. These sets have the format
736
737 all | odd | even | number[-[number]]
738
739 <number>> must be a number between 1 and 32 or 64, depending on the machine's
Davor Ocelice9ed2812017-12-25 17:49:28 +0100740 word size. Any process IDs above nbproc and any thread IDs above nbthread are
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100741 ignored. It is possible to specify a range with two such number delimited by
742 a dash ('-'). It also is possible to specify all processes at once using
Christopher Faulet1dcb9cb2017-11-22 10:24:40 +0100743 "all", only odd numbers using "odd" or even numbers using "even", just like
744 with the "bind-process" directive. The second and forthcoming arguments are
Davor Ocelice9ed2812017-12-25 17:49:28 +0100745 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 +0100746 range with two such numbers delimited by a dash ('-'). Multiple CPU numbers
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100747 or ranges may be specified, and the processes or threads will be allowed to
Davor Ocelice9ed2812017-12-25 17:49:28 +0100748 bind to all of them. Obviously, multiple "cpu-map" directives may be
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100749 specified. Each "cpu-map" directive will replace the previous ones when they
750 overlap. A thread will be bound on the intersection of its mapping and the
751 one of the process on which it is attached. If the intersection is null, no
752 specific binding will be set for the thread.
Willy Tarreaufc6c0322012-11-16 16:12:27 +0100753
Christopher Fauletff4121f2017-11-22 16:38:49 +0100754 Ranges can be partially defined. The higher bound can be omitted. In such
755 case, it is replaced by the corresponding maximum value, 32 or 64 depending
756 on the machine's word size.
757
Christopher Faulet26028f62017-11-22 15:01:51 +0100758 The prefix "auto:" can be added before the process set to let HAProxy
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100759 automatically bind a process or a thread to a CPU by incrementing
760 process/thread and CPU sets. To be valid, both sets must have the same
761 size. No matter the declaration order of the CPU sets, it will be bound from
762 the lowest to the highest bound. Having a process and a thread range with the
763 "auto:" prefix is not supported. Only one range is supported, the other one
764 must be a fixed number.
Christopher Faulet26028f62017-11-22 15:01:51 +0100765
766 Examples:
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100767 cpu-map 1-4 0-3 # bind processes 1 to 4 on the first 4 CPUs
768
769 cpu-map 1/all 0-3 # bind all threads of the first process on the
770 # first 4 CPUs
771
772 cpu-map 1- 0- # will be replaced by "cpu-map 1-64 0-63"
773 # or "cpu-map 1-32 0-31" depending on the machine's
774 # word size.
775
Christopher Faulet26028f62017-11-22 15:01:51 +0100776 # all these lines bind the process 1 to the cpu 0, the process 2 to cpu 1
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100777 # and so on.
Christopher Faulet26028f62017-11-22 15:01:51 +0100778 cpu-map auto:1-4 0-3
779 cpu-map auto:1-4 0-1 2-3
780 cpu-map auto:1-4 3 2 1 0
781
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100782 # all these lines bind the thread 1 to the cpu 0, the thread 2 to cpu 1
783 # and so on.
784 cpu-map auto:1/1-4 0-3
785 cpu-map auto:1/1-4 0-1 2-3
786 cpu-map auto:1/1-4 3 2 1 0
787
Davor Ocelice9ed2812017-12-25 17:49:28 +0100788 # bind each process to exactly one CPU using all/odd/even keyword
Christopher Faulet26028f62017-11-22 15:01:51 +0100789 cpu-map auto:all 0-63
790 cpu-map auto:even 0-31
791 cpu-map auto:odd 32-63
792
793 # invalid cpu-map because process and CPU sets have different sizes.
794 cpu-map auto:1-4 0 # invalid
795 cpu-map auto:1 0-3 # invalid
796
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100797 # invalid cpu-map because automatic binding is used with a process range
798 # and a thread range.
799 cpu-map auto:all/all 0 # invalid
800 cpu-map auto:all/1-4 0 # invalid
801 cpu-map auto:1-4/all 0 # invalid
802
Emeric Brunc8e8d122012-10-02 18:42:10 +0200803crt-base <dir>
804 Assigns a default directory to fetch SSL certificates from when a relative
William Dauchy238ea3b2020-01-11 13:09:12 +0100805 path is used with "crtfile" or "crt" directives. Absolute locations specified
806 prevail and ignore "crt-base".
Emeric Brunc8e8d122012-10-02 18:42:10 +0200807
Willy Tarreau6a06a402007-07-15 20:15:28 +0200808daemon
809 Makes the process fork into background. This is the recommended mode of
810 operation. It is equivalent to the command line "-D" argument. It can be
Lukas Tribusf46bf952017-11-21 12:39:34 +0100811 disabled by the command line "-db" argument. This option is ignored in
812 systemd mode.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200813
David Carlier8167f302015-06-01 13:50:06 +0200814deviceatlas-json-file <path>
815 Sets the path of the DeviceAtlas JSON data file to be loaded by the API.
Davor Ocelice9ed2812017-12-25 17:49:28 +0100816 The path must be a valid JSON data file and accessible by HAProxy process.
David Carlier8167f302015-06-01 13:50:06 +0200817
818deviceatlas-log-level <value>
Davor Ocelice9ed2812017-12-25 17:49:28 +0100819 Sets the level of information returned by the API. This directive is
David Carlier8167f302015-06-01 13:50:06 +0200820 optional and set to 0 by default if not set.
821
822deviceatlas-separator <char>
823 Sets the character separator for the API properties results. This directive
824 is optional and set to | by default if not set.
825
Cyril Bonté0306c4a2015-10-26 22:37:38 +0100826deviceatlas-properties-cookie <name>
Cyril Bonté307ee1e2015-09-28 23:16:06 +0200827 Sets the client cookie's name used for the detection if the DeviceAtlas
828 Client-side component was used during the request. This directive is optional
829 and set to DAPROPS by default if not set.
David Carlier29b3ca32015-09-25 14:09:21 +0100830
Simon Horman98637e52014-06-20 12:30:16 +0900831external-check
Willy Tarreaud96f1122019-12-03 07:07:36 +0100832 Allows the use of an external agent to perform health checks. This is
833 disabled by default as a security precaution, and even when enabled, checks
Willy Tarreaua45a8b52019-12-06 16:31:45 +0100834 may still fail unless "insecure-fork-wanted" is enabled as well. If the
835 program launched makes use of a setuid executable (it should really not),
836 you may also need to set "insecure-setuid-wanted" in the global section.
837 See "option external-check", and "insecure-fork-wanted", and
838 "insecure-setuid-wanted".
Simon Horman98637e52014-06-20 12:30:16 +0900839
Willy Tarreau6a06a402007-07-15 20:15:28 +0200840gid <number>
841 Changes the process' group ID to <number>. It is recommended that the group
842 ID is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
843 be started with a user belonging to this group, or with superuser privileges.
Michael Schererab012dd2013-01-12 18:35:19 +0100844 Note that if haproxy is started from a user having supplementary groups, it
845 will only be able to drop these groups if started with superuser privileges.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200846 See also "group" and "uid".
Willy Tarreaud72758d2010-01-12 10:42:19 +0100847
Willy Tarreau11770ce2019-12-03 08:29:22 +0100848group <group name>
849 Similar to "gid" but uses the GID of group name <group name> from /etc/group.
850 See also "gid" and "user".
851
Cyril Bonté203ec5a2017-03-23 22:44:13 +0100852hard-stop-after <time>
853 Defines the maximum time allowed to perform a clean soft-stop.
854
855 Arguments :
856 <time> is the maximum time (by default in milliseconds) for which the
857 instance will remain alive when a soft-stop is received via the
858 SIGUSR1 signal.
859
860 This may be used to ensure that the instance will quit even if connections
861 remain opened during a soft-stop (for example with long timeouts for a proxy
862 in tcp mode). It applies both in TCP and HTTP mode.
863
864 Example:
865 global
866 hard-stop-after 30s
867
Christopher Faulet98fbe952019-07-22 16:18:24 +0200868h1-case-adjust <from> <to>
869 Defines the case adjustment to apply, when enabled, to the header name
870 <from>, to change it to <to> before sending it to HTTP/1 clients or
871 servers. <from> must be in lower case, and <from> and <to> must not differ
872 except for their case. It may be repeated if several header names need to be
873 ajusted. Duplicate entries are not allowed. If a lot of header names have to
874 be adjusted, it might be more convenient to use "h1-case-adjust-file".
875 Please note that no transformation will be applied unless "option
876 h1-case-adjust-bogus-client" or "option h1-case-adjust-bogus-server" is
877 specified in a proxy.
878
879 There is no standard case for header names because, as stated in RFC7230,
880 they are case-insensitive. So applications must handle them in a case-
881 insensitive manner. But some bogus applications violate the standards and
882 erroneously rely on the cases most commonly used by browsers. This problem
883 becomes critical with HTTP/2 because all header names must be exchanged in
884 lower case, and HAProxy follows the same convention. All header names are
885 sent in lower case to clients and servers, regardless of the HTTP version.
886
887 Applications which fail to properly process requests or responses may require
888 to temporarily use such workarounds to adjust header names sent to them for
889 the time it takes the application to be fixed. Please note that an
890 application which requires such workarounds might be vulnerable to content
891 smuggling attacks and must absolutely be fixed.
892
893 Example:
894 global
895 h1-case-adjust content-length Content-Length
896
897 See "h1-case-adjust-file", "option h1-case-adjust-bogus-client" and
898 "option h1-case-adjust-bogus-server".
899
900h1-case-adjust-file <hdrs-file>
901 Defines a file containing a list of key/value pairs used to adjust the case
902 of some header names before sending them to HTTP/1 clients or servers. The
903 file <hdrs-file> must contain 2 header names per line. The first one must be
904 in lower case and both must not differ except for their case. Lines which
905 start with '#' are ignored, just like empty lines. Leading and trailing tabs
906 and spaces are stripped. Duplicate entries are not allowed. Please note that
907 no transformation will be applied unless "option h1-case-adjust-bogus-client"
908 or "option h1-case-adjust-bogus-server" is specified in a proxy.
909
910 If this directive is repeated, only the last one will be processed. It is an
911 alternative to the directive "h1-case-adjust" if a lot of header names need
912 to be adjusted. Please read the risks associated with using this.
913
914 See "h1-case-adjust", "option h1-case-adjust-bogus-client" and
915 "option h1-case-adjust-bogus-server".
916
Willy Tarreaud96f1122019-12-03 07:07:36 +0100917insecure-fork-wanted
918 By default haproxy tries hard to prevent any thread and process creation
919 after it starts. Doing so is particularly important when using Lua files of
920 uncertain origin, and when experimenting with development versions which may
921 still contain bugs whose exploitability is uncertain. And generally speaking
922 it's good hygiene to make sure that no unexpected background activity can be
923 triggered by traffic. But this prevents external checks from working, and may
924 break some very specific Lua scripts which actively rely on the ability to
925 fork. This option is there to disable this protection. Note that it is a bad
926 idea to disable it, as a vulnerability in a library or within haproxy itself
927 will be easier to exploit once disabled. In addition, forking from Lua or
928 anywhere else is not reliable as the forked process may randomly embed a lock
929 set by another thread and never manage to finish an operation. As such it is
930 highly recommended that this option is never used and that any workload
931 requiring such a fork be reconsidered and moved to a safer solution (such as
932 agents instead of external checks). This option supports the "no" prefix to
933 disable it.
934
Willy Tarreaua45a8b52019-12-06 16:31:45 +0100935insecure-setuid-wanted
936 HAProxy doesn't need to call executables at run time (except when using
937 external checks which are strongly recommended against), and is even expected
938 to isolate itself into an empty chroot. As such, there basically is no valid
939 reason to allow a setuid executable to be called without the user being fully
940 aware of the risks. In a situation where haproxy would need to call external
941 checks and/or disable chroot, exploiting a vulnerability in a library or in
942 haproxy itself could lead to the execution of an external program. On Linux
943 it is possible to lock the process so that any setuid bit present on such an
944 executable is ignored. This significantly reduces the risk of privilege
945 escalation in such a situation. This is what haproxy does by default. In case
946 this causes a problem to an external check (for example one which would need
947 the "ping" command), then it is possible to disable this protection by
948 explicitly adding this directive in the global section. If enabled, it is
949 possible to turn it back off by prefixing it with the "no" keyword.
950
Frédéric Lécailled690dfa2019-04-25 10:52:17 +0200951log <address> [len <length>] [format <format>] [sample <ranges>:<smp_size>]
952 <facility> [max level [min level]]
Cyril Bonté3e954872018-03-20 23:30:27 +0100953 Adds a global syslog server. Several global servers can be defined. They
Davor Ocelice9ed2812017-12-25 17:49:28 +0100954 will receive logs for starts and exits, as well as all logs from proxies
Robert Tsai81ae1952007-12-05 10:47:29 +0100955 configured with "log global".
956
957 <address> can be one of:
958
Willy Tarreau2769aa02007-12-27 18:26:09 +0100959 - An IPv4 address optionally followed by a colon and a UDP port. If
Robert Tsai81ae1952007-12-05 10:47:29 +0100960 no port is specified, 514 is used by default (the standard syslog
961 port).
962
David du Colombier24bb5f52011-03-17 10:40:23 +0100963 - An IPv6 address followed by a colon and optionally a UDP port. If
964 no port is specified, 514 is used by default (the standard syslog
965 port).
966
Willy Tarreau5a32ecc2018-11-12 07:34:59 +0100967 - A filesystem path to a datagram UNIX domain socket, keeping in mind
Robert Tsai81ae1952007-12-05 10:47:29 +0100968 considerations for chroot (be sure the path is accessible inside
969 the chroot) and uid/gid (be sure the path is appropriately
Davor Ocelice9ed2812017-12-25 17:49:28 +0100970 writable).
Robert Tsai81ae1952007-12-05 10:47:29 +0100971
Willy Tarreau5a32ecc2018-11-12 07:34:59 +0100972 - A file descriptor number in the form "fd@<number>", which may point
973 to a pipe, terminal, or socket. In this case unbuffered logs are used
974 and one writev() call per log is performed. This is a bit expensive
975 but acceptable for most workloads. Messages sent this way will not be
976 truncated but may be dropped, in which case the DroppedLogs counter
977 will be incremented. The writev() call is atomic even on pipes for
978 messages up to PIPE_BUF size, which POSIX recommends to be at least
979 512 and which is 4096 bytes on most modern operating systems. Any
980 larger message may be interleaved with messages from other processes.
981 Exceptionally for debugging purposes the file descriptor may also be
982 directed to a file, but doing so will significantly slow haproxy down
983 as non-blocking calls will be ignored. Also there will be no way to
984 purge nor rotate this file without restarting the process. Note that
985 the configured syslog format is preserved, so the output is suitable
Willy Tarreauc1b06452018-11-12 11:57:56 +0100986 for use with a TCP syslog server. See also the "short" and "raw"
987 format below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +0100988
989 - "stdout" / "stderr", which are respectively aliases for "fd@1" and
990 "fd@2", see above.
991
Willy Tarreauc046d162019-08-30 15:24:59 +0200992 - A ring buffer in the form "ring@<name>", which will correspond to an
993 in-memory ring buffer accessible over the CLI using the "show events"
994 command, which will also list existing rings and their sizes. Such
995 buffers are lost on reload or restart but when used as a complement
996 this can help troubleshooting by having the logs instantly available.
997
William Lallemandb2f07452015-05-12 14:27:13 +0200998 You may want to reference some environment variables in the address
999 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01001000
Willy Tarreau18324f52014-06-27 18:10:07 +02001001 <length> is an optional maximum line length. Log lines larger than this value
1002 will be truncated before being sent. The reason is that syslog
1003 servers act differently on log line length. All servers support the
1004 default value of 1024, but some servers simply drop larger lines
1005 while others do log them. If a server supports long lines, it may
1006 make sense to set this value here in order to avoid truncating long
1007 lines. Similarly, if a server drops long lines, it is preferable to
1008 truncate them before sending them. Accepted values are 80 to 65535
1009 inclusive. The default value of 1024 is generally fine for all
1010 standard usages. Some specific cases of long captures or
Davor Ocelice9ed2812017-12-25 17:49:28 +01001011 JSON-formatted logs may require larger values. You may also need to
1012 increase "tune.http.logurilen" if your request URIs are truncated.
Willy Tarreau18324f52014-06-27 18:10:07 +02001013
Dragan Dosen7ad31542015-09-28 17:16:47 +02001014 <format> is the log format used when generating syslog messages. It may be
1015 one of the following :
1016
1017 rfc3164 The RFC3164 syslog message format. This is the default.
1018 (https://tools.ietf.org/html/rfc3164)
1019
1020 rfc5424 The RFC5424 syslog message format.
1021 (https://tools.ietf.org/html/rfc5424)
1022
Willy Tarreaue8746a02018-11-12 08:45:00 +01001023 short A message containing only a level between angle brackets such as
1024 '<3>', followed by the text. The PID, date, time, process name
1025 and system name are omitted. This is designed to be used with a
1026 local log server. This format is compatible with what the systemd
1027 logger consumes.
1028
Willy Tarreauc1b06452018-11-12 11:57:56 +01001029 raw A message containing only the text. The level, PID, date, time,
1030 process name and system name are omitted. This is designed to be
1031 used in containers or during development, where the severity only
1032 depends on the file descriptor used (stdout/stderr).
1033
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02001034 <ranges> A list of comma-separated ranges to identify the logs to sample.
1035 This is used to balance the load of the logs to send to the log
1036 server. The limits of the ranges cannot be null. They are numbered
1037 from 1. The size or period (in number of logs) of the sample must be
1038 set with <sample_size> parameter.
1039
1040 <sample_size>
1041 The size of the sample in number of logs to consider when balancing
1042 their logging loads. It is used to balance the load of the logs to
1043 send to the syslog server. This size must be greater or equal to the
1044 maximum of the high limits of the ranges.
1045 (see also <ranges> parameter).
1046
Robert Tsai81ae1952007-12-05 10:47:29 +01001047 <facility> must be one of the 24 standard syslog facilities :
Willy Tarreau6a06a402007-07-15 20:15:28 +02001048
Willy Tarreaue8746a02018-11-12 08:45:00 +01001049 kern user mail daemon auth syslog lpr news
1050 uucp cron auth2 ftp ntp audit alert cron2
1051 local0 local1 local2 local3 local4 local5 local6 local7
1052
Willy Tarreauc1b06452018-11-12 11:57:56 +01001053 Note that the facility is ignored for the "short" and "raw"
1054 formats, but still required as a positional field. It is
1055 recommended to use "daemon" in this case to make it clear that
1056 it's only supposed to be used locally.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001057
1058 An optional level can be specified to filter outgoing messages. By default,
Willy Tarreauf7edefa2009-05-10 17:20:05 +02001059 all messages are sent. If a maximum level is specified, only messages with a
1060 severity at least as important as this level will be sent. An optional minimum
1061 level can be specified. If it is set, logs emitted with a more severe level
1062 than this one will be capped to this level. This is used to avoid sending
1063 "emerg" messages on all terminals on some default syslog configurations.
1064 Eight levels are known :
Willy Tarreau6a06a402007-07-15 20:15:28 +02001065
Cyril Bontédc4d9032012-04-08 21:57:39 +02001066 emerg alert crit err warning notice info debug
Willy Tarreau6a06a402007-07-15 20:15:28 +02001067
Joe Williamsdf5b38f2010-12-29 17:05:48 +01001068log-send-hostname [<string>]
1069 Sets the hostname field in the syslog header. If optional "string" parameter
1070 is set the header is set to the string contents, otherwise uses the hostname
1071 of the system. Generally used if one is not relaying logs through an
1072 intermediate syslog server or for simply customizing the hostname printed in
1073 the logs.
1074
Kevinm48936af2010-12-22 16:08:21 +00001075log-tag <string>
1076 Sets the tag field in the syslog header to this string. It defaults to the
1077 program name as launched from the command line, which usually is "haproxy".
1078 Sometimes it can be useful to differentiate between multiple processes
Willy Tarreau094af4e2015-01-07 15:03:42 +01001079 running on the same host. See also the per-proxy "log-tag" directive.
Kevinm48936af2010-12-22 16:08:21 +00001080
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001081lua-load <file>
1082 This global directive loads and executes a Lua file. This directive can be
1083 used multiple times.
1084
William Lallemand4cfede82017-11-24 22:02:34 +01001085master-worker [no-exit-on-failure]
William Lallemande202b1e2017-06-01 17:38:56 +02001086 Master-worker mode. It is equivalent to the command line "-W" argument.
1087 This mode will launch a "master" which will monitor the "workers". Using
1088 this mode, you can reload HAProxy directly by sending a SIGUSR2 signal to
Davor Ocelice9ed2812017-12-25 17:49:28 +01001089 the master. The master-worker mode is compatible either with the foreground
William Lallemande202b1e2017-06-01 17:38:56 +02001090 or daemon mode. It is recommended to use this mode with multiprocess and
1091 systemd.
William Lallemand4cfede82017-11-24 22:02:34 +01001092 By default, if a worker exits with a bad return code, in the case of a
1093 segfault for example, all workers will be killed, and the master will leave.
1094 It is convenient to combine this behavior with Restart=on-failure in a
1095 systemd unit file in order to relaunch the whole process. If you don't want
1096 this behavior, you must use the keyword "no-exit-on-failure".
William Lallemande202b1e2017-06-01 17:38:56 +02001097
William Lallemand4cfede82017-11-24 22:02:34 +01001098 See also "-W" in the management guide.
William Lallemande202b1e2017-06-01 17:38:56 +02001099
William Lallemand27edc4b2019-05-07 17:49:33 +02001100mworker-max-reloads <number>
1101 In master-worker mode, this option limits the number of time a worker can
John Roeslerfb2fce12019-07-10 15:45:51 -05001102 survive to a reload. If the worker did not leave after a reload, once its
William Lallemand27edc4b2019-05-07 17:49:33 +02001103 number of reloads is greater than this number, the worker will receive a
1104 SIGTERM. This option helps to keep under control the number of workers.
1105 See also "show proc" in the Management Guide.
1106
Willy Tarreau6a06a402007-07-15 20:15:28 +02001107nbproc <number>
1108 Creates <number> processes when going daemon. This requires the "daemon"
1109 mode. By default, only one process is created, which is the recommended mode
1110 of operation. For systems limited to small sets of file descriptors per
Willy Tarreau149ab772019-01-26 14:27:06 +01001111 process, it may be needed to fork multiple daemons. When set to a value
1112 larger than 1, threads are automatically disabled. USING MULTIPLE PROCESSES
Willy Tarreau1f672a82019-01-26 14:20:55 +01001113 IS HARDER TO DEBUG AND IS REALLY DISCOURAGED. See also "daemon" and
1114 "nbthread".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001115
Christopher Fauletbe0faa22017-08-29 15:37:10 +02001116nbthread <number>
1117 This setting is only available when support for threads was built in. It
Willy Tarreau26f6ae12019-02-02 12:56:15 +01001118 makes haproxy run on <number> threads. This is exclusive with "nbproc". While
1119 "nbproc" historically used to be the only way to use multiple processors, it
1120 also involved a number of shortcomings related to the lack of synchronization
1121 between processes (health-checks, peers, stick-tables, stats, ...) which do
1122 not affect threads. As such, any modern configuration is strongly encouraged
Willy Tarreau149ab772019-01-26 14:27:06 +01001123 to migrate away from "nbproc" to "nbthread". "nbthread" also works when
1124 HAProxy is started in foreground. On some platforms supporting CPU affinity,
1125 when nbproc is not used, the default "nbthread" value is automatically set to
1126 the number of CPUs the process is bound to upon startup. This means that the
1127 thread count can easily be adjusted from the calling process using commands
1128 like "taskset" or "cpuset". Otherwise, this value defaults to 1. The default
1129 value is reported in the output of "haproxy -vv". See also "nbproc".
Christopher Fauletbe0faa22017-08-29 15:37:10 +02001130
Willy Tarreau6a06a402007-07-15 20:15:28 +02001131pidfile <pidfile>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001132 Writes PIDs of all daemons into file <pidfile>. This option is equivalent to
Willy Tarreau6a06a402007-07-15 20:15:28 +02001133 the "-p" command line argument. The file must be accessible to the user
1134 starting the process. See also "daemon".
1135
Willy Tarreau1d549722016-02-16 12:41:57 +01001136presetenv <name> <value>
1137 Sets environment variable <name> to value <value>. If the variable exists, it
1138 is NOT overwritten. The changes immediately take effect so that the next line
1139 in the configuration file sees the new value. See also "setenv", "resetenv",
1140 and "unsetenv".
1141
1142resetenv [<name> ...]
1143 Removes all environment variables except the ones specified in argument. It
1144 allows to use a clean controlled environment before setting new values with
1145 setenv or unsetenv. Please note that some internal functions may make use of
1146 some environment variables, such as time manipulation functions, but also
1147 OpenSSL or even external checks. This must be used with extreme care and only
1148 after complete validation. The changes immediately take effect so that the
1149 next line in the configuration file sees the new environment. See also
1150 "setenv", "presetenv", and "unsetenv".
1151
Christopher Fauletff4121f2017-11-22 16:38:49 +01001152stats bind-process [ all | odd | even | <process_num>[-[process_num>]] ] ...
Willy Tarreau35b7b162012-10-22 23:17:18 +02001153 Limits the stats socket to a certain set of processes numbers. By default the
1154 stats socket is bound to all processes, causing a warning to be emitted when
1155 nbproc is greater than 1 because there is no way to select the target process
1156 when connecting. However, by using this setting, it becomes possible to pin
1157 the stats socket to a specific set of processes, typically the first one. The
1158 warning will automatically be disabled when this setting is used, whatever
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001159 the number of processes used. The maximum process ID depends on the machine's
Christopher Fauletff4121f2017-11-22 16:38:49 +01001160 word size (32 or 64). Ranges can be partially defined. The higher bound can
1161 be omitted. In such case, it is replaced by the corresponding maximum
1162 value. A better option consists in using the "process" setting of the "stats
1163 socket" line to force the process on each line.
Willy Tarreau35b7b162012-10-22 23:17:18 +02001164
Baptiste Assmann5626f482015-08-23 10:00:10 +02001165server-state-base <directory>
1166 Specifies the directory prefix to be prepended in front of all servers state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001167 file names which do not start with a '/'. See also "server-state-file",
1168 "load-server-state-from-file" and "server-state-file-name".
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +02001169
1170server-state-file <file>
1171 Specifies the path to the file containing state of servers. If the path starts
1172 with a slash ('/'), it is considered absolute, otherwise it is considered
1173 relative to the directory specified using "server-state-base" (if set) or to
1174 the current directory. Before reloading HAProxy, it is possible to save the
1175 servers' current state using the stats command "show servers state". The
1176 output of this command must be written in the file pointed by <file>. When
1177 starting up, before handling traffic, HAProxy will read, load and apply state
1178 for each server found in the file and available in its current running
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001179 configuration. See also "server-state-base" and "show servers state",
1180 "load-server-state-from-file" and "server-state-file-name"
Baptiste Assmann5626f482015-08-23 10:00:10 +02001181
Willy Tarreau1d549722016-02-16 12:41:57 +01001182setenv <name> <value>
1183 Sets environment variable <name> to value <value>. If the variable exists, it
1184 is overwritten. The changes immediately take effect so that the next line in
1185 the configuration file sees the new value. See also "presetenv", "resetenv",
1186 and "unsetenv".
1187
Willy Tarreau636848a2019-04-15 19:38:50 +02001188set-dumpable
1189 This option is better left disabled by default and enabled only upon a
William Dauchyec730982019-10-27 20:08:10 +01001190 developer's request. If it has been enabled, it may still be forcibly
1191 disabled by prefixing it with the "no" keyword. It has no impact on
1192 performance nor stability but will try hard to re-enable core dumps that were
1193 possibly disabled by file size limitations (ulimit -f), core size limitations
1194 (ulimit -c), or "dumpability" of a process after changing its UID/GID (such
1195 as /proc/sys/fs/suid_dumpable on Linux). Core dumps might still be limited by
1196 the current directory's permissions (check what directory the file is started
1197 from), the chroot directory's permission (it may be needed to temporarily
1198 disable the chroot directive or to move it to a dedicated writable location),
1199 or any other system-specific constraint. For example, some Linux flavours are
1200 notorious for replacing the default core file with a path to an executable
1201 not even installed on the system (check /proc/sys/kernel/core_pattern). Often,
1202 simply writing "core", "core.%p" or "/var/log/core/core.%p" addresses the
1203 issue. When trying to enable this option waiting for a rare issue to
1204 re-appear, it's often a good idea to first try to obtain such a dump by
1205 issuing, for example, "kill -11" to the haproxy process and verify that it
1206 leaves a core where expected when dying.
Willy Tarreau636848a2019-04-15 19:38:50 +02001207
Willy Tarreau610f04b2014-02-13 11:36:41 +01001208ssl-default-bind-ciphers <ciphers>
1209 This setting is only available when support for OpenSSL was built in. It sets
1210 the default string describing the list of cipher algorithms ("cipher suite")
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +00001211 that are negotiated during the SSL/TLS handshake up to TLSv1.2 for all
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001212 "bind" lines which do not explicitly define theirs. The format of the string
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001213 is defined in "man 1 ciphers" from OpenSSL man pages. For background
1214 information and recommendations see e.g.
1215 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
1216 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
1217 cipher configuration, please check the "ssl-default-bind-ciphersuites" keyword.
1218 Please check the "bind" keyword for more information.
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001219
1220ssl-default-bind-ciphersuites <ciphersuites>
1221 This setting is only available when support for OpenSSL was built in and
1222 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default string
1223 describing the list of cipher algorithms ("cipher suite") that are negotiated
1224 during the TLSv1.3 handshake for all "bind" lines which do not explicitly define
1225 theirs. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001226 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites". For
1227 cipher configuration for TLSv1.2 and earlier, please check the
1228 "ssl-default-bind-ciphers" keyword. Please check the "bind" keyword for more
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001229 information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001230
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001231ssl-default-bind-options [<option>]...
1232 This setting is only available when support for OpenSSL was built in. It sets
1233 default ssl-options to force on all "bind" lines. Please check the "bind"
1234 keyword to see available options.
1235
1236 Example:
1237 global
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +02001238 ssl-default-bind-options ssl-min-ver TLSv1.0 no-tls-tickets
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001239
Willy Tarreau610f04b2014-02-13 11:36:41 +01001240ssl-default-server-ciphers <ciphers>
1241 This setting is only available when support for OpenSSL was built in. It
1242 sets the default string describing the list of cipher algorithms that are
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +00001243 negotiated during the SSL/TLS handshake up to TLSv1.2 with the server,
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001244 for all "server" lines which do not explicitly define theirs. The format of
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001245 the string is defined in "man 1 ciphers" from OpenSSL man pages. For background
1246 information and recommendations see e.g.
1247 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
1248 (https://mozilla.github.io/server-side-tls/ssl-config-generator/).
1249 For TLSv1.3 cipher configuration, please check the
1250 "ssl-default-server-ciphersuites" keyword. Please check the "server" keyword
1251 for more information.
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001252
1253ssl-default-server-ciphersuites <ciphersuites>
1254 This setting is only available when support for OpenSSL was built in and
1255 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default
1256 string describing the list of cipher algorithms that are negotiated during
1257 the TLSv1.3 handshake with the server, for all "server" lines which do not
1258 explicitly define theirs. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001259 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites". For
1260 cipher configuration for TLSv1.2 and earlier, please check the
1261 "ssl-default-server-ciphers" keyword. Please check the "server" keyword for
1262 more information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001263
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001264ssl-default-server-options [<option>]...
1265 This setting is only available when support for OpenSSL was built in. It sets
1266 default ssl-options to force on all "server" lines. Please check the "server"
1267 keyword to see available options.
1268
Remi Gacogne47783ef2015-05-29 15:53:22 +02001269ssl-dh-param-file <file>
1270 This setting is only available when support for OpenSSL was built in. It sets
1271 the default DH parameters that are used during the SSL/TLS handshake when
1272 ephemeral Diffie-Hellman (DHE) key exchange is used, for all "bind" lines
Davor Ocelice9ed2812017-12-25 17:49:28 +01001273 which do not explicitly define theirs. It will be overridden by custom DH
Remi Gacogne47783ef2015-05-29 15:53:22 +02001274 parameters found in a bind certificate file if any. If custom DH parameters
Cyril Bonté307ee1e2015-09-28 23:16:06 +02001275 are not specified either by using ssl-dh-param-file or by setting them
1276 directly in the certificate file, pre-generated DH parameters of the size
1277 specified by tune.ssl.default-dh-param will be used. Custom parameters are
1278 known to be more secure and therefore their use is recommended.
Remi Gacogne47783ef2015-05-29 15:53:22 +02001279 Custom DH parameters may be generated by using the OpenSSL command
1280 "openssl dhparam <size>", where size should be at least 2048, as 1024-bit DH
1281 parameters should not be considered secure anymore.
1282
Emeric Brun850efd52014-01-29 12:24:34 +01001283ssl-server-verify [none|required]
1284 The default behavior for SSL verify on servers side. If specified to 'none',
1285 servers certificates are not verified. The default is 'required' except if
1286 forced using cmdline option '-dV'.
1287
Willy Tarreauabb175f2012-09-24 12:43:26 +02001288stats socket [<address:port>|<path>] [param*]
1289 Binds a UNIX socket to <path> or a TCPv4/v6 address to <address:port>.
1290 Connections to this socket will return various statistics outputs and even
1291 allow some commands to be issued to change some runtime settings. Please
Willy Tarreau1af20c72017-06-23 16:01:14 +02001292 consult section 9.3 "Unix Socket commands" of Management Guide for more
Kevin Decherf949c7202015-10-13 23:26:44 +02001293 details.
Willy Tarreau6162db22009-10-10 17:13:00 +02001294
Willy Tarreauabb175f2012-09-24 12:43:26 +02001295 All parameters supported by "bind" lines are supported, for instance to
1296 restrict access to some users or their access rights. Please consult
1297 section 5.1 for more information.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001298
1299stats timeout <timeout, in milliseconds>
1300 The default timeout on the stats socket is set to 10 seconds. It is possible
1301 to change this value with "stats timeout". The value must be passed in
Willy Tarreaubefdff12007-12-02 22:27:38 +01001302 milliseconds, or be suffixed by a time unit among { us, ms, s, m, h, d }.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001303
1304stats maxconn <connections>
1305 By default, the stats socket is limited to 10 concurrent connections. It is
1306 possible to change this value with "stats maxconn".
1307
Willy Tarreau6a06a402007-07-15 20:15:28 +02001308uid <number>
1309 Changes the process' user ID to <number>. It is recommended that the user ID
1310 is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
1311 be started with superuser privileges in order to be able to switch to another
1312 one. See also "gid" and "user".
1313
1314ulimit-n <number>
1315 Sets the maximum number of per-process file-descriptors to <number>. By
1316 default, it is automatically computed, so it is recommended not to use this
1317 option.
1318
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001319unix-bind [ prefix <prefix> ] [ mode <mode> ] [ user <user> ] [ uid <uid> ]
1320 [ group <group> ] [ gid <gid> ]
1321
1322 Fixes common settings to UNIX listening sockets declared in "bind" statements.
1323 This is mainly used to simplify declaration of those UNIX sockets and reduce
1324 the risk of errors, since those settings are most commonly required but are
1325 also process-specific. The <prefix> setting can be used to force all socket
1326 path to be relative to that directory. This might be needed to access another
1327 component's chroot. Note that those paths are resolved before haproxy chroots
1328 itself, so they are absolute. The <mode>, <user>, <uid>, <group> and <gid>
1329 all have the same meaning as their homonyms used by the "bind" statement. If
1330 both are specified, the "bind" statement has priority, meaning that the
1331 "unix-bind" settings may be seen as process-wide default settings.
1332
Willy Tarreau1d549722016-02-16 12:41:57 +01001333unsetenv [<name> ...]
1334 Removes environment variables specified in arguments. This can be useful to
1335 hide some sensitive information that are occasionally inherited from the
1336 user's environment during some operations. Variables which did not exist are
1337 silently ignored so that after the operation, it is certain that none of
1338 these variables remain. The changes immediately take effect so that the next
1339 line in the configuration file will not see these variables. See also
1340 "setenv", "presetenv", and "resetenv".
1341
Willy Tarreau6a06a402007-07-15 20:15:28 +02001342user <user name>
1343 Similar to "uid" but uses the UID of user name <user name> from /etc/passwd.
1344 See also "uid" and "group".
1345
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02001346node <name>
1347 Only letters, digits, hyphen and underscore are allowed, like in DNS names.
1348
1349 This statement is useful in HA configurations where two or more processes or
1350 servers share the same IP address. By setting a different node-name on all
1351 nodes, it becomes easy to immediately spot what server is handling the
1352 traffic.
1353
1354description <text>
1355 Add a text that describes the instance.
1356
1357 Please note that it is required to escape certain characters (# for example)
1358 and this text is inserted into a html page so you should avoid using
1359 "<" and ">" characters.
1360
Thomas Holmesdb04f192015-05-18 13:21:39 +0100136151degrees-data-file <file path>
1362 The path of the 51Degrees data file to provide device detection services. The
Davor Ocelice9ed2812017-12-25 17:49:28 +01001363 file should be unzipped and accessible by HAProxy with relevant permissions.
Thomas Holmesdb04f192015-05-18 13:21:39 +01001364
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001365 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001366 compiled with USE_51DEGREES.
1367
Ben Shillitof25e8e52016-12-02 14:25:37 +0000136851degrees-property-name-list [<string> ...]
Thomas Holmesdb04f192015-05-18 13:21:39 +01001369 A list of 51Degrees property names to be load from the dataset. A full list
1370 of names is available on the 51Degrees website:
1371 https://51degrees.com/resources/property-dictionary
1372
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001373 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001374 compiled with USE_51DEGREES.
1375
Dragan Dosen93b38d92015-06-29 16:43:25 +0200137651degrees-property-separator <char>
Thomas Holmesdb04f192015-05-18 13:21:39 +01001377 A char that will be appended to every property value in a response header
1378 containing 51Degrees results. If not set that will be set as ','.
1379
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001380 Please note that this option is only available when haproxy has been
1381 compiled with USE_51DEGREES.
1382
138351degrees-cache-size <number>
1384 Sets the size of the 51Degrees converter cache to <number> entries. This
1385 is an LRU cache which reminds previous device detections and their results.
1386 By default, this cache is disabled.
1387
1388 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001389 compiled with USE_51DEGREES.
1390
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001391wurfl-data-file <file path>
1392 The path of the WURFL data file to provide device detection services. The
1393 file should be accessible by HAProxy with relevant permissions.
1394
1395 Please note that this option is only available when haproxy has been compiled
1396 with USE_WURFL=1.
1397
1398wurfl-information-list [<capability>]*
1399 A space-delimited list of WURFL capabilities, virtual capabilities, property
1400 names we plan to use in injected headers. A full list of capability and
1401 virtual capability names is available on the Scientiamobile website :
1402
1403 https://www.scientiamobile.com/wurflCapability
1404
1405 Valid WURFL properties are:
1406 - wurfl_id Contains the device ID of the matched device.
1407
1408 - wurfl_root_id Contains the device root ID of the matched
1409 device.
1410
1411 - wurfl_isdevroot Tells if the matched device is a root device.
1412 Possible values are "TRUE" or "FALSE".
1413
1414 - wurfl_useragent The original useragent coming with this
1415 particular web request.
1416
1417 - wurfl_api_version Contains a string representing the currently
1418 used Libwurfl API version.
1419
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001420 - wurfl_info A string containing information on the parsed
1421 wurfl.xml and its full path.
1422
1423 - wurfl_last_load_time Contains the UNIX timestamp of the last time
1424 WURFL has been loaded successfully.
1425
1426 - wurfl_normalized_useragent The normalized useragent.
1427
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001428 Please note that this option is only available when haproxy has been compiled
1429 with USE_WURFL=1.
1430
1431wurfl-information-list-separator <char>
1432 A char that will be used to separate values in a response header containing
1433 WURFL results. If not set that a comma (',') will be used by default.
1434
1435 Please note that this option is only available when haproxy has been compiled
1436 with USE_WURFL=1.
1437
1438wurfl-patch-file [<file path>]
1439 A list of WURFL patch file paths. Note that patches are loaded during startup
1440 thus before the chroot.
1441
1442 Please note that this option is only available when haproxy has been compiled
1443 with USE_WURFL=1.
1444
paulborilebad132c2019-04-18 11:57:04 +02001445wurfl-cache-size <size>
1446 Sets the WURFL Useragent cache size. For faster lookups, already processed user
1447 agents are kept in a LRU cache :
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001448 - "0" : no cache is used.
paulborilebad132c2019-04-18 11:57:04 +02001449 - <size> : size of lru cache in elements.
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001450
1451 Please note that this option is only available when haproxy has been compiled
1452 with USE_WURFL=1.
1453
William Dauchy0fec3ab2019-10-27 20:08:11 +01001454strict-limits
1455 Makes process fail at startup when a setrlimit fails. Haproxy is tries to set
1456 the best setrlimit according to what has been calculated. If it fails, it
1457 will emit a warning. Use this option if you want an explicit failure of
1458 haproxy when those limits fail. This option is disabled by default. If it has
1459 been enabled, it may still be forcibly disabled by prefixing it with the "no"
1460 keyword.
1461
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014623.2. Performance tuning
Willy Tarreau6a06a402007-07-15 20:15:28 +02001463-----------------------
1464
Willy Tarreaubeb859a2018-11-22 18:07:59 +01001465busy-polling
1466 In some situations, especially when dealing with low latency on processors
1467 supporting a variable frequency or when running inside virtual machines, each
1468 time the process waits for an I/O using the poller, the processor goes back
1469 to sleep or is offered to another VM for a long time, and it causes
1470 excessively high latencies. This option provides a solution preventing the
1471 processor from sleeping by always using a null timeout on the pollers. This
1472 results in a significant latency reduction (30 to 100 microseconds observed)
1473 at the expense of a risk to overheat the processor. It may even be used with
1474 threads, in which case improperly bound threads may heavily conflict,
1475 resulting in a worse performance and high values for the CPU stolen fields
1476 in "show info" output, indicating which threads are misconfigured. It is
1477 important not to let the process run on the same processor as the network
1478 interrupts when this option is used. It is also better to avoid using it on
1479 multiple CPU threads sharing the same core. This option is disabled by
1480 default. If it has been enabled, it may still be forcibly disabled by
1481 prefixing it with the "no" keyword. It is ignored by the "select" and
1482 "poll" pollers.
1483
William Dauchy3894d972019-12-28 15:36:02 +01001484 This option is automatically disabled on old processes in the context of
1485 seamless reload; it avoids too much cpu conflicts when multiple processes
1486 stay around for some time waiting for the end of their current connections.
1487
Willy Tarreau1746eec2014-04-25 10:46:47 +02001488max-spread-checks <delay in milliseconds>
1489 By default, haproxy tries to spread the start of health checks across the
1490 smallest health check interval of all the servers in a farm. The principle is
1491 to avoid hammering services running on the same server. But when using large
1492 check intervals (10 seconds or more), the last servers in the farm take some
1493 time before starting to be tested, which can be a problem. This parameter is
1494 used to enforce an upper bound on delay between the first and the last check,
1495 even if the servers' check intervals are larger. When servers run with
1496 shorter intervals, their intervals will be respected though.
1497
Willy Tarreau6a06a402007-07-15 20:15:28 +02001498maxconn <number>
1499 Sets the maximum per-process number of concurrent connections to <number>. It
1500 is equivalent to the command-line argument "-n". Proxies will stop accepting
1501 connections when this limit is reached. The "ulimit-n" parameter is
Willy Tarreau8274e102014-06-19 15:31:25 +02001502 automatically adjusted according to this value. See also "ulimit-n". Note:
1503 the "select" poller cannot reliably use more than 1024 file descriptors on
1504 some platforms. If your platform only supports select and reports "select
1505 FAILED" on startup, you need to reduce maxconn until it works (slightly
Willy Tarreaub28f3442019-03-04 08:13:43 +01001506 below 500 in general). If this value is not set, it will automatically be
1507 calculated based on the current file descriptors limit reported by the
1508 "ulimit -n" command, possibly reduced to a lower value if a memory limit
1509 is enforced, based on the buffer size, memory allocated to compression, SSL
1510 cache size, and use or not of SSL and the associated maxsslconn (which can
1511 also be automatic).
Willy Tarreau6a06a402007-07-15 20:15:28 +02001512
Willy Tarreau81c25d02011-09-07 15:17:21 +02001513maxconnrate <number>
1514 Sets the maximum per-process number of connections per second to <number>.
1515 Proxies will stop accepting connections when this limit is reached. It can be
1516 used to limit the global capacity regardless of each frontend capacity. It is
1517 important to note that this can only be used as a service protection measure,
1518 as there will not necessarily be a fair share between frontends when the
1519 limit is reached, so it's a good idea to also limit each frontend to some
1520 value close to its expected share. Also, lowering tune.maxaccept can improve
1521 fairness.
1522
William Lallemandd85f9172012-11-09 17:05:39 +01001523maxcomprate <number>
1524 Sets the maximum per-process input compression rate to <number> kilobytes
Davor Ocelice9ed2812017-12-25 17:49:28 +01001525 per second. For each session, if the maximum is reached, the compression
William Lallemandd85f9172012-11-09 17:05:39 +01001526 level will be decreased during the session. If the maximum is reached at the
1527 beginning of a session, the session will not compress at all. If the maximum
1528 is not reached, the compression level will be increased up to
Davor Ocelice9ed2812017-12-25 17:49:28 +01001529 tune.comp.maxlevel. A value of zero means there is no limit, this is the
William Lallemandd85f9172012-11-09 17:05:39 +01001530 default value.
1531
William Lallemand072a2bf2012-11-20 17:01:01 +01001532maxcompcpuusage <number>
1533 Sets the maximum CPU usage HAProxy can reach before stopping the compression
1534 for new requests or decreasing the compression level of current requests.
1535 It works like 'maxcomprate' but measures CPU usage instead of incoming data
1536 bandwidth. The value is expressed in percent of the CPU used by haproxy. In
1537 case of multiple processes (nbproc > 1), each process manages its individual
1538 usage. A value of 100 disable the limit. The default value is 100. Setting
1539 a lower value will prevent the compression work from slowing the whole
1540 process down and from introducing high latencies.
1541
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001542maxpipes <number>
1543 Sets the maximum per-process number of pipes to <number>. Currently, pipes
1544 are only used by kernel-based tcp splicing. Since a pipe contains two file
1545 descriptors, the "ulimit-n" value will be increased accordingly. The default
1546 value is maxconn/4, which seems to be more than enough for most heavy usages.
1547 The splice code dynamically allocates and releases pipes, and can fall back
1548 to standard copy, so setting this value too low may only impact performance.
1549
Willy Tarreau93e7c002013-10-07 18:51:07 +02001550maxsessrate <number>
1551 Sets the maximum per-process number of sessions per second to <number>.
1552 Proxies will stop accepting connections when this limit is reached. It can be
1553 used to limit the global capacity regardless of each frontend capacity. It is
1554 important to note that this can only be used as a service protection measure,
1555 as there will not necessarily be a fair share between frontends when the
1556 limit is reached, so it's a good idea to also limit each frontend to some
1557 value close to its expected share. Also, lowering tune.maxaccept can improve
1558 fairness.
1559
Willy Tarreau403edff2012-09-06 11:58:37 +02001560maxsslconn <number>
1561 Sets the maximum per-process number of concurrent SSL connections to
1562 <number>. By default there is no SSL-specific limit, which means that the
1563 global maxconn setting will apply to all connections. Setting this limit
1564 avoids having openssl use too much memory and crash when malloc returns NULL
1565 (since it unfortunately does not reliably check for such conditions). Note
1566 that the limit applies both to incoming and outgoing connections, so one
1567 connection which is deciphered then ciphered accounts for 2 SSL connections.
Willy Tarreaud0256482015-01-15 21:45:22 +01001568 If this value is not set, but a memory limit is enforced, this value will be
1569 automatically computed based on the memory limit, maxconn, the buffer size,
1570 memory allocated to compression, SSL cache size, and use of SSL in either
1571 frontends, backends or both. If neither maxconn nor maxsslconn are specified
1572 when there is a memory limit, haproxy will automatically adjust these values
1573 so that 100% of the connections can be made over SSL with no risk, and will
1574 consider the sides where it is enabled (frontend, backend, both).
Willy Tarreau403edff2012-09-06 11:58:37 +02001575
Willy Tarreaue43d5322013-10-07 20:01:52 +02001576maxsslrate <number>
1577 Sets the maximum per-process number of SSL sessions per second to <number>.
1578 SSL listeners will stop accepting connections when this limit is reached. It
1579 can be used to limit the global SSL CPU usage regardless of each frontend
1580 capacity. It is important to note that this can only be used as a service
1581 protection measure, as there will not necessarily be a fair share between
1582 frontends when the limit is reached, so it's a good idea to also limit each
1583 frontend to some value close to its expected share. It is also important to
1584 note that the sessions are accounted before they enter the SSL stack and not
1585 after, which also protects the stack against bad handshakes. Also, lowering
1586 tune.maxaccept can improve fairness.
1587
William Lallemand9d5f5482012-11-07 16:12:57 +01001588maxzlibmem <number>
1589 Sets the maximum amount of RAM in megabytes per process usable by the zlib.
1590 When the maximum amount is reached, future sessions will not compress as long
1591 as RAM is unavailable. When sets to 0, there is no limit.
William Lallemande3a7d992012-11-20 11:25:20 +01001592 The default value is 0. The value is available in bytes on the UNIX socket
1593 with "show info" on the line "MaxZlibMemUsage", the memory used by zlib is
1594 "ZlibMemUsage" in bytes.
1595
Willy Tarreau6a06a402007-07-15 20:15:28 +02001596noepoll
1597 Disables the use of the "epoll" event polling system on Linux. It is
1598 equivalent to the command-line argument "-de". The next polling system
Willy Tarreaue9f49e72012-11-11 17:42:00 +01001599 used will generally be "poll". See also "nopoll".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001600
1601nokqueue
1602 Disables the use of the "kqueue" event polling system on BSD. It is
1603 equivalent to the command-line argument "-dk". The next polling system
1604 used will generally be "poll". See also "nopoll".
1605
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +00001606noevports
1607 Disables the use of the event ports event polling system on SunOS systems
1608 derived from Solaris 10 and later. It is equivalent to the command-line
1609 argument "-dv". The next polling system used will generally be "poll". See
1610 also "nopoll".
1611
Willy Tarreau6a06a402007-07-15 20:15:28 +02001612nopoll
1613 Disables the use of the "poll" event polling system. It is equivalent to the
1614 command-line argument "-dp". The next polling system used will be "select".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001615 It should never be needed to disable "poll" since it's available on all
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +00001616 platforms supported by HAProxy. See also "nokqueue", "noepoll" and
1617 "noevports".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001618
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001619nosplice
1620 Disables the use of kernel tcp splicing between sockets on Linux. It is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001621 equivalent to the command line argument "-dS". Data will then be copied
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001622 using conventional and more portable recv/send calls. Kernel tcp splicing is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001623 limited to some very recent instances of kernel 2.6. Most versions between
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001624 2.6.25 and 2.6.28 are buggy and will forward corrupted data, so they must not
1625 be used. This option makes it easier to globally disable kernel splicing in
1626 case of doubt. See also "option splice-auto", "option splice-request" and
1627 "option splice-response".
1628
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001629nogetaddrinfo
1630 Disables the use of getaddrinfo(3) for name resolving. It is equivalent to
1631 the command line argument "-dG". Deprecated gethostbyname(3) will be used.
1632
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +00001633noreuseport
1634 Disables the use of SO_REUSEPORT - see socket(7). It is equivalent to the
1635 command line argument "-dR".
1636
Willy Tarreaud2d33482019-04-25 17:09:07 +02001637profiling.tasks { auto | on | off }
1638 Enables ('on') or disables ('off') per-task CPU profiling. When set to 'auto'
1639 the profiling automatically turns on a thread when it starts to suffer from
1640 an average latency of 1000 microseconds or higher as reported in the
1641 "avg_loop_us" activity field, and automatically turns off when the latency
John Roeslerfb2fce12019-07-10 15:45:51 -05001642 returns below 990 microseconds (this value is an average over the last 1024
Willy Tarreaud2d33482019-04-25 17:09:07 +02001643 loops so it does not vary quickly and tends to significantly smooth short
1644 spikes). It may also spontaneously trigger from time to time on overloaded
1645 systems, containers, or virtual machines, or when the system swaps (which
1646 must absolutely never happen on a load balancer).
1647
1648 CPU profiling per task can be very convenient to report where the time is
1649 spent and which requests have what effect on which other request. Enabling
1650 it will typically affect the overall's performance by less than 1%, thus it
1651 is recommended to leave it to the default 'auto' value so that it only
1652 operates when a problem is identified. This feature requires a system
Willy Tarreau75c62c22018-11-22 11:02:09 +01001653 supporting the clock_gettime(2) syscall with clock identifiers
1654 CLOCK_MONOTONIC and CLOCK_THREAD_CPUTIME_ID, otherwise the reported time will
1655 be zero. This option may be changed at run time using "set profiling" on the
1656 CLI.
1657
Willy Tarreaufe255b72007-10-14 23:09:26 +02001658spread-checks <0..50, in percent>
Simon Hormand60d6912013-11-25 10:46:36 +09001659 Sometimes it is desirable to avoid sending agent and health checks to
1660 servers at exact intervals, for instance when many logical servers are
1661 located on the same physical server. With the help of this parameter, it
1662 becomes possible to add some randomness in the check interval between 0
1663 and +/- 50%. A value between 2 and 5 seems to show good results. The
1664 default value remains at 0.
Willy Tarreaufe255b72007-10-14 23:09:26 +02001665
Davor Ocelice9ed2812017-12-25 17:49:28 +01001666ssl-engine <name> [algo <comma-separated list of algorithms>]
Grant Zhang872f9c22017-01-21 01:10:18 +00001667 Sets the OpenSSL engine to <name>. List of valid values for <name> may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01001668 obtained using the command "openssl engine". This statement may be used
Grant Zhang872f9c22017-01-21 01:10:18 +00001669 multiple times, it will simply enable multiple crypto engines. Referencing an
1670 unsupported engine will prevent haproxy from starting. Note that many engines
1671 will lead to lower HTTPS performance than pure software with recent
1672 processors. The optional command "algo" sets the default algorithms an ENGINE
1673 will supply using the OPENSSL function ENGINE_set_default_string(). A value
Davor Ocelice9ed2812017-12-25 17:49:28 +01001674 of "ALL" uses the engine for all cryptographic operations. If no list of
1675 algo is specified then the value of "ALL" is used. A comma-separated list
Grant Zhang872f9c22017-01-21 01:10:18 +00001676 of different algorithms may be specified, including: RSA, DSA, DH, EC, RAND,
1677 CIPHERS, DIGESTS, PKEY, PKEY_CRYPTO, PKEY_ASN1. This is the same format that
1678 openssl configuration file uses:
1679 https://www.openssl.org/docs/man1.0.2/apps/config.html
1680
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00001681ssl-mode-async
1682 Adds SSL_MODE_ASYNC mode to the SSL context. This enables asynchronous TLS
Emeric Brun3854e012017-05-17 20:42:48 +02001683 I/O operations if asynchronous capable SSL engines are used. The current
Emeric Brunb5e42a82017-06-06 12:35:14 +00001684 implementation supports a maximum of 32 engines. The Openssl ASYNC API
1685 doesn't support moving read/write buffers and is not compliant with
1686 haproxy's buffer management. So the asynchronous mode is disabled on
John Roeslerfb2fce12019-07-10 15:45:51 -05001687 read/write operations (it is only enabled during initial and renegotiation
Emeric Brunb5e42a82017-06-06 12:35:14 +00001688 handshakes).
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00001689
Willy Tarreau33cb0652014-12-23 22:52:37 +01001690tune.buffers.limit <number>
1691 Sets a hard limit on the number of buffers which may be allocated per process.
1692 The default value is zero which means unlimited. The minimum non-zero value
1693 will always be greater than "tune.buffers.reserve" and should ideally always
1694 be about twice as large. Forcing this value can be particularly useful to
1695 limit the amount of memory a process may take, while retaining a sane
Davor Ocelice9ed2812017-12-25 17:49:28 +01001696 behavior. When this limit is reached, sessions which need a buffer wait for
Willy Tarreau33cb0652014-12-23 22:52:37 +01001697 another one to be released by another session. Since buffers are dynamically
1698 allocated and released, the waiting time is very short and not perceptible
1699 provided that limits remain reasonable. In fact sometimes reducing the limit
1700 may even increase performance by increasing the CPU cache's efficiency. Tests
1701 have shown good results on average HTTP traffic with a limit to 1/10 of the
1702 expected global maxconn setting, which also significantly reduces memory
1703 usage. The memory savings come from the fact that a number of connections
1704 will not allocate 2*tune.bufsize. It is best not to touch this value unless
1705 advised to do so by an haproxy core developer.
1706
Willy Tarreau1058ae72014-12-23 22:40:40 +01001707tune.buffers.reserve <number>
1708 Sets the number of buffers which are pre-allocated and reserved for use only
1709 during memory shortage conditions resulting in failed memory allocations. The
1710 minimum value is 2 and is also the default. There is no reason a user would
1711 want to change this value, it's mostly aimed at haproxy core developers.
1712
Willy Tarreau27a674e2009-08-17 07:23:33 +02001713tune.bufsize <number>
1714 Sets the buffer size to this size (in bytes). Lower values allow more
1715 sessions to coexist in the same amount of RAM, and higher values allow some
1716 applications with very large cookies to work. The default value is 16384 and
1717 can be changed at build time. It is strongly recommended not to change this
1718 from the default value, as very low values will break some services such as
1719 statistics, and values larger than default size will increase memory usage,
1720 possibly causing the system to run out of memory. At least the global maxconn
Willy Tarreau45a66cc2017-11-24 11:28:00 +01001721 parameter should be decreased by the same factor as this one is increased. In
1722 addition, use of HTTP/2 mandates that this value must be 16384 or more. If an
1723 HTTP request is larger than (tune.bufsize - tune.maxrewrite), haproxy will
Dmitry Sivachenkof6f4f7b2012-10-21 18:10:25 +04001724 return HTTP 400 (Bad Request) error. Similarly if an HTTP response is larger
Willy Tarreauc77d3642018-12-12 06:19:42 +01001725 than this size, haproxy will return HTTP 502 (Bad Gateway). Note that the
1726 value set using this parameter will automatically be rounded up to the next
1727 multiple of 8 on 32-bit machines and 16 on 64-bit machines.
Willy Tarreau27a674e2009-08-17 07:23:33 +02001728
Willy Tarreau43961d52010-10-04 20:39:20 +02001729tune.chksize <number>
1730 Sets the check buffer size to this size (in bytes). Higher values may help
1731 find string or regex patterns in very large pages, though doing so may imply
1732 more memory and CPU usage. The default value is 16384 and can be changed at
1733 build time. It is not recommended to change this value, but to use better
1734 checks whenever possible.
1735
William Lallemandf3747832012-11-09 12:33:10 +01001736tune.comp.maxlevel <number>
1737 Sets the maximum compression level. The compression level affects CPU
1738 usage during compression. This value affects CPU usage during compression.
1739 Each session using compression initializes the compression algorithm with
1740 this value. The default value is 1.
1741
Willy Tarreauc299e1e2019-02-27 11:35:12 +01001742tune.fail-alloc
1743 If compiled with DEBUG_FAIL_ALLOC, gives the percentage of chances an
1744 allocation attempt fails. Must be between 0 (no failure) and 100 (no
1745 success). This is useful to debug and make sure memory failures are handled
1746 gracefully.
1747
Willy Tarreaufe20e5b2017-07-27 11:42:14 +02001748tune.h2.header-table-size <number>
1749 Sets the HTTP/2 dynamic header table size. It defaults to 4096 bytes and
1750 cannot be larger than 65536 bytes. A larger value may help certain clients
1751 send more compact requests, depending on their capabilities. This amount of
1752 memory is consumed for each HTTP/2 connection. It is recommended not to
1753 change it.
1754
Willy Tarreaue6baec02017-07-27 11:45:11 +02001755tune.h2.initial-window-size <number>
1756 Sets the HTTP/2 initial window size, which is the number of bytes the client
Davor Ocelice9ed2812017-12-25 17:49:28 +01001757 can upload before waiting for an acknowledgment from haproxy. This setting
1758 only affects payload contents (i.e. the body of POST requests), not headers.
Willy Tarreaue6baec02017-07-27 11:45:11 +02001759 The default value is 65535, which roughly allows up to 5 Mbps of upload
1760 bandwidth per client over a network showing a 100 ms ping time, or 500 Mbps
1761 over a 1-ms local network. It can make sense to increase this value to allow
1762 faster uploads, or to reduce it to increase fairness when dealing with many
1763 clients. It doesn't affect resource usage.
1764
Willy Tarreau5242ef82017-07-27 11:47:28 +02001765tune.h2.max-concurrent-streams <number>
1766 Sets the HTTP/2 maximum number of concurrent streams per connection (ie the
1767 number of outstanding requests on a single connection). The default value is
1768 100. A larger one may slightly improve page load time for complex sites when
1769 visited over high latency networks, but increases the amount of resources a
1770 single client may allocate. A value of zero disables the limit so a single
1771 client may create as many streams as allocatable by haproxy. It is highly
1772 recommended not to change this value.
1773
Willy Tarreaua24b35c2019-02-21 13:24:36 +01001774tune.h2.max-frame-size <number>
1775 Sets the HTTP/2 maximum frame size that haproxy announces it is willing to
1776 receive to its peers. The default value is the largest between 16384 and the
1777 buffer size (tune.bufsize). In any case, haproxy will not announce support
1778 for frame sizes larger than buffers. The main purpose of this setting is to
1779 allow to limit the maximum frame size setting when using large buffers. Too
1780 large frame sizes might have performance impact or cause some peers to
1781 misbehave. It is highly recommended not to change this value.
1782
Willy Tarreau193b8c62012-11-22 00:17:38 +01001783tune.http.cookielen <number>
1784 Sets the maximum length of captured cookies. This is the maximum value that
1785 the "capture cookie xxx len yyy" will be allowed to take, and any upper value
1786 will automatically be truncated to this one. It is important not to set too
1787 high a value because all cookie captures still allocate this size whatever
1788 their configured value (they share a same pool). This value is per request
1789 per response, so the memory allocated is twice this value per connection.
1790 When not specified, the limit is set to 63 characters. It is recommended not
1791 to change this value.
1792
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001793tune.http.logurilen <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001794 Sets the maximum length of request URI in logs. This prevents truncating long
1795 request URIs with valuable query strings in log lines. This is not related
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001796 to syslog limits. If you increase this limit, you may also increase the
Davor Ocelice9ed2812017-12-25 17:49:28 +01001797 'log ... len yyy' parameter. Your syslog daemon may also need specific
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001798 configuration directives too.
1799 The default value is 1024.
1800
Willy Tarreauac1932d2011-10-24 19:14:41 +02001801tune.http.maxhdr <number>
1802 Sets the maximum number of headers in a request. When a request comes with a
1803 number of headers greater than this value (including the first line), it is
1804 rejected with a "400 Bad Request" status code. Similarly, too large responses
1805 are blocked with "502 Bad Gateway". The default value is 101, which is enough
1806 for all usages, considering that the widely deployed Apache server uses the
1807 same limit. It can be useful to push this limit further to temporarily allow
Christopher Faulet50174f32017-06-21 16:31:35 +02001808 a buggy application to work by the time it gets fixed. The accepted range is
1809 1..32767. Keep in mind that each new header consumes 32bits of memory for
1810 each session, so don't push this limit too high.
Willy Tarreauac1932d2011-10-24 19:14:41 +02001811
Willy Tarreau7e312732014-02-12 16:35:14 +01001812tune.idletimer <timeout>
1813 Sets the duration after which haproxy will consider that an empty buffer is
1814 probably associated with an idle stream. This is used to optimally adjust
1815 some packet sizes while forwarding large and small data alternatively. The
1816 decision to use splice() or to send large buffers in SSL is modulated by this
1817 parameter. The value is in milliseconds between 0 and 65535. A value of zero
1818 means that haproxy will not try to detect idle streams. The default is 1000,
Davor Ocelice9ed2812017-12-25 17:49:28 +01001819 which seems to correctly detect end user pauses (e.g. read a page before
John Roeslerfb2fce12019-07-10 15:45:51 -05001820 clicking). There should be no reason for changing this value. Please check
Willy Tarreau7e312732014-02-12 16:35:14 +01001821 tune.ssl.maxrecord below.
1822
Willy Tarreau7ac908b2019-02-27 12:02:18 +01001823tune.listener.multi-queue { on | off }
1824 Enables ('on') or disables ('off') the listener's multi-queue accept which
1825 spreads the incoming traffic to all threads a "bind" line is allowed to run
1826 on instead of taking them for itself. This provides a smoother traffic
1827 distribution and scales much better, especially in environments where threads
1828 may be unevenly loaded due to external activity (network interrupts colliding
1829 with one thread for example). This option is enabled by default, but it may
1830 be forcefully disabled for troubleshooting or for situations where it is
1831 estimated that the operating system already provides a good enough
1832 distribution and connections are extremely short-lived.
1833
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001834tune.lua.forced-yield <number>
1835 This directive forces the Lua engine to execute a yield each <number> of
Tim Düsterhus4896c442016-11-29 02:15:19 +01001836 instructions executed. This permits interrupting a long script and allows the
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001837 HAProxy scheduler to process other tasks like accepting connections or
1838 forwarding traffic. The default value is 10000 instructions. If HAProxy often
Davor Ocelice9ed2812017-12-25 17:49:28 +01001839 executes some Lua code but more responsiveness is required, this value can be
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001840 lowered. If the Lua code is quite long and its result is absolutely required
1841 to process the data, the <number> can be increased.
1842
Willy Tarreau32f61e22015-03-18 17:54:59 +01001843tune.lua.maxmem
1844 Sets the maximum amount of RAM in megabytes per process usable by Lua. By
1845 default it is zero which means unlimited. It is important to set a limit to
1846 ensure that a bug in a script will not result in the system running out of
1847 memory.
1848
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001849tune.lua.session-timeout <timeout>
1850 This is the execution timeout for the Lua sessions. This is useful for
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001851 preventing infinite loops or spending too much time in Lua. This timeout
1852 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001853 not taken in account. The default timeout is 4s.
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001854
1855tune.lua.task-timeout <timeout>
1856 Purpose is the same as "tune.lua.session-timeout", but this timeout is
1857 dedicated to the tasks. By default, this timeout isn't set because a task may
1858 remain alive during of the lifetime of HAProxy. For example, a task used to
1859 check servers.
1860
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001861tune.lua.service-timeout <timeout>
1862 This is the execution timeout for the Lua services. This is useful for
1863 preventing infinite loops or spending too much time in Lua. This timeout
1864 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001865 not taken in account. The default timeout is 4s.
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001866
Willy Tarreaua0250ba2008-01-06 11:22:57 +01001867tune.maxaccept <number>
Willy Tarreau16a21472012-11-19 12:39:59 +01001868 Sets the maximum number of consecutive connections a process may accept in a
1869 row before switching to other work. In single process mode, higher numbers
1870 give better performance at high connection rates. However in multi-process
1871 modes, keeping a bit of fairness between processes generally is better to
1872 increase performance. This value applies individually to each listener, so
1873 that the number of processes a listener is bound to is taken into account.
1874 This value defaults to 64. In multi-process mode, it is divided by twice
1875 the number of processes the listener is bound to. Setting this value to -1
1876 completely disables the limitation. It should normally not be needed to tweak
1877 this value.
Willy Tarreaua0250ba2008-01-06 11:22:57 +01001878
1879tune.maxpollevents <number>
1880 Sets the maximum amount of events that can be processed at once in a call to
1881 the polling system. The default value is adapted to the operating system. It
1882 has been noticed that reducing it below 200 tends to slightly decrease
1883 latency at the expense of network bandwidth, and increasing it above 200
1884 tends to trade latency for slightly increased bandwidth.
1885
Willy Tarreau27a674e2009-08-17 07:23:33 +02001886tune.maxrewrite <number>
1887 Sets the reserved buffer space to this size in bytes. The reserved space is
1888 used for header rewriting or appending. The first reads on sockets will never
1889 fill more than bufsize-maxrewrite. Historically it has defaulted to half of
1890 bufsize, though that does not make much sense since there are rarely large
1891 numbers of headers to add. Setting it too high prevents processing of large
1892 requests or responses. Setting it too low prevents addition of new headers
1893 to already large requests or to POST requests. It is generally wise to set it
1894 to about 1024. It is automatically readjusted to half of bufsize if it is
1895 larger than that. This means you don't have to worry about it when changing
1896 bufsize.
1897
Willy Tarreauf3045d22015-04-29 16:24:50 +02001898tune.pattern.cache-size <number>
1899 Sets the size of the pattern lookup cache to <number> entries. This is an LRU
1900 cache which reminds previous lookups and their results. It is used by ACLs
1901 and maps on slow pattern lookups, namely the ones using the "sub", "reg",
1902 "dir", "dom", "end", "bin" match methods as well as the case-insensitive
1903 strings. It applies to pattern expressions which means that it will be able
1904 to memorize the result of a lookup among all the patterns specified on a
1905 configuration line (including all those loaded from files). It automatically
1906 invalidates entries which are updated using HTTP actions or on the CLI. The
1907 default cache size is set to 10000 entries, which limits its footprint to
Willy Tarreau403bfbb2019-10-23 06:59:31 +02001908 about 5 MB per process/thread on 32-bit systems and 8 MB per process/thread
1909 on 64-bit systems, as caches are thread/process local. There is a very low
Willy Tarreauf3045d22015-04-29 16:24:50 +02001910 risk of collision in this cache, which is in the order of the size of the
1911 cache divided by 2^64. Typically, at 10000 requests per second with the
1912 default cache size of 10000 entries, there's 1% chance that a brute force
1913 attack could cause a single collision after 60 years, or 0.1% after 6 years.
1914 This is considered much lower than the risk of a memory corruption caused by
1915 aging components. If this is not acceptable, the cache can be disabled by
1916 setting this parameter to 0.
1917
Willy Tarreaubd9a0a72011-10-23 21:14:29 +02001918tune.pipesize <number>
1919 Sets the kernel pipe buffer size to this size (in bytes). By default, pipes
1920 are the default size for the system. But sometimes when using TCP splicing,
1921 it can improve performance to increase pipe sizes, especially if it is
1922 suspected that pipes are not filled and that many calls to splice() are
1923 performed. This has an impact on the kernel's memory footprint, so this must
1924 not be changed if impacts are not understood.
1925
Olivier Houchard88698d92019-04-16 19:07:22 +02001926tune.pool-low-fd-ratio <number>
1927 This setting sets the max number of file descriptors (in percentage) used by
1928 haproxy globally against the maximum number of file descriptors haproxy can
1929 use before we stop putting connection into the idle pool for reuse. The
1930 default is 20.
1931
1932tune.pool-high-fd-ratio <number>
1933 This setting sets the max number of file descriptors (in percentage) used by
1934 haproxy globally against the maximum number of file descriptors haproxy can
1935 use before we start killing idle connections when we can't reuse a connection
1936 and we have to create a new one. The default is 25 (one quarter of the file
1937 descriptor will mean that roughly half of the maximum front connections can
1938 keep an idle connection behind, anything beyond this probably doesn't make
John Roeslerfb2fce12019-07-10 15:45:51 -05001939 much sense in the general case when targeting connection reuse).
Olivier Houchard88698d92019-04-16 19:07:22 +02001940
Willy Tarreaue803de22010-01-21 17:43:04 +01001941tune.rcvbuf.client <number>
1942tune.rcvbuf.server <number>
1943 Forces the kernel socket receive buffer size on the client or the server side
1944 to the specified value in bytes. This value applies to all TCP/HTTP frontends
1945 and backends. It should normally never be set, and the default size (0) lets
John Roeslerfb2fce12019-07-10 15:45:51 -05001946 the kernel auto-tune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01001947 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01001948 order to save kernel memory by preventing it from buffering too large amounts
1949 of received data. Lower values will significantly increase CPU usage though.
1950
Willy Tarreaub22fc302015-12-14 12:04:35 +01001951tune.recv_enough <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001952 HAProxy uses some hints to detect that a short read indicates the end of the
Willy Tarreaub22fc302015-12-14 12:04:35 +01001953 socket buffers. One of them is that a read returns more than <recv_enough>
1954 bytes, which defaults to 10136 (7 segments of 1448 each). This default value
1955 may be changed by this setting to better deal with workloads involving lots
1956 of short messages such as telnet or SSH sessions.
1957
Olivier Houchard1599b802018-05-24 18:59:04 +02001958tune.runqueue-depth <number>
John Roeslerfb2fce12019-07-10 15:45:51 -05001959 Sets the maximum amount of task that can be processed at once when running
Olivier Houchard1599b802018-05-24 18:59:04 +02001960 tasks. The default value is 200. Increasing it may incur latency when
1961 dealing with I/Os, making it too small can incur extra overhead.
1962
Willy Tarreaue803de22010-01-21 17:43:04 +01001963tune.sndbuf.client <number>
1964tune.sndbuf.server <number>
1965 Forces the kernel socket send buffer size on the client or the server side to
1966 the specified value in bytes. This value applies to all TCP/HTTP frontends
1967 and backends. It should normally never be set, and the default size (0) lets
John Roeslerfb2fce12019-07-10 15:45:51 -05001968 the kernel auto-tune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01001969 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01001970 order to save kernel memory by preventing it from buffering too large amounts
1971 of received data. Lower values will significantly increase CPU usage though.
1972 Another use case is to prevent write timeouts with extremely slow clients due
1973 to the kernel waiting for a large part of the buffer to be read before
1974 notifying haproxy again.
1975
Willy Tarreau6ec58db2012-11-16 16:32:15 +01001976tune.ssl.cachesize <number>
Emeric Brunaf9619d2012-11-28 18:47:52 +01001977 Sets the size of the global SSL session cache, in a number of blocks. A block
1978 is large enough to contain an encoded session without peer certificate.
1979 An encoded session with peer certificate is stored in multiple blocks
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001980 depending on the size of the peer certificate. A block uses approximately
Emeric Brunaf9619d2012-11-28 18:47:52 +01001981 200 bytes of memory. The default value may be forced at build time, otherwise
Davor Ocelice9ed2812017-12-25 17:49:28 +01001982 defaults to 20000. When the cache is full, the most idle entries are purged
Emeric Brunaf9619d2012-11-28 18:47:52 +01001983 and reassigned. Higher values reduce the occurrence of such a purge, hence
1984 the number of CPU-intensive SSL handshakes by ensuring that all users keep
1985 their session as long as possible. All entries are pre-allocated upon startup
Emeric Brun22890a12012-12-28 14:41:32 +01001986 and are shared between all processes if "nbproc" is greater than 1. Setting
1987 this value to 0 disables the SSL session cache.
Willy Tarreau6ec58db2012-11-16 16:32:15 +01001988
Emeric Brun8dc60392014-05-09 13:52:00 +02001989tune.ssl.force-private-cache
Lukas Tribus27935782018-10-01 02:00:16 +02001990 This option disables SSL session cache sharing between all processes. It
Emeric Brun8dc60392014-05-09 13:52:00 +02001991 should normally not be used since it will force many renegotiations due to
1992 clients hitting a random process. But it may be required on some operating
1993 systems where none of the SSL cache synchronization method may be used. In
1994 this case, adding a first layer of hash-based load balancing before the SSL
1995 layer might limit the impact of the lack of session sharing.
1996
Emeric Brun4f65bff2012-11-16 15:11:00 +01001997tune.ssl.lifetime <timeout>
1998 Sets how long a cached SSL session may remain valid. This time is expressed
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001999 in seconds and defaults to 300 (5 min). It is important to understand that it
Emeric Brun4f65bff2012-11-16 15:11:00 +01002000 does not guarantee that sessions will last that long, because if the cache is
2001 full, the longest idle sessions will be purged despite their configured
2002 lifetime. The real usefulness of this setting is to prevent sessions from
2003 being used for too long.
2004
Willy Tarreaubfd59462013-02-21 07:46:09 +01002005tune.ssl.maxrecord <number>
2006 Sets the maximum amount of bytes passed to SSL_write() at a time. Default
2007 value 0 means there is no limit. Over SSL/TLS, the client can decipher the
2008 data only once it has received a full record. With large records, it means
2009 that clients might have to download up to 16kB of data before starting to
2010 process them. Limiting the value can improve page load times on browsers
2011 located over high latency or low bandwidth networks. It is suggested to find
2012 optimal values which fit into 1 or 2 TCP segments (generally 1448 bytes over
2013 Ethernet with TCP timestamps enabled, or 1460 when timestamps are disabled),
2014 keeping in mind that SSL/TLS add some overhead. Typical values of 1419 and
2015 2859 gave good results during tests. Use "strace -e trace=write" to find the
Davor Ocelice9ed2812017-12-25 17:49:28 +01002016 best value. HAProxy will automatically switch to this setting after an idle
Willy Tarreau7e312732014-02-12 16:35:14 +01002017 stream has been detected (see tune.idletimer above).
Willy Tarreaubfd59462013-02-21 07:46:09 +01002018
Remi Gacognef46cd6e2014-06-12 14:58:40 +02002019tune.ssl.default-dh-param <number>
2020 Sets the maximum size of the Diffie-Hellman parameters used for generating
2021 the ephemeral/temporary Diffie-Hellman key in case of DHE key exchange. The
2022 final size will try to match the size of the server's RSA (or DSA) key (e.g,
2023 a 2048 bits temporary DH key for a 2048 bits RSA key), but will not exceed
2024 this maximum value. Default value if 1024. Only 1024 or higher values are
2025 allowed. Higher values will increase the CPU load, and values greater than
2026 1024 bits are not supported by Java 7 and earlier clients. This value is not
Remi Gacogne47783ef2015-05-29 15:53:22 +02002027 used if static Diffie-Hellman parameters are supplied either directly
2028 in the certificate file or by using the ssl-dh-param-file parameter.
Remi Gacognef46cd6e2014-06-12 14:58:40 +02002029
Christopher Faulet31af49d2015-06-09 17:29:50 +02002030tune.ssl.ssl-ctx-cache-size <number>
2031 Sets the size of the cache used to store generated certificates to <number>
2032 entries. This is a LRU cache. Because generating a SSL certificate
2033 dynamically is expensive, they are cached. The default cache size is set to
2034 1000 entries.
2035
Thierry FOURNIER5bf77322017-02-25 12:45:22 +01002036tune.ssl.capture-cipherlist-size <number>
2037 Sets the maximum size of the buffer used for capturing client-hello cipher
2038 list. If the value is 0 (default value) the capture is disabled, otherwise
2039 a buffer is allocated for each SSL/TLS connection.
2040
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002041tune.vars.global-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01002042tune.vars.proc-max-size <size>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002043tune.vars.reqres-max-size <size>
2044tune.vars.sess-max-size <size>
2045tune.vars.txn-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01002046 These five tunes help to manage the maximum amount of memory used by the
2047 variables system. "global" limits the overall amount of memory available for
2048 all scopes. "proc" limits the memory for the process scope, "sess" limits the
2049 memory for the session scope, "txn" for the transaction scope, and "reqres"
2050 limits the memory for each request or response processing.
2051 Memory accounting is hierarchical, meaning more coarse grained limits include
2052 the finer grained ones: "proc" includes "sess", "sess" includes "txn", and
2053 "txn" includes "reqres".
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002054
Daniel Schneller0b547052016-03-21 20:46:57 +01002055 For example, when "tune.vars.sess-max-size" is limited to 100,
2056 "tune.vars.txn-max-size" and "tune.vars.reqres-max-size" cannot exceed
2057 100 either. If we create a variable "txn.var" that contains 100 bytes,
2058 all available space is consumed.
2059 Notice that exceeding the limits at runtime will not result in an error
2060 message, but values might be cut off or corrupted. So make sure to accurately
2061 plan for the amount of space needed to store all your variables.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002062
William Lallemanda509e4c2012-11-07 16:54:34 +01002063tune.zlib.memlevel <number>
2064 Sets the memLevel parameter in zlib initialization for each session. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002065 defines how much memory should be allocated for the internal compression
William Lallemanda509e4c2012-11-07 16:54:34 +01002066 state. A value of 1 uses minimum memory but is slow and reduces compression
Davor Ocelice9ed2812017-12-25 17:49:28 +01002067 ratio, a value of 9 uses maximum memory for optimal speed. Can be a value
William Lallemanda509e4c2012-11-07 16:54:34 +01002068 between 1 and 9. The default value is 8.
2069
2070tune.zlib.windowsize <number>
2071 Sets the window size (the size of the history buffer) as a parameter of the
2072 zlib initialization for each session. Larger values of this parameter result
Davor Ocelice9ed2812017-12-25 17:49:28 +01002073 in better compression at the expense of memory usage. Can be a value between
2074 8 and 15. The default value is 15.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002075
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020763.3. Debugging
2077--------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02002078
2079debug
2080 Enables debug mode which dumps to stdout all exchanges, and disables forking
2081 into background. It is the equivalent of the command-line argument "-d". It
2082 should never be used in a production configuration since it may prevent full
2083 system startup.
2084
2085quiet
2086 Do not display any message during startup. It is equivalent to the command-
2087 line argument "-q".
2088
Emeric Brunf099e792010-09-27 12:05:28 +02002089
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010020903.4. Userlists
2091--------------
2092It is possible to control access to frontend/backend/listen sections or to
2093http stats by allowing only authenticated and authorized users. To do this,
2094it is required to create at least one userlist and to define users.
2095
2096userlist <listname>
Cyril Bonté78caf842010-03-10 22:41:43 +01002097 Creates new userlist with name <listname>. Many independent userlists can be
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002098 used to store authentication & authorization data for independent customers.
2099
2100group <groupname> [users <user>,<user>,(...)]
Cyril Bonté78caf842010-03-10 22:41:43 +01002101 Adds group <groupname> to the current userlist. It is also possible to
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002102 attach users to this group by using a comma separated list of names
2103 proceeded by "users" keyword.
2104
Cyril Bontéf0c60612010-02-06 14:44:47 +01002105user <username> [password|insecure-password <password>]
2106 [groups <group>,<group>,(...)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002107 Adds user <username> to the current userlist. Both secure (encrypted) and
2108 insecure (unencrypted) passwords can be used. Encrypted passwords are
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01002109 evaluated using the crypt(3) function, so depending on the system's
2110 capabilities, different algorithms are supported. For example, modern Glibc
2111 based Linux systems support MD5, SHA-256, SHA-512, and, of course, the
2112 classic DES-based method of encrypting passwords.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002113
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01002114 Attention: Be aware that using encrypted passwords might cause significantly
2115 increased CPU usage, depending on the number of requests, and the algorithm
2116 used. For any of the hashed variants, the password for each request must
2117 be processed through the chosen algorithm, before it can be compared to the
2118 value specified in the config file. Most current algorithms are deliberately
2119 designed to be expensive to compute to achieve resistance against brute
2120 force attacks. They do not simply salt/hash the clear text password once,
2121 but thousands of times. This can quickly become a major factor in haproxy's
2122 overall CPU consumption!
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002123
2124 Example:
Cyril Bontéf0c60612010-02-06 14:44:47 +01002125 userlist L1
2126 group G1 users tiger,scott
2127 group G2 users xdb,scott
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002128
Cyril Bontéf0c60612010-02-06 14:44:47 +01002129 user tiger password $6$k6y3o.eP$JlKBx9za9667qe4(...)xHSwRv6J.C0/D7cV91
2130 user scott insecure-password elgato
2131 user xdb insecure-password hello
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002132
Cyril Bontéf0c60612010-02-06 14:44:47 +01002133 userlist L2
2134 group G1
2135 group G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002136
Cyril Bontéf0c60612010-02-06 14:44:47 +01002137 user tiger password $6$k6y3o.eP$JlKBx(...)xHSwRv6J.C0/D7cV91 groups G1
2138 user scott insecure-password elgato groups G1,G2
2139 user xdb insecure-password hello groups G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002140
2141 Please note that both lists are functionally identical.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002142
Emeric Brunf099e792010-09-27 12:05:28 +02002143
21443.5. Peers
Cyril Bontédc4d9032012-04-08 21:57:39 +02002145----------
Emeric Brun94900952015-06-11 18:25:54 +02002146It is possible to propagate entries of any data-types in stick-tables between
2147several haproxy instances over TCP connections in a multi-master fashion. Each
2148instance pushes its local updates and insertions to remote peers. The pushed
2149values overwrite remote ones without aggregation. Interrupted exchanges are
2150automatically detected and recovered from the last known point.
2151In addition, during a soft restart, the old process connects to the new one
2152using such a TCP connection to push all its entries before the new process
2153tries to connect to other peers. That ensures very fast replication during a
2154reload, it typically takes a fraction of a second even for large tables.
2155Note that Server IDs are used to identify servers remotely, so it is important
2156that configurations look similar or at least that the same IDs are forced on
2157each server on all participants.
Emeric Brunf099e792010-09-27 12:05:28 +02002158
2159peers <peersect>
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002160 Creates a new peer list with name <peersect>. It is an independent section,
Emeric Brunf099e792010-09-27 12:05:28 +02002161 which is referenced by one or more stick-tables.
2162
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002163bind [<address>]:<port_range> [, ...] [param*]
2164 Defines the binding parameters of the local peer of this "peers" section.
2165 Such lines are not supported with "peer" line in the same "peers" section.
2166
Willy Tarreau77e4bd12015-05-01 20:02:17 +02002167disabled
2168 Disables a peers section. It disables both listening and any synchronization
2169 related to this section. This is provided to disable synchronization of stick
2170 tables without having to comment out all "peers" references.
2171
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002172default-bind [param*]
2173 Defines the binding parameters for the local peer, excepted its address.
2174
2175default-server [param*]
2176 Change default options for a server in a "peers" section.
2177
2178 Arguments:
2179 <param*> is a list of parameters for this server. The "default-server"
2180 keyword accepts an important number of options and has a complete
2181 section dedicated to it. Please refer to section 5 for more
2182 details.
2183
2184
2185 See also: "server" and section 5 about server options
2186
Willy Tarreau77e4bd12015-05-01 20:02:17 +02002187enable
2188 This re-enables a disabled peers section which was previously disabled.
2189
Frédéric Lécailleb6f759b2019-11-05 09:57:45 +01002190log <address> [len <length>] [format <format>] [sample <ranges>:<smp_size>]
2191 <facility> [<level> [<minlevel>]]
2192 "peers" sections support the same "log" keyword as for the proxies to
2193 log information about the "peers" listener. See "log" option for proxies for
2194 more details.
2195
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002196peer <peername> <ip>:<port> [param*]
Emeric Brunf099e792010-09-27 12:05:28 +02002197 Defines a peer inside a peers section.
2198 If <peername> is set to the local peer name (by default hostname, or forced
2199 using "-L" command line option), haproxy will listen for incoming remote peer
2200 connection on <ip>:<port>. Otherwise, <ip>:<port> defines where to connect to
2201 to join the remote peer, and <peername> is used at the protocol level to
2202 identify and validate the remote peer on the server side.
2203
2204 During a soft restart, local peer <ip>:<port> is used by the old instance to
2205 connect the new one and initiate a complete replication (teaching process).
2206
2207 It is strongly recommended to have the exact same peers declaration on all
2208 peers and to only rely on the "-L" command line argument to change the local
2209 peer name. This makes it easier to maintain coherent configuration files
2210 across all peers.
2211
William Lallemandb2f07452015-05-12 14:27:13 +02002212 You may want to reference some environment variables in the address
2213 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01002214
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002215 Note: "peer" keyword may transparently be replaced by "server" keyword (see
2216 "server" keyword explanation below).
2217
2218server <peername> [<ip>:<port>] [param*]
Michael Prokop4438c602019-05-24 10:25:45 +02002219 As previously mentioned, "peer" keyword may be replaced by "server" keyword
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002220 with a support for all "server" parameters found in 5.2 paragraph.
2221 If the underlying peer is local, <ip>:<port> parameters must not be present.
2222 These parameters must be provided on a "bind" line (see "bind" keyword
2223 of this "peers" section).
2224 Some of these parameters are irrelevant for "peers" sections.
2225
2226
Cyril Bontédc4d9032012-04-08 21:57:39 +02002227 Example:
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002228 # The old way.
Emeric Brunf099e792010-09-27 12:05:28 +02002229 peers mypeers
Willy Tarreauf7b30a92010-12-06 22:59:17 +01002230 peer haproxy1 192.168.0.1:1024
2231 peer haproxy2 192.168.0.2:1024
2232 peer haproxy3 10.2.0.1:1024
Emeric Brunf099e792010-09-27 12:05:28 +02002233
2234 backend mybackend
2235 mode tcp
2236 balance roundrobin
2237 stick-table type ip size 20k peers mypeers
2238 stick on src
2239
Willy Tarreauf7b30a92010-12-06 22:59:17 +01002240 server srv1 192.168.0.30:80
2241 server srv2 192.168.0.31:80
Emeric Brunf099e792010-09-27 12:05:28 +02002242
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002243 Example:
2244 peers mypeers
2245 bind 127.0.0.11:10001 ssl crt mycerts/pem
2246 default-server ssl verify none
2247 server hostA 127.0.0.10:10000
2248 server hostB #local peer
Emeric Brunf099e792010-09-27 12:05:28 +02002249
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01002250
2251table <tablename> type {ip | integer | string [len <length>] | binary [len <length>]}
2252 size <size> [expire <expire>] [nopurge] [store <data_type>]*
2253
2254 Configure a stickiness table for the current section. This line is parsed
2255 exactly the same way as the "stick-table" keyword in others section, except
John Roeslerfb2fce12019-07-10 15:45:51 -05002256 for the "peers" argument which is not required here and with an additional
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01002257 mandatory first parameter to designate the stick-table. Contrary to others
2258 sections, there may be several "table" lines in "peers" sections (see also
2259 "stick-table" keyword).
2260
2261 Also be aware of the fact that "peers" sections have their own stick-table
2262 namespaces to avoid collisions between stick-table names identical in
2263 different "peers" section. This is internally handled prepending the "peers"
2264 sections names to the name of the stick-tables followed by a '/' character.
2265 If somewhere else in the configuration file you have to refer to such
2266 stick-tables declared in "peers" sections you must use the prefixed version
2267 of the stick-table name as follows:
2268
2269 peers mypeers
2270 peer A ...
2271 peer B ...
2272 table t1 ...
2273
2274 frontend fe1
2275 tcp-request content track-sc0 src table mypeers/t1
2276
2277 This is also this prefixed version of the stick-table names which must be
2278 used to refer to stick-tables through the CLI.
2279
2280 About "peers" protocol, as only "peers" belonging to the same section may
2281 communicate with each others, there is no need to do such a distinction.
2282 Several "peers" sections may declare stick-tables with the same name.
2283 This is shorter version of the stick-table name which is sent over the network.
2284 There is only a '/' character as prefix to avoid stick-table name collisions between
2285 stick-tables declared as backends and stick-table declared in "peers" sections
2286 as follows in this weird but supported configuration:
2287
2288 peers mypeers
2289 peer A ...
2290 peer B ...
2291 table t1 type string size 10m store gpc0
2292
2293 backend t1
2294 stick-table type string size 10m store gpc0 peers mypeers
2295
2296 Here "t1" table declared in "mypeeers" section has "mypeers/t1" as global name.
2297 "t1" table declared as a backend as "t1" as global name. But at peer protocol
2298 level the former table is named "/t1", the latter is again named "t1".
2299
Simon Horman51a1cf62015-02-03 13:00:44 +090023003.6. Mailers
2301------------
2302It is possible to send email alerts when the state of servers changes.
2303If configured email alerts are sent to each mailer that is configured
2304in a mailers section. Email is sent to mailers using SMTP.
2305
Pieter Baauw386a1272015-08-16 15:26:24 +02002306mailers <mailersect>
Simon Horman51a1cf62015-02-03 13:00:44 +09002307 Creates a new mailer list with the name <mailersect>. It is an
2308 independent section which is referenced by one or more proxies.
2309
2310mailer <mailername> <ip>:<port>
2311 Defines a mailer inside a mailers section.
2312
2313 Example:
2314 mailers mymailers
2315 mailer smtp1 192.168.0.1:587
2316 mailer smtp2 192.168.0.2:587
2317
2318 backend mybackend
2319 mode tcp
2320 balance roundrobin
2321
2322 email-alert mailers mymailers
2323 email-alert from test1@horms.org
2324 email-alert to test2@horms.org
2325
2326 server srv1 192.168.0.30:80
2327 server srv2 192.168.0.31:80
2328
Pieter Baauw235fcfc2016-02-13 15:33:40 +01002329timeout mail <time>
2330 Defines the time available for a mail/connection to be made and send to
2331 the mail-server. If not defined the default value is 10 seconds. To allow
2332 for at least two SYN-ACK packets to be send during initial TCP handshake it
2333 is advised to keep this value above 4 seconds.
2334
2335 Example:
2336 mailers mymailers
2337 timeout mail 20s
2338 mailer smtp1 192.168.0.1:587
Simon Horman51a1cf62015-02-03 13:00:44 +09002339
William Lallemandc9515522019-06-12 16:32:11 +020023403.7. Programs
2341-------------
2342In master-worker mode, it is possible to launch external binaries with the
2343master, these processes are called programs. These programs are launched and
2344managed the same way as the workers.
2345
2346During a reload of HAProxy, those processes are dealing with the same
2347sequence as a worker:
2348
2349 - the master is re-executed
2350 - the master sends a SIGUSR1 signal to the program
2351 - if "option start-on-reload" is not disabled, the master launches a new
2352 instance of the program
2353
2354During a stop, or restart, a SIGTERM is sent to the programs.
2355
2356program <name>
2357 This is a new program section, this section will create an instance <name>
2358 which is visible in "show proc" on the master CLI. (See "9.4. Master CLI" in
2359 the management guide).
2360
2361command <command> [arguments*]
2362 Define the command to start with optional arguments. The command is looked
2363 up in the current PATH if it does not include an absolute path. This is a
2364 mandatory option of the program section. Arguments containing spaces must
2365 be enclosed in quotes or double quotes or be prefixed by a backslash.
2366
Andrew Heberle97236962019-07-12 11:50:26 +08002367user <user name>
2368 Changes the executed command user ID to the <user name> from /etc/passwd.
2369 See also "group".
2370
2371group <group name>
2372 Changes the executed command group ID to the <group name> from /etc/group.
2373 See also "user".
2374
William Lallemandc9515522019-06-12 16:32:11 +02002375option start-on-reload
2376no option start-on-reload
2377 Start (or not) a new instance of the program upon a reload of the master.
2378 The default is to start a new instance. This option may only be used in a
2379 program section.
2380
2381
Christopher Faulet76edc0f2020-01-13 15:52:01 +010023823.8. HTTP-errors
2383----------------
2384
2385It is possible to globally declare several groups of HTTP errors, to be
2386imported afterwards in any proxy section. Same group may be referenced at
2387several places and can be fully or partially imported.
2388
2389http-errors <name>
2390 Create a new http-errors group with the name <name>. It is an independent
2391 section that may be referenced by one or more proxies using its name.
2392
2393errorfile <code> <file>
2394 Associate a file contents to an HTTP error code
2395
2396 Arguments :
2397 <code> is the HTTP status code. Currently, HAProxy is capable of
2398 generating codes 200, 400, 403, 404, 405, 408, 410, 425, 429,
2399 500, 502, 503, and 504.
2400
2401 <file> designates a file containing the full HTTP response. It is
2402 recommended to follow the common practice of appending ".http" to
2403 the filename so that people do not confuse the response with HTML
2404 error pages, and to use absolute paths, since files are read
2405 before any chroot is performed.
2406
2407 Please referrers to "errorfile" keyword in section 4 for details.
2408
2409 Example:
2410 http-errors website-1
2411 errorfile 400 /etc/haproxy/errorfiles/site1/400.http
2412 errorfile 404 /etc/haproxy/errorfiles/site1/404.http
2413 errorfile 408 /dev/null # work around Chrome pre-connect bug
2414
2415 http-errors website-2
2416 errorfile 400 /etc/haproxy/errorfiles/site2/400.http
2417 errorfile 404 /etc/haproxy/errorfiles/site2/404.http
2418 errorfile 408 /dev/null # work around Chrome pre-connect bug
2419
2420
Willy Tarreauc57f0e22009-05-10 13:12:33 +020024214. Proxies
Willy Tarreau6a06a402007-07-15 20:15:28 +02002422----------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002423
Willy Tarreau6a06a402007-07-15 20:15:28 +02002424Proxy configuration can be located in a set of sections :
William Lallemand6e62fb62015-04-28 16:55:23 +02002425 - defaults [<name>]
Willy Tarreau6a06a402007-07-15 20:15:28 +02002426 - frontend <name>
2427 - backend <name>
2428 - listen <name>
2429
2430A "defaults" section sets default parameters for all other sections following
2431its declaration. Those default parameters are reset by the next "defaults"
2432section. See below for the list of parameters which can be set in a "defaults"
Willy Tarreau0ba27502007-12-24 16:55:16 +01002433section. The name is optional but its use is encouraged for better readability.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002434
2435A "frontend" section describes a set of listening sockets accepting client
2436connections.
2437
2438A "backend" section describes a set of servers to which the proxy will connect
2439to forward incoming connections.
2440
2441A "listen" section defines a complete proxy with its frontend and backend
2442parts combined in one section. It is generally useful for TCP-only traffic.
2443
Willy Tarreau0ba27502007-12-24 16:55:16 +01002444All proxy names must be formed from upper and lower case letters, digits,
2445'-' (dash), '_' (underscore) , '.' (dot) and ':' (colon). ACL names are
2446case-sensitive, which means that "www" and "WWW" are two different proxies.
2447
2448Historically, all proxy names could overlap, it just caused troubles in the
2449logs. Since the introduction of content switching, it is mandatory that two
2450proxies with overlapping capabilities (frontend/backend) have different names.
2451However, it is still permitted that a frontend and a backend share the same
2452name, as this configuration seems to be commonly encountered.
2453
2454Right now, two major proxy modes are supported : "tcp", also known as layer 4,
2455and "http", also known as layer 7. In layer 4 mode, HAProxy simply forwards
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002456bidirectional traffic between two sides. In layer 7 mode, HAProxy analyzes the
Willy Tarreau0ba27502007-12-24 16:55:16 +01002457protocol, and can interact with it by allowing, blocking, switching, adding,
2458modifying, or removing arbitrary contents in requests or responses, based on
2459arbitrary criteria.
2460
Willy Tarreau70dffda2014-01-30 03:07:23 +01002461In HTTP mode, the processing applied to requests and responses flowing over
2462a connection depends in the combination of the frontend's HTTP options and
Julien Pivotto21ad3152019-12-10 13:11:17 +01002463the backend's. HAProxy supports 3 connection modes :
Willy Tarreau70dffda2014-01-30 03:07:23 +01002464
2465 - KAL : keep alive ("option http-keep-alive") which is the default mode : all
2466 requests and responses are processed, and connections remain open but idle
2467 between responses and new requests.
2468
Willy Tarreau70dffda2014-01-30 03:07:23 +01002469 - SCL: server close ("option http-server-close") : the server-facing
2470 connection is closed after the end of the response is received, but the
2471 client-facing connection remains open.
2472
Christopher Faulet315b39c2018-09-21 16:26:19 +02002473 - CLO: close ("option httpclose"): the connection is closed after the end of
2474 the response and "Connection: close" appended in both directions.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002475
2476The effective mode that will be applied to a connection passing through a
2477frontend and a backend can be determined by both proxy modes according to the
2478following matrix, but in short, the modes are symmetric, keep-alive is the
Christopher Faulet315b39c2018-09-21 16:26:19 +02002479weakest option and close is the strongest.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002480
Christopher Faulet315b39c2018-09-21 16:26:19 +02002481 Backend mode
Willy Tarreau70dffda2014-01-30 03:07:23 +01002482
Christopher Faulet315b39c2018-09-21 16:26:19 +02002483 | KAL | SCL | CLO
2484 ----+-----+-----+----
2485 KAL | KAL | SCL | CLO
2486 ----+-----+-----+----
Christopher Faulet315b39c2018-09-21 16:26:19 +02002487 mode SCL | SCL | SCL | CLO
2488 ----+-----+-----+----
2489 CLO | CLO | CLO | CLO
Willy Tarreau70dffda2014-01-30 03:07:23 +01002490
Willy Tarreau0ba27502007-12-24 16:55:16 +01002491
Willy Tarreau70dffda2014-01-30 03:07:23 +01002492
Willy Tarreauc57f0e22009-05-10 13:12:33 +020024934.1. Proxy keywords matrix
2494--------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002495
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002496The following list of keywords is supported. Most of them may only be used in a
2497limited set of section types. Some of them are marked as "deprecated" because
2498they are inherited from an old syntax which may be confusing or functionally
2499limited, and there are new recommended keywords to replace them. Keywords
Davor Ocelice9ed2812017-12-25 17:49:28 +01002500marked with "(*)" can be optionally inverted using the "no" prefix, e.g. "no
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002501option contstats". This makes sense when the option has been enabled by default
Willy Tarreau3842f002009-06-14 11:39:52 +02002502and must be disabled for a specific instance. Such options may also be prefixed
2503with "default" in order to restore default settings regardless of what has been
2504specified in a previous "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002505
Willy Tarreau6a06a402007-07-15 20:15:28 +02002506
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002507 keyword defaults frontend listen backend
2508------------------------------------+----------+----------+---------+---------
2509acl - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002510backlog X X X -
2511balance X - X X
2512bind - X X -
2513bind-process X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002514capture cookie - X X -
2515capture request header - X X -
2516capture response header - X X -
William Lallemand82fe75c2012-10-23 10:25:10 +02002517compression X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002518cookie X - X X
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02002519declare capture - X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002520default-server X - X X
2521default_backend X X X -
2522description - X X X
2523disabled X X X X
2524dispatch - - X X
Simon Horman51a1cf62015-02-03 13:00:44 +09002525email-alert from X X X X
Simon Horman64e34162015-02-06 11:11:57 +09002526email-alert level X X X X
Simon Horman51a1cf62015-02-03 13:00:44 +09002527email-alert mailers X X X X
2528email-alert myhostname X X X X
2529email-alert to X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002530enabled X X X X
2531errorfile X X X X
Christopher Faulet76edc0f2020-01-13 15:52:01 +01002532errorfiles X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002533errorloc X X X X
2534errorloc302 X X X X
2535-- keyword -------------------------- defaults - frontend - listen -- backend -
2536errorloc303 X X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01002537force-persist - - X X
Christopher Fauletc3fe5332016-04-07 15:30:10 +02002538filter - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002539fullconn X - X X
2540grace X X X X
2541hash-type X - X X
2542http-check disable-on-404 X - X X
Willy Tarreaubd741542010-03-16 18:46:54 +01002543http-check expect - - X X
Willy Tarreau7ab6aff2010-10-12 06:30:16 +02002544http-check send-state X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002545http-request - X X X
Willy Tarreaue365c0b2013-06-11 16:06:12 +02002546http-response - X X X
Willy Tarreau30631952015-08-06 15:05:24 +02002547http-reuse X - X X
Baptiste Assmann2c42ef52013-10-09 21:57:02 +02002548http-send-name-header - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002549id - X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01002550ignore-persist - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02002551load-server-state-from-file X - X X
William Lallemand0f99e342011-10-12 17:50:54 +02002552log (*) X X X X
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01002553log-format X X X -
Dragan Dosen7ad31542015-09-28 17:16:47 +02002554log-format-sd X X X -
Willy Tarreau094af4e2015-01-07 15:03:42 +01002555log-tag X X X X
Willy Tarreauc35362a2014-04-25 13:58:37 +02002556max-keep-alive-queue X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002557maxconn X X X -
2558mode X X X X
2559monitor fail - X X -
2560monitor-net X X X -
2561monitor-uri X X X -
2562option abortonclose (*) X - X X
2563option accept-invalid-http-request (*) X X X -
2564option accept-invalid-http-response (*) X - X X
2565option allbackups (*) X - X X
2566option checkcache (*) X - X X
2567option clitcpka (*) X X X -
2568option contstats (*) X X X -
2569option dontlog-normal (*) X X X -
2570option dontlognull (*) X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002571-- keyword -------------------------- defaults - frontend - listen -- backend -
2572option forwardfor X X X X
Christopher Faulet98fbe952019-07-22 16:18:24 +02002573option h1-case-adjust-bogus-client (*) X X X -
2574option h1-case-adjust-bogus-server (*) X - X X
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02002575option http-buffer-request (*) X X X X
Willy Tarreau82649f92015-05-01 22:40:51 +02002576option http-ignore-probes (*) X X X -
Willy Tarreau16bfb022010-01-16 19:48:41 +01002577option http-keep-alive (*) X X X X
Willy Tarreau96e31212011-05-30 18:10:30 +02002578option http-no-delay (*) X X X X
Christopher Faulet98db9762018-09-21 10:25:19 +02002579option http-pretend-keepalive (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002580option http-server-close (*) X X X X
2581option http-use-proxy-header (*) X X X -
2582option httpchk X - X X
2583option httpclose (*) X X X X
Freddy Spierenburge88b7732019-03-25 14:35:17 +01002584option httplog X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002585option http_proxy (*) X X X X
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002586option independent-streams (*) X X X X
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02002587option ldap-check X - X X
Simon Horman98637e52014-06-20 12:30:16 +09002588option external-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002589option log-health-checks (*) X - X X
2590option log-separate-errors (*) X X X -
2591option logasap (*) X X X -
2592option mysql-check X - X X
2593option nolinger (*) X X X X
2594option originalto X X X X
2595option persist (*) X - X X
Baptiste Assmann809e22a2015-10-12 20:22:55 +02002596option pgsql-check X - X X
2597option prefer-last-server (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002598option redispatch (*) X - X X
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02002599option redis-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002600option smtpchk X - X X
2601option socket-stats (*) X X X -
2602option splice-auto (*) X X X X
2603option splice-request (*) X X X X
2604option splice-response (*) X X X X
Christopher Fauletba7bc162016-11-07 21:07:38 +01002605option spop-check - - - X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002606option srvtcpka (*) X - X X
2607option ssl-hello-chk X - X X
2608-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreaued179852013-12-16 01:07:00 +01002609option tcp-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002610option tcp-smart-accept (*) X X X -
2611option tcp-smart-connect (*) X - X X
2612option tcpka X X X X
2613option tcplog X X X X
2614option transparent (*) X - X X
Simon Horman98637e52014-06-20 12:30:16 +09002615external-check command X - X X
2616external-check path X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002617persist rdp-cookie X - X X
2618rate-limit sessions X X X -
2619redirect - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002620-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002621retries X - X X
Olivier Houcharda254a372019-04-05 15:30:12 +02002622retry-on X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002623server - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02002624server-state-file-name X - X X
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02002625server-template - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002626source X - X X
Baptiste Assmann5a549212015-10-12 20:30:24 +02002627stats admin - X X X
2628stats auth X X X X
2629stats enable X X X X
2630stats hide-version X X X X
2631stats http-request - X X X
2632stats realm X X X X
2633stats refresh X X X X
2634stats scope X X X X
2635stats show-desc X X X X
2636stats show-legends X X X X
2637stats show-node X X X X
2638stats uri X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002639-- keyword -------------------------- defaults - frontend - listen -- backend -
2640stick match - - X X
2641stick on - - X X
2642stick store-request - - X X
Willy Tarreaud8dc99f2011-07-01 11:33:25 +02002643stick store-response - - X X
Adam Spiers68af3c12017-04-06 16:31:39 +01002644stick-table - X X X
Willy Tarreau938c7fe2014-04-25 14:21:39 +02002645tcp-check connect - - X X
2646tcp-check expect - - X X
2647tcp-check send - - X X
2648tcp-check send-binary - - X X
Willy Tarreaue9656522010-08-17 15:40:09 +02002649tcp-request connection - X X -
2650tcp-request content - X X X
Willy Tarreaua56235c2010-09-14 11:31:36 +02002651tcp-request inspect-delay - X X X
Willy Tarreau4f614292016-10-21 17:49:36 +02002652tcp-request session - X X -
Emeric Brun0a3b67f2010-09-24 15:34:53 +02002653tcp-response content - - X X
2654tcp-response inspect-delay - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002655timeout check X - X X
2656timeout client X X X -
Willy Tarreau05cdd962014-05-10 14:30:07 +02002657timeout client-fin X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002658timeout connect X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002659timeout http-keep-alive X X X X
2660timeout http-request X X X X
2661timeout queue X - X X
2662timeout server X - X X
Willy Tarreau05cdd962014-05-10 14:30:07 +02002663timeout server-fin X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002664timeout tarpit X X X X
Willy Tarreauce887fd2012-05-12 12:50:00 +02002665timeout tunnel X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002666transparent (deprecated) X - X X
William Lallemanda73203e2012-03-12 12:48:57 +01002667unique-id-format X X X -
2668unique-id-header X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002669use_backend - X X -
Christopher Fauletb30b3102019-09-12 23:03:09 +02002670use-fcgi-app - - X X
Willy Tarreau4a5cade2012-04-05 21:09:48 +02002671use-server - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002672------------------------------------+----------+----------+---------+---------
2673 keyword defaults frontend listen backend
Willy Tarreau6a06a402007-07-15 20:15:28 +02002674
Willy Tarreau0ba27502007-12-24 16:55:16 +01002675
Willy Tarreauc57f0e22009-05-10 13:12:33 +020026764.2. Alphabetically sorted keywords reference
2677---------------------------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002678
2679This section provides a description of each keyword and its usage.
2680
2681
2682acl <aclname> <criterion> [flags] [operator] <value> ...
2683 Declare or complete an access list.
2684 May be used in sections : defaults | frontend | listen | backend
2685 no | yes | yes | yes
2686 Example:
2687 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
2688 acl invalid_src src_port 0:1023
2689 acl local_dst hdr(host) -i localhost
2690
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002691 See section 7 about ACL usage.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002692
2693
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01002694backlog <conns>
2695 Give hints to the system about the approximate listen backlog desired size
2696 May be used in sections : defaults | frontend | listen | backend
2697 yes | yes | yes | no
2698 Arguments :
2699 <conns> is the number of pending connections. Depending on the operating
2700 system, it may represent the number of already acknowledged
Cyril Bontédc4d9032012-04-08 21:57:39 +02002701 connections, of non-acknowledged ones, or both.
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01002702
2703 In order to protect against SYN flood attacks, one solution is to increase
2704 the system's SYN backlog size. Depending on the system, sometimes it is just
2705 tunable via a system parameter, sometimes it is not adjustable at all, and
2706 sometimes the system relies on hints given by the application at the time of
2707 the listen() syscall. By default, HAProxy passes the frontend's maxconn value
2708 to the listen() syscall. On systems which can make use of this value, it can
2709 sometimes be useful to be able to specify a different value, hence this
2710 backlog parameter.
2711
2712 On Linux 2.4, the parameter is ignored by the system. On Linux 2.6, it is
2713 used as a hint and the system accepts up to the smallest greater power of
2714 two, and never more than some limits (usually 32768).
2715
2716 See also : "maxconn" and the target operating system's tuning guide.
2717
2718
Willy Tarreau0ba27502007-12-24 16:55:16 +01002719balance <algorithm> [ <arguments> ]
Willy Tarreau226071e2014-04-10 11:55:45 +02002720balance url_param <param> [check_post]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002721 Define the load balancing algorithm to be used in a backend.
2722 May be used in sections : defaults | frontend | listen | backend
2723 yes | no | yes | yes
2724 Arguments :
2725 <algorithm> is the algorithm used to select a server when doing load
2726 balancing. This only applies when no persistence information
2727 is available, or when a connection is redispatched to another
2728 server. <algorithm> may be one of the following :
2729
2730 roundrobin Each server is used in turns, according to their weights.
2731 This is the smoothest and fairest algorithm when the server's
2732 processing time remains equally distributed. This algorithm
2733 is dynamic, which means that server weights may be adjusted
Willy Tarreau9757a382009-10-03 12:56:50 +02002734 on the fly for slow starts for instance. It is limited by
Godbacha34bdc02013-07-22 07:44:53 +08002735 design to 4095 active servers per backend. Note that in some
Willy Tarreau9757a382009-10-03 12:56:50 +02002736 large farms, when a server becomes up after having been down
2737 for a very short time, it may sometimes take a few hundreds
2738 requests for it to be re-integrated into the farm and start
2739 receiving traffic. This is normal, though very rare. It is
2740 indicated here in case you would have the chance to observe
2741 it, so that you don't worry.
2742
2743 static-rr Each server is used in turns, according to their weights.
2744 This algorithm is as similar to roundrobin except that it is
2745 static, which means that changing a server's weight on the
2746 fly will have no effect. On the other hand, it has no design
2747 limitation on the number of servers, and when a server goes
2748 up, it is always immediately reintroduced into the farm, once
2749 the full map is recomputed. It also uses slightly less CPU to
2750 run (around -1%).
Willy Tarreau0ba27502007-12-24 16:55:16 +01002751
Willy Tarreau2d2a7f82008-03-17 12:07:56 +01002752 leastconn The server with the lowest number of connections receives the
2753 connection. Round-robin is performed within groups of servers
2754 of the same load to ensure that all servers will be used. Use
2755 of this algorithm is recommended where very long sessions are
2756 expected, such as LDAP, SQL, TSE, etc... but is not very well
2757 suited for protocols using short sessions such as HTTP. This
2758 algorithm is dynamic, which means that server weights may be
2759 adjusted on the fly for slow starts for instance.
2760
Willy Tarreauf09c6602012-02-13 17:12:08 +01002761 first The first server with available connection slots receives the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002762 connection. The servers are chosen from the lowest numeric
Willy Tarreauf09c6602012-02-13 17:12:08 +01002763 identifier to the highest (see server parameter "id"), which
2764 defaults to the server's position in the farm. Once a server
Willy Tarreau64559c52012-04-07 09:08:45 +02002765 reaches its maxconn value, the next server is used. It does
Willy Tarreauf09c6602012-02-13 17:12:08 +01002766 not make sense to use this algorithm without setting maxconn.
2767 The purpose of this algorithm is to always use the smallest
2768 number of servers so that extra servers can be powered off
2769 during non-intensive hours. This algorithm ignores the server
2770 weight, and brings more benefit to long session such as RDP
Willy Tarreau64559c52012-04-07 09:08:45 +02002771 or IMAP than HTTP, though it can be useful there too. In
2772 order to use this algorithm efficiently, it is recommended
2773 that a cloud controller regularly checks server usage to turn
2774 them off when unused, and regularly checks backend queue to
2775 turn new servers on when the queue inflates. Alternatively,
2776 using "http-check send-state" may inform servers on the load.
Willy Tarreauf09c6602012-02-13 17:12:08 +01002777
Willy Tarreau0ba27502007-12-24 16:55:16 +01002778 source The source IP address is hashed and divided by the total
2779 weight of the running servers to designate which server will
2780 receive the request. This ensures that the same client IP
2781 address will always reach the same server as long as no
2782 server goes down or up. If the hash result changes due to the
2783 number of running servers changing, many clients will be
2784 directed to a different server. This algorithm is generally
2785 used in TCP mode where no cookie may be inserted. It may also
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002786 be used on the Internet to provide a best-effort stickiness
Willy Tarreau0ba27502007-12-24 16:55:16 +01002787 to clients which refuse session cookies. This algorithm is
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002788 static by default, which means that changing a server's
2789 weight on the fly will have no effect, but this can be
2790 changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002791
Oskar Stolc8dc41842012-05-19 10:19:54 +01002792 uri This algorithm hashes either the left part of the URI (before
2793 the question mark) or the whole URI (if the "whole" parameter
2794 is present) and divides the hash value by the total weight of
2795 the running servers. The result designates which server will
2796 receive the request. This ensures that the same URI will
2797 always be directed to the same server as long as no server
2798 goes up or down. This is used with proxy caches and
2799 anti-virus proxies in order to maximize the cache hit rate.
2800 Note that this algorithm may only be used in an HTTP backend.
2801 This algorithm is static by default, which means that
2802 changing a server's weight on the fly will have no effect,
2803 but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002804
Oskar Stolc8dc41842012-05-19 10:19:54 +01002805 This algorithm supports two optional parameters "len" and
Marek Majkowski9c30fc12008-04-27 23:25:55 +02002806 "depth", both followed by a positive integer number. These
2807 options may be helpful when it is needed to balance servers
2808 based on the beginning of the URI only. The "len" parameter
2809 indicates that the algorithm should only consider that many
2810 characters at the beginning of the URI to compute the hash.
2811 Note that having "len" set to 1 rarely makes sense since most
2812 URIs start with a leading "/".
2813
2814 The "depth" parameter indicates the maximum directory depth
2815 to be used to compute the hash. One level is counted for each
2816 slash in the request. If both parameters are specified, the
2817 evaluation stops when either is reached.
2818
Willy Tarreau0ba27502007-12-24 16:55:16 +01002819 url_param The URL parameter specified in argument will be looked up in
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002820 the query string of each HTTP GET request.
2821
2822 If the modifier "check_post" is used, then an HTTP POST
Cyril Bontédc4d9032012-04-08 21:57:39 +02002823 request entity will be searched for the parameter argument,
2824 when it is not found in a query string after a question mark
Willy Tarreau226071e2014-04-10 11:55:45 +02002825 ('?') in the URL. The message body will only start to be
2826 analyzed once either the advertised amount of data has been
2827 received or the request buffer is full. In the unlikely event
2828 that chunked encoding is used, only the first chunk is
Cyril Bontédc4d9032012-04-08 21:57:39 +02002829 scanned. Parameter values separated by a chunk boundary, may
Willy Tarreau226071e2014-04-10 11:55:45 +02002830 be randomly balanced if at all. This keyword used to support
2831 an optional <max_wait> parameter which is now ignored.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002832
2833 If the parameter is found followed by an equal sign ('=') and
2834 a value, then the value is hashed and divided by the total
2835 weight of the running servers. The result designates which
2836 server will receive the request.
2837
2838 This is used to track user identifiers in requests and ensure
2839 that a same user ID will always be sent to the same server as
2840 long as no server goes up or down. If no value is found or if
2841 the parameter is not found, then a round robin algorithm is
2842 applied. Note that this algorithm may only be used in an HTTP
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002843 backend. This algorithm is static by default, which means
2844 that changing a server's weight on the fly will have no
2845 effect, but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002846
Cyril Bontédc4d9032012-04-08 21:57:39 +02002847 hdr(<name>) The HTTP header <name> will be looked up in each HTTP
2848 request. Just as with the equivalent ACL 'hdr()' function,
2849 the header name in parenthesis is not case sensitive. If the
2850 header is absent or if it does not contain any value, the
2851 roundrobin algorithm is applied instead.
Benoitaffb4812009-03-25 13:02:10 +01002852
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002853 An optional 'use_domain_only' parameter is available, for
Benoitaffb4812009-03-25 13:02:10 +01002854 reducing the hash algorithm to the main domain part with some
2855 specific headers such as 'Host'. For instance, in the Host
2856 value "haproxy.1wt.eu", only "1wt" will be considered.
2857
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002858 This algorithm is static by default, which means that
2859 changing a server's weight on the fly will have no effect,
2860 but this can be changed using "hash-type".
2861
Willy Tarreau21c741a2019-01-14 18:14:27 +01002862 random
2863 random(<draws>)
2864 A random number will be used as the key for the consistent
Willy Tarreau760e81d2018-05-03 07:20:40 +02002865 hashing function. This means that the servers' weights are
2866 respected, dynamic weight changes immediately take effect, as
2867 well as new server additions. Random load balancing can be
2868 useful with large farms or when servers are frequently added
Willy Tarreau21c741a2019-01-14 18:14:27 +01002869 or removed as it may avoid the hammering effect that could
2870 result from roundrobin or leastconn in this situation. The
2871 hash-balance-factor directive can be used to further improve
2872 fairness of the load balancing, especially in situations
2873 where servers show highly variable response times. When an
2874 argument <draws> is present, it must be an integer value one
2875 or greater, indicating the number of draws before selecting
2876 the least loaded of these servers. It was indeed demonstrated
2877 that picking the least loaded of two servers is enough to
2878 significantly improve the fairness of the algorithm, by
2879 always avoiding to pick the most loaded server within a farm
2880 and getting rid of any bias that could be induced by the
2881 unfair distribution of the consistent list. Higher values N
2882 will take away N-1 of the highest loaded servers at the
2883 expense of performance. With very high values, the algorithm
2884 will converge towards the leastconn's result but much slower.
2885 The default value is 2, which generally shows very good
2886 distribution and performance. This algorithm is also known as
2887 the Power of Two Random Choices and is described here :
2888 http://www.eecs.harvard.edu/~michaelm/postscripts/handbook2001.pdf
Willy Tarreau760e81d2018-05-03 07:20:40 +02002889
Emeric Brun736aa232009-06-30 17:56:00 +02002890 rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02002891 rdp-cookie(<name>)
Emeric Brun736aa232009-06-30 17:56:00 +02002892 The RDP cookie <name> (or "mstshash" if omitted) will be
2893 looked up and hashed for each incoming TCP request. Just as
2894 with the equivalent ACL 'req_rdp_cookie()' function, the name
2895 is not case-sensitive. This mechanism is useful as a degraded
2896 persistence mode, as it makes it possible to always send the
2897 same user (or the same session ID) to the same server. If the
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002898 cookie is not found, the normal roundrobin algorithm is
Emeric Brun736aa232009-06-30 17:56:00 +02002899 used instead.
2900
2901 Note that for this to work, the frontend must ensure that an
2902 RDP cookie is already present in the request buffer. For this
2903 you must use 'tcp-request content accept' rule combined with
2904 a 'req_rdp_cookie_cnt' ACL.
2905
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002906 This algorithm is static by default, which means that
2907 changing a server's weight on the fly will have no effect,
2908 but this can be changed using "hash-type".
2909
Cyril Bontédc4d9032012-04-08 21:57:39 +02002910 See also the rdp_cookie pattern fetch function.
Simon Hormanab814e02011-06-24 14:50:20 +09002911
Willy Tarreau0ba27502007-12-24 16:55:16 +01002912 <arguments> is an optional list of arguments which may be needed by some
Marek Majkowski9c30fc12008-04-27 23:25:55 +02002913 algorithms. Right now, only "url_param" and "uri" support an
2914 optional argument.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002915
Willy Tarreau3cd9af22009-03-15 14:06:41 +01002916 The load balancing algorithm of a backend is set to roundrobin when no other
2917 algorithm, mode nor option have been set. The algorithm may only be set once
2918 for each backend.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002919
Lukas Tribus80512b12018-10-27 20:07:40 +02002920 With authentication schemes that require the same connection like NTLM, URI
John Roeslerfb2fce12019-07-10 15:45:51 -05002921 based algorithms must not be used, as they would cause subsequent requests
Lukas Tribus80512b12018-10-27 20:07:40 +02002922 to be routed to different backend servers, breaking the invalid assumptions
2923 NTLM relies on.
2924
Willy Tarreau0ba27502007-12-24 16:55:16 +01002925 Examples :
2926 balance roundrobin
2927 balance url_param userid
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002928 balance url_param session_id check_post 64
Benoitaffb4812009-03-25 13:02:10 +01002929 balance hdr(User-Agent)
2930 balance hdr(host)
2931 balance hdr(Host) use_domain_only
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002932
2933 Note: the following caveats and limitations on using the "check_post"
2934 extension with "url_param" must be considered :
2935
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002936 - all POST requests are eligible for consideration, because there is no way
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002937 to determine if the parameters will be found in the body or entity which
2938 may contain binary data. Therefore another method may be required to
2939 restrict consideration of POST requests that have no URL parameters in
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02002940 the body. (see acl http_end)
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002941
2942 - using a <max_wait> value larger than the request buffer size does not
2943 make sense and is useless. The buffer size is set at build time, and
2944 defaults to 16 kB.
2945
2946 - Content-Encoding is not supported, the parameter search will probably
2947 fail; and load balancing will fall back to Round Robin.
2948
2949 - Expect: 100-continue is not supported, load balancing will fall back to
2950 Round Robin.
2951
Lukas Tribus23953682017-04-28 13:24:30 +00002952 - Transfer-Encoding (RFC7230 3.3.1) is only supported in the first chunk.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002953 If the entire parameter value is not present in the first chunk, the
2954 selection of server is undefined (actually, defined by how little
2955 actually appeared in the first chunk).
2956
2957 - This feature does not support generation of a 100, 411 or 501 response.
2958
2959 - In some cases, requesting "check_post" MAY attempt to scan the entire
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002960 contents of a message body. Scanning normally terminates when linear
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002961 white space or control characters are found, indicating the end of what
2962 might be a URL parameter list. This is probably not a concern with SGML
2963 type message bodies.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002964
Willy Tarreau294d0f02015-08-10 19:40:12 +02002965 See also : "dispatch", "cookie", "transparent", "hash-type" and "http_proxy".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002966
2967
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002968bind [<address>]:<port_range> [, ...] [param*]
2969bind /<path> [, ...] [param*]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002970 Define one or several listening addresses and/or ports in a frontend.
2971 May be used in sections : defaults | frontend | listen | backend
2972 no | yes | yes | no
2973 Arguments :
Willy Tarreaub1e52e82008-01-13 14:49:51 +01002974 <address> is optional and can be a host name, an IPv4 address, an IPv6
2975 address, or '*'. It designates the address the frontend will
2976 listen on. If unset, all IPv4 addresses of the system will be
2977 listened on. The same will apply for '*' or the system's
David du Colombier9c938da2011-03-17 10:40:27 +01002978 special address "0.0.0.0". The IPv6 equivalent is '::'.
Willy Tarreau24709282013-03-10 21:32:12 +01002979 Optionally, an address family prefix may be used before the
2980 address to force the family regardless of the address format,
2981 which can be useful to specify a path to a unix socket with
2982 no slash ('/'). Currently supported prefixes are :
2983 - 'ipv4@' -> address is always IPv4
2984 - 'ipv6@' -> address is always IPv6
2985 - 'unix@' -> address is a path to a local unix socket
Willy Tarreau70f72e02014-07-08 00:37:50 +02002986 - 'abns@' -> address is in abstract namespace (Linux only).
2987 Note: since abstract sockets are not "rebindable", they
2988 do not cope well with multi-process mode during
2989 soft-restart, so it is better to avoid them if
2990 nbproc is greater than 1. The effect is that if the
2991 new process fails to start, only one of the old ones
2992 will be able to rebind to the socket.
Willy Tarreau40aa0702013-03-10 23:51:38 +01002993 - 'fd@<n>' -> use file descriptor <n> inherited from the
2994 parent. The fd must be bound and may or may not already
2995 be listening.
William Lallemand2fe7dd02018-09-11 16:51:29 +02002996 - 'sockpair@<n>'-> like fd@ but you must use the fd of a
2997 connected unix socket or of a socketpair. The bind waits
2998 to receive a FD over the unix socket and uses it as if it
2999 was the FD of an accept(). Should be used carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02003000 You may want to reference some environment variables in the
3001 address parameter, see section 2.3 about environment
3002 variables.
Willy Tarreaub1e52e82008-01-13 14:49:51 +01003003
Willy Tarreauc5011ca2010-03-22 11:53:56 +01003004 <port_range> is either a unique TCP port, or a port range for which the
3005 proxy will accept connections for the IP address specified
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003006 above. The port is mandatory for TCP listeners. Note that in
3007 the case of an IPv6 address, the port is always the number
3008 after the last colon (':'). A range can either be :
Willy Tarreauc5011ca2010-03-22 11:53:56 +01003009 - a numerical port (ex: '80')
3010 - a dash-delimited ports range explicitly stating the lower
3011 and upper bounds (ex: '2000-2100') which are included in
3012 the range.
3013
3014 Particular care must be taken against port ranges, because
3015 every <address:port> couple consumes one socket (= a file
3016 descriptor), so it's easy to consume lots of descriptors
3017 with a simple range, and to run out of sockets. Also, each
3018 <address:port> couple must be used only once among all
3019 instances running on a same system. Please note that binding
3020 to ports lower than 1024 generally require particular
Jamie Gloudon801a0a32012-08-25 00:18:33 -04003021 privileges to start the program, which are independent of
Willy Tarreauc5011ca2010-03-22 11:53:56 +01003022 the 'uid' parameter.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003023
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003024 <path> is a UNIX socket path beginning with a slash ('/'). This is
Davor Ocelice9ed2812017-12-25 17:49:28 +01003025 alternative to the TCP listening port. HAProxy will then
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003026 receive UNIX connections on the socket located at this place.
3027 The path must begin with a slash and by default is absolute.
3028 It can be relative to the prefix defined by "unix-bind" in
3029 the global section. Note that the total length of the prefix
3030 followed by the socket path cannot exceed some system limits
3031 for UNIX sockets, which commonly are set to 107 characters.
3032
Willy Tarreaub6205fd2012-09-24 12:27:33 +02003033 <param*> is a list of parameters common to all sockets declared on the
3034 same line. These numerous parameters depend on OS and build
3035 options and have a complete section dedicated to them. Please
3036 refer to section 5 to for more details.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02003037
Willy Tarreau0ba27502007-12-24 16:55:16 +01003038 It is possible to specify a list of address:port combinations delimited by
3039 commas. The frontend will then listen on all of these addresses. There is no
3040 fixed limit to the number of addresses and ports which can be listened on in
3041 a frontend, as well as there is no limit to the number of "bind" statements
3042 in a frontend.
3043
3044 Example :
3045 listen http_proxy
3046 bind :80,:443
3047 bind 10.0.0.1:10080,10.0.0.1:10443
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003048 bind /var/run/ssl-frontend.sock user root mode 600 accept-proxy
Willy Tarreau0ba27502007-12-24 16:55:16 +01003049
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02003050 listen http_https_proxy
3051 bind :80
Cyril Bonté0d44fc62012-10-09 22:45:33 +02003052 bind :443 ssl crt /etc/haproxy/site.pem
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02003053
Willy Tarreau24709282013-03-10 21:32:12 +01003054 listen http_https_proxy_explicit
3055 bind ipv6@:80
3056 bind ipv4@public_ssl:443 ssl crt /etc/haproxy/site.pem
3057 bind unix@ssl-frontend.sock user root mode 600 accept-proxy
3058
Willy Tarreaudad36a32013-03-11 01:20:04 +01003059 listen external_bind_app1
William Lallemandb2f07452015-05-12 14:27:13 +02003060 bind "fd@${FD_APP1}"
Willy Tarreaudad36a32013-03-11 01:20:04 +01003061
Willy Tarreau55dcaf62015-09-27 15:03:15 +02003062 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
3063 sun_path length is used for the address length. Some other programs
3064 such as socat use the string length only by default. Pass the option
3065 ",unix-tightsocklen=0" to any abstract socket definition in socat to
3066 make it compatible with HAProxy's.
3067
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003068 See also : "source", "option forwardfor", "unix-bind" and the PROXY protocol
Willy Tarreaub6205fd2012-09-24 12:27:33 +02003069 documentation, and section 5 about bind options.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003070
3071
Christopher Fauletff4121f2017-11-22 16:38:49 +01003072bind-process [ all | odd | even | <process_num>[-[<process_num>]] ] ...
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003073 Limit visibility of an instance to a certain set of processes numbers.
3074 May be used in sections : defaults | frontend | listen | backend
3075 yes | yes | yes | yes
3076 Arguments :
3077 all All process will see this instance. This is the default. It
3078 may be used to override a default value.
3079
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003080 odd This instance will be enabled on processes 1,3,5,...63. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003081 option may be combined with other numbers.
3082
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003083 even This instance will be enabled on processes 2,4,6,...64. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003084 option may be combined with other numbers. Do not use it
3085 with less than 2 processes otherwise some instances might be
3086 missing from all processes.
3087
Christopher Fauletff4121f2017-11-22 16:38:49 +01003088 process_num The instance will be enabled on this process number or range,
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003089 whose values must all be between 1 and 32 or 64 depending on
Christopher Fauletff4121f2017-11-22 16:38:49 +01003090 the machine's word size. Ranges can be partially defined. The
3091 higher bound can be omitted. In such case, it is replaced by
3092 the corresponding maximum value. If a proxy is bound to
3093 process numbers greater than the configured global.nbproc, it
3094 will either be forced to process #1 if a single process was
Willy Tarreau102df612014-05-07 23:56:38 +02003095 specified, or to all processes otherwise.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003096
3097 This keyword limits binding of certain instances to certain processes. This
3098 is useful in order not to have too many processes listening to the same
3099 ports. For instance, on a dual-core machine, it might make sense to set
3100 'nbproc 2' in the global section, then distributes the listeners among 'odd'
3101 and 'even' instances.
3102
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003103 At the moment, it is not possible to reference more than 32 or 64 processes
3104 using this keyword, but this should be more than enough for most setups.
3105 Please note that 'all' really means all processes regardless of the machine's
3106 word size, and is not limited to the first 32 or 64.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003107
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02003108 Each "bind" line may further be limited to a subset of the proxy's processes,
3109 please consult the "process" bind keyword in section 5.1.
3110
Willy Tarreaub369a042014-09-16 13:21:03 +02003111 When a frontend has no explicit "bind-process" line, it tries to bind to all
3112 the processes referenced by its "bind" lines. That means that frontends can
3113 easily adapt to their listeners' processes.
3114
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003115 If some backends are referenced by frontends bound to other processes, the
3116 backend automatically inherits the frontend's processes.
3117
3118 Example :
3119 listen app_ip1
3120 bind 10.0.0.1:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02003121 bind-process odd
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003122
3123 listen app_ip2
3124 bind 10.0.0.2:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02003125 bind-process even
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003126
3127 listen management
3128 bind 10.0.0.3:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02003129 bind-process 1 2 3 4
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003130
Willy Tarreau110ecc12012-11-15 17:50:01 +01003131 listen management
3132 bind 10.0.0.4:80
3133 bind-process 1-4
3134
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02003135 See also : "nbproc" in global section, and "process" in section 5.1.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003136
3137
Willy Tarreau0ba27502007-12-24 16:55:16 +01003138capture cookie <name> len <length>
3139 Capture and log a cookie in the request and in the response.
3140 May be used in sections : defaults | frontend | listen | backend
3141 no | yes | yes | no
3142 Arguments :
3143 <name> is the beginning of the name of the cookie to capture. In order
3144 to match the exact name, simply suffix the name with an equal
3145 sign ('='). The full name will appear in the logs, which is
3146 useful with application servers which adjust both the cookie name
Davor Ocelice9ed2812017-12-25 17:49:28 +01003147 and value (e.g. ASPSESSIONXXX).
Willy Tarreau0ba27502007-12-24 16:55:16 +01003148
3149 <length> is the maximum number of characters to report in the logs, which
3150 include the cookie name, the equal sign and the value, all in the
3151 standard "name=value" form. The string will be truncated on the
3152 right if it exceeds <length>.
3153
3154 Only the first cookie is captured. Both the "cookie" request headers and the
3155 "set-cookie" response headers are monitored. This is particularly useful to
3156 check for application bugs causing session crossing or stealing between
3157 users, because generally the user's cookies can only change on a login page.
3158
3159 When the cookie was not presented by the client, the associated log column
3160 will report "-". When a request does not cause a cookie to be assigned by the
3161 server, a "-" is reported in the response column.
3162
3163 The capture is performed in the frontend only because it is necessary that
3164 the log format does not change for a given frontend depending on the
3165 backends. This may change in the future. Note that there can be only one
Willy Tarreau193b8c62012-11-22 00:17:38 +01003166 "capture cookie" statement in a frontend. The maximum capture length is set
3167 by the global "tune.http.cookielen" setting and defaults to 63 characters. It
3168 is not possible to specify a capture in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003169
3170 Example:
3171 capture cookie ASPSESSION len 32
3172
3173 See also : "capture request header", "capture response header" as well as
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003174 section 8 about logging.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003175
3176
3177capture request header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01003178 Capture and log the last occurrence of the specified request header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003179 May be used in sections : defaults | frontend | listen | backend
3180 no | yes | yes | no
3181 Arguments :
3182 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003183 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01003184 appear in the requests, with the first letter of each word in
3185 upper case. The header name will not appear in the logs, only the
3186 value is reported, but the position in the logs is respected.
3187
3188 <length> is the maximum number of characters to extract from the value and
3189 report in the logs. The string will be truncated on the right if
3190 it exceeds <length>.
3191
Willy Tarreau4460d032012-11-21 23:37:37 +01003192 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01003193 value will be added to the logs between braces ('{}'). If multiple headers
3194 are captured, they will be delimited by a vertical bar ('|') and will appear
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003195 in the same order they were declared in the configuration. Non-existent
3196 headers will be logged just as an empty string. Common uses for request
3197 header captures include the "Host" field in virtual hosting environments, the
3198 "Content-length" when uploads are supported, "User-agent" to quickly
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003199 differentiate between real users and robots, and "X-Forwarded-For" in proxied
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003200 environments to find where the request came from.
3201
3202 Note that when capturing headers such as "User-agent", some spaces may be
3203 logged, making the log analysis more difficult. Thus be careful about what
3204 you log if you know your log parser is not smart enough to rely on the
3205 braces.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003206
Willy Tarreau0900abb2012-11-22 00:21:46 +01003207 There is no limit to the number of captured request headers nor to their
3208 length, though it is wise to keep them low to limit memory usage per session.
3209 In order to keep log format consistent for a same frontend, header captures
3210 can only be declared in a frontend. It is not possible to specify a capture
3211 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003212
3213 Example:
3214 capture request header Host len 15
3215 capture request header X-Forwarded-For len 15
Cyril Bontéd1b0f7c2015-10-26 22:37:39 +01003216 capture request header Referer len 15
Willy Tarreau0ba27502007-12-24 16:55:16 +01003217
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003218 See also : "capture cookie", "capture response header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01003219 about logging.
3220
3221
3222capture response header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01003223 Capture and log the last occurrence of the specified response header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003224 May be used in sections : defaults | frontend | listen | backend
3225 no | yes | yes | no
3226 Arguments :
3227 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003228 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01003229 appear in the response, with the first letter of each word in
3230 upper case. The header name will not appear in the logs, only the
3231 value is reported, but the position in the logs is respected.
3232
3233 <length> is the maximum number of characters to extract from the value and
3234 report in the logs. The string will be truncated on the right if
3235 it exceeds <length>.
3236
Willy Tarreau4460d032012-11-21 23:37:37 +01003237 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01003238 result will be added to the logs between braces ('{}') after the captured
3239 request headers. If multiple headers are captured, they will be delimited by
3240 a vertical bar ('|') and will appear in the same order they were declared in
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003241 the configuration. Non-existent headers will be logged just as an empty
3242 string. Common uses for response header captures include the "Content-length"
3243 header which indicates how many bytes are expected to be returned, the
3244 "Location" header to track redirections.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003245
Willy Tarreau0900abb2012-11-22 00:21:46 +01003246 There is no limit to the number of captured response headers nor to their
3247 length, though it is wise to keep them low to limit memory usage per session.
3248 In order to keep log format consistent for a same frontend, header captures
3249 can only be declared in a frontend. It is not possible to specify a capture
3250 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003251
3252 Example:
3253 capture response header Content-length len 9
3254 capture response header Location len 15
3255
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003256 See also : "capture cookie", "capture request header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01003257 about logging.
3258
3259
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003260compression algo <algorithm> ...
3261compression type <mime type> ...
Willy Tarreau70737d12012-10-27 00:34:28 +02003262compression offload
William Lallemand82fe75c2012-10-23 10:25:10 +02003263 Enable HTTP compression.
3264 May be used in sections : defaults | frontend | listen | backend
3265 yes | yes | yes | yes
3266 Arguments :
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003267 algo is followed by the list of supported compression algorithms.
3268 type is followed by the list of MIME types that will be compressed.
3269 offload makes haproxy work as a compression offloader only (see notes).
3270
3271 The currently supported algorithms are :
Willy Tarreauc91840a2015-03-28 17:00:39 +01003272 identity this is mostly for debugging, and it was useful for developing
3273 the compression feature. Identity does not apply any change on
3274 data.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003275
Willy Tarreauc91840a2015-03-28 17:00:39 +01003276 gzip applies gzip compression. This setting is only available when
Baptiste Assmannf085d632015-12-21 17:57:32 +01003277 support for zlib or libslz was built in.
Willy Tarreauc91840a2015-03-28 17:00:39 +01003278
3279 deflate same as "gzip", but with deflate algorithm and zlib format.
3280 Note that this algorithm has ambiguous support on many
3281 browsers and no support at all from recent ones. It is
3282 strongly recommended not to use it for anything else than
3283 experimentation. This setting is only available when support
Baptiste Assmannf085d632015-12-21 17:57:32 +01003284 for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003285
Willy Tarreauc91840a2015-03-28 17:00:39 +01003286 raw-deflate same as "deflate" without the zlib wrapper, and used as an
3287 alternative when the browser wants "deflate". All major
3288 browsers understand it and despite violating the standards,
3289 it is known to work better than "deflate", at least on MSIE
3290 and some versions of Safari. Do not use it in conjunction
3291 with "deflate", use either one or the other since both react
3292 to the same Accept-Encoding token. This setting is only
Baptiste Assmannf085d632015-12-21 17:57:32 +01003293 available when support for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003294
Dmitry Sivachenko87c208b2012-11-22 20:03:26 +04003295 Compression will be activated depending on the Accept-Encoding request
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003296 header. With identity, it does not take care of that header.
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04003297 If backend servers support HTTP compression, these directives
3298 will be no-op: haproxy will see the compressed response and will not
3299 compress again. If backend servers do not support HTTP compression and
3300 there is Accept-Encoding header in request, haproxy will compress the
3301 matching response.
Willy Tarreau70737d12012-10-27 00:34:28 +02003302
3303 The "offload" setting makes haproxy remove the Accept-Encoding header to
3304 prevent backend servers from compressing responses. It is strongly
3305 recommended not to do this because this means that all the compression work
3306 will be done on the single point where haproxy is located. However in some
3307 deployment scenarios, haproxy may be installed in front of a buggy gateway
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04003308 with broken HTTP compression implementation which can't be turned off.
3309 In that case haproxy can be used to prevent that gateway from emitting
3310 invalid payloads. In this case, simply removing the header in the
3311 configuration does not work because it applies before the header is parsed,
3312 so that prevents haproxy from compressing. The "offload" setting should
Willy Tarreauffea9fd2014-07-12 16:37:02 +02003313 then be used for such scenarios. Note: for now, the "offload" setting is
3314 ignored when set in a defaults section.
William Lallemand82fe75c2012-10-23 10:25:10 +02003315
William Lallemand05097442012-11-20 12:14:28 +01003316 Compression is disabled when:
Baptiste Assmann650d53d2013-01-05 15:44:44 +01003317 * the request does not advertise a supported compression algorithm in the
3318 "Accept-Encoding" header
3319 * the response message is not HTTP/1.1
Tim Duesterhusbb48c9a2019-01-30 23:46:04 +01003320 * HTTP status code is not one of 200, 201, 202, or 203
Baptiste Assmann650d53d2013-01-05 15:44:44 +01003321 * response contain neither a "Content-Length" header nor a
3322 "Transfer-Encoding" whose last value is "chunked"
3323 * response contains a "Content-Type" header whose first value starts with
3324 "multipart"
3325 * the response contains the "no-transform" value in the "Cache-control"
3326 header
3327 * User-Agent matches "Mozilla/4" unless it is MSIE 6 with XP SP2, or MSIE 7
3328 and later
3329 * The response contains a "Content-Encoding" header, indicating that the
3330 response is already compressed (see compression offload)
Tim Duesterhusbb48c9a2019-01-30 23:46:04 +01003331 * The response contains an invalid "ETag" header or multiple ETag headers
William Lallemand05097442012-11-20 12:14:28 +01003332
Tim Duesterhusb229f012019-01-29 16:38:56 +01003333 Note: The compression does not emit the Warning header.
William Lallemand05097442012-11-20 12:14:28 +01003334
William Lallemand82fe75c2012-10-23 10:25:10 +02003335 Examples :
3336 compression algo gzip
3337 compression type text/html text/plain
Willy Tarreau0ba27502007-12-24 16:55:16 +01003338
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003339
Willy Tarreau55165fe2009-05-10 12:02:55 +02003340cookie <name> [ rewrite | insert | prefix ] [ indirect ] [ nocache ]
Willy Tarreau4992dd22012-05-31 21:02:17 +02003341 [ postonly ] [ preserve ] [ httponly ] [ secure ]
3342 [ domain <domain> ]* [ maxidle <idle> ] [ maxlife <life> ]
Christopher Faulet2f533902020-01-21 11:06:48 +01003343 [ dynamic ] [ attr <value> ]*
Willy Tarreau0ba27502007-12-24 16:55:16 +01003344 Enable cookie-based persistence in a backend.
3345 May be used in sections : defaults | frontend | listen | backend
3346 yes | no | yes | yes
3347 Arguments :
3348 <name> is the name of the cookie which will be monitored, modified or
3349 inserted in order to bring persistence. This cookie is sent to
3350 the client via a "Set-Cookie" header in the response, and is
3351 brought back by the client in a "Cookie" header in all requests.
3352 Special care should be taken to choose a name which does not
3353 conflict with any likely application cookie. Also, if the same
Davor Ocelice9ed2812017-12-25 17:49:28 +01003354 backends are subject to be used by the same clients (e.g.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003355 HTTP/HTTPS), care should be taken to use different cookie names
3356 between all backends if persistence between them is not desired.
3357
3358 rewrite This keyword indicates that the cookie will be provided by the
3359 server and that haproxy will have to modify its value to set the
3360 server's identifier in it. This mode is handy when the management
3361 of complex combinations of "Set-cookie" and "Cache-control"
3362 headers is left to the application. The application can then
3363 decide whether or not it is appropriate to emit a persistence
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003364 cookie. Since all responses should be monitored, this mode
3365 doesn't work in HTTP tunnel mode. Unless the application
Davor Ocelice9ed2812017-12-25 17:49:28 +01003366 behavior is very complex and/or broken, it is advised not to
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003367 start with this mode for new deployments. This keyword is
3368 incompatible with "insert" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003369
3370 insert This keyword indicates that the persistence cookie will have to
Willy Tarreaua79094d2010-08-31 22:54:15 +02003371 be inserted by haproxy in server responses if the client did not
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003372
Willy Tarreaua79094d2010-08-31 22:54:15 +02003373 already have a cookie that would have permitted it to access this
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003374 server. When used without the "preserve" option, if the server
Michael Prokop4438c602019-05-24 10:25:45 +02003375 emits a cookie with the same name, it will be removed before
Davor Ocelice9ed2812017-12-25 17:49:28 +01003376 processing. For this reason, this mode can be used to upgrade
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003377 existing configurations running in the "rewrite" mode. The cookie
3378 will only be a session cookie and will not be stored on the
3379 client's disk. By default, unless the "indirect" option is added,
3380 the server will see the cookies emitted by the client. Due to
3381 caching effects, it is generally wise to add the "nocache" or
3382 "postonly" keywords (see below). The "insert" keyword is not
3383 compatible with "rewrite" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003384
3385 prefix This keyword indicates that instead of relying on a dedicated
3386 cookie for the persistence, an existing one will be completed.
3387 This may be needed in some specific environments where the client
3388 does not support more than one single cookie and the application
3389 already needs it. In this case, whenever the server sets a cookie
3390 named <name>, it will be prefixed with the server's identifier
3391 and a delimiter. The prefix will be removed from all client
3392 requests so that the server still finds the cookie it emitted.
3393 Since all requests and responses are subject to being modified,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003394 this mode doesn't work with tunnel mode. The "prefix" keyword is
Willy Tarreau37229df2011-10-17 12:24:55 +02003395 not compatible with "rewrite" and "insert". Note: it is highly
3396 recommended not to use "indirect" with "prefix", otherwise server
3397 cookie updates would not be sent to clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003398
Willy Tarreaua79094d2010-08-31 22:54:15 +02003399 indirect When this option is specified, no cookie will be emitted to a
3400 client which already has a valid one for the server which has
3401 processed the request. If the server sets such a cookie itself,
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003402 it will be removed, unless the "preserve" option is also set. In
3403 "insert" mode, this will additionally remove cookies from the
3404 requests transmitted to the server, making the persistence
3405 mechanism totally transparent from an application point of view.
Willy Tarreau37229df2011-10-17 12:24:55 +02003406 Note: it is highly recommended not to use "indirect" with
3407 "prefix", otherwise server cookie updates would not be sent to
3408 clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003409
3410 nocache This option is recommended in conjunction with the insert mode
3411 when there is a cache between the client and HAProxy, as it
3412 ensures that a cacheable response will be tagged non-cacheable if
3413 a cookie needs to be inserted. This is important because if all
3414 persistence cookies are added on a cacheable home page for
3415 instance, then all customers will then fetch the page from an
3416 outer cache and will all share the same persistence cookie,
3417 leading to one server receiving much more traffic than others.
3418 See also the "insert" and "postonly" options.
3419
3420 postonly This option ensures that cookie insertion will only be performed
3421 on responses to POST requests. It is an alternative to the
3422 "nocache" option, because POST responses are not cacheable, so
3423 this ensures that the persistence cookie will never get cached.
3424 Since most sites do not need any sort of persistence before the
3425 first POST which generally is a login request, this is a very
3426 efficient method to optimize caching without risking to find a
3427 persistence cookie in the cache.
3428 See also the "insert" and "nocache" options.
3429
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003430 preserve This option may only be used with "insert" and/or "indirect". It
3431 allows the server to emit the persistence cookie itself. In this
3432 case, if a cookie is found in the response, haproxy will leave it
3433 untouched. This is useful in order to end persistence after a
3434 logout request for instance. For this, the server just has to
Davor Ocelice9ed2812017-12-25 17:49:28 +01003435 emit a cookie with an invalid value (e.g. empty) or with a date in
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003436 the past. By combining this mechanism with the "disable-on-404"
3437 check option, it is possible to perform a completely graceful
3438 shutdown because users will definitely leave the server after
3439 they logout.
3440
Willy Tarreau4992dd22012-05-31 21:02:17 +02003441 httponly This option tells haproxy to add an "HttpOnly" cookie attribute
3442 when a cookie is inserted. This attribute is used so that a
3443 user agent doesn't share the cookie with non-HTTP components.
3444 Please check RFC6265 for more information on this attribute.
3445
3446 secure This option tells haproxy to add a "Secure" cookie attribute when
3447 a cookie is inserted. This attribute is used so that a user agent
3448 never emits this cookie over non-secure channels, which means
3449 that a cookie learned with this flag will be presented only over
3450 SSL/TLS connections. Please check RFC6265 for more information on
3451 this attribute.
3452
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02003453 domain This option allows to specify the domain at which a cookie is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003454 inserted. It requires exactly one parameter: a valid domain
Willy Tarreau68a897b2009-12-03 23:28:34 +01003455 name. If the domain begins with a dot, the browser is allowed to
3456 use it for any host ending with that name. It is also possible to
3457 specify several domain names by invoking this option multiple
3458 times. Some browsers might have small limits on the number of
3459 domains, so be careful when doing that. For the record, sending
3460 10 domains to MSIE 6 or Firefox 2 works as expected.
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02003461
Willy Tarreau996a92c2010-10-13 19:30:47 +02003462 maxidle This option allows inserted cookies to be ignored after some idle
3463 time. It only works with insert-mode cookies. When a cookie is
3464 sent to the client, the date this cookie was emitted is sent too.
3465 Upon further presentations of this cookie, if the date is older
3466 than the delay indicated by the parameter (in seconds), it will
3467 be ignored. Otherwise, it will be refreshed if needed when the
3468 response is sent to the client. This is particularly useful to
3469 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01003470 too long on the same server (e.g. after a farm size change). When
Willy Tarreau996a92c2010-10-13 19:30:47 +02003471 this option is set and a cookie has no date, it is always
3472 accepted, but gets refreshed in the response. This maintains the
3473 ability for admins to access their sites. Cookies that have a
3474 date in the future further than 24 hours are ignored. Doing so
3475 lets admins fix timezone issues without risking kicking users off
3476 the site.
3477
3478 maxlife This option allows inserted cookies to be ignored after some life
3479 time, whether they're in use or not. It only works with insert
3480 mode cookies. When a cookie is first sent to the client, the date
3481 this cookie was emitted is sent too. Upon further presentations
3482 of this cookie, if the date is older than the delay indicated by
3483 the parameter (in seconds), it will be ignored. If the cookie in
3484 the request has no date, it is accepted and a date will be set.
3485 Cookies that have a date in the future further than 24 hours are
3486 ignored. Doing so lets admins fix timezone issues without risking
3487 kicking users off the site. Contrary to maxidle, this value is
3488 not refreshed, only the first visit date counts. Both maxidle and
3489 maxlife may be used at the time. This is particularly useful to
3490 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01003491 too long on the same server (e.g. after a farm size change). This
Willy Tarreau996a92c2010-10-13 19:30:47 +02003492 is stronger than the maxidle method in that it forces a
3493 redispatch after some absolute delay.
3494
Olivier Houchard4e694042017-03-14 20:01:29 +01003495 dynamic Activate dynamic cookies. When used, a session cookie is
3496 dynamically created for each server, based on the IP and port
3497 of the server, and a secret key, specified in the
3498 "dynamic-cookie-key" backend directive.
3499 The cookie will be regenerated each time the IP address change,
3500 and is only generated for IPv4/IPv6.
3501
Christopher Faulet2f533902020-01-21 11:06:48 +01003502 attr This option tells haproxy to add an extra attribute when a
3503 cookie is inserted. The attribute value can contain any
3504 characters except control ones or ";". This option may be
3505 repeated.
3506
Willy Tarreau0ba27502007-12-24 16:55:16 +01003507 There can be only one persistence cookie per HTTP backend, and it can be
3508 declared in a defaults section. The value of the cookie will be the value
3509 indicated after the "cookie" keyword in a "server" statement. If no cookie
3510 is declared for a given server, the cookie is not set.
Willy Tarreau6a06a402007-07-15 20:15:28 +02003511
Willy Tarreau0ba27502007-12-24 16:55:16 +01003512 Examples :
3513 cookie JSESSIONID prefix
3514 cookie SRV insert indirect nocache
3515 cookie SRV insert postonly indirect
Willy Tarreau996a92c2010-10-13 19:30:47 +02003516 cookie SRV insert indirect nocache maxidle 30m maxlife 8h
Willy Tarreau0ba27502007-12-24 16:55:16 +01003517
Willy Tarreau294d0f02015-08-10 19:40:12 +02003518 See also : "balance source", "capture cookie", "server" and "ignore-persist".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003519
Willy Tarreau983e01e2010-01-11 18:42:06 +01003520
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02003521declare capture [ request | response ] len <length>
3522 Declares a capture slot.
3523 May be used in sections : defaults | frontend | listen | backend
3524 no | yes | yes | no
3525 Arguments:
3526 <length> is the length allowed for the capture.
3527
3528 This declaration is only available in the frontend or listen section, but the
3529 reserved slot can be used in the backends. The "request" keyword allocates a
3530 capture slot for use in the request, and "response" allocates a capture slot
3531 for use in the response.
3532
3533 See also: "capture-req", "capture-res" (sample converters),
Baptiste Assmann5ac425c2015-10-21 23:13:46 +02003534 "capture.req.hdr", "capture.res.hdr" (sample fetches),
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02003535 "http-request capture" and "http-response capture".
3536
3537
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003538default-server [param*]
3539 Change default options for a server in a backend
3540 May be used in sections : defaults | frontend | listen | backend
3541 yes | no | yes | yes
3542 Arguments:
Willy Tarreau983e01e2010-01-11 18:42:06 +01003543 <param*> is a list of parameters for this server. The "default-server"
3544 keyword accepts an important number of options and has a complete
3545 section dedicated to it. Please refer to section 5 for more
3546 details.
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003547
Willy Tarreau983e01e2010-01-11 18:42:06 +01003548 Example :
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003549 default-server inter 1000 weight 13
3550
3551 See also: "server" and section 5 about server options
Willy Tarreau0ba27502007-12-24 16:55:16 +01003552
Willy Tarreau983e01e2010-01-11 18:42:06 +01003553
Willy Tarreau0ba27502007-12-24 16:55:16 +01003554default_backend <backend>
3555 Specify the backend to use when no "use_backend" rule has been matched.
3556 May be used in sections : defaults | frontend | listen | backend
3557 yes | yes | yes | no
3558 Arguments :
3559 <backend> is the name of the backend to use.
3560
3561 When doing content-switching between frontend and backends using the
3562 "use_backend" keyword, it is often useful to indicate which backend will be
3563 used when no rule has matched. It generally is the dynamic backend which
3564 will catch all undetermined requests.
3565
Willy Tarreau0ba27502007-12-24 16:55:16 +01003566 Example :
3567
3568 use_backend dynamic if url_dyn
3569 use_backend static if url_css url_img extension_img
3570 default_backend dynamic
3571
Willy Tarreau98d04852015-05-26 12:18:29 +02003572 See also : "use_backend"
Willy Tarreau2769aa02007-12-27 18:26:09 +01003573
Willy Tarreau0ba27502007-12-24 16:55:16 +01003574
Baptiste Assmann27f51342013-10-09 06:51:49 +02003575description <string>
3576 Describe a listen, frontend or backend.
3577 May be used in sections : defaults | frontend | listen | backend
3578 no | yes | yes | yes
3579 Arguments : string
3580
3581 Allows to add a sentence to describe the related object in the HAProxy HTML
3582 stats page. The description will be printed on the right of the object name
3583 it describes.
3584 No need to backslash spaces in the <string> arguments.
3585
3586
Willy Tarreau0ba27502007-12-24 16:55:16 +01003587disabled
3588 Disable a proxy, frontend or backend.
3589 May be used in sections : defaults | frontend | listen | backend
3590 yes | yes | yes | yes
3591 Arguments : none
3592
3593 The "disabled" keyword is used to disable an instance, mainly in order to
3594 liberate a listening port or to temporarily disable a service. The instance
3595 will still be created and its configuration will be checked, but it will be
3596 created in the "stopped" state and will appear as such in the statistics. It
3597 will not receive any traffic nor will it send any health-checks or logs. It
3598 is possible to disable many instances at once by adding the "disabled"
3599 keyword in a "defaults" section.
3600
3601 See also : "enabled"
3602
3603
Willy Tarreau5ce94572010-06-07 14:35:41 +02003604dispatch <address>:<port>
3605 Set a default server address
3606 May be used in sections : defaults | frontend | listen | backend
3607 no | no | yes | yes
Cyril Bonté108cf6e2012-04-21 23:30:29 +02003608 Arguments :
Willy Tarreau5ce94572010-06-07 14:35:41 +02003609
3610 <address> is the IPv4 address of the default server. Alternatively, a
3611 resolvable hostname is supported, but this name will be resolved
3612 during start-up.
3613
3614 <ports> is a mandatory port specification. All connections will be sent
3615 to this port, and it is not permitted to use port offsets as is
3616 possible with normal servers.
3617
Willy Tarreau787aed52011-04-15 06:45:37 +02003618 The "dispatch" keyword designates a default server for use when no other
Willy Tarreau5ce94572010-06-07 14:35:41 +02003619 server can take the connection. In the past it was used to forward non
3620 persistent connections to an auxiliary load balancer. Due to its simple
3621 syntax, it has also been used for simple TCP relays. It is recommended not to
3622 use it for more clarity, and to use the "server" directive instead.
3623
3624 See also : "server"
3625
Olivier Houchard4e694042017-03-14 20:01:29 +01003626
3627dynamic-cookie-key <string>
3628 Set the dynamic cookie secret key for a backend.
3629 May be used in sections : defaults | frontend | listen | backend
3630 yes | no | yes | yes
3631 Arguments : The secret key to be used.
3632
3633 When dynamic cookies are enabled (see the "dynamic" directive for cookie),
Davor Ocelice9ed2812017-12-25 17:49:28 +01003634 a dynamic cookie is created for each server (unless one is explicitly
Olivier Houchard4e694042017-03-14 20:01:29 +01003635 specified on the "server" line), using a hash of the IP address of the
3636 server, the TCP port, and the secret key.
Davor Ocelice9ed2812017-12-25 17:49:28 +01003637 That way, we can ensure session persistence across multiple load-balancers,
Olivier Houchard4e694042017-03-14 20:01:29 +01003638 even if servers are dynamically added or removed.
Willy Tarreau5ce94572010-06-07 14:35:41 +02003639
Willy Tarreau0ba27502007-12-24 16:55:16 +01003640enabled
3641 Enable a proxy, frontend or backend.
3642 May be used in sections : defaults | frontend | listen | backend
3643 yes | yes | yes | yes
3644 Arguments : none
3645
3646 The "enabled" keyword is used to explicitly enable an instance, when the
3647 defaults has been set to "disabled". This is very rarely used.
3648
3649 See also : "disabled"
3650
3651
3652errorfile <code> <file>
3653 Return a file contents instead of errors generated by HAProxy
3654 May be used in sections : defaults | frontend | listen | backend
3655 yes | yes | yes | yes
3656 Arguments :
3657 <code> is the HTTP status code. Currently, HAProxy is capable of
Florian Tham9205fea2020-01-08 13:35:30 +01003658 generating codes 200, 400, 403, 404, 405, 408, 410, 425, 429, 500,
3659 502, 503, and 504.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003660
3661 <file> designates a file containing the full HTTP response. It is
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003662 recommended to follow the common practice of appending ".http" to
Willy Tarreau0ba27502007-12-24 16:55:16 +01003663 the filename so that people do not confuse the response with HTML
Willy Tarreau59140a22009-02-22 12:02:30 +01003664 error pages, and to use absolute paths, since files are read
3665 before any chroot is performed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003666
3667 It is important to understand that this keyword is not meant to rewrite
3668 errors returned by the server, but errors detected and returned by HAProxy.
3669 This is why the list of supported errors is limited to a small set.
3670
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003671 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3672
Willy Tarreau0ba27502007-12-24 16:55:16 +01003673 The files are returned verbatim on the TCP socket. This allows any trick such
3674 as redirections to another URL or site, as well as tricks to clean cookies,
3675 force enable or disable caching, etc... The package provides default error
3676 files returning the same contents as default errors.
3677
Willy Tarreau59140a22009-02-22 12:02:30 +01003678 The files should not exceed the configured buffer size (BUFSIZE), which
3679 generally is 8 or 16 kB, otherwise they will be truncated. It is also wise
Davor Ocelice9ed2812017-12-25 17:49:28 +01003680 not to put any reference to local contents (e.g. images) in order to avoid
Willy Tarreau59140a22009-02-22 12:02:30 +01003681 loops between the client and HAProxy when all servers are down, causing an
3682 error to be returned instead of an image. For better HTTP compliance, it is
3683 recommended that all header lines end with CR-LF and not LF alone.
3684
Willy Tarreau0ba27502007-12-24 16:55:16 +01003685 The files are read at the same time as the configuration and kept in memory.
3686 For this reason, the errors continue to be returned even when the process is
3687 chrooted, and no file change is considered while the process is running. A
Willy Tarreauc27debf2008-01-06 08:57:02 +01003688 simple method for developing those files consists in associating them to the
Willy Tarreau0ba27502007-12-24 16:55:16 +01003689 403 status code and interrogating a blocked URL.
3690
3691 See also : "errorloc", "errorloc302", "errorloc303"
3692
Willy Tarreau59140a22009-02-22 12:02:30 +01003693 Example :
3694 errorfile 400 /etc/haproxy/errorfiles/400badreq.http
Willy Tarreau989222a2016-01-15 10:26:26 +01003695 errorfile 408 /dev/null # work around Chrome pre-connect bug
Willy Tarreau59140a22009-02-22 12:02:30 +01003696 errorfile 403 /etc/haproxy/errorfiles/403forbid.http
3697 errorfile 503 /etc/haproxy/errorfiles/503sorry.http
3698
Willy Tarreau2769aa02007-12-27 18:26:09 +01003699
Christopher Faulet76edc0f2020-01-13 15:52:01 +01003700errorfiles <name> [<code> ...]
3701 Import, fully or partially, the error files defined in the <name> http-errors
3702 section.
3703 May be used in sections : defaults | frontend | listen | backend
3704 yes | yes | yes | yes
3705 Arguments :
3706 <name> is the name of an existing http-errors section.
3707
3708 <code> is a HTTP status code. Several status code may be listed.
3709 Currently, HAProxy is capable of generating codes 200, 400, 403,
3710 404, 405, 408, 410, 425, 429, 500, 502, 503, and 504.
3711
3712 Errors defined in the http-errors section with the name <name> are imported
3713 in the current proxy. If no status code is specified, all error files of the
3714 http-errors section are imported. Otherwise, only error files associated to
3715 the listed status code are imported. Those error files override the already
3716 defined custom errors for the proxy. And they may be overridden by following
3717 ones. Fonctionnly, it is exactly the same than declaring all error files by
3718 hand using "errorfile" directives.
3719
3720 See also : "errorfile", "errorloc", "errorloc302" , "errorloc303" and section
3721 3.8 about http-errors.
3722
3723 Example :
3724 errorfiles generic
3725 errorfiles site-1 403 404
3726
3727
Willy Tarreau2769aa02007-12-27 18:26:09 +01003728errorloc <code> <url>
3729errorloc302 <code> <url>
3730 Return an HTTP redirection to a URL instead of errors generated by HAProxy
3731 May be used in sections : defaults | frontend | listen | backend
3732 yes | yes | yes | yes
3733 Arguments :
3734 <code> is the HTTP status code. Currently, HAProxy is capable of
Florian Tham9205fea2020-01-08 13:35:30 +01003735 generating codes 200, 400, 403, 404, 405, 408, 410, 425, 429, 500,
3736 502, 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003737
3738 <url> it is the exact contents of the "Location" header. It may contain
3739 either a relative URI to an error page hosted on the same site,
3740 or an absolute URI designating an error page on another site.
3741 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01003742 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01003743
3744 It is important to understand that this keyword is not meant to rewrite
3745 errors returned by the server, but errors detected and returned by HAProxy.
3746 This is why the list of supported errors is limited to a small set.
3747
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003748 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3749
Willy Tarreau2769aa02007-12-27 18:26:09 +01003750 Note that both keyword return the HTTP 302 status code, which tells the
3751 client to fetch the designated URL using the same HTTP method. This can be
3752 quite problematic in case of non-GET methods such as POST, because the URL
3753 sent to the client might not be allowed for something other than GET. To
Willy Tarreau989222a2016-01-15 10:26:26 +01003754 work around this problem, please use "errorloc303" which send the HTTP 303
Willy Tarreau2769aa02007-12-27 18:26:09 +01003755 status code, indicating to the client that the URL must be fetched with a GET
3756 request.
3757
3758 See also : "errorfile", "errorloc303"
3759
3760
3761errorloc303 <code> <url>
3762 Return an HTTP redirection to a URL instead of errors generated by HAProxy
3763 May be used in sections : defaults | frontend | listen | backend
3764 yes | yes | yes | yes
3765 Arguments :
3766 <code> is the HTTP status code. Currently, HAProxy is capable of
Florian Tham9205fea2020-01-08 13:35:30 +01003767 generating codes 200, 400, 403, 404, 405, 408, 410, 425, 429, 500,
3768 502, 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003769
3770 <url> it is the exact contents of the "Location" header. It may contain
3771 either a relative URI to an error page hosted on the same site,
3772 or an absolute URI designating an error page on another site.
3773 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01003774 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01003775
3776 It is important to understand that this keyword is not meant to rewrite
3777 errors returned by the server, but errors detected and returned by HAProxy.
3778 This is why the list of supported errors is limited to a small set.
3779
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003780 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3781
Willy Tarreau2769aa02007-12-27 18:26:09 +01003782 Note that both keyword return the HTTP 303 status code, which tells the
3783 client to fetch the designated URL using the same HTTP GET method. This
3784 solves the usual problems associated with "errorloc" and the 302 code. It is
3785 possible that some very old browsers designed before HTTP/1.1 do not support
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003786 it, but no such problem has been reported till now.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003787
3788 See also : "errorfile", "errorloc", "errorloc302"
3789
3790
Simon Horman51a1cf62015-02-03 13:00:44 +09003791email-alert from <emailaddr>
3792 Declare the from email address to be used in both the envelope and header
Davor Ocelice9ed2812017-12-25 17:49:28 +01003793 of email alerts. This is the address that email alerts are sent from.
Simon Horman51a1cf62015-02-03 13:00:44 +09003794 May be used in sections: defaults | frontend | listen | backend
3795 yes | yes | yes | yes
3796
3797 Arguments :
3798
3799 <emailaddr> is the from email address to use when sending email alerts
3800
3801 Also requires "email-alert mailers" and "email-alert to" to be set
3802 and if so sending email alerts is enabled for the proxy.
3803
Simon Horman64e34162015-02-06 11:11:57 +09003804 See also : "email-alert level", "email-alert mailers",
Cyril Bonté307ee1e2015-09-28 23:16:06 +02003805 "email-alert myhostname", "email-alert to", section 3.6 about
3806 mailers.
Simon Horman64e34162015-02-06 11:11:57 +09003807
3808
3809email-alert level <level>
3810 Declare the maximum log level of messages for which email alerts will be
3811 sent. This acts as a filter on the sending of email alerts.
3812 May be used in sections: defaults | frontend | listen | backend
3813 yes | yes | yes | yes
3814
3815 Arguments :
3816
3817 <level> One of the 8 syslog levels:
3818 emerg alert crit err warning notice info debug
3819 The above syslog levels are ordered from lowest to highest.
3820
3821 By default level is alert
3822
3823 Also requires "email-alert from", "email-alert mailers" and
3824 "email-alert to" to be set and if so sending email alerts is enabled
3825 for the proxy.
3826
Simon Horman1421e212015-04-30 13:10:35 +09003827 Alerts are sent when :
3828
3829 * An un-paused server is marked as down and <level> is alert or lower
3830 * A paused server is marked as down and <level> is notice or lower
3831 * A server is marked as up or enters the drain state and <level>
3832 is notice or lower
3833 * "option log-health-checks" is enabled, <level> is info or lower,
3834 and a health check status update occurs
3835
Simon Horman64e34162015-02-06 11:11:57 +09003836 See also : "email-alert from", "email-alert mailers",
3837 "email-alert myhostname", "email-alert to",
Simon Horman51a1cf62015-02-03 13:00:44 +09003838 section 3.6 about mailers.
3839
3840
3841email-alert mailers <mailersect>
3842 Declare the mailers to be used when sending email alerts
3843 May be used in sections: defaults | frontend | listen | backend
3844 yes | yes | yes | yes
3845
3846 Arguments :
3847
3848 <mailersect> is the name of the mailers section to send email alerts.
3849
3850 Also requires "email-alert from" and "email-alert to" to be set
3851 and if so sending email alerts is enabled for the proxy.
3852
Simon Horman64e34162015-02-06 11:11:57 +09003853 See also : "email-alert from", "email-alert level", "email-alert myhostname",
3854 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09003855
3856
3857email-alert myhostname <hostname>
3858 Declare the to hostname address to be used when communicating with
3859 mailers.
3860 May be used in sections: defaults | frontend | listen | backend
3861 yes | yes | yes | yes
3862
3863 Arguments :
3864
Baptiste Assmann738bad92015-12-21 15:27:53 +01003865 <hostname> is the hostname to use when communicating with mailers
Simon Horman51a1cf62015-02-03 13:00:44 +09003866
3867 By default the systems hostname is used.
3868
3869 Also requires "email-alert from", "email-alert mailers" and
3870 "email-alert to" to be set and if so sending email alerts is enabled
3871 for the proxy.
3872
Simon Horman64e34162015-02-06 11:11:57 +09003873 See also : "email-alert from", "email-alert level", "email-alert mailers",
3874 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09003875
3876
3877email-alert to <emailaddr>
Davor Ocelice9ed2812017-12-25 17:49:28 +01003878 Declare both the recipient address in the envelope and to address in the
Simon Horman51a1cf62015-02-03 13:00:44 +09003879 header of email alerts. This is the address that email alerts are sent to.
3880 May be used in sections: defaults | frontend | listen | backend
3881 yes | yes | yes | yes
3882
3883 Arguments :
3884
3885 <emailaddr> is the to email address to use when sending email alerts
3886
3887 Also requires "email-alert mailers" and "email-alert to" to be set
3888 and if so sending email alerts is enabled for the proxy.
3889
Simon Horman64e34162015-02-06 11:11:57 +09003890 See also : "email-alert from", "email-alert level", "email-alert mailers",
Simon Horman51a1cf62015-02-03 13:00:44 +09003891 "email-alert myhostname", section 3.6 about mailers.
3892
3893
Willy Tarreau4de91492010-01-22 19:10:05 +01003894force-persist { if | unless } <condition>
3895 Declare a condition to force persistence on down servers
3896 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01003897 no | no | yes | yes
Willy Tarreau4de91492010-01-22 19:10:05 +01003898
3899 By default, requests are not dispatched to down servers. It is possible to
3900 force this using "option persist", but it is unconditional and redispatches
3901 to a valid server if "option redispatch" is set. That leaves with very little
3902 possibilities to force some requests to reach a server which is artificially
3903 marked down for maintenance operations.
3904
3905 The "force-persist" statement allows one to declare various ACL-based
3906 conditions which, when met, will cause a request to ignore the down status of
3907 a server and still try to connect to it. That makes it possible to start a
3908 server, still replying an error to the health checks, and run a specially
3909 configured browser to test the service. Among the handy methods, one could
3910 use a specific source IP address, or a specific cookie. The cookie also has
3911 the advantage that it can easily be added/removed on the browser from a test
3912 page. Once the service is validated, it is then possible to open the service
3913 to the world by returning a valid response to health checks.
3914
3915 The forced persistence is enabled when an "if" condition is met, or unless an
3916 "unless" condition is met. The final redispatch is always disabled when this
3917 is used.
3918
Cyril Bonté0d4bf012010-04-25 23:21:46 +02003919 See also : "option redispatch", "ignore-persist", "persist",
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +02003920 and section 7 about ACL usage.
Willy Tarreau4de91492010-01-22 19:10:05 +01003921
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003922
3923filter <name> [param*]
3924 Add the filter <name> in the filter list attached to the proxy.
3925 May be used in sections : defaults | frontend | listen | backend
3926 no | yes | yes | yes
3927 Arguments :
3928 <name> is the name of the filter. Officially supported filters are
3929 referenced in section 9.
3930
Tim Düsterhus4896c442016-11-29 02:15:19 +01003931 <param*> is a list of parameters accepted by the filter <name>. The
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003932 parsing of these parameters are the responsibility of the
Tim Düsterhus4896c442016-11-29 02:15:19 +01003933 filter. Please refer to the documentation of the corresponding
3934 filter (section 9) for all details on the supported parameters.
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003935
3936 Multiple occurrences of the filter line can be used for the same proxy. The
3937 same filter can be referenced many times if needed.
3938
3939 Example:
3940 listen
3941 bind *:80
3942
3943 filter trace name BEFORE-HTTP-COMP
3944 filter compression
3945 filter trace name AFTER-HTTP-COMP
3946
3947 compression algo gzip
3948 compression offload
3949
3950 server srv1 192.168.0.1:80
3951
3952 See also : section 9.
3953
Willy Tarreau4de91492010-01-22 19:10:05 +01003954
Willy Tarreau2769aa02007-12-27 18:26:09 +01003955fullconn <conns>
3956 Specify at what backend load the servers will reach their maxconn
3957 May be used in sections : defaults | frontend | listen | backend
3958 yes | no | yes | yes
3959 Arguments :
3960 <conns> is the number of connections on the backend which will make the
3961 servers use the maximal number of connections.
3962
Willy Tarreau198a7442008-01-17 12:05:32 +01003963 When a server has a "maxconn" parameter specified, it means that its number
Willy Tarreau2769aa02007-12-27 18:26:09 +01003964 of concurrent connections will never go higher. Additionally, if it has a
Willy Tarreau198a7442008-01-17 12:05:32 +01003965 "minconn" parameter, it indicates a dynamic limit following the backend's
Willy Tarreau2769aa02007-12-27 18:26:09 +01003966 load. The server will then always accept at least <minconn> connections,
3967 never more than <maxconn>, and the limit will be on the ramp between both
3968 values when the backend has less than <conns> concurrent connections. This
3969 makes it possible to limit the load on the servers during normal loads, but
3970 push it further for important loads without overloading the servers during
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003971 exceptional loads.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003972
Willy Tarreaufbb78422011-06-05 15:38:35 +02003973 Since it's hard to get this value right, haproxy automatically sets it to
3974 10% of the sum of the maxconns of all frontends that may branch to this
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01003975 backend (based on "use_backend" and "default_backend" rules). That way it's
3976 safe to leave it unset. However, "use_backend" involving dynamic names are
3977 not counted since there is no way to know if they could match or not.
Willy Tarreaufbb78422011-06-05 15:38:35 +02003978
Willy Tarreau2769aa02007-12-27 18:26:09 +01003979 Example :
3980 # The servers will accept between 100 and 1000 concurrent connections each
3981 # and the maximum of 1000 will be reached when the backend reaches 10000
3982 # connections.
3983 backend dynamic
3984 fullconn 10000
3985 server srv1 dyn1:80 minconn 100 maxconn 1000
3986 server srv2 dyn2:80 minconn 100 maxconn 1000
3987
3988 See also : "maxconn", "server"
3989
3990
3991grace <time>
3992 Maintain a proxy operational for some time after a soft stop
3993 May be used in sections : defaults | frontend | listen | backend
Cyril Bonté99ed3272010-01-24 23:29:44 +01003994 yes | yes | yes | yes
Willy Tarreau2769aa02007-12-27 18:26:09 +01003995 Arguments :
3996 <time> is the time (by default in milliseconds) for which the instance
3997 will remain operational with the frontend sockets still listening
3998 when a soft-stop is received via the SIGUSR1 signal.
3999
4000 This may be used to ensure that the services disappear in a certain order.
4001 This was designed so that frontends which are dedicated to monitoring by an
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004002 external equipment fail immediately while other ones remain up for the time
Willy Tarreau2769aa02007-12-27 18:26:09 +01004003 needed by the equipment to detect the failure.
4004
4005 Note that currently, there is very little benefit in using this parameter,
4006 and it may in fact complicate the soft-reconfiguration process more than
4007 simplify it.
4008
Willy Tarreau0ba27502007-12-24 16:55:16 +01004009
Andrew Rodland17be45e2016-10-25 17:04:12 -04004010hash-balance-factor <factor>
4011 Specify the balancing factor for bounded-load consistent hashing
4012 May be used in sections : defaults | frontend | listen | backend
4013 yes | no | no | yes
4014 Arguments :
4015 <factor> is the control for the maximum number of concurrent requests to
4016 send to a server, expressed as a percentage of the average number
Frédéric Lécaille93d33162019-03-06 09:35:59 +01004017 of concurrent requests across all of the active servers.
Andrew Rodland17be45e2016-10-25 17:04:12 -04004018
4019 Specifying a "hash-balance-factor" for a server with "hash-type consistent"
4020 enables an algorithm that prevents any one server from getting too many
4021 requests at once, even if some hash buckets receive many more requests than
4022 others. Setting <factor> to 0 (the default) disables the feature. Otherwise,
4023 <factor> is a percentage greater than 100. For example, if <factor> is 150,
4024 then no server will be allowed to have a load more than 1.5 times the average.
4025 If server weights are used, they will be respected.
4026
4027 If the first-choice server is disqualified, the algorithm will choose another
4028 server based on the request hash, until a server with additional capacity is
4029 found. A higher <factor> allows more imbalance between the servers, while a
4030 lower <factor> means that more servers will be checked on average, affecting
4031 performance. Reasonable values are from 125 to 200.
4032
Willy Tarreau760e81d2018-05-03 07:20:40 +02004033 This setting is also used by "balance random" which internally relies on the
4034 consistent hashing mechanism.
4035
Andrew Rodland17be45e2016-10-25 17:04:12 -04004036 See also : "balance" and "hash-type".
4037
4038
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004039hash-type <method> <function> <modifier>
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004040 Specify a method to use for mapping hashes to servers
4041 May be used in sections : defaults | frontend | listen | backend
4042 yes | no | yes | yes
4043 Arguments :
Bhaskar98634f02013-10-29 23:30:51 -04004044 <method> is the method used to select a server from the hash computed by
4045 the <function> :
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004046
Bhaskar98634f02013-10-29 23:30:51 -04004047 map-based the hash table is a static array containing all alive servers.
4048 The hashes will be very smooth, will consider weights, but
4049 will be static in that weight changes while a server is up
4050 will be ignored. This means that there will be no slow start.
4051 Also, since a server is selected by its position in the array,
4052 most mappings are changed when the server count changes. This
4053 means that when a server goes up or down, or when a server is
4054 added to a farm, most connections will be redistributed to
4055 different servers. This can be inconvenient with caches for
4056 instance.
Willy Tarreau798a39c2010-11-24 15:04:29 +01004057
Bhaskar98634f02013-10-29 23:30:51 -04004058 consistent the hash table is a tree filled with many occurrences of each
4059 server. The hash key is looked up in the tree and the closest
4060 server is chosen. This hash is dynamic, it supports changing
4061 weights while the servers are up, so it is compatible with the
4062 slow start feature. It has the advantage that when a server
4063 goes up or down, only its associations are moved. When a
4064 server is added to the farm, only a few part of the mappings
4065 are redistributed, making it an ideal method for caches.
4066 However, due to its principle, the distribution will never be
4067 very smooth and it may sometimes be necessary to adjust a
4068 server's weight or its ID to get a more balanced distribution.
4069 In order to get the same distribution on multiple load
4070 balancers, it is important that all servers have the exact
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004071 same IDs. Note: consistent hash uses sdbm and avalanche if no
4072 hash function is specified.
Bhaskar98634f02013-10-29 23:30:51 -04004073
4074 <function> is the hash function to be used :
4075
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03004076 sdbm this function was created initially for sdbm (a public-domain
Bhaskar98634f02013-10-29 23:30:51 -04004077 reimplementation of ndbm) database library. It was found to do
4078 well in scrambling bits, causing better distribution of the keys
4079 and fewer splits. It also happens to be a good general hashing
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004080 function with good distribution, unless the total server weight
4081 is a multiple of 64, in which case applying the avalanche
4082 modifier may help.
Bhaskar98634f02013-10-29 23:30:51 -04004083
4084 djb2 this function was first proposed by Dan Bernstein many years ago
4085 on comp.lang.c. Studies have shown that for certain workload this
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004086 function provides a better distribution than sdbm. It generally
4087 works well with text-based inputs though it can perform extremely
4088 poorly with numeric-only input or when the total server weight is
4089 a multiple of 33, unless the avalanche modifier is also used.
4090
Willy Tarreaua0f42712013-11-14 14:30:35 +01004091 wt6 this function was designed for haproxy while testing other
4092 functions in the past. It is not as smooth as the other ones, but
4093 is much less sensible to the input data set or to the number of
4094 servers. It can make sense as an alternative to sdbm+avalanche or
4095 djb2+avalanche for consistent hashing or when hashing on numeric
4096 data such as a source IP address or a visitor identifier in a URL
4097 parameter.
4098
Willy Tarreau324f07f2015-01-20 19:44:50 +01004099 crc32 this is the most common CRC32 implementation as used in Ethernet,
4100 gzip, PNG, etc. It is slower than the other ones but may provide
4101 a better distribution or less predictable results especially when
4102 used on strings.
4103
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004104 <modifier> indicates an optional method applied after hashing the key :
4105
4106 avalanche This directive indicates that the result from the hash
4107 function above should not be used in its raw form but that
4108 a 4-byte full avalanche hash must be applied first. The
4109 purpose of this step is to mix the resulting bits from the
4110 previous hash in order to avoid any undesired effect when
4111 the input contains some limited values or when the number of
4112 servers is a multiple of one of the hash's components (64
4113 for SDBM, 33 for DJB2). Enabling avalanche tends to make the
4114 result less predictable, but it's also not as smooth as when
4115 using the original function. Some testing might be needed
4116 with some workloads. This hash is one of the many proposed
4117 by Bob Jenkins.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004118
Bhaskar98634f02013-10-29 23:30:51 -04004119 The default hash type is "map-based" and is recommended for most usages. The
4120 default function is "sdbm", the selection of a function should be based on
4121 the range of the values being hashed.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004122
Andrew Rodland17be45e2016-10-25 17:04:12 -04004123 See also : "balance", "hash-balance-factor", "server"
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004124
4125
Willy Tarreau0ba27502007-12-24 16:55:16 +01004126http-check disable-on-404
4127 Enable a maintenance mode upon HTTP/404 response to health-checks
4128 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau2769aa02007-12-27 18:26:09 +01004129 yes | no | yes | yes
Willy Tarreau0ba27502007-12-24 16:55:16 +01004130 Arguments : none
4131
4132 When this option is set, a server which returns an HTTP code 404 will be
4133 excluded from further load-balancing, but will still receive persistent
4134 connections. This provides a very convenient method for Web administrators
4135 to perform a graceful shutdown of their servers. It is also important to note
4136 that a server which is detected as failed while it was in this mode will not
4137 generate an alert, just a notice. If the server responds 2xx or 3xx again, it
4138 will immediately be reinserted into the farm. The status on the stats page
4139 reports "NOLB" for a server in this mode. It is important to note that this
Willy Tarreaubd741542010-03-16 18:46:54 +01004140 option only works in conjunction with the "httpchk" option. If this option
4141 is used with "http-check expect", then it has precedence over it so that 404
4142 responses will still be considered as soft-stop.
4143
4144 See also : "option httpchk", "http-check expect"
4145
4146
4147http-check expect [!] <match> <pattern>
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004148 Make HTTP health checks consider response contents or specific status codes
Willy Tarreaubd741542010-03-16 18:46:54 +01004149 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau1ee51a62011-08-19 20:04:17 +02004150 yes | no | yes | yes
Willy Tarreaubd741542010-03-16 18:46:54 +01004151 Arguments :
4152 <match> is a keyword indicating how to look for a specific pattern in the
4153 response. The keyword may be one of "status", "rstatus",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004154 "string", or "rstring". The keyword may be preceded by an
Willy Tarreaubd741542010-03-16 18:46:54 +01004155 exclamation mark ("!") to negate the match. Spaces are allowed
4156 between the exclamation mark and the keyword. See below for more
4157 details on the supported keywords.
4158
4159 <pattern> is the pattern to look for. It may be a string or a regular
4160 expression. If the pattern contains spaces, they must be escaped
4161 with the usual backslash ('\').
4162
4163 By default, "option httpchk" considers that response statuses 2xx and 3xx
4164 are valid, and that others are invalid. When "http-check expect" is used,
4165 it defines what is considered valid or invalid. Only one "http-check"
4166 statement is supported in a backend. If a server fails to respond or times
4167 out, the check obviously fails. The available matches are :
4168
4169 status <string> : test the exact string match for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004170 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01004171 response's status code is exactly this string. If the
4172 "status" keyword is prefixed with "!", then the response
4173 will be considered invalid if the status code matches.
4174
4175 rstatus <regex> : test a regular expression for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004176 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01004177 response's status code matches the expression. If the
4178 "rstatus" keyword is prefixed with "!", then the response
4179 will be considered invalid if the status code matches.
4180 This is mostly used to check for multiple codes.
4181
4182 string <string> : test the exact string match in the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004183 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01004184 response's body contains this exact string. If the
4185 "string" keyword is prefixed with "!", then the response
4186 will be considered invalid if the body contains this
4187 string. This can be used to look for a mandatory word at
4188 the end of a dynamic page, or to detect a failure when a
Davor Ocelice9ed2812017-12-25 17:49:28 +01004189 specific error appears on the check page (e.g. a stack
Willy Tarreaubd741542010-03-16 18:46:54 +01004190 trace).
4191
4192 rstring <regex> : test a regular expression on the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004193 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01004194 response's body matches this expression. If the "rstring"
4195 keyword is prefixed with "!", then the response will be
4196 considered invalid if the body matches the expression.
4197 This can be used to look for a mandatory word at the end
4198 of a dynamic page, or to detect a failure when a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01004199 error appears on the check page (e.g. a stack trace).
Willy Tarreaubd741542010-03-16 18:46:54 +01004200
4201 It is important to note that the responses will be limited to a certain size
4202 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
4203 Thus, too large responses may not contain the mandatory pattern when using
4204 "string" or "rstring". If a large response is absolutely required, it is
4205 possible to change the default max size by setting the global variable.
4206 However, it is worth keeping in mind that parsing very large responses can
4207 waste some CPU cycles, especially when regular expressions are used, and that
4208 it is always better to focus the checks on smaller resources.
4209
Cyril Bonté32602d22015-01-30 00:07:07 +01004210 Also "http-check expect" doesn't support HTTP keep-alive. Keep in mind that it
4211 will automatically append a "Connection: close" header, meaning that this
4212 header should not be present in the request provided by "option httpchk".
4213
Willy Tarreaubd741542010-03-16 18:46:54 +01004214 Last, if "http-check expect" is combined with "http-check disable-on-404",
4215 then this last one has precedence when the server responds with 404.
4216
4217 Examples :
4218 # only accept status 200 as valid
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01004219 http-check expect status 200
Willy Tarreaubd741542010-03-16 18:46:54 +01004220
4221 # consider SQL errors as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01004222 http-check expect ! string SQL\ Error
Willy Tarreaubd741542010-03-16 18:46:54 +01004223
4224 # consider status 5xx only as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01004225 http-check expect ! rstatus ^5
Willy Tarreaubd741542010-03-16 18:46:54 +01004226
4227 # check that we have a correct hexadecimal tag before /html
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03004228 http-check expect rstring <!--tag:[0-9a-f]*--></html>
Willy Tarreau0ba27502007-12-24 16:55:16 +01004229
Willy Tarreaubd741542010-03-16 18:46:54 +01004230 See also : "option httpchk", "http-check disable-on-404"
Willy Tarreau2769aa02007-12-27 18:26:09 +01004231
4232
Willy Tarreauef781042010-01-27 11:53:01 +01004233http-check send-state
4234 Enable emission of a state header with HTTP health checks
4235 May be used in sections : defaults | frontend | listen | backend
4236 yes | no | yes | yes
4237 Arguments : none
4238
4239 When this option is set, haproxy will systematically send a special header
4240 "X-Haproxy-Server-State" with a list of parameters indicating to each server
4241 how they are seen by haproxy. This can be used for instance when a server is
4242 manipulated without access to haproxy and the operator needs to know whether
4243 haproxy still sees it up or not, or if the server is the last one in a farm.
4244
4245 The header is composed of fields delimited by semi-colons, the first of which
4246 is a word ("UP", "DOWN", "NOLB"), possibly followed by a number of valid
4247 checks on the total number before transition, just as appears in the stats
4248 interface. Next headers are in the form "<variable>=<value>", indicating in
4249 no specific order some values available in the stats interface :
Joseph Lynch514061c2015-01-15 17:52:59 -08004250 - a variable "address", containing the address of the backend server.
4251 This corresponds to the <address> field in the server declaration. For
4252 unix domain sockets, it will read "unix".
4253
4254 - a variable "port", containing the port of the backend server. This
4255 corresponds to the <port> field in the server declaration. For unix
4256 domain sockets, it will read "unix".
4257
Willy Tarreauef781042010-01-27 11:53:01 +01004258 - a variable "name", containing the name of the backend followed by a slash
4259 ("/") then the name of the server. This can be used when a server is
4260 checked in multiple backends.
4261
4262 - a variable "node" containing the name of the haproxy node, as set in the
4263 global "node" variable, otherwise the system's hostname if unspecified.
4264
4265 - a variable "weight" indicating the weight of the server, a slash ("/")
4266 and the total weight of the farm (just counting usable servers). This
4267 helps to know if other servers are available to handle the load when this
4268 one fails.
4269
4270 - a variable "scur" indicating the current number of concurrent connections
4271 on the server, followed by a slash ("/") then the total number of
4272 connections on all servers of the same backend.
4273
4274 - a variable "qcur" indicating the current number of requests in the
4275 server's queue.
4276
4277 Example of a header received by the application server :
4278 >>> X-Haproxy-Server-State: UP 2/3; name=bck/srv2; node=lb1; weight=1/2; \
4279 scur=13/22; qcur=0
4280
4281 See also : "option httpchk", "http-check disable-on-404"
4282
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004283
4284http-request <action> [options...] [ { if | unless } <condition> ]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004285 Access control for Layer 7 requests
4286
4287 May be used in sections: defaults | frontend | listen | backend
4288 no | yes | yes | yes
4289
Willy Tarreau20b0de52012-12-24 15:45:22 +01004290 The http-request statement defines a set of rules which apply to layer 7
4291 processing. The rules are evaluated in their declaration order when they are
4292 met in a frontend, listen or backend section. Any rule may optionally be
4293 followed by an ACL-based condition, in which case it will only be evaluated
4294 if the condition is true.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004295
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004296 The first keyword is the rule's action. The supported actions are described
4297 below.
Willy Tarreau20b0de52012-12-24 15:45:22 +01004298
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004299 There is no limit to the number of http-request statements per instance.
Willy Tarreau20b0de52012-12-24 15:45:22 +01004300
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004301 Example:
4302 acl nagios src 192.168.129.3
4303 acl local_net src 192.168.0.0/16
4304 acl auth_ok http_auth(L1)
Willy Tarreau20b0de52012-12-24 15:45:22 +01004305
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004306 http-request allow if nagios
4307 http-request allow if local_net auth_ok
4308 http-request auth realm Gimme if local_net auth_ok
4309 http-request deny
Willy Tarreau81499eb2012-12-27 12:19:02 +01004310
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004311 Example:
4312 acl key req.hdr(X-Add-Acl-Key) -m found
4313 acl add path /addacl
4314 acl del path /delacl
Willy Tarreau20b0de52012-12-24 15:45:22 +01004315
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004316 acl myhost hdr(Host) -f myhost.lst
Willy Tarreau20b0de52012-12-24 15:45:22 +01004317
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004318 http-request add-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key add
4319 http-request del-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key del
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02004320
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004321 Example:
4322 acl value req.hdr(X-Value) -m found
4323 acl setmap path /setmap
4324 acl delmap path /delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06004325
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004326 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06004327
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004328 http-request set-map(map.lst) %[src] %[req.hdr(X-Value)] if setmap value
4329 http-request del-map(map.lst) %[src] if delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06004330
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004331 See also : "stats http-request", section 3.4 about userlists and section 7
4332 about ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06004333
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004334http-request add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004335
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004336 This is used to add a new entry into an ACL. The ACL must be loaded from a
4337 file (even a dummy empty file). The file name of the ACL to be updated is
4338 passed between parentheses. It takes one argument: <key fmt>, which follows
4339 log-format rules, to collect content of the new entry. It performs a lookup
4340 in the ACL before insertion, to avoid duplicated (or more) values. This
4341 lookup is done by a linear search and can be expensive with large lists!
4342 It is the equivalent of the "add acl" command from the stats socket, but can
4343 be triggered by an HTTP request.
Sasha Pachev218f0642014-06-16 12:05:59 -06004344
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004345http-request add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004346
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004347 This appends an HTTP header field whose name is specified in <name> and
4348 whose value is defined by <fmt> which follows the log-format rules (see
4349 Custom Log Format in section 8.2.4). This is particularly useful to pass
4350 connection-specific information to the server (e.g. the client's SSL
4351 certificate), or to combine several headers into one. This rule is not
4352 final, so it is possible to add other similar rules. Note that header
4353 addition is performed immediately, so one rule might reuse the resulting
4354 header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06004355
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004356http-request allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004357
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004358 This stops the evaluation of the rules and lets the request pass the check.
4359 No further "http-request" rules are evaluated.
Sasha Pachev218f0642014-06-16 12:05:59 -06004360
Sasha Pachev218f0642014-06-16 12:05:59 -06004361
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004362http-request auth [realm <realm>] [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004363
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004364 This stops the evaluation of the rules and immediately responds with an
4365 HTTP 401 or 407 error code to invite the user to present a valid user name
4366 and password. No further "http-request" rules are evaluated. An optional
4367 "realm" parameter is supported, it sets the authentication realm that is
4368 returned with the response (typically the application's name).
Sasha Pachev218f0642014-06-16 12:05:59 -06004369
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004370 Example:
4371 acl auth_ok http_auth_group(L1) G1
4372 http-request auth unless auth_ok
Sasha Pachev218f0642014-06-16 12:05:59 -06004373
Jarno Huuskonen251a6b72019-01-04 14:05:02 +02004374http-request cache-use <name> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004375
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02004376 See section 6.2 about cache setup.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004377
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004378http-request capture <sample> [ len <length> | id <id> ]
4379 [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004380
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004381 This captures sample expression <sample> from the request buffer, and
4382 converts it to a string of at most <len> characters. The resulting string is
4383 stored into the next request "capture" slot, so it will possibly appear next
4384 to some captured HTTP headers. It will then automatically appear in the logs,
4385 and it will be possible to extract it using sample fetch rules to feed it
4386 into headers or anything. The length should be limited given that this size
4387 will be allocated for each capture during the whole session life.
4388 Please check section 7.3 (Fetching samples) and "capture request header" for
4389 more information.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004390
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004391 If the keyword "id" is used instead of "len", the action tries to store the
4392 captured string in a previously declared capture slot. This is useful to run
4393 captures in backends. The slot id can be declared by a previous directive
Baptiste Assmann19a69b32020-01-16 14:34:22 +01004394 "http-request capture" or with the "declare capture" keyword.
4395
4396 When using this action in a backend, double check that the relevant
4397 frontend(s) have the required capture slots otherwise, this rule will be
4398 ignored at run time. This can't be detected at configuration parsing time
4399 due to HAProxy's ability to dynamically resolve backend name at runtime.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004400
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004401http-request del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004402
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004403 This is used to delete an entry from an ACL. The ACL must be loaded from a
4404 file (even a dummy empty file). The file name of the ACL to be updated is
4405 passed between parentheses. It takes one argument: <key fmt>, which follows
4406 log-format rules, to collect content of the entry to delete.
4407 It is the equivalent of the "del acl" command from the stats socket, but can
4408 be triggered by an HTTP request.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004409
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004410http-request del-header <name> [ { if | unless } <condition> ]
Willy Tarreauf4c43c12013-06-11 17:01:13 +02004411
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004412 This removes all HTTP header fields whose name is specified in <name>.
Willy Tarreau9a355ec2013-06-11 17:45:46 +02004413
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004414http-request del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau42cf39e2013-06-11 18:51:32 +02004415
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004416 This is used to delete an entry from a MAP. The MAP must be loaded from a
4417 file (even a dummy empty file). The file name of the MAP to be updated is
4418 passed between parentheses. It takes one argument: <key fmt>, which follows
4419 log-format rules, to collect content of the entry to delete.
4420 It takes one argument: "file name" It is the equivalent of the "del map"
4421 command from the stats socket, but can be triggered by an HTTP request.
Willy Tarreau51347ed2013-06-11 19:34:13 +02004422
Christopher Faulet554c0eb2020-01-14 12:00:28 +01004423http-request deny [deny_status <status>] [ { errorfile | errorfiles } <err> ]
4424 [ { if | unless } <condition> ]
Patrick Hemmer268a7072018-05-11 12:52:31 -04004425
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004426 This stops the evaluation of the rules and immediately rejects the request
4427 and emits an HTTP 403 error, or optionally the status code specified as an
4428 argument to "deny_status". The list of permitted status codes is limited to
Christopher Faulet554c0eb2020-01-14 12:00:28 +01004429 those that can be overridden by the "errorfile" directive. A specific error
4430 message may be specified. It may be an error file, using the "errorfile"
4431 keyword followed by the file containing the full HTTP response. It may also
4432 be an error from an http-errors section, using the "errorfiles" keyword
4433 followed by the section name.
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004434 No further "http-request" rules are evaluated.
Patrick Hemmer268a7072018-05-11 12:52:31 -04004435
Olivier Houchard602bf7d2019-05-10 13:59:15 +02004436http-request disable-l7-retry [ { if | unless } <condition> ]
4437 This disables any attempt to retry the request if it fails for any other
4438 reason than a connection failure. This can be useful for example to make
4439 sure POST requests aren't retried on failure.
4440
Baptiste Assmann333939c2019-01-21 08:34:50 +01004441http-request do-resolve(<var>,<resolvers>,[ipv4,ipv6]) <expr> :
4442
4443 This action performs a DNS resolution of the output of <expr> and stores
4444 the result in the variable <var>. It uses the DNS resolvers section
4445 pointed by <resolvers>.
4446 It is possible to choose a resolution preference using the optional
4447 arguments 'ipv4' or 'ipv6'.
4448 When performing the DNS resolution, the client side connection is on
4449 pause waiting till the end of the resolution.
4450 If an IP address can be found, it is stored into <var>. If any kind of
4451 error occurs, then <var> is not set.
4452 One can use this action to discover a server IP address at run time and
4453 based on information found in the request (IE a Host header).
4454 If this action is used to find the server's IP address (using the
4455 "set-dst" action), then the server IP address in the backend must be set
4456 to 0.0.0.0.
4457
4458 Example:
4459 resolvers mydns
4460 nameserver local 127.0.0.53:53
4461 nameserver google 8.8.8.8:53
4462 timeout retry 1s
4463 hold valid 10s
4464 hold nx 3s
4465 hold other 3s
4466 hold obsolete 0s
4467 accepted_payload_size 8192
4468
4469 frontend fe
4470 bind 10.42.0.1:80
4471 http-request do-resolve(txn.myip,mydns,ipv4) hdr(Host),lower
4472 http-request capture var(txn.myip) len 40
4473
4474 # return 503 when the variable is not set,
4475 # which mean DNS resolution error
4476 use_backend b_503 unless { var(txn.myip) -m found }
4477
4478 default_backend be
4479
4480 backend b_503
4481 # dummy backend used to return 503.
4482 # one can use the errorfile directive to send a nice
4483 # 503 error page to end users
4484
4485 backend be
4486 # rule to prevent HAProxy from reconnecting to services
4487 # on the local network (forged DNS name used to scan the network)
4488 http-request deny if { var(txn.myip) -m ip 127.0.0.0/8 10.0.0.0/8 }
4489 http-request set-dst var(txn.myip)
4490 server clear 0.0.0.0:0
4491
4492 NOTE: Don't forget to set the "protection" rules to ensure HAProxy won't
4493 be used to scan the network or worst won't loop over itself...
4494
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01004495http-request early-hint <name> <fmt> [ { if | unless } <condition> ]
4496
4497 This is used to build an HTTP 103 Early Hints response prior to any other one.
4498 This appends an HTTP header field to this response whose name is specified in
4499 <name> and whose value is defined by <fmt> which follows the log-format rules
4500 (see Custom Log Format in section 8.2.4). This is particularly useful to pass
Frédéric Lécaille3aac1062018-11-13 09:42:13 +01004501 to the client some Link headers to preload resources required to render the
4502 HTML documents.
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01004503
4504 See RFC 8297 for more information.
4505
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004506http-request redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004507
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004508 This performs an HTTP redirection based on a redirect rule. This is exactly
4509 the same as the "redirect" statement except that it inserts a redirect rule
4510 which can be processed in the middle of other "http-request" rules and that
4511 these rules use the "log-format" strings. See the "redirect" keyword for the
4512 rule's syntax.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004513
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004514http-request reject [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004515
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004516 This stops the evaluation of the rules and immediately closes the connection
4517 without sending any response. It acts similarly to the
4518 "tcp-request content reject" rules. It can be useful to force an immediate
4519 connection closure on HTTP/2 connections.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004520
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004521http-request replace-header <name> <match-regex> <replace-fmt>
4522 [ { if | unless } <condition> ]
Willy Tarreaua9083d02015-05-08 15:27:59 +02004523
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004524 This matches the value of all occurences of header field <name> against
4525 <match-regex>. Matching is performed case-sensitively. Matching values are
4526 completely replaced by <replace-fmt>. Format characters are allowed in
4527 <replace-fmt> and work like <fmt> arguments in "http-request add-header".
4528 Standard back-references using the backslash ('\') followed by a number are
4529 supported.
Thierry FOURNIER82bf70d2015-05-26 17:58:29 +02004530
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004531 This action acts on whole header lines, regardless of the number of values
4532 they may contain. Thus it is well-suited to process headers naturally
4533 containing commas in their value, such as If-Modified-Since. Headers that
4534 contain a comma-separated list of values, such as Accept, should be processed
4535 using "http-request replace-value".
William Lallemand86d0df02017-11-24 21:36:45 +01004536
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004537 Example:
4538 http-request replace-header Cookie foo=([^;]*);(.*) foo=\1;ip=%bi;\2
4539
4540 # applied to:
4541 Cookie: foo=foobar; expires=Tue, 14-Jun-2016 01:40:45 GMT;
4542
4543 # outputs:
4544 Cookie: foo=foobar;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT;
4545
4546 # assuming the backend IP is 192.168.1.20
Willy Tarreau09448f72014-06-25 18:12:15 +02004547
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004548 http-request replace-header User-Agent curl foo
4549
4550 # applied to:
4551 User-Agent: curl/7.47.0
Willy Tarreau09448f72014-06-25 18:12:15 +02004552
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004553 # outputs:
4554 User-Agent: foo
Willy Tarreau09448f72014-06-25 18:12:15 +02004555
Willy Tarreau262c3f12019-12-17 06:52:51 +01004556http-request replace-path <match-regex> <replace-fmt>
4557 [ { if | unless } <condition> ]
4558
4559 This works like "replace-header" except that it works on the request's path
4560 component instead of a header. The path component starts at the first '/'
4561 after an optional scheme+authority. It does contain the query string if any
4562 is present. The replacement does not modify the scheme nor authority.
4563
4564 It is worth noting that regular expressions may be more expensive to evaluate
4565 than certain ACLs, so rare replacements may benefit from a condition to avoid
4566 performing the evaluation at all if it does not match.
4567
4568 Example:
4569 # prefix /foo : turn /bar?q=1 into /foo/bar?q=1 :
4570 http-request replace-path (.*) /foo\1
4571
4572 # suffix /foo : turn /bar?q=1 into /bar/foo?q=1 :
4573 http-request replace-path ([^?]*)(\?(.*))? \1/foo\2
4574
4575 # strip /foo : turn /foo/bar?q=1 into /bar?q=1
4576 http-request replace-path /foo/(.*) /\1
4577 # or more efficient if only some requests match :
4578 http-request replace-path /foo/(.*) /\1 if { url_beg /foo/ }
4579
Willy Tarreau33810222019-06-12 17:44:02 +02004580http-request replace-uri <match-regex> <replace-fmt>
4581 [ { if | unless } <condition> ]
4582
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004583 This works like "replace-header" except that it works on the request's URI part
4584 instead of a header. The URI part may contain an optional scheme, authority or
4585 query string. These are considered to be part of the value that is matched
4586 against.
4587
4588 It is worth noting that regular expressions may be more expensive to evaluate
4589 than certain ACLs, so rare replacements may benefit from a condition to avoid
4590 performing the evaluation at all if it does not match.
Willy Tarreau33810222019-06-12 17:44:02 +02004591
Willy Tarreau62b59132019-12-17 06:51:20 +01004592 IMPORTANT NOTE: historically in HTTP/1.x, the vast majority of requests sent
4593 by browsers use the "origin form", which differs from the "absolute form" in
4594 that they do not contain a scheme nor authority in the URI portion. Mostly
4595 only requests sent to proxies, those forged by hand and some emitted by
4596 certain applications use the absolute form. As such, "replace-uri" usually
4597 works fine most of the time in HTTP/1.x with rules starting with a "/". But
4598 with HTTP/2, clients are encouraged to send absolute URIs only, which look
4599 like the ones HTTP/1 clients use to talk to proxies. Such partial replace-uri
4600 rules may then fail in HTTP/2 when they work in HTTP/1. Either the rules need
Willy Tarreau262c3f12019-12-17 06:52:51 +01004601 to be adapted to optionally match a scheme and authority, or replace-path
4602 should be used.
Willy Tarreau33810222019-06-12 17:44:02 +02004603
Willy Tarreau62b59132019-12-17 06:51:20 +01004604 Example:
4605 # rewrite all "http" absolute requests to "https":
4606 http-request replace-uri ^http://(.*) https://\1
Willy Tarreau33810222019-06-12 17:44:02 +02004607
Willy Tarreau62b59132019-12-17 06:51:20 +01004608 # prefix /foo : turn /bar?q=1 into /foo/bar?q=1 :
4609 http-request replace-uri ([^/:]*://[^/]*)?(.*) \1/foo\2
Willy Tarreau33810222019-06-12 17:44:02 +02004610
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004611http-request replace-value <name> <match-regex> <replace-fmt>
4612 [ { if | unless } <condition> ]
Willy Tarreau09448f72014-06-25 18:12:15 +02004613
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004614 This works like "replace-header" except that it matches the regex against
4615 every comma-delimited value of the header field <name> instead of the
4616 entire header. This is suited for all headers which are allowed to carry
4617 more than one value. An example could be the Accept header.
Willy Tarreau09448f72014-06-25 18:12:15 +02004618
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004619 Example:
4620 http-request replace-value X-Forwarded-For ^192\.168\.(.*)$ 172.16.\1
Thierry FOURNIER236657b2015-08-19 08:25:14 +02004621
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004622 # applied to:
4623 X-Forwarded-For: 192.168.10.1, 192.168.13.24, 10.0.0.37
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02004624
Tim Duesterhus2252beb2019-10-29 00:05:13 +01004625 # outputs:
4626 X-Forwarded-For: 172.16.10.1, 172.16.13.24, 10.0.0.37
Frédéric Lécaille6778b272018-01-29 15:22:53 +01004627
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004628http-request sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
4629http-request sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004630
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004631 This actions increments the GPC0 or GPC1 counter according with the sticky
4632 counter designated by <sc-id>. If an error occurs, this action silently fails
4633 and the actions evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004634
Cédric Dufour0d7712d2019-11-06 18:38:53 +01004635http-request sc-set-gpt0(<sc-id>) { <int> | <expr> }
4636 [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004637
Cédric Dufour0d7712d2019-11-06 18:38:53 +01004638 This action sets the 32-bit unsigned GPT0 tag according to the sticky counter
4639 designated by <sc-id> and the value of <int>/<expr>. The expected result is a
4640 boolean. If an error occurs, this action silently fails and the actions
4641 evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004642
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004643http-request set-dst <expr> [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004644
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004645 This is used to set the destination IP address to the value of specified
4646 expression. Useful when a proxy in front of HAProxy rewrites destination IP,
4647 but provides the correct IP in a HTTP header; or you want to mask the IP for
4648 privacy. If you want to connect to the new address/port, use '0.0.0.0:0' as a
4649 server address in the backend.
Christopher Faulet85d79c92016-11-09 16:54:56 +01004650
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004651 Arguments:
4652 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
4653 by some converters.
Christopher Faulet85d79c92016-11-09 16:54:56 +01004654
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004655 Example:
4656 http-request set-dst hdr(x-dst)
4657 http-request set-dst dst,ipmask(24)
Christopher Faulet85d79c92016-11-09 16:54:56 +01004658
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004659 When possible, set-dst preserves the original destination port as long as the
4660 address family allows it, otherwise the destination port is set to 0.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004661
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004662http-request set-dst-port <expr> [ { if | unless } <condition> ]
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004663
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004664 This is used to set the destination port address to the value of specified
4665 expression. If you want to connect to the new address/port, use '0.0.0.0:0'
4666 as a server address in the backend.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004667
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004668 Arguments:
4669 <expr> Is a standard HAProxy expression formed by a sample-fetch
4670 followed by some converters.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004671
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004672 Example:
4673 http-request set-dst-port hdr(x-port)
4674 http-request set-dst-port int(4000)
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004675
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004676 When possible, set-dst-port preserves the original destination address as
4677 long as the address family supports a port, otherwise it forces the
4678 destination address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +02004679
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004680http-request set-header <name> <fmt> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02004681
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004682 This does the same as "http-request add-header" except that the header name
4683 is first removed if it existed. This is useful when passing security
4684 information to the server, where the header must not be manipulated by
4685 external users. Note that the new value is computed before the removal so it
4686 is possible to concatenate a value to an existing header.
William Lallemand44be6402016-05-25 01:51:35 +02004687
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004688 Example:
4689 http-request set-header X-Haproxy-Current-Date %T
4690 http-request set-header X-SSL %[ssl_fc]
4691 http-request set-header X-SSL-Session_ID %[ssl_fc_session_id,hex]
4692 http-request set-header X-SSL-Client-Verify %[ssl_c_verify]
4693 http-request set-header X-SSL-Client-DN %{+Q}[ssl_c_s_dn]
4694 http-request set-header X-SSL-Client-CN %{+Q}[ssl_c_s_dn(cn)]
4695 http-request set-header X-SSL-Issuer %{+Q}[ssl_c_i_dn]
4696 http-request set-header X-SSL-Client-NotBefore %{+Q}[ssl_c_notbefore]
4697 http-request set-header X-SSL-Client-NotAfter %{+Q}[ssl_c_notafter]
William Lallemand44be6402016-05-25 01:51:35 +02004698
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004699http-request set-log-level <level> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02004700
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004701 This is used to change the log level of the current request when a certain
4702 condition is met. Valid levels are the 8 syslog levels (see the "log"
4703 keyword) plus the special level "silent" which disables logging for this
4704 request. This rule is not final so the last matching rule wins. This rule
4705 can be useful to disable health checks coming from another equipment.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004706
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004707http-request set-map(<file-name>) <key fmt> <value fmt>
4708 [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004709
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004710 This is used to add a new entry into a MAP. The MAP must be loaded from a
4711 file (even a dummy empty file). The file name of the MAP to be updated is
4712 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
4713 log-format rules, used to collect MAP key, and <value fmt>, which follows
4714 log-format rules, used to collect content for the new entry.
4715 It performs a lookup in the MAP before insertion, to avoid duplicated (or
4716 more) values. This lookup is done by a linear search and can be expensive
4717 with large lists! It is the equivalent of the "set map" command from the
4718 stats socket, but can be triggered by an HTTP request.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004719
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004720http-request set-mark <mark> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004721
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004722 This is used to set the Netfilter MARK on all packets sent to the client to
4723 the value passed in <mark> on platforms which support it. This value is an
4724 unsigned 32 bit value which can be matched by netfilter and by the routing
4725 table. It can be expressed both in decimal or hexadecimal format (prefixed by
4726 "0x"). This can be useful to force certain packets to take a different route
4727 (for example a cheaper network path for bulk downloads). This works on Linux
4728 kernels 2.6.32 and above and requires admin privileges.
Willy Tarreau00005ce2016-10-21 15:07:45 +02004729
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004730http-request set-method <fmt> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004731
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004732 This rewrites the request method with the result of the evaluation of format
4733 string <fmt>. There should be very few valid reasons for having to do so as
4734 this is more likely to break something than to fix it.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004735
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004736http-request set-nice <nice> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004737
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004738 This sets the "nice" factor of the current request being processed. It only
4739 has effect against the other requests being processed at the same time.
4740 The default value is 0, unless altered by the "nice" setting on the "bind"
4741 line. The accepted range is -1024..1024. The higher the value, the nicest
4742 the request will be. Lower values will make the request more important than
4743 other ones. This can be useful to improve the speed of some requests, or
4744 lower the priority of non-important requests. Using this setting without
4745 prior experimentation can cause some major slowdown.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004746
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004747http-request set-path <fmt> [ { if | unless } <condition> ]
Willy Tarreau00005ce2016-10-21 15:07:45 +02004748
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004749 This rewrites the request path with the result of the evaluation of format
4750 string <fmt>. The query string, if any, is left intact. If a scheme and
4751 authority is found before the path, they are left intact as well. If the
4752 request doesn't have a path ("*"), this one is replaced with the format.
4753 This can be used to prepend a directory component in front of a path for
4754 example. See also "http-request set-query" and "http-request set-uri".
Willy Tarreau2d392c22015-08-24 01:43:45 +02004755
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004756 Example :
4757 # prepend the host name before the path
4758 http-request set-path /%[hdr(host)]%[path]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004759
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004760http-request set-priority-class <expr> [ { if | unless } <condition> ]
Olivier Houchardccaa7de2017-10-02 11:51:03 +02004761
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004762 This is used to set the queue priority class of the current request.
4763 The value must be a sample expression which converts to an integer in the
4764 range -2047..2047. Results outside this range will be truncated.
4765 The priority class determines the order in which queued requests are
4766 processed. Lower values have higher priority.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004767
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004768http-request set-priority-offset <expr> [ { if | unless } <condition> ]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004769
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004770 This is used to set the queue priority timestamp offset of the current
4771 request. The value must be a sample expression which converts to an integer
4772 in the range -524287..524287. Results outside this range will be truncated.
4773 When a request is queued, it is ordered first by the priority class, then by
4774 the current timestamp adjusted by the given offset in milliseconds. Lower
4775 values have higher priority.
4776 Note that the resulting timestamp is is only tracked with enough precision
4777 for 524,287ms (8m44s287ms). If the request is queued long enough to where the
4778 adjusted timestamp exceeds this value, it will be misidentified as highest
4779 priority. Thus it is important to set "timeout queue" to a value, where when
4780 combined with the offset, does not exceed this limit.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004781
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004782http-request set-query <fmt> [ { if | unless } <condition> ]
Willy Tarreau20b0de52012-12-24 15:45:22 +01004783
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004784 This rewrites the request's query string which appears after the first
4785 question mark ("?") with the result of the evaluation of format string <fmt>.
4786 The part prior to the question mark is left intact. If the request doesn't
4787 contain a question mark and the new value is not empty, then one is added at
4788 the end of the URI, followed by the new value. If a question mark was
4789 present, it will never be removed even if the value is empty. This can be
4790 used to add or remove parameters from the query string.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08004791
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004792 See also "http-request set-query" and "http-request set-uri".
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004793
4794 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004795 # replace "%3D" with "=" in the query string
4796 http-request set-query %[query,regsub(%3D,=,g)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004797
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004798http-request set-src <expr> [ { if | unless } <condition> ]
4799 This is used to set the source IP address to the value of specified
4800 expression. Useful when a proxy in front of HAProxy rewrites source IP, but
4801 provides the correct IP in a HTTP header; or you want to mask source IP for
4802 privacy.
4803
4804 Arguments :
4805 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
4806 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004807
Cyril Bonté78caf842010-03-10 22:41:43 +01004808 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004809 http-request set-src hdr(x-forwarded-for)
4810 http-request set-src src,ipmask(24)
4811
4812 When possible, set-src preserves the original source port as long as the
4813 address family allows it, otherwise the source port is set to 0.
4814
4815http-request set-src-port <expr> [ { if | unless } <condition> ]
4816
4817 This is used to set the source port address to the value of specified
4818 expression.
4819
4820 Arguments:
4821 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
4822 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004823
Willy Tarreau20b0de52012-12-24 15:45:22 +01004824 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004825 http-request set-src-port hdr(x-port)
4826 http-request set-src-port int(4000)
4827
4828 When possible, set-src-port preserves the original source address as long as
4829 the address family supports a port, otherwise it forces the source address to
4830 IPv4 "0.0.0.0" before rewriting the port.
4831
4832http-request set-tos <tos> [ { if | unless } <condition> ]
4833
4834 This is used to set the TOS or DSCP field value of packets sent to the client
4835 to the value passed in <tos> on platforms which support this. This value
4836 represents the whole 8 bits of the IP TOS field, and can be expressed both in
4837 decimal or hexadecimal format (prefixed by "0x"). Note that only the 6 higher
4838 bits are used in DSCP or TOS, and the two lower bits are always 0. This can
4839 be used to adjust some routing behavior on border routers based on some
4840 information from the request.
4841
4842 See RFC 2474, 2597, 3260 and 4594 for more information.
4843
4844http-request set-uri <fmt> [ { if | unless } <condition> ]
4845
4846 This rewrites the request URI with the result of the evaluation of format
4847 string <fmt>. The scheme, authority, path and query string are all replaced
4848 at once. This can be used to rewrite hosts in front of proxies, or to
4849 perform complex modifications to the URI such as moving parts between the
4850 path and the query string.
4851 See also "http-request set-path" and "http-request set-query".
4852
4853http-request set-var(<var-name>) <expr> [ { if | unless } <condition> ]
4854
4855 This is used to set the contents of a variable. The variable is declared
4856 inline.
4857
4858 Arguments:
4859 <var-name> The name of the variable starts with an indication about its
4860 scope. The scopes allowed are:
4861 "proc" : the variable is shared with the whole process
4862 "sess" : the variable is shared with the whole session
4863 "txn" : the variable is shared with the transaction
4864 (request and response)
4865 "req" : the variable is shared only during request
4866 processing
4867 "res" : the variable is shared only during response
4868 processing
4869 This prefix is followed by a name. The separator is a '.'.
4870 The name may only contain characters 'a-z', 'A-Z', '0-9'
4871 and '_'.
4872
4873 <expr> Is a standard HAProxy expression formed by a sample-fetch
4874 followed by some converters.
Willy Tarreau20b0de52012-12-24 15:45:22 +01004875
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004876 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004877 http-request set-var(req.my_var) req.fhdr(user-agent),lower
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004878
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004879http-request send-spoe-group <engine-name> <group-name>
4880 [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004881
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004882 This action is used to trigger sending of a group of SPOE messages. To do so,
4883 the SPOE engine used to send messages must be defined, as well as the SPOE
4884 group to send. Of course, the SPOE engine must refer to an existing SPOE
4885 filter. If not engine name is provided on the SPOE filter line, the SPOE
4886 agent name must be used.
4887
4888 Arguments:
4889 <engine-name> The SPOE engine name.
4890
4891 <group-name> The SPOE group name as specified in the engine
4892 configuration.
4893
4894http-request silent-drop [ { if | unless } <condition> ]
4895
4896 This stops the evaluation of the rules and makes the client-facing connection
4897 suddenly disappear using a system-dependent way that tries to prevent the
4898 client from being notified. The effect it then that the client still sees an
4899 established connection while there's none on HAProxy. The purpose is to
4900 achieve a comparable effect to "tarpit" except that it doesn't use any local
4901 resource at all on the machine running HAProxy. It can resist much higher
4902 loads than "tarpit", and slow down stronger attackers. It is important to
4903 understand the impact of using this mechanism. All stateful equipment placed
4904 between the client and HAProxy (firewalls, proxies, load balancers) will also
4905 keep the established connection for a long time and may suffer from this
4906 action.
4907 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
4908 option is used to block the emission of a TCP reset. On other systems, the
4909 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
4910 router, though it's still delivered to local networks. Do not use it unless
4911 you fully understand how it works.
4912
Christopher Faulet46f95542019-12-20 10:07:22 +01004913http-request strict-mode { on | off }
4914
4915 This enables or disables the strict rewriting mode for following rules. It
4916 does not affect rules declared before it and it is only applicable on rules
4917 performing a rewrite on the requests. When the strict mode is enabled, any
4918 rewrite failure triggers an internal error. Otherwise, such errors are
4919 silently ignored. The purpose of the strict rewriting mode is to make some
4920 rewrites optionnal while others must be performed to continue the request
4921 processing.
4922
Christopher Faulet1aea50e2020-01-17 16:03:53 +01004923 By default, the strict rewriting mode is enabled. Its value is also reset
Christopher Faulet46f95542019-12-20 10:07:22 +01004924 when a ruleset evaluation ends. So, for instance, if you change the mode on
4925 the frontend, the default mode is restored when HAProxy starts the backend
4926 rules evaluation.
4927
Christopher Faulet554c0eb2020-01-14 12:00:28 +01004928http-request tarpit [deny_status <status>] [ { errorfile | errorfiles } <err> ]
4929 [ { if | unless } <condition> ]
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004930
4931 This stops the evaluation of the rules and immediately blocks the request
4932 without responding for a delay specified by "timeout tarpit" or
4933 "timeout connect" if the former is not set. After that delay, if the client
4934 is still connected, an HTTP error 500 (or optionally the status code
4935 specified as an argument to "deny_status") is returned so that the client
4936 does not suspect it has been tarpitted. Logs will report the flags "PT".
4937 The goal of the tarpit rule is to slow down robots during an attack when
4938 they're limited on the number of concurrent requests. It can be very
4939 efficient against very dumb robots, and will significantly reduce the load
4940 on firewalls compared to a "deny" rule. But when facing "correctly"
4941 developed robots, it can make things worse by forcing haproxy and the front
Christopher Faulet554c0eb2020-01-14 12:00:28 +01004942 firewall to support insane number of concurrent connections. A specific error
4943 message may be specified. It may be an error file, using the "errorfile"
4944 keyword followed by the file containing the full HTTP response. It may also
4945 be an error from an http-errors section, using the "errorfiles" keyword
4946 followed by the section name.
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004947 See also the "silent-drop" action.
4948
4949http-request track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
4950http-request track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
4951http-request track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
4952
4953 This enables tracking of sticky counters from current request. These rules do
4954 not stop evaluation and do not change default action. The number of counters
4955 that may be simultaneously tracked by the same connection is set in
4956 MAX_SESS_STKCTR at build time (reported in haproxy -vv) which defaults to 3,
4957 so the track-sc number is between 0 and (MAX_SESS_STCKTR-1). The first
4958 "track-sc0" rule executed enables tracking of the counters of the specified
4959 table as the first set. The first "track-sc1" rule executed enables tracking
4960 of the counters of the specified table as the second set. The first
4961 "track-sc2" rule executed enables tracking of the counters of the specified
4962 table as the third set. It is a recommended practice to use the first set of
4963 counters for the per-frontend counters and the second set for the per-backend
4964 ones. But this is just a guideline, all may be used everywhere.
4965
4966 Arguments :
4967 <key> is mandatory, and is a sample expression rule as described in
4968 section 7.3. It describes what elements of the incoming request or
4969 connection will be analyzed, extracted, combined, and used to
4970 select which table entry to update the counters.
4971
4972 <table> is an optional table to be used instead of the default one, which
4973 is the stick-table declared in the current proxy. All the counters
4974 for the matches and updates for the key will then be performed in
4975 that table until the session ends.
4976
4977 Once a "track-sc*" rule is executed, the key is looked up in the table and if
4978 it is not found, an entry is allocated for it. Then a pointer to that entry
4979 is kept during all the session's life, and this entry's counters are updated
4980 as often as possible, every time the session's counters are updated, and also
4981 systematically when the session ends. Counters are only updated for events
4982 that happen after the tracking has been started. As an exception, connection
4983 counters and request counters are systematically updated so that they reflect
4984 useful information.
4985
4986 If the entry tracks concurrent connection counters, one connection is counted
4987 for as long as the entry is tracked, and the entry will not expire during
4988 that time. Tracking counters also provides a performance advantage over just
4989 checking the keys, because only one table lookup is performed for all ACL
4990 checks that make use of it.
4991
4992http-request unset-var(<var-name>) [ { if | unless } <condition> ]
4993
4994 This is used to unset a variable. See above for details about <var-name>.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004995
4996 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004997 http-request unset-var(req.my_var)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004998
Christopher Faulet579d83b2019-11-22 15:34:17 +01004999http-request use-service <service-name> [ { if | unless } <condition> ]
5000
5001 This directive executes the configured HTTP service to reply to the request
5002 and stops the evaluation of the rules. An HTTP service may choose to reply by
5003 sending any valid HTTP response or it may immediately close the connection
5004 without sending any response. Outside natives services, for instance the
5005 Prometheus exporter, it is possible to write your own services in Lua. No
5006 further "http-request" rules are evaluated.
5007
5008 Arguments :
5009 <service-name> is mandatory. It is the service to call
5010
5011 Example:
5012 http-request use-service prometheus-exporter if { path /metrics }
5013
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005014http-request wait-for-handshake [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005015
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005016 This will delay the processing of the request until the SSL handshake
5017 happened. This is mostly useful to delay processing early data until we're
5018 sure they are valid.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005019
Willy Tarreauef781042010-01-27 11:53:01 +01005020
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005021http-response <action> <options...> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02005022 Access control for Layer 7 responses
5023
5024 May be used in sections: defaults | frontend | listen | backend
5025 no | yes | yes | yes
5026
5027 The http-response statement defines a set of rules which apply to layer 7
5028 processing. The rules are evaluated in their declaration order when they are
5029 met in a frontend, listen or backend section. Any rule may optionally be
5030 followed by an ACL-based condition, in which case it will only be evaluated
5031 if the condition is true. Since these rules apply on responses, the backend
5032 rules are applied first, followed by the frontend's rules.
5033
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005034 The first keyword is the rule's action. The supported actions are described
5035 below.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02005036
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005037 There is no limit to the number of http-response statements per instance.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02005038
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005039 Example:
5040 acl key_acl res.hdr(X-Acl-Key) -m found
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02005041
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005042 acl myhost hdr(Host) -f myhost.lst
Sasha Pachev218f0642014-06-16 12:05:59 -06005043
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005044 http-response add-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
5045 http-response del-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
Sasha Pachev218f0642014-06-16 12:05:59 -06005046
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005047 Example:
5048 acl value res.hdr(X-Value) -m found
Sasha Pachev218f0642014-06-16 12:05:59 -06005049
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005050 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06005051
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005052 http-response set-map(map.lst) %[src] %[res.hdr(X-Value)] if value
5053 http-response del-map(map.lst) %[src] if ! value
Sasha Pachev218f0642014-06-16 12:05:59 -06005054
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005055 See also : "http-request", section 3.4 about userlists and section 7 about
5056 ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06005057
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005058http-response add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005059
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005060 This is used to add a new entry into an ACL. The ACL must be loaded from a
5061 file (even a dummy empty file). The file name of the ACL to be updated is
5062 passed between parentheses. It takes one argument: <key fmt>, which follows
5063 log-format rules, to collect content of the new entry. It performs a lookup
5064 in the ACL before insertion, to avoid duplicated (or more) values.
5065 This lookup is done by a linear search and can be expensive with large lists!
5066 It is the equivalent of the "add acl" command from the stats socket, but can
5067 be triggered by an HTTP response.
Sasha Pachev218f0642014-06-16 12:05:59 -06005068
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005069http-response add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005070
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005071 This appends an HTTP header field whose name is specified in <name> and whose
5072 value is defined by <fmt> which follows the log-format rules (see Custom Log
5073 Format in section 8.2.4). This may be used to send a cookie to a client for
5074 example, or to pass some internal information.
5075 This rule is not final, so it is possible to add other similar rules.
5076 Note that header addition is performed immediately, so one rule might reuse
5077 the resulting header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06005078
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005079http-response allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005080
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005081 This stops the evaluation of the rules and lets the response pass the check.
5082 No further "http-response" rules are evaluated for the current section.
Sasha Pachev218f0642014-06-16 12:05:59 -06005083
Jarno Huuskonen251a6b72019-01-04 14:05:02 +02005084http-response cache-store <name> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005085
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02005086 See section 6.2 about cache setup.
Sasha Pachev218f0642014-06-16 12:05:59 -06005087
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005088http-response capture <sample> id <id> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005089
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005090 This captures sample expression <sample> from the response buffer, and
5091 converts it to a string. The resulting string is stored into the next request
5092 "capture" slot, so it will possibly appear next to some captured HTTP
5093 headers. It will then automatically appear in the logs, and it will be
5094 possible to extract it using sample fetch rules to feed it into headers or
5095 anything. Please check section 7.3 (Fetching samples) and
5096 "capture response header" for more information.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02005097
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005098 The keyword "id" is the id of the capture slot which is used for storing the
5099 string. The capture slot must be defined in an associated frontend.
5100 This is useful to run captures in backends. The slot id can be declared by a
5101 previous directive "http-response capture" or with the "declare capture"
5102 keyword.
Baptiste Assmann19a69b32020-01-16 14:34:22 +01005103
5104 When using this action in a backend, double check that the relevant
5105 frontend(s) have the required capture slots otherwise, this rule will be
5106 ignored at run time. This can't be detected at configuration parsing time
5107 due to HAProxy's ability to dynamically resolve backend name at runtime.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02005108
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005109http-response del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02005110
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005111 This is used to delete an entry from an ACL. The ACL must be loaded from a
5112 file (even a dummy empty file). The file name of the ACL to be updated is
5113 passed between parentheses. It takes one argument: <key fmt>, which follows
5114 log-format rules, to collect content of the entry to delete.
5115 It is the equivalent of the "del acl" command from the stats socket, but can
5116 be triggered by an HTTP response.
Willy Tarreauf4c43c12013-06-11 17:01:13 +02005117
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005118http-response del-header <name> [ { if | unless } <condition> ]
Willy Tarreau9a355ec2013-06-11 17:45:46 +02005119
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005120 This removes all HTTP header fields whose name is specified in <name>.
Willy Tarreau42cf39e2013-06-11 18:51:32 +02005121
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005122http-response del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau51347ed2013-06-11 19:34:13 +02005123
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005124 This is used to delete an entry from a MAP. The MAP must be loaded from a
5125 file (even a dummy empty file). The file name of the MAP to be updated is
5126 passed between parentheses. It takes one argument: <key fmt>, which follows
5127 log-format rules, to collect content of the entry to delete.
5128 It takes one argument: "file name" It is the equivalent of the "del map"
5129 command from the stats socket, but can be triggered by an HTTP response.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005130
Christopher Faulet554c0eb2020-01-14 12:00:28 +01005131http-response deny [deny_status <status>] [ { errorfile | errorfiles } <err> ]
5132 [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005133
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005134 This stops the evaluation of the rules and immediately rejects the response
Christopher Faulet040c8cd2020-01-13 16:43:45 +01005135 and emits an HTTP 502 error, or optionally the status code specified as an
5136 argument to "deny_status". The list of permitted status codes is limited to
Christopher Faulet554c0eb2020-01-14 12:00:28 +01005137 those that can be overridden by the "errorfile" directive. A specific error
5138 message may be specified. It may be an error file, using the "errorfile"
5139 keyword followed by the file containing the full HTTP response. It may also
5140 be an error from an http-errors section, using the "errorfiles" keyword
5141 followed by the section name.
Christopher Faulet040c8cd2020-01-13 16:43:45 +01005142 No further "http-response" rules are evaluated.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005143
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005144http-response redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005145
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005146 This performs an HTTP redirection based on a redirect rule.
5147 This supports a format string similarly to "http-request redirect" rules,
5148 with the exception that only the "location" type of redirect is possible on
5149 the response. See the "redirect" keyword for the rule's syntax. When a
5150 redirect rule is applied during a response, connections to the server are
5151 closed so that no data can be forwarded from the server to the client.
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02005152
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005153http-response replace-header <name> <regex-match> <replace-fmt>
5154 [ { if | unless } <condition> ]
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02005155
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005156 This works like "http-request replace-header" except that it works on the
5157 server's response instead of the client's request.
William Lallemand86d0df02017-11-24 21:36:45 +01005158
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005159 Example:
5160 http-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2
Willy Tarreau51d861a2015-05-22 17:30:48 +02005161
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005162 # applied to:
5163 Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02005164
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005165 # outputs:
5166 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 +02005167
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005168 # assuming the backend IP is 192.168.1.20.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02005169
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005170http-response replace-value <name> <regex-match> <replace-fmt>
5171 [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02005172
Tim Duesterhus6bd909b2020-01-17 15:53:18 +01005173 This works like "http-request replace-value" except that it works on the
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005174 server's response instead of the client's request.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02005175
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005176 Example:
5177 http-response replace-value Cache-control ^public$ private
Christopher Faulet85d79c92016-11-09 16:54:56 +01005178
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005179 # applied to:
5180 Cache-Control: max-age=3600, public
Christopher Faulet85d79c92016-11-09 16:54:56 +01005181
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005182 # outputs:
5183 Cache-Control: max-age=3600, private
Christopher Faulet85d79c92016-11-09 16:54:56 +01005184
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005185http-response sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
5186http-response sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Ruoshan Huange4edc6b2016-07-14 15:07:45 +08005187
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005188 This action increments the GPC0 or GPC1 counter according with the sticky
5189 counter designated by <sc-id>. If an error occurs, this action silently fails
5190 and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +02005191
Cédric Dufour0d7712d2019-11-06 18:38:53 +01005192http-response sc-set-gpt0(<sc-id>) { <int> | <expr> }
5193 [ { if | unless } <condition> ]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02005194
Cédric Dufour0d7712d2019-11-06 18:38:53 +01005195 This action sets the 32-bit unsigned GPT0 tag according to the sticky counter
5196 designated by <sc-id> and the value of <int>/<expr>. The expected result is a
5197 boolean. If an error occurs, this action silently fails and the actions
5198 evaluation continues.
Frédéric Lécaille6778b272018-01-29 15:22:53 +01005199
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005200http-response send-spoe-group [ { if | unless } <condition> ]
Willy Tarreau2d392c22015-08-24 01:43:45 +02005201
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005202 This action is used to trigger sending of a group of SPOE messages. To do so,
5203 the SPOE engine used to send messages must be defined, as well as the SPOE
5204 group to send. Of course, the SPOE engine must refer to an existing SPOE
5205 filter. If not engine name is provided on the SPOE filter line, the SPOE
5206 agent name must be used.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02005207
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005208 Arguments:
5209 <engine-name> The SPOE engine name.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02005210
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005211 <group-name> The SPOE group name as specified in the engine
5212 configuration.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02005213
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005214http-response set-header <name> <fmt> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02005215
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005216 This does the same as "add-header" except that the header name is first
5217 removed if it existed. This is useful when passing security information to
5218 the server, where the header must not be manipulated by external users.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02005219
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005220http-response set-log-level <level> [ { if | unless } <condition> ]
5221
5222 This is used to change the log level of the current request when a certain
5223 condition is met. Valid levels are the 8 syslog levels (see the "log"
5224 keyword) plus the special level "silent" which disables logging for this
5225 request. This rule is not final so the last matching rule wins. This rule can
5226 be useful to disable health checks coming from another equipment.
5227
5228http-response set-map(<file-name>) <key fmt> <value fmt>
5229
5230 This is used to add a new entry into a MAP. The MAP must be loaded from a
5231 file (even a dummy empty file). The file name of the MAP to be updated is
5232 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
5233 log-format rules, used to collect MAP key, and <value fmt>, which follows
5234 log-format rules, used to collect content for the new entry. It performs a
5235 lookup in the MAP before insertion, to avoid duplicated (or more) values.
5236 This lookup is done by a linear search and can be expensive with large lists!
5237 It is the equivalent of the "set map" command from the stats socket, but can
5238 be triggered by an HTTP response.
5239
5240http-response set-mark <mark> [ { if | unless } <condition> ]
5241
5242 This is used to set the Netfilter MARK on all packets sent to the client to
5243 the value passed in <mark> on platforms which support it. This value is an
5244 unsigned 32 bit value which can be matched by netfilter and by the routing
5245 table. It can be expressed both in decimal or hexadecimal format (prefixed
5246 by "0x"). This can be useful to force certain packets to take a different
5247 route (for example a cheaper network path for bulk downloads). This works on
5248 Linux kernels 2.6.32 and above and requires admin privileges.
5249
5250http-response set-nice <nice> [ { if | unless } <condition> ]
5251
5252 This sets the "nice" factor of the current request being processed.
5253 It only has effect against the other requests being processed at the same
5254 time. The default value is 0, unless altered by the "nice" setting on the
5255 "bind" line. The accepted range is -1024..1024. The higher the value, the
5256 nicest the request will be. Lower values will make the request more important
5257 than other ones. This can be useful to improve the speed of some requests, or
5258 lower the priority of non-important requests. Using this setting without
5259 prior experimentation can cause some major slowdown.
5260
5261http-response set-status <status> [reason <str>]
5262 [ { if | unless } <condition> ]
5263
5264 This replaces the response status code with <status> which must be an integer
5265 between 100 and 999. Optionally, a custom reason text can be provided defined
5266 by <str>, or the default reason for the specified code will be used as a
5267 fallback.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08005268
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005269 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005270 # return "431 Request Header Fields Too Large"
5271 http-response set-status 431
5272 # return "503 Slow Down", custom reason
5273 http-response set-status 503 reason "Slow Down".
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005274
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005275http-response set-tos <tos> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005276
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005277 This is used to set the TOS or DSCP field value of packets sent to the client
5278 to the value passed in <tos> on platforms which support this.
5279 This value represents the whole 8 bits of the IP TOS field, and can be
5280 expressed both in decimal or hexadecimal format (prefixed by "0x"). Note that
5281 only the 6 higher bits are used in DSCP or TOS, and the two lower bits are
5282 always 0. This can be used to adjust some routing behavior on border routers
5283 based on some information from the request.
5284
5285 See RFC 2474, 2597, 3260 and 4594 for more information.
5286
5287http-response set-var(<var-name>) <expr> [ { if | unless } <condition> ]
5288
5289 This is used to set the contents of a variable. The variable is declared
5290 inline.
5291
5292 Arguments:
5293 <var-name> The name of the variable starts with an indication about its
5294 scope. The scopes allowed are:
5295 "proc" : the variable is shared with the whole process
5296 "sess" : the variable is shared with the whole session
5297 "txn" : the variable is shared with the transaction
5298 (request and response)
5299 "req" : the variable is shared only during request
5300 processing
5301 "res" : the variable is shared only during response
5302 processing
5303 This prefix is followed by a name. The separator is a '.'.
5304 The name may only contain characters 'a-z', 'A-Z', '0-9', '.'
5305 and '_'.
5306
5307 <expr> Is a standard HAProxy expression formed by a sample-fetch
5308 followed by some converters.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005309
5310 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005311 http-response set-var(sess.last_redir) res.hdr(location)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005312
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005313http-response silent-drop [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005314
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005315 This stops the evaluation of the rules and makes the client-facing connection
5316 suddenly disappear using a system-dependent way that tries to prevent the
5317 client from being notified. The effect it then that the client still sees an
5318 established connection while there's none on HAProxy. The purpose is to
5319 achieve a comparable effect to "tarpit" except that it doesn't use any local
5320 resource at all on the machine running HAProxy. It can resist much higher
5321 loads than "tarpit", and slow down stronger attackers. It is important to
5322 understand the impact of using this mechanism. All stateful equipment placed
5323 between the client and HAProxy (firewalls, proxies, load balancers) will also
5324 keep the established connection for a long time and may suffer from this
5325 action.
5326 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
5327 option is used to block the emission of a TCP reset. On other systems, the
5328 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
5329 router, though it's still delivered to local networks. Do not use it unless
5330 you fully understand how it works.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005331
Christopher Faulet46f95542019-12-20 10:07:22 +01005332http-response strict-mode { on | off }
5333
5334 This enables or disables the strict rewriting mode for following rules. It
5335 does not affect rules declared before it and it is only applicable on rules
5336 performing a rewrite on the responses. When the strict mode is enabled, any
5337 rewrite failure triggers an internal error. Otherwise, such errors are
5338 silently ignored. The purpose of the strict rewriting mode is to make some
5339 rewrites optionnal while others must be performed to continue the response
5340 processing.
5341
Christopher Faulet1aea50e2020-01-17 16:03:53 +01005342 By default, the strict rewriting mode is enabled. Its value is also reset
Christopher Faulet46f95542019-12-20 10:07:22 +01005343 when a ruleset evaluation ends. So, for instance, if you change the mode on
5344 the bacnkend, the default mode is restored when HAProxy starts the frontend
5345 rules evaluation.
5346
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005347http-response track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
5348http-response track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
5349http-response track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02005350
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005351 This enables tracking of sticky counters from current response. Please refer
5352 to "http-request track-sc" for a complete description. The only difference
5353 from "http-request track-sc" is the <key> sample expression can only make use
5354 of samples in response (e.g. res.*, status etc.) and samples below Layer 6
5355 (e.g. SSL-related samples, see section 7.3.4). If the sample is not
5356 supported, haproxy will fail and warn while parsing the config.
5357
5358http-response unset-var(<var-name>) [ { if | unless } <condition> ]
5359
5360 This is used to unset a variable. See "http-response set-var" for details
5361 about <var-name>.
5362
5363 Example:
5364 http-response unset-var(sess.last_redir)
5365
Baptiste Assmann5ecb77f2013-10-06 23:24:13 +02005366
Willy Tarreau30631952015-08-06 15:05:24 +02005367http-reuse { never | safe | aggressive | always }
5368 Declare how idle HTTP connections may be shared between requests
5369
5370 May be used in sections: defaults | frontend | listen | backend
5371 yes | no | yes | yes
5372
5373 By default, a connection established between haproxy and the backend server
Olivier Houchard86006a52018-12-14 19:37:49 +01005374 which is considered safe for reuse is moved back to the server's idle
5375 connections pool so that any other request can make use of it. This is the
5376 "safe" strategy below.
Willy Tarreau30631952015-08-06 15:05:24 +02005377
5378 The argument indicates the desired connection reuse strategy :
5379
Olivier Houchard86006a52018-12-14 19:37:49 +01005380 - "never" : idle connections are never shared between sessions. This mode
5381 may be enforced to cancel a different strategy inherited from
5382 a defaults section or for troubleshooting. For example, if an
5383 old bogus application considers that multiple requests over
5384 the same connection come from the same client and it is not
5385 possible to fix the application, it may be desirable to
5386 disable connection sharing in a single backend. An example of
5387 such an application could be an old haproxy using cookie
5388 insertion in tunnel mode and not checking any request past the
5389 first one.
Willy Tarreau30631952015-08-06 15:05:24 +02005390
Olivier Houchard86006a52018-12-14 19:37:49 +01005391 - "safe" : this is the default and the recommended strategy. The first
5392 request of a session is always sent over its own connection,
5393 and only subsequent requests may be dispatched over other
5394 existing connections. This ensures that in case the server
5395 closes the connection when the request is being sent, the
5396 browser can decide to silently retry it. Since it is exactly
5397 equivalent to regular keep-alive, there should be no side
5398 effects.
Willy Tarreau30631952015-08-06 15:05:24 +02005399
5400 - "aggressive" : this mode may be useful in webservices environments where
5401 all servers are not necessarily known and where it would be
5402 appreciable to deliver most first requests over existing
5403 connections. In this case, first requests are only delivered
5404 over existing connections that have been reused at least once,
5405 proving that the server correctly supports connection reuse.
5406 It should only be used when it's sure that the client can
5407 retry a failed request once in a while and where the benefit
Michael Prokop4438c602019-05-24 10:25:45 +02005408 of aggressive connection reuse significantly outweighs the
Willy Tarreau30631952015-08-06 15:05:24 +02005409 downsides of rare connection failures.
5410
5411 - "always" : this mode is only recommended when the path to the server is
5412 known for never breaking existing connections quickly after
5413 releasing them. It allows the first request of a session to be
5414 sent to an existing connection. This can provide a significant
5415 performance increase over the "safe" strategy when the backend
5416 is a cache farm, since such components tend to show a
Davor Ocelice9ed2812017-12-25 17:49:28 +01005417 consistent behavior and will benefit from the connection
Willy Tarreau30631952015-08-06 15:05:24 +02005418 sharing. It is recommended that the "http-keep-alive" timeout
5419 remains low in this mode so that no dead connections remain
5420 usable. In most cases, this will lead to the same performance
5421 gains as "aggressive" but with more risks. It should only be
5422 used when it improves the situation over "aggressive".
5423
5424 When http connection sharing is enabled, a great care is taken to respect the
Davor Ocelice9ed2812017-12-25 17:49:28 +01005425 connection properties and compatibility. Specifically :
5426 - connections made with "usesrc" followed by a client-dependent value
5427 ("client", "clientip", "hdr_ip") are marked private and never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02005428
5429 - connections sent to a server with a TLS SNI extension are marked private
Davor Ocelice9ed2812017-12-25 17:49:28 +01005430 and are never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02005431
Lukas Tribusfd9b68c2018-10-27 20:06:59 +02005432 - connections with certain bogus authentication schemes (relying on the
5433 connection) like NTLM are detected, marked private and are never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02005434
Lukas Tribuse8adfeb2019-11-06 11:50:25 +01005435 A connection pool is involved and configurable with "pool-max-conn".
Willy Tarreau30631952015-08-06 15:05:24 +02005436
5437 Note: connection reuse improves the accuracy of the "server maxconn" setting,
5438 because almost no new connection will be established while idle connections
5439 remain available. This is particularly true with the "always" strategy.
5440
5441 See also : "option http-keep-alive", "server maxconn"
5442
5443
Mark Lamourinec2247f02012-01-04 13:02:01 -05005444http-send-name-header [<header>]
5445 Add the server name to a request. Use the header string given by <header>
Mark Lamourinec2247f02012-01-04 13:02:01 -05005446 May be used in sections: defaults | frontend | listen | backend
5447 yes | no | yes | yes
Mark Lamourinec2247f02012-01-04 13:02:01 -05005448 Arguments :
Mark Lamourinec2247f02012-01-04 13:02:01 -05005449 <header> The header string to use to send the server name
5450
Willy Tarreau81bef7e2019-10-07 14:58:02 +02005451 The "http-send-name-header" statement causes the header field named <header>
5452 to be set to the name of the target server at the moment the request is about
5453 to be sent on the wire. Any existing occurrences of this header are removed.
5454 Upon retries and redispatches, the header field is updated to always reflect
5455 the server being attempted to connect to. Given that this header is modified
5456 very late in the connection setup, it may have unexpected effects on already
5457 modified headers. For example using it with transport-level header such as
5458 connection, content-length, transfer-encoding and so on will likely result in
5459 invalid requests being sent to the server. Additionally it has been reported
5460 that this directive is currently being used as a way to overwrite the Host
5461 header field in outgoing requests; while this trick has been known to work
5462 as a side effect of the feature for some time, it is not officially supported
5463 and might possibly not work anymore in a future version depending on the
5464 technical difficulties this feature induces. A long-term solution instead
5465 consists in fixing the application which required this trick so that it binds
5466 to the correct host name.
Mark Lamourinec2247f02012-01-04 13:02:01 -05005467
5468 See also : "server"
5469
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01005470id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +02005471 Set a persistent ID to a proxy.
5472 May be used in sections : defaults | frontend | listen | backend
5473 no | yes | yes | yes
5474 Arguments : none
5475
5476 Set a persistent ID for the proxy. This ID must be unique and positive.
5477 An unused ID will automatically be assigned if unset. The first assigned
5478 value will be 1. This ID is currently only returned in statistics.
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01005479
5480
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005481ignore-persist { if | unless } <condition>
5482 Declare a condition to ignore persistence
5483 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01005484 no | no | yes | yes
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005485
5486 By default, when cookie persistence is enabled, every requests containing
5487 the cookie are unconditionally persistent (assuming the target server is up
5488 and running).
5489
5490 The "ignore-persist" statement allows one to declare various ACL-based
5491 conditions which, when met, will cause a request to ignore persistence.
5492 This is sometimes useful to load balance requests for static files, which
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005493 often don't require persistence. This can also be used to fully disable
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005494 persistence for a specific User-Agent (for example, some web crawler bots).
5495
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005496 The persistence is ignored when an "if" condition is met, or unless an
5497 "unless" condition is met.
5498
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03005499 Example:
5500 acl url_static path_beg /static /images /img /css
5501 acl url_static path_end .gif .png .jpg .css .js
5502 ignore-persist if url_static
5503
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005504 See also : "force-persist", "cookie", and section 7 about ACL usage.
5505
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005506load-server-state-from-file { global | local | none }
5507 Allow seamless reload of HAProxy
5508 May be used in sections: defaults | frontend | listen | backend
5509 yes | no | yes | yes
5510
5511 This directive points HAProxy to a file where server state from previous
5512 running process has been saved. That way, when starting up, before handling
5513 traffic, the new process can apply old states to servers exactly has if no
Davor Ocelice9ed2812017-12-25 17:49:28 +01005514 reload occurred. The purpose of the "load-server-state-from-file" directive is
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005515 to tell haproxy which file to use. For now, only 2 arguments to either prevent
5516 loading state or load states from a file containing all backends and servers.
5517 The state file can be generated by running the command "show servers state"
5518 over the stats socket and redirect output.
5519
Davor Ocelice9ed2812017-12-25 17:49:28 +01005520 The format of the file is versioned and is very specific. To understand it,
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005521 please read the documentation of the "show servers state" command (chapter
Willy Tarreau1af20c72017-06-23 16:01:14 +02005522 9.3 of Management Guide).
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005523
5524 Arguments:
5525 global load the content of the file pointed by the global directive
5526 named "server-state-file".
5527
5528 local load the content of the file pointed by the directive
5529 "server-state-file-name" if set. If not set, then the backend
5530 name is used as a file name.
5531
5532 none don't load any stat for this backend
5533
5534 Notes:
Willy Tarreaue5a60682016-11-09 14:54:53 +01005535 - server's IP address is preserved across reloads by default, but the
5536 order can be changed thanks to the server's "init-addr" setting. This
5537 means that an IP address change performed on the CLI at run time will
Davor Ocelice9ed2812017-12-25 17:49:28 +01005538 be preserved, and that any change to the local resolver (e.g. /etc/hosts)
Willy Tarreaue5a60682016-11-09 14:54:53 +01005539 will possibly not have any effect if the state file is in use.
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005540
5541 - server's weight is applied from previous running process unless it has
5542 has changed between previous and new configuration files.
5543
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005544 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005545
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005546 global
5547 stats socket /tmp/socket
5548 server-state-file /tmp/server_state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005549
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005550 defaults
5551 load-server-state-from-file global
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005552
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005553 backend bk
5554 server s1 127.0.0.1:22 check weight 11
5555 server s2 127.0.0.1:22 check weight 12
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005556
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005557
5558 Then one can run :
5559
5560 socat /tmp/socket - <<< "show servers state" > /tmp/server_state
5561
5562 Content of the file /tmp/server_state would be like this:
5563
5564 1
5565 # <field names skipped for the doc example>
5566 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
5567 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
5568
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005569 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005570
5571 global
5572 stats socket /tmp/socket
5573 server-state-base /etc/haproxy/states
5574
5575 defaults
5576 load-server-state-from-file local
5577
5578 backend bk
5579 server s1 127.0.0.1:22 check weight 11
5580 server s2 127.0.0.1:22 check weight 12
5581
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005582
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005583 Then one can run :
5584
5585 socat /tmp/socket - <<< "show servers state bk" > /etc/haproxy/states/bk
5586
5587 Content of the file /etc/haproxy/states/bk would be like this:
5588
5589 1
5590 # <field names skipped for the doc example>
5591 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
5592 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
5593
5594 See also: "server-state-file", "server-state-file-name", and
5595 "show servers state"
5596
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005597
Willy Tarreau2769aa02007-12-27 18:26:09 +01005598log global
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02005599log <address> [len <length>] [format <format>] [sample <ranges>:<smp_size>]
5600 <facility> [<level> [<minlevel>]]
William Lallemand0f99e342011-10-12 17:50:54 +02005601no log
Willy Tarreau2769aa02007-12-27 18:26:09 +01005602 Enable per-instance logging of events and traffic.
5603 May be used in sections : defaults | frontend | listen | backend
5604 yes | yes | yes | yes
William Lallemand0f99e342011-10-12 17:50:54 +02005605
5606 Prefix :
5607 no should be used when the logger list must be flushed. For example,
5608 if you don't want to inherit from the default logger list. This
5609 prefix does not allow arguments.
5610
Willy Tarreau2769aa02007-12-27 18:26:09 +01005611 Arguments :
5612 global should be used when the instance's logging parameters are the
5613 same as the global ones. This is the most common usage. "global"
5614 replaces <address>, <facility> and <level> with those of the log
5615 entries found in the "global" section. Only one "log global"
5616 statement may be used per instance, and this form takes no other
5617 parameter.
5618
5619 <address> indicates where to send the logs. It takes the same format as
5620 for the "global" section's logs, and can be one of :
5621
5622 - An IPv4 address optionally followed by a colon (':') and a UDP
5623 port. If no port is specified, 514 is used by default (the
5624 standard syslog port).
5625
David du Colombier24bb5f52011-03-17 10:40:23 +01005626 - An IPv6 address followed by a colon (':') and optionally a UDP
5627 port. If no port is specified, 514 is used by default (the
5628 standard syslog port).
5629
Willy Tarreau2769aa02007-12-27 18:26:09 +01005630 - A filesystem path to a UNIX domain socket, keeping in mind
5631 considerations for chroot (be sure the path is accessible
5632 inside the chroot) and uid/gid (be sure the path is
Davor Ocelice9ed2812017-12-25 17:49:28 +01005633 appropriately writable).
Willy Tarreau2769aa02007-12-27 18:26:09 +01005634
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01005635 - A file descriptor number in the form "fd@<number>", which may
5636 point to a pipe, terminal, or socket. In this case unbuffered
5637 logs are used and one writev() call per log is performed. This
5638 is a bit expensive but acceptable for most workloads. Messages
5639 sent this way will not be truncated but may be dropped, in
5640 which case the DroppedLogs counter will be incremented. The
5641 writev() call is atomic even on pipes for messages up to
5642 PIPE_BUF size, which POSIX recommends to be at least 512 and
5643 which is 4096 bytes on most modern operating systems. Any
5644 larger message may be interleaved with messages from other
5645 processes. Exceptionally for debugging purposes the file
5646 descriptor may also be directed to a file, but doing so will
5647 significantly slow haproxy down as non-blocking calls will be
5648 ignored. Also there will be no way to purge nor rotate this
5649 file without restarting the process. Note that the configured
5650 syslog format is preserved, so the output is suitable for use
Willy Tarreauc1b06452018-11-12 11:57:56 +01005651 with a TCP syslog server. See also the "short" and "raw"
5652 formats below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01005653
5654 - "stdout" / "stderr", which are respectively aliases for "fd@1"
5655 and "fd@2", see above.
5656
Willy Tarreauc046d162019-08-30 15:24:59 +02005657 - A ring buffer in the form "ring@<name>", which will correspond
5658 to an in-memory ring buffer accessible over the CLI using the
5659 "show events" command, which will also list existing rings and
5660 their sizes. Such buffers are lost on reload or restart but
5661 when used as a complement this can help troubleshooting by
5662 having the logs instantly available.
5663
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01005664 You may want to reference some environment variables in the
5665 address parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01005666
Willy Tarreau18324f52014-06-27 18:10:07 +02005667 <length> is an optional maximum line length. Log lines larger than this
5668 value will be truncated before being sent. The reason is that
5669 syslog servers act differently on log line length. All servers
5670 support the default value of 1024, but some servers simply drop
5671 larger lines while others do log them. If a server supports long
5672 lines, it may make sense to set this value here in order to avoid
5673 truncating long lines. Similarly, if a server drops long lines,
5674 it is preferable to truncate them before sending them. Accepted
5675 values are 80 to 65535 inclusive. The default value of 1024 is
5676 generally fine for all standard usages. Some specific cases of
Davor Ocelice9ed2812017-12-25 17:49:28 +01005677 long captures or JSON-formatted logs may require larger values.
Willy Tarreau18324f52014-06-27 18:10:07 +02005678
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02005679 <ranges> A list of comma-separated ranges to identify the logs to sample.
5680 This is used to balance the load of the logs to send to the log
5681 server. The limits of the ranges cannot be null. They are numbered
5682 from 1. The size or period (in number of logs) of the sample must
5683 be set with <sample_size> parameter.
5684
5685 <sample_size>
5686 The size of the sample in number of logs to consider when balancing
5687 their logging loads. It is used to balance the load of the logs to
5688 send to the syslog server. This size must be greater or equal to the
5689 maximum of the high limits of the ranges.
5690 (see also <ranges> parameter).
5691
Willy Tarreauadb345d2018-11-12 07:56:13 +01005692 <format> is the log format used when generating syslog messages. It may be
5693 one of the following :
5694
5695 rfc3164 The RFC3164 syslog message format. This is the default.
5696 (https://tools.ietf.org/html/rfc3164)
5697
5698 rfc5424 The RFC5424 syslog message format.
5699 (https://tools.ietf.org/html/rfc5424)
5700
Willy Tarreaue8746a02018-11-12 08:45:00 +01005701 short A message containing only a level between angle brackets such as
5702 '<3>', followed by the text. The PID, date, time, process name
5703 and system name are omitted. This is designed to be used with a
5704 local log server. This format is compatible with what the
5705 systemd logger consumes.
5706
Willy Tarreauc1b06452018-11-12 11:57:56 +01005707 raw A message containing only the text. The level, PID, date, time,
5708 process name and system name are omitted. This is designed to
5709 be used in containers or during development, where the severity
5710 only depends on the file descriptor used (stdout/stderr).
5711
Willy Tarreau2769aa02007-12-27 18:26:09 +01005712 <facility> must be one of the 24 standard syslog facilities :
5713
Willy Tarreaue8746a02018-11-12 08:45:00 +01005714 kern user mail daemon auth syslog lpr news
5715 uucp cron auth2 ftp ntp audit alert cron2
5716 local0 local1 local2 local3 local4 local5 local6 local7
5717
Willy Tarreauc1b06452018-11-12 11:57:56 +01005718 Note that the facility is ignored for the "short" and "raw"
5719 formats, but still required as a positional field. It is
5720 recommended to use "daemon" in this case to make it clear that
5721 it's only supposed to be used locally.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005722
5723 <level> is optional and can be specified to filter outgoing messages. By
5724 default, all messages are sent. If a level is specified, only
5725 messages with a severity at least as important as this level
Willy Tarreauf7edefa2009-05-10 17:20:05 +02005726 will be sent. An optional minimum level can be specified. If it
5727 is set, logs emitted with a more severe level than this one will
5728 be capped to this level. This is used to avoid sending "emerg"
5729 messages on all terminals on some default syslog configurations.
5730 Eight levels are known :
Willy Tarreau2769aa02007-12-27 18:26:09 +01005731
5732 emerg alert crit err warning notice info debug
5733
William Lallemand0f99e342011-10-12 17:50:54 +02005734 It is important to keep in mind that it is the frontend which decides what to
5735 log from a connection, and that in case of content switching, the log entries
5736 from the backend will be ignored. Connections are logged at level "info".
Willy Tarreaucc6c8912009-02-22 10:53:55 +01005737
5738 However, backend log declaration define how and where servers status changes
5739 will be logged. Level "notice" will be used to indicate a server going up,
5740 "warning" will be used for termination signals and definitive service
5741 termination, and "alert" will be used for when a server goes down.
5742
5743 Note : According to RFC3164, messages are truncated to 1024 bytes before
5744 being emitted.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005745
5746 Example :
5747 log global
Willy Tarreauc1b06452018-11-12 11:57:56 +01005748 log stdout format short daemon # send log to systemd
5749 log stdout format raw daemon # send everything to stdout
5750 log stderr format raw daemon notice # send important events to stderr
Willy Tarreauf7edefa2009-05-10 17:20:05 +02005751 log 127.0.0.1:514 local0 notice # only send important events
5752 log 127.0.0.1:514 local0 notice notice # same but limit output level
William Lallemandb2f07452015-05-12 14:27:13 +02005753 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
Willy Tarreaudad36a32013-03-11 01:20:04 +01005754
Willy Tarreau2769aa02007-12-27 18:26:09 +01005755
William Lallemand48940402012-01-30 16:47:22 +01005756log-format <string>
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01005757 Specifies the log format string to use for traffic logs
5758 May be used in sections: defaults | frontend | listen | backend
5759 yes | yes | yes | no
William Lallemand48940402012-01-30 16:47:22 +01005760
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01005761 This directive specifies the log format string that will be used for all logs
5762 resulting from traffic passing through the frontend using this line. If the
5763 directive is used in a defaults section, all subsequent frontends will use
5764 the same log format. Please see section 8.2.4 which covers the log format
5765 string in depth.
William Lallemand48940402012-01-30 16:47:22 +01005766
Guillaume de Lafond29f45602017-03-31 19:52:15 +02005767 "log-format" directive overrides previous "option tcplog", "log-format" and
5768 "option httplog" directives.
5769
Dragan Dosen7ad31542015-09-28 17:16:47 +02005770log-format-sd <string>
5771 Specifies the RFC5424 structured-data log format string
5772 May be used in sections: defaults | frontend | listen | backend
5773 yes | yes | yes | no
5774
5775 This directive specifies the RFC5424 structured-data log format string that
5776 will be used for all logs resulting from traffic passing through the frontend
5777 using this line. If the directive is used in a defaults section, all
5778 subsequent frontends will use the same log format. Please see section 8.2.4
5779 which covers the log format string in depth.
5780
5781 See https://tools.ietf.org/html/rfc5424#section-6.3 for more information
5782 about the RFC5424 structured-data part.
5783
5784 Note : This log format string will be used only for loggers that have set
5785 log format to "rfc5424".
5786
5787 Example :
5788 log-format-sd [exampleSDID@1234\ bytes=\"%B\"\ status=\"%ST\"]
5789
5790
Willy Tarreau094af4e2015-01-07 15:03:42 +01005791log-tag <string>
5792 Specifies the log tag to use for all outgoing logs
5793 May be used in sections: defaults | frontend | listen | backend
5794 yes | yes | yes | yes
5795
5796 Sets the tag field in the syslog header to this string. It defaults to the
5797 log-tag set in the global section, otherwise the program name as launched
5798 from the command line, which usually is "haproxy". Sometimes it can be useful
5799 to differentiate between multiple processes running on the same host, or to
5800 differentiate customer instances running in the same process. In the backend,
5801 logs about servers up/down will use this tag. As a hint, it can be convenient
5802 to set a log-tag related to a hosted customer in a defaults section then put
5803 all the frontends and backends for that customer, then start another customer
5804 in a new defaults section. See also the global "log-tag" directive.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005805
Willy Tarreauc35362a2014-04-25 13:58:37 +02005806max-keep-alive-queue <value>
5807 Set the maximum server queue size for maintaining keep-alive connections
5808 May be used in sections: defaults | frontend | listen | backend
5809 yes | no | yes | yes
5810
5811 HTTP keep-alive tries to reuse the same server connection whenever possible,
5812 but sometimes it can be counter-productive, for example if a server has a lot
5813 of connections while other ones are idle. This is especially true for static
5814 servers.
5815
5816 The purpose of this setting is to set a threshold on the number of queued
5817 connections at which haproxy stops trying to reuse the same server and prefers
5818 to find another one. The default value, -1, means there is no limit. A value
5819 of zero means that keep-alive requests will never be queued. For very close
5820 servers which can be reached with a low latency and which are not sensible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01005821 breaking keep-alive, a low value is recommended (e.g. local static server can
Willy Tarreauc35362a2014-04-25 13:58:37 +02005822 use a value of 10 or less). For remote servers suffering from a high latency,
5823 higher values might be needed to cover for the latency and/or the cost of
5824 picking a different server.
5825
5826 Note that this has no impact on responses which are maintained to the same
5827 server consecutively to a 401 response. They will still go to the same server
5828 even if they have to be queued.
5829
5830 See also : "option http-server-close", "option prefer-last-server", server
5831 "maxconn" and cookie persistence.
5832
Olivier Houcharda4d4fdf2018-12-14 19:27:06 +01005833max-session-srv-conns <nb>
5834 Set the maximum number of outgoing connections we can keep idling for a given
5835 client session. The default is 5 (it precisely equals MAX_SRV_LIST which is
5836 defined at build time).
Willy Tarreauc35362a2014-04-25 13:58:37 +02005837
Willy Tarreau2769aa02007-12-27 18:26:09 +01005838maxconn <conns>
5839 Fix the maximum number of concurrent connections on a frontend
5840 May be used in sections : defaults | frontend | listen | backend
5841 yes | yes | yes | no
5842 Arguments :
5843 <conns> is the maximum number of concurrent connections the frontend will
5844 accept to serve. Excess connections will be queued by the system
5845 in the socket's listen queue and will be served once a connection
5846 closes.
5847
5848 If the system supports it, it can be useful on big sites to raise this limit
5849 very high so that haproxy manages connection queues, instead of leaving the
5850 clients with unanswered connection attempts. This value should not exceed the
5851 global maxconn. Also, keep in mind that a connection contains two buffers
Baptiste Assmann79fb45d2016-03-06 23:34:31 +01005852 of tune.bufsize (16kB by default) each, as well as some other data resulting
5853 in about 33 kB of RAM being consumed per established connection. That means
5854 that a medium system equipped with 1GB of RAM can withstand around
5855 20000-25000 concurrent connections if properly tuned.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005856
5857 Also, when <conns> is set to large values, it is possible that the servers
5858 are not sized to accept such loads, and for this reason it is generally wise
5859 to assign them some reasonable connection limits.
5860
Willy Tarreauc8d5b952019-02-27 17:25:52 +01005861 When this value is set to zero, which is the default, the global "maxconn"
5862 value is used.
Vincent Bernat6341be52012-06-27 17:18:30 +02005863
Willy Tarreau2769aa02007-12-27 18:26:09 +01005864 See also : "server", global section's "maxconn", "fullconn"
5865
5866
5867mode { tcp|http|health }
5868 Set the running mode or protocol of the instance
5869 May be used in sections : defaults | frontend | listen | backend
5870 yes | yes | yes | yes
5871 Arguments :
5872 tcp The instance will work in pure TCP mode. A full-duplex connection
5873 will be established between clients and servers, and no layer 7
5874 examination will be performed. This is the default mode. It
5875 should be used for SSL, SSH, SMTP, ...
5876
5877 http The instance will work in HTTP mode. The client request will be
5878 analyzed in depth before connecting to any server. Any request
5879 which is not RFC-compliant will be rejected. Layer 7 filtering,
5880 processing and switching will be possible. This is the mode which
5881 brings HAProxy most of its value.
5882
5883 health The instance will work in "health" mode. It will just reply "OK"
Willy Tarreau82569f92012-09-27 23:48:56 +02005884 to incoming connections and close the connection. Alternatively,
5885 If the "httpchk" option is set, "HTTP/1.0 200 OK" will be sent
5886 instead. Nothing will be logged in either case. This mode is used
5887 to reply to external components health checks. This mode is
5888 deprecated and should not be used anymore as it is possible to do
5889 the same and even better by combining TCP or HTTP modes with the
5890 "monitor" keyword.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005891
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005892 When doing content switching, it is mandatory that the frontend and the
5893 backend are in the same mode (generally HTTP), otherwise the configuration
5894 will be refused.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005895
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005896 Example :
Willy Tarreau2769aa02007-12-27 18:26:09 +01005897 defaults http_instances
5898 mode http
5899
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005900 See also : "monitor", "monitor-net"
Willy Tarreau2769aa02007-12-27 18:26:09 +01005901
Willy Tarreau0ba27502007-12-24 16:55:16 +01005902
Cyril Bontéf0c60612010-02-06 14:44:47 +01005903monitor fail { if | unless } <condition>
Willy Tarreau2769aa02007-12-27 18:26:09 +01005904 Add a condition to report a failure to a monitor HTTP request.
Willy Tarreau0ba27502007-12-24 16:55:16 +01005905 May be used in sections : defaults | frontend | listen | backend
5906 no | yes | yes | no
Willy Tarreau0ba27502007-12-24 16:55:16 +01005907 Arguments :
5908 if <cond> the monitor request will fail if the condition is satisfied,
5909 and will succeed otherwise. The condition should describe a
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005910 combined test which must induce a failure if all conditions
Willy Tarreau0ba27502007-12-24 16:55:16 +01005911 are met, for instance a low number of servers both in a
5912 backend and its backup.
5913
5914 unless <cond> the monitor request will succeed only if the condition is
5915 satisfied, and will fail otherwise. Such a condition may be
5916 based on a test on the presence of a minimum number of active
5917 servers in a list of backends.
5918
5919 This statement adds a condition which can force the response to a monitor
5920 request to report a failure. By default, when an external component queries
5921 the URI dedicated to monitoring, a 200 response is returned. When one of the
5922 conditions above is met, haproxy will return 503 instead of 200. This is
5923 very useful to report a site failure to an external component which may base
5924 routing advertisements between multiple sites on the availability reported by
5925 haproxy. In this case, one would rely on an ACL involving the "nbsrv"
Willy Tarreauae94d4d2011-05-11 16:28:49 +02005926 criterion. Note that "monitor fail" only works in HTTP mode. Both status
5927 messages may be tweaked using "errorfile" or "errorloc" if needed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01005928
5929 Example:
5930 frontend www
Willy Tarreau2769aa02007-12-27 18:26:09 +01005931 mode http
Willy Tarreau0ba27502007-12-24 16:55:16 +01005932 acl site_dead nbsrv(dynamic) lt 2
5933 acl site_dead nbsrv(static) lt 2
5934 monitor-uri /site_alive
5935 monitor fail if site_dead
5936
Willy Tarreauae94d4d2011-05-11 16:28:49 +02005937 See also : "monitor-net", "monitor-uri", "errorfile", "errorloc"
Willy Tarreau2769aa02007-12-27 18:26:09 +01005938
5939
5940monitor-net <source>
5941 Declare a source network which is limited to monitor requests
5942 May be used in sections : defaults | frontend | listen | backend
5943 yes | yes | yes | no
5944 Arguments :
5945 <source> is the source IPv4 address or network which will only be able to
5946 get monitor responses to any request. It can be either an IPv4
5947 address, a host name, or an address followed by a slash ('/')
5948 followed by a mask.
5949
5950 In TCP mode, any connection coming from a source matching <source> will cause
5951 the connection to be immediately closed without any log. This allows another
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005952 equipment to probe the port and verify that it is still listening, without
Willy Tarreau2769aa02007-12-27 18:26:09 +01005953 forwarding the connection to a remote server.
5954
5955 In HTTP mode, a connection coming from a source matching <source> will be
5956 accepted, the following response will be sent without waiting for a request,
5957 then the connection will be closed : "HTTP/1.0 200 OK". This is normally
5958 enough for any front-end HTTP probe to detect that the service is UP and
Willy Tarreau82569f92012-09-27 23:48:56 +02005959 running without forwarding the request to a backend server. Note that this
5960 response is sent in raw format, without any transformation. This is important
5961 as it means that it will not be SSL-encrypted on SSL listeners.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005962
Willy Tarreau82569f92012-09-27 23:48:56 +02005963 Monitor requests are processed very early, just after tcp-request connection
5964 ACLs which are the only ones able to block them. These connections are short
5965 lived and never wait for any data from the client. They cannot be logged, and
5966 it is the intended purpose. They are only used to report HAProxy's health to
5967 an upper component, nothing more. Please note that "monitor fail" rules do
5968 not apply to connections intercepted by "monitor-net".
Willy Tarreau2769aa02007-12-27 18:26:09 +01005969
Willy Tarreau95cd2832010-03-04 23:36:33 +01005970 Last, please note that only one "monitor-net" statement can be specified in
5971 a frontend. If more than one is found, only the last one will be considered.
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005972
Willy Tarreau2769aa02007-12-27 18:26:09 +01005973 Example :
5974 # addresses .252 and .253 are just probing us.
5975 frontend www
5976 monitor-net 192.168.0.252/31
5977
5978 See also : "monitor fail", "monitor-uri"
5979
5980
5981monitor-uri <uri>
5982 Intercept a URI used by external components' monitor requests
5983 May be used in sections : defaults | frontend | listen | backend
5984 yes | yes | yes | no
5985 Arguments :
5986 <uri> is the exact URI which we want to intercept to return HAProxy's
5987 health status instead of forwarding the request.
5988
5989 When an HTTP request referencing <uri> will be received on a frontend,
5990 HAProxy will not forward it nor log it, but instead will return either
5991 "HTTP/1.0 200 OK" or "HTTP/1.0 503 Service unavailable", depending on failure
5992 conditions defined with "monitor fail". This is normally enough for any
5993 front-end HTTP probe to detect that the service is UP and running without
5994 forwarding the request to a backend server. Note that the HTTP method, the
5995 version and all headers are ignored, but the request must at least be valid
5996 at the HTTP level. This keyword may only be used with an HTTP-mode frontend.
5997
Willy Tarreau721d8e02017-12-01 18:25:08 +01005998 Monitor requests are processed very early, just after the request is parsed
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02005999 and even before any "http-request". The only rulesets applied before are the
6000 tcp-request ones. They cannot be logged either, and it is the intended
6001 purpose. They are only used to report HAProxy's health to an upper component,
6002 nothing more. However, it is possible to add any number of conditions using
6003 "monitor fail" and ACLs so that the result can be adjusted to whatever check
6004 can be imagined (most often the number of available servers in a backend).
Willy Tarreau2769aa02007-12-27 18:26:09 +01006005
6006 Example :
6007 # Use /haproxy_test to report haproxy's status
6008 frontend www
6009 mode http
6010 monitor-uri /haproxy_test
6011
6012 See also : "monitor fail", "monitor-net"
6013
Willy Tarreau0ba27502007-12-24 16:55:16 +01006014
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006015option abortonclose
6016no option abortonclose
6017 Enable or disable early dropping of aborted requests pending in queues.
6018 May be used in sections : defaults | frontend | listen | backend
6019 yes | no | yes | yes
6020 Arguments : none
6021
6022 In presence of very high loads, the servers will take some time to respond.
6023 The per-instance connection queue will inflate, and the response time will
6024 increase respective to the size of the queue times the average per-session
6025 response time. When clients will wait for more than a few seconds, they will
Willy Tarreau198a7442008-01-17 12:05:32 +01006026 often hit the "STOP" button on their browser, leaving a useless request in
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006027 the queue, and slowing down other users, and the servers as well, because the
6028 request will eventually be served, then aborted at the first error
6029 encountered while delivering the response.
6030
6031 As there is no way to distinguish between a full STOP and a simple output
6032 close on the client side, HTTP agents should be conservative and consider
6033 that the client might only have closed its output channel while waiting for
6034 the response. However, this introduces risks of congestion when lots of users
6035 do the same, and is completely useless nowadays because probably no client at
6036 all will close the session while waiting for the response. Some HTTP agents
Davor Ocelice9ed2812017-12-25 17:49:28 +01006037 support this behavior (Squid, Apache, HAProxy), and others do not (TUX, most
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006038 hardware-based load balancers). So the probability for a closed input channel
Willy Tarreau198a7442008-01-17 12:05:32 +01006039 to represent a user hitting the "STOP" button is close to 100%, and the risk
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006040 of being the single component to break rare but valid traffic is extremely
6041 low, which adds to the temptation to be able to abort a session early while
6042 still not served and not pollute the servers.
6043
Davor Ocelice9ed2812017-12-25 17:49:28 +01006044 In HAProxy, the user can choose the desired behavior using the option
6045 "abortonclose". By default (without the option) the behavior is HTTP
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006046 compliant and aborted requests will be served. But when the option is
6047 specified, a session with an incoming channel closed will be aborted while
6048 it is still possible, either pending in the queue for a connection slot, or
6049 during the connection establishment if the server has not yet acknowledged
6050 the connection request. This considerably reduces the queue size and the load
6051 on saturated servers when users are tempted to click on STOP, which in turn
Willy Tarreaud72758d2010-01-12 10:42:19 +01006052 reduces the response time for other users.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006053
6054 If this option has been enabled in a "defaults" section, it can be disabled
6055 in a specific instance by prepending the "no" keyword before it.
6056
6057 See also : "timeout queue" and server's "maxconn" and "maxqueue" parameters
6058
6059
Willy Tarreau4076a152009-04-02 15:18:36 +02006060option accept-invalid-http-request
6061no option accept-invalid-http-request
6062 Enable or disable relaxing of HTTP request parsing
6063 May be used in sections : defaults | frontend | listen | backend
6064 yes | yes | yes | no
6065 Arguments : none
6066
Willy Tarreau91852eb2015-05-01 13:26:00 +02006067 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02006068 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01006069 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02006070 forbidden characters are essentially used to build attacks exploiting server
6071 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
6072 server will emit invalid header names for whatever reason (configuration,
6073 implementation) and the issue will not be immediately fixed. In such a case,
6074 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau422246e2012-01-07 23:54:13 +01006075 even if that does not make sense, by specifying this option. Similarly, the
6076 list of characters allowed to appear in a URI is well defined by RFC3986, and
6077 chars 0-31, 32 (space), 34 ('"'), 60 ('<'), 62 ('>'), 92 ('\'), 94 ('^'), 96
6078 ('`'), 123 ('{'), 124 ('|'), 125 ('}'), 127 (delete) and anything above are
Davor Ocelice9ed2812017-12-25 17:49:28 +01006079 not allowed at all. HAProxy always blocks a number of them (0..32, 127). The
Willy Tarreau91852eb2015-05-01 13:26:00 +02006080 remaining ones are blocked by default unless this option is enabled. This
Willy Tarreau13317662015-05-01 13:47:08 +02006081 option also relaxes the test on the HTTP version, it allows HTTP/0.9 requests
6082 to pass through (no version specified) and multiple digits for both the major
6083 and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02006084
6085 This option should never be enabled by default as it hides application bugs
6086 and open security breaches. It should only be deployed after a problem has
6087 been confirmed.
6088
6089 When this option is enabled, erroneous header names will still be accepted in
6090 requests, but the complete request will be captured in order to permit later
Willy Tarreau422246e2012-01-07 23:54:13 +01006091 analysis using the "show errors" request on the UNIX stats socket. Similarly,
6092 requests containing invalid chars in the URI part will be logged. Doing this
Willy Tarreau4076a152009-04-02 15:18:36 +02006093 also helps confirming that the issue has been solved.
6094
6095 If this option has been enabled in a "defaults" section, it can be disabled
6096 in a specific instance by prepending the "no" keyword before it.
6097
6098 See also : "option accept-invalid-http-response" and "show errors" on the
6099 stats socket.
6100
6101
6102option accept-invalid-http-response
6103no option accept-invalid-http-response
6104 Enable or disable relaxing of HTTP response parsing
6105 May be used in sections : defaults | frontend | listen | backend
6106 yes | no | yes | yes
6107 Arguments : none
6108
Willy Tarreau91852eb2015-05-01 13:26:00 +02006109 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02006110 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01006111 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02006112 forbidden characters are essentially used to build attacks exploiting server
6113 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
6114 server will emit invalid header names for whatever reason (configuration,
6115 implementation) and the issue will not be immediately fixed. In such a case,
6116 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau91852eb2015-05-01 13:26:00 +02006117 even if that does not make sense, by specifying this option. This option also
6118 relaxes the test on the HTTP version format, it allows multiple digits for
6119 both the major and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02006120
6121 This option should never be enabled by default as it hides application bugs
6122 and open security breaches. It should only be deployed after a problem has
6123 been confirmed.
6124
6125 When this option is enabled, erroneous header names will still be accepted in
6126 responses, but the complete response will be captured in order to permit
6127 later analysis using the "show errors" request on the UNIX stats socket.
6128 Doing this also helps confirming that the issue has been solved.
6129
6130 If this option has been enabled in a "defaults" section, it can be disabled
6131 in a specific instance by prepending the "no" keyword before it.
6132
6133 See also : "option accept-invalid-http-request" and "show errors" on the
6134 stats socket.
6135
6136
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006137option allbackups
6138no option allbackups
6139 Use either all backup servers at a time or only the first one
6140 May be used in sections : defaults | frontend | listen | backend
6141 yes | no | yes | yes
6142 Arguments : none
6143
6144 By default, the first operational backup server gets all traffic when normal
6145 servers are all down. Sometimes, it may be preferred to use multiple backups
6146 at once, because one will not be enough. When "option allbackups" is enabled,
6147 the load balancing will be performed among all backup servers when all normal
6148 ones are unavailable. The same load balancing algorithm will be used and the
6149 servers' weights will be respected. Thus, there will not be any priority
6150 order between the backup servers anymore.
6151
6152 This option is mostly used with static server farms dedicated to return a
6153 "sorry" page when an application is completely offline.
6154
6155 If this option has been enabled in a "defaults" section, it can be disabled
6156 in a specific instance by prepending the "no" keyword before it.
6157
6158
6159option checkcache
6160no option checkcache
Godbach7056a352013-12-11 20:01:07 +08006161 Analyze all server responses and block responses with cacheable cookies
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006162 May be used in sections : defaults | frontend | listen | backend
6163 yes | no | yes | yes
6164 Arguments : none
6165
6166 Some high-level frameworks set application cookies everywhere and do not
6167 always let enough control to the developer to manage how the responses should
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006168 be cached. When a session cookie is returned on a cacheable object, there is a
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006169 high risk of session crossing or stealing between users traversing the same
6170 caches. In some situations, it is better to block the response than to let
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02006171 some sensitive session information go in the wild.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006172
6173 The option "checkcache" enables deep inspection of all server responses for
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006174 strict compliance with HTTP specification in terms of cacheability. It
Willy Tarreau198a7442008-01-17 12:05:32 +01006175 carefully checks "Cache-control", "Pragma" and "Set-cookie" headers in server
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006176 response to check if there's a risk of caching a cookie on a client-side
6177 proxy. When this option is enabled, the only responses which can be delivered
Willy Tarreau198a7442008-01-17 12:05:32 +01006178 to the client are :
Davor Ocelice9ed2812017-12-25 17:49:28 +01006179 - all those without "Set-Cookie" header;
Willy Tarreauc55ddce2017-12-21 11:41:38 +01006180 - all those with a return code other than 200, 203, 204, 206, 300, 301,
6181 404, 405, 410, 414, 501, provided that the server has not set a
Davor Ocelice9ed2812017-12-25 17:49:28 +01006182 "Cache-control: public" header field;
Willy Tarreau24ea0bc2017-12-21 11:32:55 +01006183 - all those that result from a request using a method other than GET, HEAD,
6184 OPTIONS, TRACE, provided that the server has not set a 'Cache-Control:
Davor Ocelice9ed2812017-12-25 17:49:28 +01006185 public' header field;
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006186 - those with a 'Pragma: no-cache' header
6187 - those with a 'Cache-control: private' header
6188 - those with a 'Cache-control: no-store' header
6189 - those with a 'Cache-control: max-age=0' header
6190 - those with a 'Cache-control: s-maxage=0' header
6191 - those with a 'Cache-control: no-cache' header
6192 - those with a 'Cache-control: no-cache="set-cookie"' header
6193 - those with a 'Cache-control: no-cache="set-cookie,' header
6194 (allowing other fields after set-cookie)
6195
6196 If a response doesn't respect these requirements, then it will be blocked
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02006197 just as if it was from an "http-response deny" rule, with an "HTTP 502 bad
6198 gateway". The session state shows "PH--" meaning that the proxy blocked the
6199 response during headers processing. Additionally, an alert will be sent in
6200 the logs so that admins are informed that there's something to be fixed.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006201
6202 Due to the high impact on the application, the application should be tested
6203 in depth with the option enabled before going to production. It is also a
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01006204 good practice to always activate it during tests, even if it is not used in
Davor Ocelice9ed2812017-12-25 17:49:28 +01006205 production, as it will report potentially dangerous application behaviors.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006206
6207 If this option has been enabled in a "defaults" section, it can be disabled
6208 in a specific instance by prepending the "no" keyword before it.
6209
6210
6211option clitcpka
6212no option clitcpka
6213 Enable or disable the sending of TCP keepalive packets on the client side
6214 May be used in sections : defaults | frontend | listen | backend
6215 yes | yes | yes | no
6216 Arguments : none
6217
6218 When there is a firewall or any session-aware component between a client and
6219 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01006220 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006221 components decides to expire a session which has remained idle for too long.
6222
6223 Enabling socket-level TCP keep-alives makes the system regularly send packets
6224 to the other end of the connection, leaving it active. The delay between
6225 keep-alive probes is controlled by the system only and depends both on the
6226 operating system and its tuning parameters.
6227
6228 It is important to understand that keep-alive packets are neither emitted nor
6229 received at the application level. It is only the network stacks which sees
6230 them. For this reason, even if one side of the proxy already uses keep-alives
6231 to maintain its connection alive, those keep-alive packets will not be
6232 forwarded to the other side of the proxy.
6233
6234 Please note that this has nothing to do with HTTP keep-alive.
6235
6236 Using option "clitcpka" enables the emission of TCP keep-alive probes on the
6237 client side of a connection, which should help when session expirations are
6238 noticed between HAProxy and a client.
6239
6240 If this option has been enabled in a "defaults" section, it can be disabled
6241 in a specific instance by prepending the "no" keyword before it.
6242
6243 See also : "option srvtcpka", "option tcpka"
6244
6245
Willy Tarreau0ba27502007-12-24 16:55:16 +01006246option contstats
6247 Enable continuous traffic statistics updates
6248 May be used in sections : defaults | frontend | listen | backend
6249 yes | yes | yes | no
6250 Arguments : none
6251
6252 By default, counters used for statistics calculation are incremented
6253 only when a session finishes. It works quite well when serving small
6254 objects, but with big ones (for example large images or archives) or
6255 with A/V streaming, a graph generated from haproxy counters looks like
Willy Tarreaudef0d222016-11-08 22:03:00 +01006256 a hedgehog. With this option enabled counters get incremented frequently
6257 along the session, typically every 5 seconds, which is often enough to
6258 produce clean graphs. Recounting touches a hotpath directly so it is not
6259 not enabled by default, as it can cause a lot of wakeups for very large
6260 session counts and cause a small performance drop.
Willy Tarreau0ba27502007-12-24 16:55:16 +01006261
6262
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006263option dontlog-normal
6264no option dontlog-normal
6265 Enable or disable logging of normal, successful connections
6266 May be used in sections : defaults | frontend | listen | backend
6267 yes | yes | yes | no
6268 Arguments : none
6269
6270 There are large sites dealing with several thousand connections per second
6271 and for which logging is a major pain. Some of them are even forced to turn
6272 logs off and cannot debug production issues. Setting this option ensures that
6273 normal connections, those which experience no error, no timeout, no retry nor
6274 redispatch, will not be logged. This leaves disk space for anomalies. In HTTP
6275 mode, the response status code is checked and return codes 5xx will still be
6276 logged.
6277
6278 It is strongly discouraged to use this option as most of the time, the key to
6279 complex issues is in the normal logs which will not be logged here. If you
6280 need to separate logs, see the "log-separate-errors" option instead.
6281
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006282 See also : "log", "dontlognull", "log-separate-errors" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006283 logging.
6284
6285
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006286option dontlognull
6287no option dontlognull
6288 Enable or disable logging of null connections
6289 May be used in sections : defaults | frontend | listen | backend
6290 yes | yes | yes | no
6291 Arguments : none
6292
6293 In certain environments, there are components which will regularly connect to
6294 various systems to ensure that they are still alive. It can be the case from
6295 another load balancer as well as from monitoring systems. By default, even a
6296 simple port probe or scan will produce a log. If those connections pollute
6297 the logs too much, it is possible to enable option "dontlognull" to indicate
6298 that a connection on which no data has been transferred will not be logged,
Willy Tarreau0f228a02015-05-01 15:37:53 +02006299 which typically corresponds to those probes. Note that errors will still be
6300 returned to the client and accounted for in the stats. If this is not what is
6301 desired, option http-ignore-probes can be used instead.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006302
6303 It is generally recommended not to use this option in uncontrolled
Davor Ocelice9ed2812017-12-25 17:49:28 +01006304 environments (e.g. internet), otherwise scans and other malicious activities
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006305 would not be logged.
6306
6307 If this option has been enabled in a "defaults" section, it can be disabled
6308 in a specific instance by prepending the "no" keyword before it.
6309
Willy Tarreau0f228a02015-05-01 15:37:53 +02006310 See also : "log", "http-ignore-probes", "monitor-net", "monitor-uri", and
6311 section 8 about logging.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006312
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006313
Willy Tarreau87cf5142011-08-19 22:57:24 +02006314option forwardfor [ except <network> ] [ header <name> ] [ if-none ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01006315 Enable insertion of the X-Forwarded-For header to requests sent to servers
6316 May be used in sections : defaults | frontend | listen | backend
6317 yes | yes | yes | yes
6318 Arguments :
6319 <network> is an optional argument used to disable this option for sources
6320 matching <network>
Ross Westaf72a1d2008-08-03 10:51:45 +02006321 <name> an optional argument to specify a different "X-Forwarded-For"
Willy Tarreaud72758d2010-01-12 10:42:19 +01006322 header name.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006323
6324 Since HAProxy works in reverse-proxy mode, the servers see its IP address as
6325 their client address. This is sometimes annoying when the client's IP address
6326 is expected in server logs. To solve this problem, the well-known HTTP header
6327 "X-Forwarded-For" may be added by HAProxy to all requests sent to the server.
6328 This header contains a value representing the client's IP address. Since this
6329 header is always appended at the end of the existing header list, the server
6330 must be configured to always use the last occurrence of this header only. See
Ross Westaf72a1d2008-08-03 10:51:45 +02006331 the server's manual to find how to enable use of this standard header. Note
6332 that only the last occurrence of the header must be used, since it is really
6333 possible that the client has already brought one.
6334
Willy Tarreaud72758d2010-01-12 10:42:19 +01006335 The keyword "header" may be used to supply a different header name to replace
Ross Westaf72a1d2008-08-03 10:51:45 +02006336 the default "X-Forwarded-For". This can be useful where you might already
Davor Ocelice9ed2812017-12-25 17:49:28 +01006337 have a "X-Forwarded-For" header from a different application (e.g. stunnel),
Willy Tarreaud72758d2010-01-12 10:42:19 +01006338 and you need preserve it. Also if your backend server doesn't use the
Davor Ocelice9ed2812017-12-25 17:49:28 +01006339 "X-Forwarded-For" header and requires different one (e.g. Zeus Web Servers
Ross Westaf72a1d2008-08-03 10:51:45 +02006340 require "X-Cluster-Client-IP").
Willy Tarreauc27debf2008-01-06 08:57:02 +01006341
6342 Sometimes, a same HAProxy instance may be shared between a direct client
6343 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
6344 used to decrypt HTTPS traffic). It is possible to disable the addition of the
6345 header for a known source address or network by adding the "except" keyword
6346 followed by the network address. In this case, any source IP matching the
6347 network will not cause an addition of this header. Most common uses are with
6348 private networks or 127.0.0.1.
6349
Willy Tarreau87cf5142011-08-19 22:57:24 +02006350 Alternatively, the keyword "if-none" states that the header will only be
6351 added if it is not present. This should only be used in perfectly trusted
6352 environment, as this might cause a security issue if headers reaching haproxy
6353 are under the control of the end-user.
6354
Willy Tarreauc27debf2008-01-06 08:57:02 +01006355 This option may be specified either in the frontend or in the backend. If at
Ross Westaf72a1d2008-08-03 10:51:45 +02006356 least one of them uses it, the header will be added. Note that the backend's
6357 setting of the header subargument takes precedence over the frontend's if
Willy Tarreau87cf5142011-08-19 22:57:24 +02006358 both are defined. In the case of the "if-none" argument, if at least one of
6359 the frontend or the backend does not specify it, it wants the addition to be
6360 mandatory, so it wins.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006361
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02006362 Example :
Willy Tarreauc27debf2008-01-06 08:57:02 +01006363 # Public HTTP address also used by stunnel on the same machine
6364 frontend www
6365 mode http
6366 option forwardfor except 127.0.0.1 # stunnel already adds the header
6367
Ross Westaf72a1d2008-08-03 10:51:45 +02006368 # Those servers want the IP Address in X-Client
6369 backend www
6370 mode http
6371 option forwardfor header X-Client
6372
Willy Tarreau87cf5142011-08-19 22:57:24 +02006373 See also : "option httpclose", "option http-server-close",
Christopher Faulet315b39c2018-09-21 16:26:19 +02006374 "option http-keep-alive"
Willy Tarreauc27debf2008-01-06 08:57:02 +01006375
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006376
Christopher Faulet98fbe952019-07-22 16:18:24 +02006377option h1-case-adjust-bogus-client
6378no option h1-case-adjust-bogus-client
6379 Enable or disable the case adjustment of HTTP/1 headers sent to bogus clients
6380 May be used in sections : defaults | frontend | listen | backend
6381 yes | yes | yes | no
6382 Arguments : none
6383
6384 There is no standard case for header names because, as stated in RFC7230,
6385 they are case-insensitive. So applications must handle them in a case-
6386 insensitive manner. But some bogus applications violate the standards and
6387 erroneously rely on the cases most commonly used by browsers. This problem
6388 becomes critical with HTTP/2 because all header names must be exchanged in
6389 lower case, and HAProxy follows the same convention. All header names are
6390 sent in lower case to clients and servers, regardless of the HTTP version.
6391
6392 When HAProxy receives an HTTP/1 response, its header names are converted to
6393 lower case and manipulated and sent this way to the clients. If a client is
6394 known to violate the HTTP standards and to fail to process a response coming
6395 from HAProxy, it is possible to transform the lower case header names to a
6396 different format when the response is formatted and sent to the client, by
6397 enabling this option and specifying the list of headers to be reformatted
6398 using the global directives "h1-case-adjust" or "h1-case-adjust-file". This
6399 must only be a temporary workaround for the time it takes the client to be
6400 fixed, because clients which require such workarounds might be vulnerable to
6401 content smuggling attacks and must absolutely be fixed.
6402
6403 Please note that this option will not affect standards-compliant clients.
6404
6405 If this option has been enabled in a "defaults" section, it can be disabled
6406 in a specific instance by prepending the "no" keyword before it.
6407
6408 See also: "option h1-case-adjust-bogus-server", "h1-case-adjust",
6409 "h1-case-adjust-file".
6410
6411
6412option h1-case-adjust-bogus-server
6413no option h1-case-adjust-bogus-server
6414 Enable or disable the case adjustment of HTTP/1 headers sent to bogus servers
6415 May be used in sections : defaults | frontend | listen | backend
6416 yes | no | yes | yes
6417 Arguments : none
6418
6419 There is no standard case for header names because, as stated in RFC7230,
6420 they are case-insensitive. So applications must handle them in a case-
6421 insensitive manner. But some bogus applications violate the standards and
6422 erroneously rely on the cases most commonly used by browsers. This problem
6423 becomes critical with HTTP/2 because all header names must be exchanged in
6424 lower case, and HAProxy follows the same convention. All header names are
6425 sent in lower case to clients and servers, regardless of the HTTP version.
6426
6427 When HAProxy receives an HTTP/1 request, its header names are converted to
6428 lower case and manipulated and sent this way to the servers. If a server is
6429 known to violate the HTTP standards and to fail to process a request coming
6430 from HAProxy, it is possible to transform the lower case header names to a
6431 different format when the request is formatted and sent to the server, by
6432 enabling this option and specifying the list of headers to be reformatted
6433 using the global directives "h1-case-adjust" or "h1-case-adjust-file". This
6434 must only be a temporary workaround for the time it takes the server to be
6435 fixed, because servers which require such workarounds might be vulnerable to
6436 content smuggling attacks and must absolutely be fixed.
6437
6438 Please note that this option will not affect standards-compliant servers.
6439
6440 If this option has been enabled in a "defaults" section, it can be disabled
6441 in a specific instance by prepending the "no" keyword before it.
6442
6443 See also: "option h1-case-adjust-bogus-client", "h1-case-adjust",
6444 "h1-case-adjust-file".
6445
6446
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02006447option http-buffer-request
6448no option http-buffer-request
6449 Enable or disable waiting for whole HTTP request body before proceeding
6450 May be used in sections : defaults | frontend | listen | backend
6451 yes | yes | yes | yes
6452 Arguments : none
6453
6454 It is sometimes desirable to wait for the body of an HTTP request before
6455 taking a decision. This is what is being done by "balance url_param" for
6456 example. The first use case is to buffer requests from slow clients before
6457 connecting to the server. Another use case consists in taking the routing
6458 decision based on the request body's contents. This option placed in a
6459 frontend or backend forces the HTTP processing to wait until either the whole
Christopher Faulet6db8a2e2019-11-19 16:27:25 +01006460 body is received or the request buffer is full. It can have undesired side
6461 effects with some applications abusing HTTP by expecting unbuffered
6462 transmissions between the frontend and the backend, so this should definitely
6463 not be used by default.
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02006464
Baptiste Assmanneccdf432015-10-28 13:49:01 +01006465 See also : "option http-no-delay", "timeout http-request"
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02006466
6467
Willy Tarreau0f228a02015-05-01 15:37:53 +02006468option http-ignore-probes
6469no option http-ignore-probes
6470 Enable or disable logging of null connections and request timeouts
6471 May be used in sections : defaults | frontend | listen | backend
6472 yes | yes | yes | no
6473 Arguments : none
6474
6475 Recently some browsers started to implement a "pre-connect" feature
6476 consisting in speculatively connecting to some recently visited web sites
6477 just in case the user would like to visit them. This results in many
6478 connections being established to web sites, which end up in 408 Request
6479 Timeout if the timeout strikes first, or 400 Bad Request when the browser
6480 decides to close them first. These ones pollute the log and feed the error
6481 counters. There was already "option dontlognull" but it's insufficient in
6482 this case. Instead, this option does the following things :
6483 - prevent any 400/408 message from being sent to the client if nothing
Davor Ocelice9ed2812017-12-25 17:49:28 +01006484 was received over a connection before it was closed;
6485 - prevent any log from being emitted in this situation;
Willy Tarreau0f228a02015-05-01 15:37:53 +02006486 - prevent any error counter from being incremented
6487
6488 That way the empty connection is silently ignored. Note that it is better
6489 not to use this unless it is clear that it is needed, because it will hide
6490 real problems. The most common reason for not receiving a request and seeing
6491 a 408 is due to an MTU inconsistency between the client and an intermediary
6492 element such as a VPN, which blocks too large packets. These issues are
6493 generally seen with POST requests as well as GET with large cookies. The logs
6494 are often the only way to detect them.
6495
6496 If this option has been enabled in a "defaults" section, it can be disabled
6497 in a specific instance by prepending the "no" keyword before it.
6498
6499 See also : "log", "dontlognull", "errorfile", and section 8 about logging.
6500
6501
Willy Tarreau16bfb022010-01-16 19:48:41 +01006502option http-keep-alive
6503no option http-keep-alive
6504 Enable or disable HTTP keep-alive from client to server
6505 May be used in sections : defaults | frontend | listen | backend
6506 yes | yes | yes | yes
6507 Arguments : none
6508
Willy Tarreau70dffda2014-01-30 03:07:23 +01006509 By default HAProxy operates in keep-alive mode with regards to persistent
6510 connections: for each connection it processes each request and response, and
Christopher Faulet315b39c2018-09-21 16:26:19 +02006511 leaves the connection idle on both sides between the end of a response and
6512 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02006513 as "option http-server-close" or "option httpclose". This option allows to
6514 set back the keep-alive mode, which can be useful when another mode was used
6515 in a defaults section.
Willy Tarreau70dffda2014-01-30 03:07:23 +01006516
6517 Setting "option http-keep-alive" enables HTTP keep-alive mode on the client-
6518 and server- sides. This provides the lowest latency on the client side (slow
Willy Tarreau16bfb022010-01-16 19:48:41 +01006519 network) and the fastest session reuse on the server side at the expense
6520 of maintaining idle connections to the servers. In general, it is possible
6521 with this option to achieve approximately twice the request rate that the
6522 "http-server-close" option achieves on small objects. There are mainly two
6523 situations where this option may be useful :
6524
6525 - when the server is non-HTTP compliant and authenticates the connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01006526 instead of requests (e.g. NTLM authentication)
Willy Tarreau16bfb022010-01-16 19:48:41 +01006527
6528 - when the cost of establishing the connection to the server is significant
6529 compared to the cost of retrieving the associated object from the server.
6530
6531 This last case can happen when the server is a fast static server of cache.
6532 In this case, the server will need to be properly tuned to support high enough
6533 connection counts because connections will last until the client sends another
6534 request.
6535
6536 If the client request has to go to another backend or another server due to
6537 content switching or the load balancing algorithm, the idle connection will
Willy Tarreau9420b122013-12-15 18:58:25 +01006538 immediately be closed and a new one re-opened. Option "prefer-last-server" is
6539 available to try optimize server selection so that if the server currently
6540 attached to an idle connection is usable, it will be used.
Willy Tarreau16bfb022010-01-16 19:48:41 +01006541
Willy Tarreau16bfb022010-01-16 19:48:41 +01006542 At the moment, logs will not indicate whether requests came from the same
6543 session or not. The accept date reported in the logs corresponds to the end
6544 of the previous request, and the request time corresponds to the time spent
6545 waiting for a new request. The keep-alive request time is still bound to the
6546 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
6547 not set.
6548
Christopher Faulet159e6672019-07-16 15:09:52 +02006549 This option disables and replaces any previous "option httpclose" or "option
6550 http-server-close". When backend and frontend options differ, all of these 4
6551 options have precedence over "option http-keep-alive".
Willy Tarreau16bfb022010-01-16 19:48:41 +01006552
Christopher Faulet315b39c2018-09-21 16:26:19 +02006553 See also : "option httpclose",, "option http-server-close",
Willy Tarreau9420b122013-12-15 18:58:25 +01006554 "option prefer-last-server", "option http-pretend-keepalive",
Frédéric Lécaille93d33162019-03-06 09:35:59 +01006555 and "1.1. The HTTP transaction model".
Willy Tarreau16bfb022010-01-16 19:48:41 +01006556
6557
Willy Tarreau96e31212011-05-30 18:10:30 +02006558option http-no-delay
6559no option http-no-delay
6560 Instruct the system to favor low interactive delays over performance in HTTP
6561 May be used in sections : defaults | frontend | listen | backend
6562 yes | yes | yes | yes
6563 Arguments : none
6564
6565 In HTTP, each payload is unidirectional and has no notion of interactivity.
6566 Any agent is expected to queue data somewhat for a reasonably low delay.
6567 There are some very rare server-to-server applications that abuse the HTTP
6568 protocol and expect the payload phase to be highly interactive, with many
6569 interleaved data chunks in both directions within a single request. This is
6570 absolutely not supported by the HTTP specification and will not work across
6571 most proxies or servers. When such applications attempt to do this through
6572 haproxy, it works but they will experience high delays due to the network
6573 optimizations which favor performance by instructing the system to wait for
6574 enough data to be available in order to only send full packets. Typical
6575 delays are around 200 ms per round trip. Note that this only happens with
6576 abnormal uses. Normal uses such as CONNECT requests nor WebSockets are not
6577 affected.
6578
6579 When "option http-no-delay" is present in either the frontend or the backend
6580 used by a connection, all such optimizations will be disabled in order to
6581 make the exchanges as fast as possible. Of course this offers no guarantee on
6582 the functionality, as it may break at any other place. But if it works via
6583 HAProxy, it will work as fast as possible. This option should never be used
6584 by default, and should never be used at all unless such a buggy application
6585 is discovered. The impact of using this option is an increase of bandwidth
6586 usage and CPU usage, which may significantly lower performance in high
6587 latency environments.
6588
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02006589 See also : "option http-buffer-request"
6590
Willy Tarreau96e31212011-05-30 18:10:30 +02006591
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006592option http-pretend-keepalive
6593no option http-pretend-keepalive
6594 Define whether haproxy will announce keepalive to the server or not
6595 May be used in sections : defaults | frontend | listen | backend
Christopher Faulet98db9762018-09-21 10:25:19 +02006596 yes | no | yes | yes
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006597 Arguments : none
6598
Christopher Faulet315b39c2018-09-21 16:26:19 +02006599 When running with "option http-server-close" or "option httpclose", haproxy
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006600 adds a "Connection: close" header to the request forwarded to the server.
6601 Unfortunately, when some servers see this header, they automatically refrain
6602 from using the chunked encoding for responses of unknown length, while this
6603 is totally unrelated. The immediate effect is that this prevents haproxy from
6604 maintaining the client connection alive. A second effect is that a client or
6605 a cache could receive an incomplete response without being aware of it, and
6606 consider the response complete.
6607
6608 By setting "option http-pretend-keepalive", haproxy will make the server
6609 believe it will keep the connection alive. The server will then not fall back
6610 to the abnormal undesired above. When haproxy gets the whole response, it
6611 will close the connection with the server just as it would do with the
Christopher Faulet315b39c2018-09-21 16:26:19 +02006612 "option httpclose". That way the client gets a normal response and the
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006613 connection is correctly closed on the server side.
6614
6615 It is recommended not to enable this option by default, because most servers
6616 will more efficiently close the connection themselves after the last packet,
6617 and release its buffers slightly earlier. Also, the added packet on the
6618 network could slightly reduce the overall peak performance. However it is
6619 worth noting that when this option is enabled, haproxy will have slightly
6620 less work to do. So if haproxy is the bottleneck on the whole architecture,
6621 enabling this option might save a few CPU cycles.
6622
Christopher Faulet98db9762018-09-21 10:25:19 +02006623 This option may be set in backend and listen sections. Using it in a frontend
6624 section will be ignored and a warning will be reported during startup. It is
6625 a backend related option, so there is no real reason to set it on a
6626 frontend. This option may be combined with "option httpclose", which will
6627 cause keepalive to be announced to the server and close to be announced to
6628 the client. This practice is discouraged though.
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006629
6630 If this option has been enabled in a "defaults" section, it can be disabled
6631 in a specific instance by prepending the "no" keyword before it.
6632
Christopher Faulet315b39c2018-09-21 16:26:19 +02006633 See also : "option httpclose", "option http-server-close", and
Willy Tarreau16bfb022010-01-16 19:48:41 +01006634 "option http-keep-alive"
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006635
Willy Tarreauc27debf2008-01-06 08:57:02 +01006636
Willy Tarreaub608feb2010-01-02 22:47:18 +01006637option http-server-close
6638no option http-server-close
6639 Enable or disable HTTP connection closing on the server side
6640 May be used in sections : defaults | frontend | listen | backend
6641 yes | yes | yes | yes
6642 Arguments : none
6643
Willy Tarreau70dffda2014-01-30 03:07:23 +01006644 By default HAProxy operates in keep-alive mode with regards to persistent
6645 connections: for each connection it processes each request and response, and
6646 leaves the connection idle on both sides between the end of a response and
6647 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02006648 as "option http-server-close" or "option httpclose". Setting "option
6649 http-server-close" enables HTTP connection-close mode on the server side
6650 while keeping the ability to support HTTP keep-alive and pipelining on the
6651 client side. This provides the lowest latency on the client side (slow
6652 network) and the fastest session reuse on the server side to save server
6653 resources, similarly to "option httpclose". It also permits non-keepalive
6654 capable servers to be served in keep-alive mode to the clients if they
6655 conform to the requirements of RFC7230. Please note that some servers do not
6656 always conform to those requirements when they see "Connection: close" in the
6657 request. The effect will be that keep-alive will never be used. A workaround
6658 consists in enabling "option http-pretend-keepalive".
Willy Tarreaub608feb2010-01-02 22:47:18 +01006659
6660 At the moment, logs will not indicate whether requests came from the same
6661 session or not. The accept date reported in the logs corresponds to the end
6662 of the previous request, and the request time corresponds to the time spent
6663 waiting for a new request. The keep-alive request time is still bound to the
Willy Tarreaub16a5742010-01-10 14:46:16 +01006664 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
6665 not set.
Willy Tarreaub608feb2010-01-02 22:47:18 +01006666
6667 This option may be set both in a frontend and in a backend. It is enabled if
6668 at least one of the frontend or backend holding a connection has it enabled.
Christopher Faulet159e6672019-07-16 15:09:52 +02006669 It disables and replaces any previous "option httpclose" or "option
6670 http-keep-alive". Please check section 4 ("Proxies") to see how this option
6671 combines with others when frontend and backend options differ.
Willy Tarreaub608feb2010-01-02 22:47:18 +01006672
6673 If this option has been enabled in a "defaults" section, it can be disabled
6674 in a specific instance by prepending the "no" keyword before it.
6675
Christopher Faulet315b39c2018-09-21 16:26:19 +02006676 See also : "option httpclose", "option http-pretend-keepalive",
6677 "option http-keep-alive", and "1.1. The HTTP transaction model".
Willy Tarreaub608feb2010-01-02 22:47:18 +01006678
Willy Tarreau88d349d2010-01-25 12:15:43 +01006679option http-use-proxy-header
Cyril Bontéf0c60612010-02-06 14:44:47 +01006680no option http-use-proxy-header
Willy Tarreau88d349d2010-01-25 12:15:43 +01006681 Make use of non-standard Proxy-Connection header instead of Connection
6682 May be used in sections : defaults | frontend | listen | backend
6683 yes | yes | yes | no
6684 Arguments : none
6685
Lukas Tribus23953682017-04-28 13:24:30 +00006686 While RFC7230 explicitly states that HTTP/1.1 agents must use the
Willy Tarreau88d349d2010-01-25 12:15:43 +01006687 Connection header to indicate their wish of persistent or non-persistent
6688 connections, both browsers and proxies ignore this header for proxied
6689 connections and make use of the undocumented, non-standard Proxy-Connection
6690 header instead. The issue begins when trying to put a load balancer between
6691 browsers and such proxies, because there will be a difference between what
6692 haproxy understands and what the client and the proxy agree on.
6693
6694 By setting this option in a frontend, haproxy can automatically switch to use
6695 that non-standard header if it sees proxied requests. A proxied request is
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01006696 defined here as one where the URI begins with neither a '/' nor a '*'. This
6697 is incompatible with the HTTP tunnel mode. Note that this option can only be
6698 specified in a frontend and will affect the request along its whole life.
Willy Tarreau88d349d2010-01-25 12:15:43 +01006699
Willy Tarreau844a7e72010-01-31 21:46:18 +01006700 Also, when this option is set, a request which requires authentication will
6701 automatically switch to use proxy authentication headers if it is itself a
6702 proxied request. That makes it possible to check or enforce authentication in
6703 front of an existing proxy.
6704
Willy Tarreau88d349d2010-01-25 12:15:43 +01006705 This option should normally never be used, except in front of a proxy.
6706
Christopher Faulet315b39c2018-09-21 16:26:19 +02006707 See also : "option httpclose", and "option http-server-close".
Willy Tarreau88d349d2010-01-25 12:15:43 +01006708
Willy Tarreaud63335a2010-02-26 12:56:52 +01006709option httpchk
6710option httpchk <uri>
6711option httpchk <method> <uri>
6712option httpchk <method> <uri> <version>
6713 Enable HTTP protocol to check on the servers health
6714 May be used in sections : defaults | frontend | listen | backend
6715 yes | no | yes | yes
6716 Arguments :
6717 <method> is the optional HTTP method used with the requests. When not set,
6718 the "OPTIONS" method is used, as it generally requires low server
6719 processing and is easy to filter out from the logs. Any method
6720 may be used, though it is not recommended to invent non-standard
6721 ones.
6722
6723 <uri> is the URI referenced in the HTTP requests. It defaults to " / "
6724 which is accessible by default on almost any server, but may be
6725 changed to any other URI. Query strings are permitted.
6726
6727 <version> is the optional HTTP version string. It defaults to "HTTP/1.0"
6728 but some servers might behave incorrectly in HTTP 1.0, so turning
6729 it to HTTP/1.1 may sometimes help. Note that the Host field is
6730 mandatory in HTTP/1.1, and as a trick, it is possible to pass it
6731 after "\r\n" following the version string.
6732
6733 By default, server health checks only consist in trying to establish a TCP
6734 connection. When "option httpchk" is specified, a complete HTTP request is
6735 sent once the TCP connection is established, and responses 2xx and 3xx are
6736 considered valid, while all other ones indicate a server failure, including
6737 the lack of any response.
6738
6739 The port and interval are specified in the server configuration.
6740
6741 This option does not necessarily require an HTTP backend, it also works with
6742 plain TCP backends. This is particularly useful to check simple scripts bound
6743 to some dedicated ports using the inetd daemon.
6744
6745 Examples :
6746 # Relay HTTPS traffic to Apache instance and check service availability
6747 # using HTTP request "OPTIONS * HTTP/1.1" on port 80.
6748 backend https_relay
6749 mode tcp
6750 option httpchk OPTIONS * HTTP/1.1\r\nHost:\ www
6751 server apache1 192.168.1.1:443 check port 80
6752
Simon Hormanafc47ee2013-11-25 10:46:35 +09006753 See also : "option ssl-hello-chk", "option smtpchk", "option mysql-check",
6754 "option pgsql-check", "http-check" and the "check", "port" and
6755 "inter" server options.
Willy Tarreaud63335a2010-02-26 12:56:52 +01006756
6757
Willy Tarreauc27debf2008-01-06 08:57:02 +01006758option httpclose
6759no option httpclose
Christopher Faulet315b39c2018-09-21 16:26:19 +02006760 Enable or disable HTTP connection closing
Willy Tarreauc27debf2008-01-06 08:57:02 +01006761 May be used in sections : defaults | frontend | listen | backend
6762 yes | yes | yes | yes
6763 Arguments : none
6764
Willy Tarreau70dffda2014-01-30 03:07:23 +01006765 By default HAProxy operates in keep-alive mode with regards to persistent
6766 connections: for each connection it processes each request and response, and
6767 leaves the connection idle on both sides between the end of a response and
6768 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02006769 as "option http-server-close" or "option httpclose".
Willy Tarreau70dffda2014-01-30 03:07:23 +01006770
Christopher Faulet315b39c2018-09-21 16:26:19 +02006771 If "option httpclose" is set, HAProxy will close connections with the server
6772 and the client as soon as the request and the response are received. It will
John Roeslerfb2fce12019-07-10 15:45:51 -05006773 also check if a "Connection: close" header is already set in each direction,
Christopher Faulet315b39c2018-09-21 16:26:19 +02006774 and will add one if missing. Any "Connection" header different from "close"
6775 will also be removed.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006776
Christopher Faulet315b39c2018-09-21 16:26:19 +02006777 This option may also be combined with "option http-pretend-keepalive", which
6778 will disable sending of the "Connection: close" header, but will still cause
6779 the connection to be closed once the whole response is received.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006780
6781 This option may be set both in a frontend and in a backend. It is enabled if
6782 at least one of the frontend or backend holding a connection has it enabled.
Christopher Faulet159e6672019-07-16 15:09:52 +02006783 It disables and replaces any previous "option http-server-close" or "option
6784 http-keep-alive". Please check section 4 ("Proxies") to see how this option
6785 combines with others when frontend and backend options differ.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006786
6787 If this option has been enabled in a "defaults" section, it can be disabled
6788 in a specific instance by prepending the "no" keyword before it.
6789
Christopher Faulet315b39c2018-09-21 16:26:19 +02006790 See also : "option http-server-close" and "1.1. The HTTP transaction model".
Willy Tarreauc27debf2008-01-06 08:57:02 +01006791
6792
Emeric Brun3a058f32009-06-30 18:26:00 +02006793option httplog [ clf ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01006794 Enable logging of HTTP request, session state and timers
6795 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01006796 yes | yes | yes | no
Emeric Brun3a058f32009-06-30 18:26:00 +02006797 Arguments :
6798 clf if the "clf" argument is added, then the output format will be
6799 the CLF format instead of HAProxy's default HTTP format. You can
6800 use this when you need to feed HAProxy's logs through a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01006801 log analyzer which only support the CLF format and which is not
Emeric Brun3a058f32009-06-30 18:26:00 +02006802 extensible.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006803
6804 By default, the log output format is very poor, as it only contains the
6805 source and destination addresses, and the instance name. By specifying
6806 "option httplog", each log line turns into a much richer format including,
6807 but not limited to, the HTTP request, the connection timers, the session
6808 status, the connections numbers, the captured headers and cookies, the
6809 frontend, backend and server name, and of course the source address and
6810 ports.
6811
PiBa-NLbd556bf2014-12-11 21:31:54 +01006812 Specifying only "option httplog" will automatically clear the 'clf' mode
6813 if it was set by default.
Emeric Brun3a058f32009-06-30 18:26:00 +02006814
Guillaume de Lafond29f45602017-03-31 19:52:15 +02006815 "option httplog" overrides any previous "log-format" directive.
6816
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006817 See also : section 8 about logging.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006818
Willy Tarreau55165fe2009-05-10 12:02:55 +02006819
6820option http_proxy
6821no option http_proxy
6822 Enable or disable plain HTTP proxy mode
6823 May be used in sections : defaults | frontend | listen | backend
6824 yes | yes | yes | yes
6825 Arguments : none
6826
6827 It sometimes happens that people need a pure HTTP proxy which understands
6828 basic proxy requests without caching nor any fancy feature. In this case,
6829 it may be worth setting up an HAProxy instance with the "option http_proxy"
6830 set. In this mode, no server is declared, and the connection is forwarded to
6831 the IP address and port found in the URL after the "http://" scheme.
6832
6833 No host address resolution is performed, so this only works when pure IP
6834 addresses are passed. Since this option's usage perimeter is rather limited,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01006835 it will probably be used only by experts who know they need exactly it. This
6836 is incompatible with the HTTP tunnel mode.
Willy Tarreau55165fe2009-05-10 12:02:55 +02006837
6838 If this option has been enabled in a "defaults" section, it can be disabled
6839 in a specific instance by prepending the "no" keyword before it.
6840
6841 Example :
6842 # this backend understands HTTP proxy requests and forwards them directly.
6843 backend direct_forward
6844 option httpclose
6845 option http_proxy
6846
6847 See also : "option httpclose"
6848
Willy Tarreau211ad242009-10-03 21:45:07 +02006849
Jamie Gloudon801a0a32012-08-25 00:18:33 -04006850option independent-streams
6851no option independent-streams
6852 Enable or disable independent timeout processing for both directions
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006853 May be used in sections : defaults | frontend | listen | backend
6854 yes | yes | yes | yes
6855 Arguments : none
6856
6857 By default, when data is sent over a socket, both the write timeout and the
6858 read timeout for that socket are refreshed, because we consider that there is
6859 activity on that socket, and we have no other means of guessing if we should
6860 receive data or not.
6861
Davor Ocelice9ed2812017-12-25 17:49:28 +01006862 While this default behavior is desirable for almost all applications, there
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006863 exists a situation where it is desirable to disable it, and only refresh the
6864 read timeout if there are incoming data. This happens on sessions with large
6865 timeouts and low amounts of exchanged data such as telnet session. If the
6866 server suddenly disappears, the output data accumulates in the system's
6867 socket buffers, both timeouts are correctly refreshed, and there is no way
6868 to know the server does not receive them, so we don't timeout. However, when
6869 the underlying protocol always echoes sent data, it would be enough by itself
6870 to detect the issue using the read timeout. Note that this problem does not
6871 happen with more verbose protocols because data won't accumulate long in the
6872 socket buffers.
6873
6874 When this option is set on the frontend, it will disable read timeout updates
6875 on data sent to the client. There probably is little use of this case. When
6876 the option is set on the backend, it will disable read timeout updates on
6877 data sent to the server. Doing so will typically break large HTTP posts from
6878 slow lines, so use it with caution.
6879
Willy Tarreauce887fd2012-05-12 12:50:00 +02006880 See also : "timeout client", "timeout server" and "timeout tunnel"
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006881
6882
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02006883option ldap-check
6884 Use LDAPv3 health checks for server testing
6885 May be used in sections : defaults | frontend | listen | backend
6886 yes | no | yes | yes
6887 Arguments : none
6888
6889 It is possible to test that the server correctly talks LDAPv3 instead of just
6890 testing that it accepts the TCP connection. When this option is set, an
6891 LDAPv3 anonymous simple bind message is sent to the server, and the response
6892 is analyzed to find an LDAPv3 bind response message.
6893
6894 The server is considered valid only when the LDAP response contains success
6895 resultCode (http://tools.ietf.org/html/rfc4511#section-4.1.9).
6896
6897 Logging of bind requests is server dependent see your documentation how to
6898 configure it.
6899
6900 Example :
6901 option ldap-check
6902
6903 See also : "option httpchk"
6904
6905
Simon Horman98637e52014-06-20 12:30:16 +09006906option external-check
6907 Use external processes for server health checks
6908 May be used in sections : defaults | frontend | listen | backend
6909 yes | no | yes | yes
6910
6911 It is possible to test the health of a server using an external command.
6912 This is achieved by running the executable set using "external-check
6913 command".
6914
6915 Requires the "external-check" global to be set.
6916
6917 See also : "external-check", "external-check command", "external-check path"
6918
6919
Willy Tarreau211ad242009-10-03 21:45:07 +02006920option log-health-checks
6921no option log-health-checks
Willy Tarreaubef1b322014-05-13 21:01:39 +02006922 Enable or disable logging of health checks status updates
Willy Tarreau211ad242009-10-03 21:45:07 +02006923 May be used in sections : defaults | frontend | listen | backend
6924 yes | no | yes | yes
6925 Arguments : none
6926
Willy Tarreaubef1b322014-05-13 21:01:39 +02006927 By default, failed health check are logged if server is UP and successful
6928 health checks are logged if server is DOWN, so the amount of additional
6929 information is limited.
Willy Tarreau211ad242009-10-03 21:45:07 +02006930
Willy Tarreaubef1b322014-05-13 21:01:39 +02006931 When this option is enabled, any change of the health check status or to
6932 the server's health will be logged, so that it becomes possible to know
6933 that a server was failing occasional checks before crashing, or exactly when
6934 it failed to respond a valid HTTP status, then when the port started to
6935 reject connections, then when the server stopped responding at all.
6936
Davor Ocelice9ed2812017-12-25 17:49:28 +01006937 Note that status changes not caused by health checks (e.g. enable/disable on
Willy Tarreaubef1b322014-05-13 21:01:39 +02006938 the CLI) are intentionally not logged by this option.
Willy Tarreau211ad242009-10-03 21:45:07 +02006939
Willy Tarreaubef1b322014-05-13 21:01:39 +02006940 See also: "option httpchk", "option ldap-check", "option mysql-check",
6941 "option pgsql-check", "option redis-check", "option smtpchk",
6942 "option tcp-check", "log" and section 8 about logging.
Willy Tarreau211ad242009-10-03 21:45:07 +02006943
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006944
6945option log-separate-errors
6946no option log-separate-errors
6947 Change log level for non-completely successful connections
6948 May be used in sections : defaults | frontend | listen | backend
6949 yes | yes | yes | no
6950 Arguments : none
6951
6952 Sometimes looking for errors in logs is not easy. This option makes haproxy
6953 raise the level of logs containing potentially interesting information such
6954 as errors, timeouts, retries, redispatches, or HTTP status codes 5xx. The
6955 level changes from "info" to "err". This makes it possible to log them
6956 separately to a different file with most syslog daemons. Be careful not to
6957 remove them from the original file, otherwise you would lose ordering which
6958 provides very important information.
6959
6960 Using this option, large sites dealing with several thousand connections per
6961 second may log normal traffic to a rotating buffer and only archive smaller
6962 error logs.
6963
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006964 See also : "log", "dontlognull", "dontlog-normal" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006965 logging.
6966
Willy Tarreauc27debf2008-01-06 08:57:02 +01006967
6968option logasap
6969no option logasap
6970 Enable or disable early logging of HTTP requests
6971 May be used in sections : defaults | frontend | listen | backend
6972 yes | yes | yes | no
6973 Arguments : none
6974
6975 By default, HTTP requests are logged upon termination so that the total
6976 transfer time and the number of bytes appear in the logs. When large objects
6977 are being transferred, it may take a while before the request appears in the
6978 logs. Using "option logasap", the request gets logged as soon as the server
6979 sends the complete headers. The only missing information in the logs will be
6980 the total number of bytes which will indicate everything except the amount
6981 of data transferred, and the total time which will not take the transfer
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01006982 time into account. In such a situation, it's a good practice to capture the
Willy Tarreauc27debf2008-01-06 08:57:02 +01006983 "Content-Length" response header so that the logs at least indicate how many
6984 bytes are expected to be transferred.
6985
Willy Tarreaucc6c8912009-02-22 10:53:55 +01006986 Examples :
6987 listen http_proxy 0.0.0.0:80
6988 mode http
6989 option httplog
6990 option logasap
6991 log 192.168.2.200 local3
6992
6993 >>> Feb 6 12:14:14 localhost \
6994 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
6995 static/srv1 9/10/7/14/+30 200 +243 - - ---- 3/1/1/1/0 1/0 \
6996 "GET /image.iso HTTP/1.0"
6997
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006998 See also : "option httplog", "capture response header", and section 8 about
Willy Tarreauc27debf2008-01-06 08:57:02 +01006999 logging.
7000
7001
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02007002option mysql-check [ user <username> [ post-41 ] ]
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02007003 Use MySQL health checks for server testing
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01007004 May be used in sections : defaults | frontend | listen | backend
7005 yes | no | yes | yes
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02007006 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +02007007 <username> This is the username which will be used when connecting to MySQL
7008 server.
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02007009 post-41 Send post v4.1 client compatible checks
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02007010
7011 If you specify a username, the check consists of sending two MySQL packet,
7012 one Client Authentication packet, and one QUIT packet, to correctly close
Davor Ocelice9ed2812017-12-25 17:49:28 +01007013 MySQL session. We then parse the MySQL Handshake Initialization packet and/or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02007014 Error packet. It is a basic but useful test which does not produce error nor
7015 aborted connect on the server. However, it requires adding an authorization
7016 in the MySQL table, like this :
7017
7018 USE mysql;
7019 INSERT INTO user (Host,User) values ('<ip_of_haproxy>','<username>');
7020 FLUSH PRIVILEGES;
7021
7022 If you don't specify a username (it is deprecated and not recommended), the
Davor Ocelice9ed2812017-12-25 17:49:28 +01007023 check only consists in parsing the Mysql Handshake Initialization packet or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02007024 Error packet, we don't send anything in this mode. It was reported that it
7025 can generate lockout if check is too frequent and/or if there is not enough
7026 traffic. In fact, you need in this case to check MySQL "max_connect_errors"
7027 value as if a connection is established successfully within fewer than MySQL
7028 "max_connect_errors" attempts after a previous connection was interrupted,
7029 the error count for the host is cleared to zero. If HAProxy's server get
7030 blocked, the "FLUSH HOSTS" statement is the only way to unblock it.
7031
7032 Remember that this does not check database presence nor database consistency.
7033 To do this, you can use an external check with xinetd for example.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01007034
Hervé COMMOWICK212f7782011-06-10 14:05:59 +02007035 The check requires MySQL >=3.22, for older version, please use TCP check.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01007036
7037 Most often, an incoming MySQL server needs to see the client's IP address for
7038 various purposes, including IP privilege matching and connection logging.
7039 When possible, it is often wise to masquerade the client's IP address when
7040 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02007041 which requires the transparent proxy feature to be compiled in, and the MySQL
7042 server to route the client via the machine hosting haproxy.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01007043
7044 See also: "option httpchk"
7045
7046
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007047option nolinger
7048no option nolinger
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007049 Enable or disable immediate session resource cleaning after close
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007050 May be used in sections: defaults | frontend | listen | backend
7051 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007052 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007053
Davor Ocelice9ed2812017-12-25 17:49:28 +01007054 When clients or servers abort connections in a dirty way (e.g. they are
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007055 physically disconnected), the session timeouts triggers and the session is
7056 closed. But it will remain in FIN_WAIT1 state for some time in the system,
7057 using some resources and possibly limiting the ability to establish newer
7058 connections.
7059
7060 When this happens, it is possible to activate "option nolinger" which forces
7061 the system to immediately remove any socket's pending data on close. Thus,
7062 the session is instantly purged from the system's tables. This usually has
7063 side effects such as increased number of TCP resets due to old retransmits
7064 getting immediately rejected. Some firewalls may sometimes complain about
7065 this too.
7066
7067 For this reason, it is not recommended to use this option when not absolutely
7068 needed. You know that you need it when you have thousands of FIN_WAIT1
7069 sessions on your system (TIME_WAIT ones do not count).
7070
7071 This option may be used both on frontends and backends, depending on the side
7072 where it is required. Use it on the frontend for clients, and on the backend
7073 for servers.
7074
7075 If this option has been enabled in a "defaults" section, it can be disabled
7076 in a specific instance by prepending the "no" keyword before it.
7077
7078
Willy Tarreau55165fe2009-05-10 12:02:55 +02007079option originalto [ except <network> ] [ header <name> ]
7080 Enable insertion of the X-Original-To header to requests sent to servers
7081 May be used in sections : defaults | frontend | listen | backend
7082 yes | yes | yes | yes
7083 Arguments :
7084 <network> is an optional argument used to disable this option for sources
7085 matching <network>
7086 <name> an optional argument to specify a different "X-Original-To"
7087 header name.
7088
7089 Since HAProxy can work in transparent mode, every request from a client can
7090 be redirected to the proxy and HAProxy itself can proxy every request to a
7091 complex SQUID environment and the destination host from SO_ORIGINAL_DST will
7092 be lost. This is annoying when you want access rules based on destination ip
7093 addresses. To solve this problem, a new HTTP header "X-Original-To" may be
7094 added by HAProxy to all requests sent to the server. This header contains a
7095 value representing the original destination IP address. Since this must be
7096 configured to always use the last occurrence of this header only. Note that
7097 only the last occurrence of the header must be used, since it is really
7098 possible that the client has already brought one.
7099
7100 The keyword "header" may be used to supply a different header name to replace
7101 the default "X-Original-To". This can be useful where you might already
7102 have a "X-Original-To" header from a different application, and you need
7103 preserve it. Also if your backend server doesn't use the "X-Original-To"
7104 header and requires different one.
7105
7106 Sometimes, a same HAProxy instance may be shared between a direct client
7107 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
7108 used to decrypt HTTPS traffic). It is possible to disable the addition of the
7109 header for a known source address or network by adding the "except" keyword
7110 followed by the network address. In this case, any source IP matching the
7111 network will not cause an addition of this header. Most common uses are with
7112 private networks or 127.0.0.1.
7113
7114 This option may be specified either in the frontend or in the backend. If at
7115 least one of them uses it, the header will be added. Note that the backend's
7116 setting of the header subargument takes precedence over the frontend's if
7117 both are defined.
7118
Willy Tarreau55165fe2009-05-10 12:02:55 +02007119 Examples :
7120 # Original Destination address
7121 frontend www
7122 mode http
7123 option originalto except 127.0.0.1
7124
7125 # Those servers want the IP Address in X-Client-Dst
7126 backend www
7127 mode http
7128 option originalto header X-Client-Dst
7129
Christopher Faulet315b39c2018-09-21 16:26:19 +02007130 See also : "option httpclose", "option http-server-close".
Willy Tarreau55165fe2009-05-10 12:02:55 +02007131
7132
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007133option persist
7134no option persist
7135 Enable or disable forced persistence on down servers
7136 May be used in sections: defaults | frontend | listen | backend
7137 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007138 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007139
7140 When an HTTP request reaches a backend with a cookie which references a dead
7141 server, by default it is redispatched to another server. It is possible to
7142 force the request to be sent to the dead server first using "option persist"
7143 if absolutely needed. A common use case is when servers are under extreme
7144 load and spend their time flapping. In this case, the users would still be
7145 directed to the server they opened the session on, in the hope they would be
7146 correctly served. It is recommended to use "option redispatch" in conjunction
7147 with this option so that in the event it would not be possible to connect to
7148 the server at all (server definitely dead), the client would finally be
7149 redirected to another valid server.
7150
7151 If this option has been enabled in a "defaults" section, it can be disabled
7152 in a specific instance by prepending the "no" keyword before it.
7153
Willy Tarreau4de91492010-01-22 19:10:05 +01007154 See also : "option redispatch", "retries", "force-persist"
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007155
7156
Willy Tarreau0c122822013-12-15 18:49:01 +01007157option pgsql-check [ user <username> ]
7158 Use PostgreSQL health checks for server testing
7159 May be used in sections : defaults | frontend | listen | backend
7160 yes | no | yes | yes
7161 Arguments :
7162 <username> This is the username which will be used when connecting to
7163 PostgreSQL server.
7164
7165 The check sends a PostgreSQL StartupMessage and waits for either
7166 Authentication request or ErrorResponse message. It is a basic but useful
7167 test which does not produce error nor aborted connect on the server.
7168 This check is identical with the "mysql-check".
7169
7170 See also: "option httpchk"
7171
7172
Willy Tarreau9420b122013-12-15 18:58:25 +01007173option prefer-last-server
7174no option prefer-last-server
7175 Allow multiple load balanced requests to remain on the same server
7176 May be used in sections: defaults | frontend | listen | backend
7177 yes | no | yes | yes
7178 Arguments : none
7179
7180 When the load balancing algorithm in use is not deterministic, and a previous
7181 request was sent to a server to which haproxy still holds a connection, it is
7182 sometimes desirable that subsequent requests on a same session go to the same
7183 server as much as possible. Note that this is different from persistence, as
7184 we only indicate a preference which haproxy tries to apply without any form
7185 of warranty. The real use is for keep-alive connections sent to servers. When
7186 this option is used, haproxy will try to reuse the same connection that is
7187 attached to the server instead of rebalancing to another server, causing a
7188 close of the connection. This can make sense for static file servers. It does
Willy Tarreau068621e2013-12-23 15:11:25 +01007189 not make much sense to use this in combination with hashing algorithms. Note,
7190 haproxy already automatically tries to stick to a server which sends a 401 or
Lukas Tribus80512b12018-10-27 20:07:40 +02007191 to a proxy which sends a 407 (authentication required), when the load
7192 balancing algorithm is not deterministic. This is mandatory for use with the
7193 broken NTLM authentication challenge, and significantly helps in
Willy Tarreau068621e2013-12-23 15:11:25 +01007194 troubleshooting some faulty applications. Option prefer-last-server might be
7195 desirable in these environments as well, to avoid redistributing the traffic
7196 after every other response.
Willy Tarreau9420b122013-12-15 18:58:25 +01007197
7198 If this option has been enabled in a "defaults" section, it can be disabled
7199 in a specific instance by prepending the "no" keyword before it.
7200
7201 See also: "option http-keep-alive"
7202
7203
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007204option redispatch
Joseph Lynch726ab712015-05-11 23:25:34 -07007205option redispatch <interval>
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007206no option redispatch
7207 Enable or disable session redistribution in case of connection failure
7208 May be used in sections: defaults | frontend | listen | backend
7209 yes | no | yes | yes
Joseph Lynch726ab712015-05-11 23:25:34 -07007210 Arguments :
7211 <interval> The optional integer value that controls how often redispatches
7212 occur when retrying connections. Positive value P indicates a
7213 redispatch is desired on every Pth retry, and negative value
Davor Ocelice9ed2812017-12-25 17:49:28 +01007214 N indicate a redispatch is desired on the Nth retry prior to the
Joseph Lynch726ab712015-05-11 23:25:34 -07007215 last retry. For example, the default of -1 preserves the
Davor Ocelice9ed2812017-12-25 17:49:28 +01007216 historical behavior of redispatching on the last retry, a
Joseph Lynch726ab712015-05-11 23:25:34 -07007217 positive value of 1 would indicate a redispatch on every retry,
7218 and a positive value of 3 would indicate a redispatch on every
7219 third retry. You can disable redispatches with a value of 0.
7220
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007221
7222 In HTTP mode, if a server designated by a cookie is down, clients may
7223 definitely stick to it because they cannot flush the cookie, so they will not
7224 be able to access the service anymore.
7225
Willy Tarreau59884a62019-01-02 14:48:31 +01007226 Specifying "option redispatch" will allow the proxy to break cookie or
7227 consistent hash based persistence and redistribute them to a working server.
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007228
Joseph Lynch726ab712015-05-11 23:25:34 -07007229 It also allows to retry connections to another server in case of multiple
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007230 connection failures. Of course, it requires having "retries" set to a nonzero
7231 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01007232
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007233 If this option has been enabled in a "defaults" section, it can be disabled
7234 in a specific instance by prepending the "no" keyword before it.
7235
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02007236 See also : "retries", "force-persist"
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007237
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007238
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02007239option redis-check
7240 Use redis health checks for server testing
7241 May be used in sections : defaults | frontend | listen | backend
7242 yes | no | yes | yes
7243 Arguments : none
7244
7245 It is possible to test that the server correctly talks REDIS protocol instead
7246 of just testing that it accepts the TCP connection. When this option is set,
7247 a PING redis command is sent to the server, and the response is analyzed to
7248 find the "+PONG" response message.
7249
7250 Example :
7251 option redis-check
7252
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03007253 See also : "option httpchk", "option tcp-check", "tcp-check expect"
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02007254
7255
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007256option smtpchk
7257option smtpchk <hello> <domain>
7258 Use SMTP health checks for server testing
7259 May be used in sections : defaults | frontend | listen | backend
7260 yes | no | yes | yes
Willy Tarreaud72758d2010-01-12 10:42:19 +01007261 Arguments :
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007262 <hello> is an optional argument. It is the "hello" command to use. It can
Lukas Tribus27935782018-10-01 02:00:16 +02007263 be either "HELO" (for SMTP) or "EHLO" (for ESMTP). All other
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007264 values will be turned into the default command ("HELO").
7265
7266 <domain> is the domain name to present to the server. It may only be
7267 specified (and is mandatory) if the hello command has been
7268 specified. By default, "localhost" is used.
7269
7270 When "option smtpchk" is set, the health checks will consist in TCP
7271 connections followed by an SMTP command. By default, this command is
7272 "HELO localhost". The server's return code is analyzed and only return codes
7273 starting with a "2" will be considered as valid. All other responses,
7274 including a lack of response will constitute an error and will indicate a
7275 dead server.
7276
7277 This test is meant to be used with SMTP servers or relays. Depending on the
7278 request, it is possible that some servers do not log each connection attempt,
Davor Ocelice9ed2812017-12-25 17:49:28 +01007279 so you may want to experiment to improve the behavior. Using telnet on port
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007280 25 is often easier than adjusting the configuration.
7281
7282 Most often, an incoming SMTP server needs to see the client's IP address for
7283 various purposes, including spam filtering, anti-spoofing and logging. When
7284 possible, it is often wise to masquerade the client's IP address when
7285 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02007286 which requires the transparent proxy feature to be compiled in.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007287
7288 Example :
7289 option smtpchk HELO mydomain.org
7290
7291 See also : "option httpchk", "source"
7292
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007293
Krzysztof Piotr Oledzkiaeebf9b2009-10-04 15:43:17 +02007294option socket-stats
7295no option socket-stats
7296
7297 Enable or disable collecting & providing separate statistics for each socket.
7298 May be used in sections : defaults | frontend | listen | backend
7299 yes | yes | yes | no
7300
7301 Arguments : none
7302
7303
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007304option splice-auto
7305no option splice-auto
7306 Enable or disable automatic kernel acceleration on sockets in both directions
7307 May be used in sections : defaults | frontend | listen | backend
7308 yes | yes | yes | yes
7309 Arguments : none
7310
7311 When this option is enabled either on a frontend or on a backend, haproxy
7312 will automatically evaluate the opportunity to use kernel tcp splicing to
Davor Ocelice9ed2812017-12-25 17:49:28 +01007313 forward data between the client and the server, in either direction. HAProxy
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007314 uses heuristics to estimate if kernel splicing might improve performance or
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007315 not. Both directions are handled independently. Note that the heuristics used
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007316 are not much aggressive in order to limit excessive use of splicing. This
7317 option requires splicing to be enabled at compile time, and may be globally
7318 disabled with the global option "nosplice". Since splice uses pipes, using it
7319 requires that there are enough spare pipes.
7320
7321 Important note: kernel-based TCP splicing is a Linux-specific feature which
7322 first appeared in kernel 2.6.25. It offers kernel-based acceleration to
7323 transfer data between sockets without copying these data to user-space, thus
7324 providing noticeable performance gains and CPU cycles savings. Since many
7325 early implementations are buggy, corrupt data and/or are inefficient, this
7326 feature is not enabled by default, and it should be used with extreme care.
7327 While it is not possible to detect the correctness of an implementation,
7328 2.6.29 is the first version offering a properly working implementation. In
7329 case of doubt, splicing may be globally disabled using the global "nosplice"
7330 keyword.
7331
7332 Example :
7333 option splice-auto
7334
7335 If this option has been enabled in a "defaults" section, it can be disabled
7336 in a specific instance by prepending the "no" keyword before it.
7337
7338 See also : "option splice-request", "option splice-response", and global
7339 options "nosplice" and "maxpipes"
7340
7341
7342option splice-request
7343no option splice-request
7344 Enable or disable automatic kernel acceleration on sockets for requests
7345 May be used in sections : defaults | frontend | listen | backend
7346 yes | yes | yes | yes
7347 Arguments : none
7348
7349 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04007350 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007351 the client to the server. It might still use the recv/send scheme if there
7352 are no spare pipes left. This option requires splicing to be enabled at
7353 compile time, and may be globally disabled with the global option "nosplice".
7354 Since splice uses pipes, using it requires that there are enough spare pipes.
7355
7356 Important note: see "option splice-auto" for usage limitations.
7357
7358 Example :
7359 option splice-request
7360
7361 If this option has been enabled in a "defaults" section, it can be disabled
7362 in a specific instance by prepending the "no" keyword before it.
7363
7364 See also : "option splice-auto", "option splice-response", and global options
7365 "nosplice" and "maxpipes"
7366
7367
7368option splice-response
7369no option splice-response
7370 Enable or disable automatic kernel acceleration on sockets for responses
7371 May be used in sections : defaults | frontend | listen | backend
7372 yes | yes | yes | yes
7373 Arguments : none
7374
7375 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04007376 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007377 the server to the client. It might still use the recv/send scheme if there
7378 are no spare pipes left. This option requires splicing to be enabled at
7379 compile time, and may be globally disabled with the global option "nosplice".
7380 Since splice uses pipes, using it requires that there are enough spare pipes.
7381
7382 Important note: see "option splice-auto" for usage limitations.
7383
7384 Example :
7385 option splice-response
7386
7387 If this option has been enabled in a "defaults" section, it can be disabled
7388 in a specific instance by prepending the "no" keyword before it.
7389
7390 See also : "option splice-auto", "option splice-request", and global options
7391 "nosplice" and "maxpipes"
7392
7393
Christopher Fauletba7bc162016-11-07 21:07:38 +01007394option spop-check
7395 Use SPOP health checks for server testing
7396 May be used in sections : defaults | frontend | listen | backend
7397 no | no | no | yes
7398 Arguments : none
7399
7400 It is possible to test that the server correctly talks SPOP protocol instead
7401 of just testing that it accepts the TCP connection. When this option is set,
7402 a HELLO handshake is performed between HAProxy and the server, and the
7403 response is analyzed to check no error is reported.
7404
7405 Example :
7406 option spop-check
7407
7408 See also : "option httpchk"
7409
7410
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007411option srvtcpka
7412no option srvtcpka
7413 Enable or disable the sending of TCP keepalive packets on the server side
7414 May be used in sections : defaults | frontend | listen | backend
7415 yes | no | yes | yes
7416 Arguments : none
7417
7418 When there is a firewall or any session-aware component between a client and
7419 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01007420 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007421 components decides to expire a session which has remained idle for too long.
7422
7423 Enabling socket-level TCP keep-alives makes the system regularly send packets
7424 to the other end of the connection, leaving it active. The delay between
7425 keep-alive probes is controlled by the system only and depends both on the
7426 operating system and its tuning parameters.
7427
7428 It is important to understand that keep-alive packets are neither emitted nor
7429 received at the application level. It is only the network stacks which sees
7430 them. For this reason, even if one side of the proxy already uses keep-alives
7431 to maintain its connection alive, those keep-alive packets will not be
7432 forwarded to the other side of the proxy.
7433
7434 Please note that this has nothing to do with HTTP keep-alive.
7435
7436 Using option "srvtcpka" enables the emission of TCP keep-alive probes on the
7437 server side of a connection, which should help when session expirations are
7438 noticed between HAProxy and a server.
7439
7440 If this option has been enabled in a "defaults" section, it can be disabled
7441 in a specific instance by prepending the "no" keyword before it.
7442
7443 See also : "option clitcpka", "option tcpka"
7444
7445
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007446option ssl-hello-chk
7447 Use SSLv3 client hello health checks for server testing
7448 May be used in sections : defaults | frontend | listen | backend
7449 yes | no | yes | yes
7450 Arguments : none
7451
7452 When some SSL-based protocols are relayed in TCP mode through HAProxy, it is
7453 possible to test that the server correctly talks SSL instead of just testing
7454 that it accepts the TCP connection. When "option ssl-hello-chk" is set, pure
7455 SSLv3 client hello messages are sent once the connection is established to
7456 the server, and the response is analyzed to find an SSL server hello message.
7457 The server is considered valid only when the response contains this server
7458 hello message.
7459
7460 All servers tested till there correctly reply to SSLv3 client hello messages,
7461 and most servers tested do not even log the requests containing only hello
7462 messages, which is appreciable.
7463
Willy Tarreau763a95b2012-10-04 23:15:39 +02007464 Note that this check works even when SSL support was not built into haproxy
7465 because it forges the SSL message. When SSL support is available, it is best
7466 to use native SSL health checks instead of this one.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007467
Willy Tarreau763a95b2012-10-04 23:15:39 +02007468 See also: "option httpchk", "check-ssl"
7469
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007470
Willy Tarreaued179852013-12-16 01:07:00 +01007471option tcp-check
7472 Perform health checks using tcp-check send/expect sequences
7473 May be used in sections: defaults | frontend | listen | backend
7474 yes | no | yes | yes
7475
7476 This health check method is intended to be combined with "tcp-check" command
7477 lists in order to support send/expect types of health check sequences.
7478
7479 TCP checks currently support 4 modes of operations :
7480 - no "tcp-check" directive : the health check only consists in a connection
7481 attempt, which remains the default mode.
7482
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007483 - "tcp-check send" or "tcp-check send-binary" only is mentioned : this is
Willy Tarreaued179852013-12-16 01:07:00 +01007484 used to send a string along with a connection opening. With some
7485 protocols, it helps sending a "QUIT" message for example that prevents
7486 the server from logging a connection error for each health check. The
7487 check result will still be based on the ability to open the connection
7488 only.
7489
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007490 - "tcp-check expect" only is mentioned : this is used to test a banner.
Willy Tarreaued179852013-12-16 01:07:00 +01007491 The connection is opened and haproxy waits for the server to present some
7492 contents which must validate some rules. The check result will be based
7493 on the matching between the contents and the rules. This is suited for
7494 POP, IMAP, SMTP, FTP, SSH, TELNET.
7495
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007496 - both "tcp-check send" and "tcp-check expect" are mentioned : this is
Davor Ocelice9ed2812017-12-25 17:49:28 +01007497 used to test a hello-type protocol. HAProxy sends a message, the server
7498 responds and its response is analyzed. the check result will be based on
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007499 the matching between the response contents and the rules. This is often
Willy Tarreaued179852013-12-16 01:07:00 +01007500 suited for protocols which require a binding or a request/response model.
7501 LDAP, MySQL, Redis and SSL are example of such protocols, though they
7502 already all have their dedicated checks with a deeper understanding of
7503 the respective protocols.
7504 In this mode, many questions may be sent and many answers may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01007505 analyzed.
Willy Tarreaued179852013-12-16 01:07:00 +01007506
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007507 A fifth mode can be used to insert comments in different steps of the
7508 script.
7509
7510 For each tcp-check rule you create, you can add a "comment" directive,
7511 followed by a string. This string will be reported in the log and stderr
7512 in debug mode. It is useful to make user-friendly error reporting.
7513 The "comment" is of course optional.
7514
7515
Willy Tarreaued179852013-12-16 01:07:00 +01007516 Examples :
Davor Ocelice9ed2812017-12-25 17:49:28 +01007517 # perform a POP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01007518 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007519 tcp-check expect string +OK\ POP3\ ready comment POP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01007520
Davor Ocelice9ed2812017-12-25 17:49:28 +01007521 # perform an IMAP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01007522 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007523 tcp-check expect string *\ OK\ IMAP4\ ready comment IMAP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01007524
7525 # look for the redis master server after ensuring it speaks well
7526 # redis protocol, then it exits properly.
Davor Ocelice9ed2812017-12-25 17:49:28 +01007527 # (send a command then analyze the response 3 times)
Willy Tarreaued179852013-12-16 01:07:00 +01007528 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007529 tcp-check comment PING\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01007530 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02007531 tcp-check expect string +PONG
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007532 tcp-check comment role\ check
Willy Tarreaued179852013-12-16 01:07:00 +01007533 tcp-check send info\ replication\r\n
7534 tcp-check expect string role:master
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007535 tcp-check comment QUIT\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01007536 tcp-check send QUIT\r\n
7537 tcp-check expect string +OK
7538
Davor Ocelice9ed2812017-12-25 17:49:28 +01007539 forge a HTTP request, then analyze the response
Willy Tarreaued179852013-12-16 01:07:00 +01007540 (send many headers before analyzing)
7541 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007542 tcp-check comment forge\ and\ send\ HTTP\ request
Willy Tarreaued179852013-12-16 01:07:00 +01007543 tcp-check send HEAD\ /\ HTTP/1.1\r\n
7544 tcp-check send Host:\ www.mydomain.com\r\n
7545 tcp-check send User-Agent:\ HAProxy\ tcpcheck\r\n
7546 tcp-check send \r\n
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007547 tcp-check expect rstring HTTP/1\..\ (2..|3..) comment check\ HTTP\ response
Willy Tarreaued179852013-12-16 01:07:00 +01007548
7549
7550 See also : "tcp-check expect", "tcp-check send"
7551
7552
Willy Tarreau9ea05a72009-06-14 12:07:01 +02007553option tcp-smart-accept
7554no option tcp-smart-accept
7555 Enable or disable the saving of one ACK packet during the accept sequence
7556 May be used in sections : defaults | frontend | listen | backend
7557 yes | yes | yes | no
7558 Arguments : none
7559
7560 When an HTTP connection request comes in, the system acknowledges it on
7561 behalf of HAProxy, then the client immediately sends its request, and the
7562 system acknowledges it too while it is notifying HAProxy about the new
7563 connection. HAProxy then reads the request and responds. This means that we
7564 have one TCP ACK sent by the system for nothing, because the request could
7565 very well be acknowledged by HAProxy when it sends its response.
7566
7567 For this reason, in HTTP mode, HAProxy automatically asks the system to avoid
7568 sending this useless ACK on platforms which support it (currently at least
7569 Linux). It must not cause any problem, because the system will send it anyway
7570 after 40 ms if the response takes more time than expected to come.
7571
7572 During complex network debugging sessions, it may be desirable to disable
7573 this optimization because delayed ACKs can make troubleshooting more complex
7574 when trying to identify where packets are delayed. It is then possible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01007575 fall back to normal behavior by specifying "no option tcp-smart-accept".
Willy Tarreau9ea05a72009-06-14 12:07:01 +02007576
7577 It is also possible to force it for non-HTTP proxies by simply specifying
7578 "option tcp-smart-accept". For instance, it can make sense with some services
7579 such as SMTP where the server speaks first.
7580
7581 It is recommended to avoid forcing this option in a defaults section. In case
7582 of doubt, consider setting it back to automatic values by prepending the
7583 "default" keyword before it, or disabling it using the "no" keyword.
7584
Willy Tarreaud88edf22009-06-14 15:48:17 +02007585 See also : "option tcp-smart-connect"
7586
7587
7588option tcp-smart-connect
7589no option tcp-smart-connect
7590 Enable or disable the saving of one ACK packet during the connect sequence
7591 May be used in sections : defaults | frontend | listen | backend
7592 yes | no | yes | yes
7593 Arguments : none
7594
7595 On certain systems (at least Linux), HAProxy can ask the kernel not to
7596 immediately send an empty ACK upon a connection request, but to directly
7597 send the buffer request instead. This saves one packet on the network and
7598 thus boosts performance. It can also be useful for some servers, because they
7599 immediately get the request along with the incoming connection.
7600
7601 This feature is enabled when "option tcp-smart-connect" is set in a backend.
7602 It is not enabled by default because it makes network troubleshooting more
7603 complex.
7604
7605 It only makes sense to enable it with protocols where the client speaks first
7606 such as HTTP. In other situations, if there is no data to send in place of
7607 the ACK, a normal ACK is sent.
7608
7609 If this option has been enabled in a "defaults" section, it can be disabled
7610 in a specific instance by prepending the "no" keyword before it.
7611
7612 See also : "option tcp-smart-accept"
7613
Willy Tarreau9ea05a72009-06-14 12:07:01 +02007614
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007615option tcpka
7616 Enable or disable the sending of TCP keepalive packets on both sides
7617 May be used in sections : defaults | frontend | listen | backend
7618 yes | yes | yes | yes
7619 Arguments : none
7620
7621 When there is a firewall or any session-aware component between a client and
7622 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01007623 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007624 components decides to expire a session which has remained idle for too long.
7625
7626 Enabling socket-level TCP keep-alives makes the system regularly send packets
7627 to the other end of the connection, leaving it active. The delay between
7628 keep-alive probes is controlled by the system only and depends both on the
7629 operating system and its tuning parameters.
7630
7631 It is important to understand that keep-alive packets are neither emitted nor
7632 received at the application level. It is only the network stacks which sees
7633 them. For this reason, even if one side of the proxy already uses keep-alives
7634 to maintain its connection alive, those keep-alive packets will not be
7635 forwarded to the other side of the proxy.
7636
7637 Please note that this has nothing to do with HTTP keep-alive.
7638
7639 Using option "tcpka" enables the emission of TCP keep-alive probes on both
7640 the client and server sides of a connection. Note that this is meaningful
7641 only in "defaults" or "listen" sections. If this option is used in a
7642 frontend, only the client side will get keep-alives, and if this option is
7643 used in a backend, only the server side will get keep-alives. For this
7644 reason, it is strongly recommended to explicitly use "option clitcpka" and
7645 "option srvtcpka" when the configuration is split between frontends and
7646 backends.
7647
7648 See also : "option clitcpka", "option srvtcpka"
7649
Willy Tarreau844e3c52008-01-11 16:28:18 +01007650
7651option tcplog
7652 Enable advanced logging of TCP connections with session state and timers
7653 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01007654 yes | yes | yes | no
Willy Tarreau844e3c52008-01-11 16:28:18 +01007655 Arguments : none
7656
7657 By default, the log output format is very poor, as it only contains the
7658 source and destination addresses, and the instance name. By specifying
7659 "option tcplog", each log line turns into a much richer format including, but
7660 not limited to, the connection timers, the session status, the connections
7661 numbers, the frontend, backend and server name, and of course the source
7662 address and ports. This option is useful for pure TCP proxies in order to
7663 find which of the client or server disconnects or times out. For normal HTTP
7664 proxies, it's better to use "option httplog" which is even more complete.
7665
Guillaume de Lafond29f45602017-03-31 19:52:15 +02007666 "option tcplog" overrides any previous "log-format" directive.
7667
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007668 See also : "option httplog", and section 8 about logging.
Willy Tarreau844e3c52008-01-11 16:28:18 +01007669
7670
Willy Tarreau844e3c52008-01-11 16:28:18 +01007671option transparent
7672no option transparent
7673 Enable client-side transparent proxying
7674 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +01007675 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +01007676 Arguments : none
7677
7678 This option was introduced in order to provide layer 7 persistence to layer 3
7679 load balancers. The idea is to use the OS's ability to redirect an incoming
7680 connection for a remote address to a local process (here HAProxy), and let
7681 this process know what address was initially requested. When this option is
7682 used, sessions without cookies will be forwarded to the original destination
7683 IP address of the incoming request (which should match that of another
7684 equipment), while requests with cookies will still be forwarded to the
7685 appropriate server.
7686
7687 Note that contrary to a common belief, this option does NOT make HAProxy
7688 present the client's IP to the server when establishing the connection.
7689
Willy Tarreaua1146052011-03-01 09:51:54 +01007690 See also: the "usesrc" argument of the "source" keyword, and the
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007691 "transparent" option of the "bind" keyword.
Willy Tarreau844e3c52008-01-11 16:28:18 +01007692
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007693
Simon Horman98637e52014-06-20 12:30:16 +09007694external-check command <command>
7695 Executable to run when performing an external-check
7696 May be used in sections : defaults | frontend | listen | backend
7697 yes | no | yes | yes
7698
7699 Arguments :
7700 <command> is the external command to run
7701
Simon Horman98637e52014-06-20 12:30:16 +09007702 The arguments passed to the to the command are:
7703
Cyril Bonté777be862014-12-02 21:21:35 +01007704 <proxy_address> <proxy_port> <server_address> <server_port>
Simon Horman98637e52014-06-20 12:30:16 +09007705
Cyril Bonté777be862014-12-02 21:21:35 +01007706 The <proxy_address> and <proxy_port> are derived from the first listener
7707 that is either IPv4, IPv6 or a UNIX socket. In the case of a UNIX socket
7708 listener the proxy_address will be the path of the socket and the
7709 <proxy_port> will be the string "NOT_USED". In a backend section, it's not
7710 possible to determine a listener, and both <proxy_address> and <proxy_port>
7711 will have the string value "NOT_USED".
Simon Horman98637e52014-06-20 12:30:16 +09007712
Cyril Bonté72cda2a2014-12-27 22:28:39 +01007713 Some values are also provided through environment variables.
7714
7715 Environment variables :
7716 HAPROXY_PROXY_ADDR The first bind address if available (or empty if not
7717 applicable, for example in a "backend" section).
7718
7719 HAPROXY_PROXY_ID The backend id.
7720
7721 HAPROXY_PROXY_NAME The backend name.
7722
7723 HAPROXY_PROXY_PORT The first bind port if available (or empty if not
7724 applicable, for example in a "backend" section or
7725 for a UNIX socket).
7726
7727 HAPROXY_SERVER_ADDR The server address.
7728
7729 HAPROXY_SERVER_CURCONN The current number of connections on the server.
7730
7731 HAPROXY_SERVER_ID The server id.
7732
7733 HAPROXY_SERVER_MAXCONN The server max connections.
7734
7735 HAPROXY_SERVER_NAME The server name.
7736
7737 HAPROXY_SERVER_PORT The server port if available (or empty for a UNIX
7738 socket).
7739
7740 PATH The PATH environment variable used when executing
7741 the command may be set using "external-check path".
7742
William Lallemand4d03e432019-06-14 15:35:37 +02007743 See also "2.3. Environment variables" for other variables.
7744
Simon Horman98637e52014-06-20 12:30:16 +09007745 If the command executed and exits with a zero status then the check is
7746 considered to have passed, otherwise the check is considered to have
7747 failed.
7748
7749 Example :
7750 external-check command /bin/true
7751
7752 See also : "external-check", "option external-check", "external-check path"
7753
7754
7755external-check path <path>
7756 The value of the PATH environment variable used when running an external-check
7757 May be used in sections : defaults | frontend | listen | backend
7758 yes | no | yes | yes
7759
7760 Arguments :
7761 <path> is the path used when executing external command to run
7762
7763 The default path is "".
7764
7765 Example :
7766 external-check path "/usr/bin:/bin"
7767
7768 See also : "external-check", "option external-check",
7769 "external-check command"
7770
7771
Emeric Brun647caf12009-06-30 17:57:00 +02007772persist rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02007773persist rdp-cookie(<name>)
Emeric Brun647caf12009-06-30 17:57:00 +02007774 Enable RDP cookie-based persistence
7775 May be used in sections : defaults | frontend | listen | backend
7776 yes | no | yes | yes
7777 Arguments :
7778 <name> is the optional name of the RDP cookie to check. If omitted, the
Willy Tarreau61e28f22010-05-16 22:31:05 +02007779 default cookie name "msts" will be used. There currently is no
7780 valid reason to change this name.
Emeric Brun647caf12009-06-30 17:57:00 +02007781
7782 This statement enables persistence based on an RDP cookie. The RDP cookie
7783 contains all information required to find the server in the list of known
Davor Ocelice9ed2812017-12-25 17:49:28 +01007784 servers. So when this option is set in the backend, the request is analyzed
Emeric Brun647caf12009-06-30 17:57:00 +02007785 and if an RDP cookie is found, it is decoded. If it matches a known server
7786 which is still UP (or if "option persist" is set), then the connection is
7787 forwarded to this server.
7788
7789 Note that this only makes sense in a TCP backend, but for this to work, the
7790 frontend must have waited long enough to ensure that an RDP cookie is present
7791 in the request buffer. This is the same requirement as with the "rdp-cookie"
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007792 load-balancing method. Thus it is highly recommended to put all statements in
Emeric Brun647caf12009-06-30 17:57:00 +02007793 a single "listen" section.
7794
Willy Tarreau61e28f22010-05-16 22:31:05 +02007795 Also, it is important to understand that the terminal server will emit this
7796 RDP cookie only if it is configured for "token redirection mode", which means
7797 that the "IP address redirection" option is disabled.
7798
Emeric Brun647caf12009-06-30 17:57:00 +02007799 Example :
7800 listen tse-farm
7801 bind :3389
7802 # wait up to 5s for an RDP cookie in the request
7803 tcp-request inspect-delay 5s
7804 tcp-request content accept if RDP_COOKIE
7805 # apply RDP cookie persistence
7806 persist rdp-cookie
7807 # if server is unknown, let's balance on the same cookie.
Cyril Bontédc4d9032012-04-08 21:57:39 +02007808 # alternatively, "balance leastconn" may be useful too.
Emeric Brun647caf12009-06-30 17:57:00 +02007809 balance rdp-cookie
7810 server srv1 1.1.1.1:3389
7811 server srv2 1.1.1.2:3389
7812
Simon Hormanab814e02011-06-24 14:50:20 +09007813 See also : "balance rdp-cookie", "tcp-request", the "req_rdp_cookie" ACL and
7814 the rdp_cookie pattern fetch function.
Emeric Brun647caf12009-06-30 17:57:00 +02007815
7816
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007817rate-limit sessions <rate>
7818 Set a limit on the number of new sessions accepted per second on a frontend
7819 May be used in sections : defaults | frontend | listen | backend
7820 yes | yes | yes | no
7821 Arguments :
7822 <rate> The <rate> parameter is an integer designating the maximum number
7823 of new sessions per second to accept on the frontend.
7824
7825 When the frontend reaches the specified number of new sessions per second, it
7826 stops accepting new connections until the rate drops below the limit again.
7827 During this time, the pending sessions will be kept in the socket's backlog
7828 (in system buffers) and haproxy will not even be aware that sessions are
7829 pending. When applying very low limit on a highly loaded service, it may make
7830 sense to increase the socket's backlog using the "backlog" keyword.
7831
7832 This feature is particularly efficient at blocking connection-based attacks
7833 or service abuse on fragile servers. Since the session rate is measured every
7834 millisecond, it is extremely accurate. Also, the limit applies immediately,
7835 no delay is needed at all to detect the threshold.
7836
7837 Example : limit the connection rate on SMTP to 10 per second max
7838 listen smtp
7839 mode tcp
7840 bind :25
7841 rate-limit sessions 10
Panagiotis Panagiotopoulos7282d8e2016-02-11 16:37:15 +02007842 server smtp1 127.0.0.1:1025
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007843
Willy Tarreaua17c2d92011-07-25 08:16:20 +02007844 Note : when the maximum rate is reached, the frontend's status is not changed
7845 but its sockets appear as "WAITING" in the statistics if the
7846 "socket-stats" option is enabled.
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007847
7848 See also : the "backlog" keyword and the "fe_sess_rate" ACL criterion.
7849
7850
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007851redirect location <loc> [code <code>] <option> [{if | unless} <condition>]
7852redirect prefix <pfx> [code <code>] <option> [{if | unless} <condition>]
7853redirect scheme <sch> [code <code>] <option> [{if | unless} <condition>]
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007854 Return an HTTP redirection if/unless a condition is matched
7855 May be used in sections : defaults | frontend | listen | backend
7856 no | yes | yes | yes
7857
7858 If/unless the condition is matched, the HTTP request will lead to a redirect
Willy Tarreauf285f542010-01-03 20:03:03 +01007859 response. If no condition is specified, the redirect applies unconditionally.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007860
Willy Tarreau0140f252008-11-19 21:07:09 +01007861 Arguments :
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007862 <loc> With "redirect location", the exact value in <loc> is placed into
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007863 the HTTP "Location" header. When used in an "http-request" rule,
7864 <loc> value follows the log-format rules and can include some
7865 dynamic values (see Custom Log Format in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007866
7867 <pfx> With "redirect prefix", the "Location" header is built from the
7868 concatenation of <pfx> and the complete URI path, including the
7869 query string, unless the "drop-query" option is specified (see
7870 below). As a special case, if <pfx> equals exactly "/", then
7871 nothing is inserted before the original URI. It allows one to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007872 redirect to the same URL (for instance, to insert a cookie). When
7873 used in an "http-request" rule, <pfx> value follows the log-format
7874 rules and can include some dynamic values (see Custom Log Format
7875 in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007876
7877 <sch> With "redirect scheme", then the "Location" header is built by
7878 concatenating <sch> with "://" then the first occurrence of the
7879 "Host" header, and then the URI path, including the query string
7880 unless the "drop-query" option is specified (see below). If no
7881 path is found or if the path is "*", then "/" is used instead. If
7882 no "Host" header is found, then an empty host component will be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007883 returned, which most recent browsers interpret as redirecting to
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007884 the same host. This directive is mostly used to redirect HTTP to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007885 HTTPS. When used in an "http-request" rule, <sch> value follows
7886 the log-format rules and can include some dynamic values (see
7887 Custom Log Format in section 8.2.4).
Willy Tarreau0140f252008-11-19 21:07:09 +01007888
7889 <code> The code is optional. It indicates which type of HTTP redirection
Willy Tarreaub67fdc42013-03-29 19:28:11 +01007890 is desired. Only codes 301, 302, 303, 307 and 308 are supported,
7891 with 302 used by default if no code is specified. 301 means
7892 "Moved permanently", and a browser may cache the Location. 302
Baptiste Assmannea849c02015-08-03 11:42:50 +02007893 means "Moved temporarily" and means that the browser should not
Willy Tarreaub67fdc42013-03-29 19:28:11 +01007894 cache the redirection. 303 is equivalent to 302 except that the
7895 browser will fetch the location with a GET method. 307 is just
7896 like 302 but makes it clear that the same method must be reused.
7897 Likewise, 308 replaces 301 if the same method must be used.
Willy Tarreau0140f252008-11-19 21:07:09 +01007898
7899 <option> There are several options which can be specified to adjust the
Davor Ocelice9ed2812017-12-25 17:49:28 +01007900 expected behavior of a redirection :
Willy Tarreau0140f252008-11-19 21:07:09 +01007901
7902 - "drop-query"
7903 When this keyword is used in a prefix-based redirection, then the
7904 location will be set without any possible query-string, which is useful
7905 for directing users to a non-secure page for instance. It has no effect
7906 with a location-type redirect.
7907
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01007908 - "append-slash"
7909 This keyword may be used in conjunction with "drop-query" to redirect
7910 users who use a URL not ending with a '/' to the same one with the '/'.
7911 It can be useful to ensure that search engines will only see one URL.
7912 For this, a return code 301 is preferred.
7913
Willy Tarreau0140f252008-11-19 21:07:09 +01007914 - "set-cookie NAME[=value]"
7915 A "Set-Cookie" header will be added with NAME (and optionally "=value")
7916 to the response. This is sometimes used to indicate that a user has
7917 been seen, for instance to protect against some types of DoS. No other
7918 cookie option is added, so the cookie will be a session cookie. Note
7919 that for a browser, a sole cookie name without an equal sign is
7920 different from a cookie with an equal sign.
7921
7922 - "clear-cookie NAME[=]"
7923 A "Set-Cookie" header will be added with NAME (and optionally "="), but
7924 with the "Max-Age" attribute set to zero. This will tell the browser to
7925 delete this cookie. It is useful for instance on logout pages. It is
7926 important to note that clearing the cookie "NAME" will not remove a
7927 cookie set with "NAME=value". You have to clear the cookie "NAME=" for
7928 that, because the browser makes the difference.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007929
7930 Example: move the login URL only to HTTPS.
7931 acl clear dst_port 80
7932 acl secure dst_port 8080
7933 acl login_page url_beg /login
Willy Tarreau0140f252008-11-19 21:07:09 +01007934 acl logout url_beg /logout
Willy Tarreau79da4692008-11-19 20:03:04 +01007935 acl uid_given url_reg /login?userid=[^&]+
Willy Tarreau0140f252008-11-19 21:07:09 +01007936 acl cookie_set hdr_sub(cookie) SEEN=1
7937
7938 redirect prefix https://mysite.com set-cookie SEEN=1 if !cookie_set
Willy Tarreau79da4692008-11-19 20:03:04 +01007939 redirect prefix https://mysite.com if login_page !secure
7940 redirect prefix http://mysite.com drop-query if login_page !uid_given
7941 redirect location http://mysite.com/ if !login_page secure
Willy Tarreau0140f252008-11-19 21:07:09 +01007942 redirect location / clear-cookie USERID= if logout
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007943
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01007944 Example: send redirects for request for articles without a '/'.
7945 acl missing_slash path_reg ^/article/[^/]*$
7946 redirect code 301 prefix / drop-query append-slash if missing_slash
7947
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007948 Example: redirect all HTTP traffic to HTTPS when SSL is handled by haproxy.
David BERARDe7153042012-11-03 00:11:31 +01007949 redirect scheme https if !{ ssl_fc }
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007950
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007951 Example: append 'www.' prefix in front of all hosts not having it
Coen Rosdorff596659b2016-04-11 11:33:49 +02007952 http-request redirect code 301 location \
7953 http://www.%[hdr(host)]%[capture.req.uri] \
7954 unless { hdr_beg(host) -i www }
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007955
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007956 See section 7 about ACL usage.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007957
Willy Tarreau303c0352008-01-17 19:01:39 +01007958
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02007959retries <value>
7960 Set the number of retries to perform on a server after a connection failure
7961 May be used in sections: defaults | frontend | listen | backend
7962 yes | no | yes | yes
7963 Arguments :
7964 <value> is the number of times a connection attempt should be retried on
7965 a server when a connection either is refused or times out. The
7966 default value is 3.
7967
7968 It is important to understand that this value applies to the number of
7969 connection attempts, not full requests. When a connection has effectively
7970 been established to a server, there will be no more retry.
7971
7972 In order to avoid immediate reconnections to a server which is restarting,
Joseph Lynch726ab712015-05-11 23:25:34 -07007973 a turn-around timer of min("timeout connect", one second) is applied before
7974 a retry occurs.
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02007975
7976 When "option redispatch" is set, the last retry may be performed on another
7977 server even if a cookie references a different server.
7978
7979 See also : "option redispatch"
7980
7981
Olivier Houcharda254a372019-04-05 15:30:12 +02007982retry-on [list of keywords]
7983 Specify when to attempt to automatically retry a failed request
7984 May be used in sections: defaults | frontend | listen | backend
7985 yes | no | yes | yes
7986 Arguments :
7987 <keywords> is a list of keywords or HTTP status codes, each representing a
7988 type of failure event on which an attempt to retry the request
7989 is desired. Please read the notes at the bottom before changing
7990 this setting. The following keywords are supported :
7991
7992 none never retry
7993
7994 conn-failure retry when the connection or the SSL handshake failed
7995 and the request could not be sent. This is the default.
7996
7997 empty-response retry when the server connection was closed after part
7998 of the request was sent, and nothing was received from
7999 the server. This type of failure may be caused by the
8000 request timeout on the server side, poor network
8001 condition, or a server crash or restart while
8002 processing the request.
8003
Olivier Houcharde3249a92019-05-03 23:01:47 +02008004 junk-response retry when the server returned something not looking
8005 like a complete HTTP response. This includes partial
8006 responses headers as well as non-HTTP contents. It
8007 usually is a bad idea to retry on such events, which
8008 may be caused a configuration issue (wrong server port)
8009 or by the request being harmful to the server (buffer
8010 overflow attack for example).
8011
Olivier Houcharda254a372019-04-05 15:30:12 +02008012 response-timeout the server timeout stroke while waiting for the server
8013 to respond to the request. This may be caused by poor
8014 network condition, the reuse of an idle connection
8015 which has expired on the path, or by the request being
8016 extremely expensive to process. It generally is a bad
8017 idea to retry on such events on servers dealing with
8018 heavy database processing (full scans, etc) as it may
8019 amplify denial of service attacks.
8020
Olivier Houchard865d8392019-05-03 22:46:27 +02008021 0rtt-rejected retry requests which were sent over early data and were
8022 rejected by the server. These requests are generally
8023 considered to be safe to retry.
8024
Olivier Houcharda254a372019-04-05 15:30:12 +02008025 <status> any HTTP status code among "404" (Not Found), "408"
8026 (Request Timeout), "425" (Too Early), "500" (Server
8027 Error), "501" (Not Implemented), "502" (Bad Gateway),
8028 "503" (Service Unavailable), "504" (Gateway Timeout).
8029
Olivier Houchardddf0e032019-05-10 18:05:40 +02008030 all-retryable-errors
8031 retry request for any error that are considered
8032 retryable. This currently activates "conn-failure",
8033 "empty-response", "junk-response", "response-timeout",
8034 "0rtt-rejected", "500", "502", "503", and "504".
8035
Olivier Houcharda254a372019-04-05 15:30:12 +02008036 Using this directive replaces any previous settings with the new ones; it is
8037 not cumulative.
8038
8039 Please note that using anything other than "none" and "conn-failure" requires
8040 to allocate a buffer and copy the whole request into it, so it has memory and
8041 performance impacts. Requests not fitting in a single buffer will never be
8042 retried (see the global tune.bufsize setting).
8043
8044 You have to make sure the application has a replay protection mechanism built
8045 in such as a unique transaction IDs passed in requests, or that replaying the
8046 same request has no consequence, or it is very dangerous to use any retry-on
8047 value beside "conn-failure" and "none". Static file servers and caches are
8048 generally considered safe against any type of retry. Using a status code can
8049 be useful to quickly leave a server showing an abnormal behavior (out of
8050 memory, file system issues, etc), but in this case it may be a good idea to
8051 immediately redispatch the connection to another server (please see "option
8052 redispatch" for this). Last, it is important to understand that most causes
8053 of failures are the requests themselves and that retrying a request causing a
8054 server to misbehave will often make the situation even worse for this server,
8055 or for the whole service in case of redispatch.
8056
8057 Unless you know exactly how the application deals with replayed requests, you
8058 should not use this directive.
8059
8060 The default is "conn-failure".
8061
8062 See also: "retries", "option redispatch", "tune.bufsize"
8063
David du Colombier486df472011-03-17 10:40:26 +01008064server <name> <address>[:[port]] [param*]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008065 Declare a server in a backend
8066 May be used in sections : defaults | frontend | listen | backend
8067 no | no | yes | yes
8068 Arguments :
8069 <name> is the internal name assigned to this server. This name will
Davor Ocelice9ed2812017-12-25 17:49:28 +01008070 appear in logs and alerts. If "http-send-name-header" is
Mark Lamourinec2247f02012-01-04 13:02:01 -05008071 set, it will be added to the request header sent to the server.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008072
David du Colombier486df472011-03-17 10:40:26 +01008073 <address> is the IPv4 or IPv6 address of the server. Alternatively, a
8074 resolvable hostname is supported, but this name will be resolved
8075 during start-up. Address "0.0.0.0" or "*" has a special meaning.
8076 It indicates that the connection will be forwarded to the same IP
Willy Tarreaud669a4f2010-07-13 14:49:50 +02008077 address as the one from the client connection. This is useful in
8078 transparent proxy architectures where the client's connection is
8079 intercepted and haproxy must forward to the original destination
8080 address. This is more or less what the "transparent" keyword does
8081 except that with a server it's possible to limit concurrency and
Willy Tarreau24709282013-03-10 21:32:12 +01008082 to report statistics. Optionally, an address family prefix may be
8083 used before the address to force the family regardless of the
8084 address format, which can be useful to specify a path to a unix
8085 socket with no slash ('/'). Currently supported prefixes are :
8086 - 'ipv4@' -> address is always IPv4
8087 - 'ipv6@' -> address is always IPv6
8088 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02008089 - 'abns@' -> address is in abstract namespace (Linux only)
William Lallemand2fe7dd02018-09-11 16:51:29 +02008090 - 'sockpair@' -> address is the FD of a connected unix
8091 socket or of a socketpair. During a connection, the
8092 backend creates a pair of connected sockets, and passes
8093 one of them over the FD. The bind part will use the
8094 received socket as the client FD. Should be used
8095 carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02008096 You may want to reference some environment variables in the
8097 address parameter, see section 2.3 about environment
Willy Tarreau6a031d12016-11-07 19:42:35 +01008098 variables. The "init-addr" setting can be used to modify the way
8099 IP addresses should be resolved upon startup.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008100
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008101 <port> is an optional port specification. If set, all connections will
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008102 be sent to this port. If unset, the same port the client
8103 connected to will be used. The port may also be prefixed by a "+"
8104 or a "-". In this case, the server's port will be determined by
8105 adding this value to the client's port.
8106
8107 <param*> is a list of parameters for this server. The "server" keywords
8108 accepts an important number of options and has a complete section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008109 dedicated to it. Please refer to section 5 for more details.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008110
8111 Examples :
8112 server first 10.1.1.1:1080 cookie first check inter 1000
8113 server second 10.1.1.2:1080 cookie second check inter 1000
Willy Tarreau24709282013-03-10 21:32:12 +01008114 server transp ipv4@
William Lallemandb2f07452015-05-12 14:27:13 +02008115 server backup "${SRV_BACKUP}:1080" backup
8116 server www1_dc1 "${LAN_DC1}.101:80"
8117 server www1_dc2 "${LAN_DC2}.101:80"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008118
Willy Tarreau55dcaf62015-09-27 15:03:15 +02008119 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
8120 sun_path length is used for the address length. Some other programs
8121 such as socat use the string length only by default. Pass the option
8122 ",unix-tightsocklen=0" to any abstract socket definition in socat to
8123 make it compatible with HAProxy's.
8124
Mark Lamourinec2247f02012-01-04 13:02:01 -05008125 See also: "default-server", "http-send-name-header" and section 5 about
8126 server options
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008127
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02008128server-state-file-name [<file>]
8129 Set the server state file to read, load and apply to servers available in
8130 this backend. It only applies when the directive "load-server-state-from-file"
8131 is set to "local". When <file> is not provided or if this directive is not
8132 set, then backend name is used. If <file> starts with a slash '/', then it is
8133 considered as an absolute path. Otherwise, <file> is concatenated to the
8134 global directive "server-state-file-base".
8135
8136 Example: the minimal configuration below would make HAProxy look for the
8137 state server file '/etc/haproxy/states/bk':
8138
8139 global
8140 server-state-file-base /etc/haproxy/states
8141
8142 backend bk
8143 load-server-state-from-file
8144
8145 See also: "server-state-file-base", "load-server-state-from-file", and
8146 "show servers state"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008147
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02008148server-template <prefix> <num | range> <fqdn>[:<port>] [params*]
8149 Set a template to initialize servers with shared parameters.
8150 The names of these servers are built from <prefix> and <num | range> parameters.
8151 May be used in sections : defaults | frontend | listen | backend
8152 no | no | yes | yes
8153
8154 Arguments:
8155 <prefix> A prefix for the server names to be built.
8156
8157 <num | range>
8158 If <num> is provided, this template initializes <num> servers
8159 with 1 up to <num> as server name suffixes. A range of numbers
8160 <num_low>-<num_high> may also be used to use <num_low> up to
8161 <num_high> as server name suffixes.
8162
8163 <fqdn> A FQDN for all the servers this template initializes.
8164
8165 <port> Same meaning as "server" <port> argument (see "server" keyword).
8166
8167 <params*>
8168 Remaining server parameters among all those supported by "server"
8169 keyword.
8170
8171 Examples:
8172 # Initializes 3 servers with srv1, srv2 and srv3 as names,
8173 # google.com as FQDN, and health-check enabled.
8174 server-template srv 1-3 google.com:80 check
8175
8176 # or
8177 server-template srv 3 google.com:80 check
8178
8179 # would be equivalent to:
8180 server srv1 google.com:80 check
8181 server srv2 google.com:80 check
8182 server srv3 google.com:80 check
8183
8184
8185
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008186source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +02008187source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreaud53f96b2009-02-04 18:46:54 +01008188source <addr>[:<port>] [interface <name>]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008189 Set the source address for outgoing connections
8190 May be used in sections : defaults | frontend | listen | backend
8191 yes | no | yes | yes
8192 Arguments :
8193 <addr> is the IPv4 address HAProxy will bind to before connecting to a
8194 server. This address is also used as a source for health checks.
Willy Tarreau24709282013-03-10 21:32:12 +01008195
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008196 The default value of 0.0.0.0 means that the system will select
Willy Tarreau24709282013-03-10 21:32:12 +01008197 the most appropriate address to reach its destination. Optionally
8198 an address family prefix may be used before the address to force
8199 the family regardless of the address format, which can be useful
8200 to specify a path to a unix socket with no slash ('/'). Currently
8201 supported prefixes are :
8202 - 'ipv4@' -> address is always IPv4
8203 - 'ipv6@' -> address is always IPv6
8204 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02008205 - 'abns@' -> address is in abstract namespace (Linux only)
Cyril Bonté307ee1e2015-09-28 23:16:06 +02008206 You may want to reference some environment variables in the
8207 address parameter, see section 2.3 about environment variables.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008208
8209 <port> is an optional port. It is normally not needed but may be useful
8210 in some very specific contexts. The default value of zero means
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02008211 the system will select a free port. Note that port ranges are not
8212 supported in the backend. If you want to force port ranges, you
8213 have to specify them on each "server" line.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008214
8215 <addr2> is the IP address to present to the server when connections are
8216 forwarded in full transparent proxy mode. This is currently only
8217 supported on some patched Linux kernels. When this address is
8218 specified, clients connecting to the server will be presented
8219 with this address, while health checks will still use the address
8220 <addr>.
8221
8222 <port2> is the optional port to present to the server when connections
8223 are forwarded in full transparent proxy mode (see <addr2> above).
8224 The default value of zero means the system will select a free
8225 port.
8226
Willy Tarreaubce70882009-09-07 11:51:47 +02008227 <hdr> is the name of a HTTP header in which to fetch the IP to bind to.
8228 This is the name of a comma-separated header list which can
8229 contain multiple IP addresses. By default, the last occurrence is
8230 used. This is designed to work with the X-Forwarded-For header
Baptiste Assmannea3e73b2013-02-02 23:47:49 +01008231 and to automatically bind to the client's IP address as seen
Willy Tarreaubce70882009-09-07 11:51:47 +02008232 by previous proxy, typically Stunnel. In order to use another
8233 occurrence from the last one, please see the <occ> parameter
8234 below. When the header (or occurrence) is not found, no binding
8235 is performed so that the proxy's default IP address is used. Also
8236 keep in mind that the header name is case insensitive, as for any
8237 HTTP header.
8238
8239 <occ> is the occurrence number of a value to be used in a multi-value
8240 header. This is to be used in conjunction with "hdr_ip(<hdr>)",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04008241 in order to specify which occurrence to use for the source IP
Willy Tarreaubce70882009-09-07 11:51:47 +02008242 address. Positive values indicate a position from the first
8243 occurrence, 1 being the first one. Negative values indicate
8244 positions relative to the last one, -1 being the last one. This
8245 is helpful for situations where an X-Forwarded-For header is set
8246 at the entry point of an infrastructure and must be used several
8247 proxy layers away. When this value is not specified, -1 is
8248 assumed. Passing a zero here disables the feature.
8249
Willy Tarreaud53f96b2009-02-04 18:46:54 +01008250 <name> is an optional interface name to which to bind to for outgoing
8251 traffic. On systems supporting this features (currently, only
8252 Linux), this allows one to bind all traffic to the server to
8253 this interface even if it is not the one the system would select
8254 based on routing tables. This should be used with extreme care.
8255 Note that using this option requires root privileges.
8256
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008257 The "source" keyword is useful in complex environments where a specific
8258 address only is allowed to connect to the servers. It may be needed when a
8259 private address must be used through a public gateway for instance, and it is
8260 known that the system cannot determine the adequate source address by itself.
8261
8262 An extension which is available on certain patched Linux kernels may be used
8263 through the "usesrc" optional keyword. It makes it possible to connect to the
8264 servers with an IP address which does not belong to the system itself. This
8265 is called "full transparent proxy mode". For this to work, the destination
8266 servers have to route their traffic back to this address through the machine
8267 running HAProxy, and IP forwarding must generally be enabled on this machine.
8268
8269 In this "full transparent proxy" mode, it is possible to force a specific IP
8270 address to be presented to the servers. This is not much used in fact. A more
8271 common use is to tell HAProxy to present the client's IP address. For this,
8272 there are two methods :
8273
8274 - present the client's IP and port addresses. This is the most transparent
8275 mode, but it can cause problems when IP connection tracking is enabled on
8276 the machine, because a same connection may be seen twice with different
8277 states. However, this solution presents the huge advantage of not
8278 limiting the system to the 64k outgoing address+port couples, because all
8279 of the client ranges may be used.
8280
8281 - present only the client's IP address and select a spare port. This
8282 solution is still quite elegant but slightly less transparent (downstream
8283 firewalls logs will not match upstream's). It also presents the downside
8284 of limiting the number of concurrent connections to the usual 64k ports.
8285 However, since the upstream and downstream ports are different, local IP
8286 connection tracking on the machine will not be upset by the reuse of the
8287 same session.
8288
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008289 This option sets the default source for all servers in the backend. It may
8290 also be specified in a "defaults" section. Finer source address specification
8291 is possible at the server level using the "source" server option. Refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008292 section 5 for more information.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008293
Baptiste Assmann91bd3372015-07-17 21:59:42 +02008294 In order to work, "usesrc" requires root privileges.
8295
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008296 Examples :
8297 backend private
8298 # Connect to the servers using our 192.168.1.200 source address
8299 source 192.168.1.200
8300
8301 backend transparent_ssl1
8302 # Connect to the SSL farm from the client's source address
8303 source 192.168.1.200 usesrc clientip
8304
8305 backend transparent_ssl2
8306 # Connect to the SSL farm from the client's source address and port
8307 # not recommended if IP conntrack is present on the local machine.
8308 source 192.168.1.200 usesrc client
8309
8310 backend transparent_ssl3
8311 # Connect to the SSL farm from the client's source address. It
8312 # is more conntrack-friendly.
8313 source 192.168.1.200 usesrc clientip
8314
8315 backend transparent_smtp
8316 # Connect to the SMTP farm from the client's source address/port
8317 # with Tproxy version 4.
8318 source 0.0.0.0 usesrc clientip
8319
Willy Tarreaubce70882009-09-07 11:51:47 +02008320 backend transparent_http
8321 # Connect to the servers using the client's IP as seen by previous
8322 # proxy.
8323 source 0.0.0.0 usesrc hdr_ip(x-forwarded-for,-1)
8324
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008325 See also : the "source" server option in section 5, the Tproxy patches for
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008326 the Linux kernel on www.balabit.com, the "bind" keyword.
8327
Willy Tarreau844e3c52008-01-11 16:28:18 +01008328
Cyril Bonté66c327d2010-10-12 00:14:37 +02008329stats admin { if | unless } <cond>
8330 Enable statistics admin level if/unless a condition is matched
8331 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008332 no | yes | yes | yes
Cyril Bonté66c327d2010-10-12 00:14:37 +02008333
8334 This statement enables the statistics admin level if/unless a condition is
8335 matched.
8336
8337 The admin level allows to enable/disable servers from the web interface. By
8338 default, statistics page is read-only for security reasons.
8339
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008340 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8341 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008342 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008343
Cyril Bonté23b39d92011-02-10 22:54:44 +01008344 Currently, the POST request is limited to the buffer size minus the reserved
8345 buffer space, which means that if the list of servers is too long, the
8346 request won't be processed. It is recommended to alter few servers at a
8347 time.
Cyril Bonté66c327d2010-10-12 00:14:37 +02008348
8349 Example :
8350 # statistics admin level only for localhost
8351 backend stats_localhost
8352 stats enable
8353 stats admin if LOCALHOST
8354
8355 Example :
8356 # statistics admin level always enabled because of the authentication
8357 backend stats_auth
8358 stats enable
8359 stats auth admin:AdMiN123
8360 stats admin if TRUE
8361
8362 Example :
8363 # statistics admin level depends on the authenticated user
8364 userlist stats-auth
8365 group admin users admin
8366 user admin insecure-password AdMiN123
8367 group readonly users haproxy
8368 user haproxy insecure-password haproxy
8369
8370 backend stats_auth
8371 stats enable
8372 acl AUTH http_auth(stats-auth)
8373 acl AUTH_ADMIN http_auth_group(stats-auth) admin
8374 stats http-request auth unless AUTH
8375 stats admin if AUTH_ADMIN
8376
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008377 See also : "stats enable", "stats auth", "stats http-request", "nbproc",
8378 "bind-process", section 3.4 about userlists and section 7 about
8379 ACL usage.
Cyril Bonté66c327d2010-10-12 00:14:37 +02008380
8381
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008382stats auth <user>:<passwd>
8383 Enable statistics with authentication and grant access to an account
8384 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008385 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008386 Arguments :
8387 <user> is a user name to grant access to
8388
8389 <passwd> is the cleartext password associated to this user
8390
8391 This statement enables statistics with default settings, and restricts access
8392 to declared users only. It may be repeated as many times as necessary to
8393 allow as many users as desired. When a user tries to access the statistics
8394 without a valid account, a "401 Forbidden" response will be returned so that
8395 the browser asks the user to provide a valid user and password. The real
8396 which will be returned to the browser is configurable using "stats realm".
8397
8398 Since the authentication method is HTTP Basic Authentication, the passwords
8399 circulate in cleartext on the network. Thus, it was decided that the
8400 configuration file would also use cleartext passwords to remind the users
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02008401 that those ones should not be sensitive and not shared with any other account.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008402
8403 It is also possible to reduce the scope of the proxies which appear in the
8404 report using "stats scope".
8405
8406 Though this statement alone is enough to enable statistics reporting, it is
8407 recommended to set all other settings in order to avoid relying on default
8408 unobvious parameters.
8409
8410 Example :
8411 # public access (limited to this backend only)
8412 backend public_www
8413 server srv1 192.168.0.1:80
8414 stats enable
8415 stats hide-version
8416 stats scope .
8417 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008418 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008419 stats auth admin1:AdMiN123
8420 stats auth admin2:AdMiN321
8421
8422 # internal monitoring access (unlimited)
8423 backend private_monitoring
8424 stats enable
8425 stats uri /admin?stats
8426 stats refresh 5s
8427
8428 See also : "stats enable", "stats realm", "stats scope", "stats uri"
8429
8430
8431stats enable
8432 Enable statistics reporting with default settings
8433 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008434 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008435 Arguments : none
8436
8437 This statement enables statistics reporting with default settings defined
8438 at build time. Unless stated otherwise, these settings are used :
8439 - stats uri : /haproxy?stats
8440 - stats realm : "HAProxy Statistics"
8441 - stats auth : no authentication
8442 - stats scope : no restriction
8443
8444 Though this statement alone is enough to enable statistics reporting, it is
8445 recommended to set all other settings in order to avoid relying on default
8446 unobvious parameters.
8447
8448 Example :
8449 # public access (limited to this backend only)
8450 backend public_www
8451 server srv1 192.168.0.1:80
8452 stats enable
8453 stats hide-version
8454 stats scope .
8455 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008456 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008457 stats auth admin1:AdMiN123
8458 stats auth admin2:AdMiN321
8459
8460 # internal monitoring access (unlimited)
8461 backend private_monitoring
8462 stats enable
8463 stats uri /admin?stats
8464 stats refresh 5s
8465
8466 See also : "stats auth", "stats realm", "stats uri"
8467
8468
Willy Tarreaud63335a2010-02-26 12:56:52 +01008469stats hide-version
8470 Enable statistics and hide HAProxy version reporting
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008471 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008472 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008473 Arguments : none
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008474
Willy Tarreaud63335a2010-02-26 12:56:52 +01008475 By default, the stats page reports some useful status information along with
8476 the statistics. Among them is HAProxy's version. However, it is generally
8477 considered dangerous to report precise version to anyone, as it can help them
8478 target known weaknesses with specific attacks. The "stats hide-version"
8479 statement removes the version from the statistics report. This is recommended
8480 for public sites or any site with a weak login/password.
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008481
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02008482 Though this statement alone is enough to enable statistics reporting, it is
8483 recommended to set all other settings in order to avoid relying on default
8484 unobvious parameters.
8485
Willy Tarreaud63335a2010-02-26 12:56:52 +01008486 Example :
8487 # public access (limited to this backend only)
8488 backend public_www
8489 server srv1 192.168.0.1:80
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02008490 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01008491 stats hide-version
8492 stats scope .
8493 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008494 stats realm HAProxy\ Statistics
Willy Tarreaud63335a2010-02-26 12:56:52 +01008495 stats auth admin1:AdMiN123
8496 stats auth admin2:AdMiN321
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008497
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008498 # internal monitoring access (unlimited)
8499 backend private_monitoring
8500 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01008501 stats uri /admin?stats
8502 stats refresh 5s
Krzysztof Piotr Oledzki15514c22010-01-04 16:03:09 +01008503
Willy Tarreaud63335a2010-02-26 12:56:52 +01008504 See also : "stats auth", "stats enable", "stats realm", "stats uri"
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008505
Willy Tarreau983e01e2010-01-11 18:42:06 +01008506
Cyril Bonté2be1b3f2010-09-30 23:46:30 +02008507stats http-request { allow | deny | auth [realm <realm>] }
8508 [ { if | unless } <condition> ]
8509 Access control for statistics
8510
8511 May be used in sections: defaults | frontend | listen | backend
8512 no | no | yes | yes
8513
8514 As "http-request", these set of options allow to fine control access to
8515 statistics. Each option may be followed by if/unless and acl.
8516 First option with matched condition (or option without condition) is final.
8517 For "deny" a 403 error will be returned, for "allow" normal processing is
8518 performed, for "auth" a 401/407 error code is returned so the client
8519 should be asked to enter a username and password.
8520
8521 There is no fixed limit to the number of http-request statements per
8522 instance.
8523
8524 See also : "http-request", section 3.4 about userlists and section 7
8525 about ACL usage.
8526
8527
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008528stats realm <realm>
8529 Enable statistics and set authentication realm
8530 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008531 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008532 Arguments :
8533 <realm> is the name of the HTTP Basic Authentication realm reported to
8534 the browser. The browser uses it to display it in the pop-up
8535 inviting the user to enter a valid username and password.
8536
8537 The realm is read as a single word, so any spaces in it should be escaped
8538 using a backslash ('\').
8539
8540 This statement is useful only in conjunction with "stats auth" since it is
8541 only related to authentication.
8542
8543 Though this statement alone is enough to enable statistics reporting, it is
8544 recommended to set all other settings in order to avoid relying on default
8545 unobvious parameters.
8546
8547 Example :
8548 # public access (limited to this backend only)
8549 backend public_www
8550 server srv1 192.168.0.1:80
8551 stats enable
8552 stats hide-version
8553 stats scope .
8554 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008555 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008556 stats auth admin1:AdMiN123
8557 stats auth admin2:AdMiN321
8558
8559 # internal monitoring access (unlimited)
8560 backend private_monitoring
8561 stats enable
8562 stats uri /admin?stats
8563 stats refresh 5s
8564
8565 See also : "stats auth", "stats enable", "stats uri"
8566
8567
8568stats refresh <delay>
8569 Enable statistics with automatic refresh
8570 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008571 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008572 Arguments :
8573 <delay> is the suggested refresh delay, specified in seconds, which will
8574 be returned to the browser consulting the report page. While the
8575 browser is free to apply any delay, it will generally respect it
8576 and refresh the page this every seconds. The refresh interval may
8577 be specified in any other non-default time unit, by suffixing the
8578 unit after the value, as explained at the top of this document.
8579
8580 This statement is useful on monitoring displays with a permanent page
8581 reporting the load balancer's activity. When set, the HTML report page will
8582 include a link "refresh"/"stop refresh" so that the user can select whether
8583 he wants automatic refresh of the page or not.
8584
8585 Though this statement alone is enough to enable statistics reporting, it is
8586 recommended to set all other settings in order to avoid relying on default
8587 unobvious parameters.
8588
8589 Example :
8590 # public access (limited to this backend only)
8591 backend public_www
8592 server srv1 192.168.0.1:80
8593 stats enable
8594 stats hide-version
8595 stats scope .
8596 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008597 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008598 stats auth admin1:AdMiN123
8599 stats auth admin2:AdMiN321
8600
8601 # internal monitoring access (unlimited)
8602 backend private_monitoring
8603 stats enable
8604 stats uri /admin?stats
8605 stats refresh 5s
8606
8607 See also : "stats auth", "stats enable", "stats realm", "stats uri"
8608
8609
8610stats scope { <name> | "." }
8611 Enable statistics and limit access scope
8612 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008613 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008614 Arguments :
8615 <name> is the name of a listen, frontend or backend section to be
8616 reported. The special name "." (a single dot) designates the
8617 section in which the statement appears.
8618
8619 When this statement is specified, only the sections enumerated with this
8620 statement will appear in the report. All other ones will be hidden. This
8621 statement may appear as many times as needed if multiple sections need to be
8622 reported. Please note that the name checking is performed as simple string
8623 comparisons, and that it is never checked that a give section name really
8624 exists.
8625
8626 Though this statement alone is enough to enable statistics reporting, it is
8627 recommended to set all other settings in order to avoid relying on default
8628 unobvious parameters.
8629
8630 Example :
8631 # public access (limited to this backend only)
8632 backend public_www
8633 server srv1 192.168.0.1:80
8634 stats enable
8635 stats hide-version
8636 stats scope .
8637 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008638 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008639 stats auth admin1:AdMiN123
8640 stats auth admin2:AdMiN321
8641
8642 # internal monitoring access (unlimited)
8643 backend private_monitoring
8644 stats enable
8645 stats uri /admin?stats
8646 stats refresh 5s
8647
8648 See also : "stats auth", "stats enable", "stats realm", "stats uri"
8649
Willy Tarreaud63335a2010-02-26 12:56:52 +01008650
Willy Tarreauc9705a12010-07-27 20:05:50 +02008651stats show-desc [ <desc> ]
Willy Tarreaud63335a2010-02-26 12:56:52 +01008652 Enable reporting of a description on the statistics page.
8653 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008654 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008655
Willy Tarreauc9705a12010-07-27 20:05:50 +02008656 <desc> is an optional description to be reported. If unspecified, the
Willy Tarreaud63335a2010-02-26 12:56:52 +01008657 description from global section is automatically used instead.
8658
8659 This statement is useful for users that offer shared services to their
8660 customers, where node or description should be different for each customer.
8661
8662 Though this statement alone is enough to enable statistics reporting, it is
8663 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +01008664 unobvious parameters. By default description is not shown.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008665
8666 Example :
8667 # internal monitoring access (unlimited)
8668 backend private_monitoring
8669 stats enable
8670 stats show-desc Master node for Europe, Asia, Africa
8671 stats uri /admin?stats
8672 stats refresh 5s
8673
8674 See also: "show-node", "stats enable", "stats uri" and "description" in
8675 global section.
8676
8677
8678stats show-legends
Willy Tarreaued2119c2014-04-24 22:10:39 +02008679 Enable reporting additional information on the statistics page
8680 May be used in sections : defaults | frontend | listen | backend
8681 yes | yes | yes | yes
8682 Arguments : none
8683
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008684 Enable reporting additional information on the statistics page :
Willy Tarreaud63335a2010-02-26 12:56:52 +01008685 - cap: capabilities (proxy)
8686 - mode: one of tcp, http or health (proxy)
8687 - id: SNMP ID (proxy, socket, server)
8688 - IP (socket, server)
8689 - cookie (backend, server)
8690
8691 Though this statement alone is enough to enable statistics reporting, it is
8692 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +01008693 unobvious parameters. Default behavior is not to show this information.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008694
8695 See also: "stats enable", "stats uri".
8696
8697
8698stats show-node [ <name> ]
8699 Enable reporting of a host name on the statistics page.
8700 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008701 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008702 Arguments:
8703 <name> is an optional name to be reported. If unspecified, the
8704 node name from global section is automatically used instead.
8705
8706 This statement is useful for users that offer shared services to their
8707 customers, where node or description might be different on a stats page
Davor Ocelice9ed2812017-12-25 17:49:28 +01008708 provided for each customer. Default behavior is not to show host name.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008709
8710 Though this statement alone is enough to enable statistics reporting, it is
8711 recommended to set all other settings in order to avoid relying on default
8712 unobvious parameters.
8713
8714 Example:
8715 # internal monitoring access (unlimited)
8716 backend private_monitoring
8717 stats enable
8718 stats show-node Europe-1
8719 stats uri /admin?stats
8720 stats refresh 5s
8721
8722 See also: "show-desc", "stats enable", "stats uri", and "node" in global
8723 section.
8724
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008725
8726stats uri <prefix>
8727 Enable statistics and define the URI prefix to access them
8728 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008729 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008730 Arguments :
8731 <prefix> is the prefix of any URI which will be redirected to stats. This
8732 prefix may contain a question mark ('?') to indicate part of a
8733 query string.
8734
8735 The statistics URI is intercepted on the relayed traffic, so it appears as a
8736 page within the normal application. It is strongly advised to ensure that the
8737 selected URI will never appear in the application, otherwise it will never be
8738 possible to reach it in the application.
8739
8740 The default URI compiled in haproxy is "/haproxy?stats", but this may be
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01008741 changed at build time, so it's better to always explicitly specify it here.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008742 It is generally a good idea to include a question mark in the URI so that
8743 intermediate proxies refrain from caching the results. Also, since any string
8744 beginning with the prefix will be accepted as a stats request, the question
8745 mark helps ensuring that no valid URI will begin with the same words.
8746
8747 It is sometimes very convenient to use "/" as the URI prefix, and put that
8748 statement in a "listen" instance of its own. That makes it easy to dedicate
8749 an address or a port to statistics only.
8750
8751 Though this statement alone is enough to enable statistics reporting, it is
8752 recommended to set all other settings in order to avoid relying on default
8753 unobvious parameters.
8754
8755 Example :
8756 # public access (limited to this backend only)
8757 backend public_www
8758 server srv1 192.168.0.1:80
8759 stats enable
8760 stats hide-version
8761 stats scope .
8762 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008763 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008764 stats auth admin1:AdMiN123
8765 stats auth admin2:AdMiN321
8766
8767 # internal monitoring access (unlimited)
8768 backend private_monitoring
8769 stats enable
8770 stats uri /admin?stats
8771 stats refresh 5s
8772
8773 See also : "stats auth", "stats enable", "stats realm"
8774
8775
Willy Tarreaud63335a2010-02-26 12:56:52 +01008776stick match <pattern> [table <table>] [{if | unless} <cond>]
8777 Define a request pattern matching condition to stick a user to a server
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008778 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaud63335a2010-02-26 12:56:52 +01008779 no | no | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008780
8781 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02008782 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008783 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01008784 will be analyzed in the hope to find a matching entry in a
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008785 stickiness table. This rule is mandatory.
8786
8787 <table> is an optional stickiness table name. If unspecified, the same
8788 backend's table is used. A stickiness table is declared using
8789 the "stick-table" statement.
8790
8791 <cond> is an optional matching condition. It makes it possible to match
8792 on a certain criterion only when other conditions are met (or
8793 not met). For instance, it could be used to match on a source IP
8794 address except when a request passes through a known proxy, in
8795 which case we'd match on a header containing that IP address.
8796
8797 Some protocols or applications require complex stickiness rules and cannot
8798 always simply rely on cookies nor hashing. The "stick match" statement
8799 describes a rule to extract the stickiness criterion from an incoming request
8800 or connection. See section 7 for a complete list of possible patterns and
8801 transformation rules.
8802
8803 The table has to be declared using the "stick-table" statement. It must be of
8804 a type compatible with the pattern. By default it is the one which is present
8805 in the same backend. It is possible to share a table with other backends by
8806 referencing it using the "table" keyword. If another table is referenced,
8807 the server's ID inside the backends are used. By default, all server IDs
8808 start at 1 in each backend, so the server ordering is enough. But in case of
8809 doubt, it is highly recommended to force server IDs using their "id" setting.
8810
8811 It is possible to restrict the conditions where a "stick match" statement
8812 will apply, using "if" or "unless" followed by a condition. See section 7 for
8813 ACL based conditions.
8814
8815 There is no limit on the number of "stick match" statements. The first that
8816 applies and matches will cause the request to be directed to the same server
8817 as was used for the request which created the entry. That way, multiple
8818 matches can be used as fallbacks.
8819
8820 The stick rules are checked after the persistence cookies, so they will not
8821 affect stickiness if a cookie has already been used to select a server. That
8822 way, it becomes very easy to insert cookies and match on IP addresses in
8823 order to maintain stickiness between HTTP and HTTPS.
8824
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008825 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8826 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008827 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008828
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008829 Example :
8830 # forward SMTP users to the same server they just used for POP in the
8831 # last 30 minutes
8832 backend pop
8833 mode tcp
8834 balance roundrobin
8835 stick store-request src
8836 stick-table type ip size 200k expire 30m
8837 server s1 192.168.1.1:110
8838 server s2 192.168.1.1:110
8839
8840 backend smtp
8841 mode tcp
8842 balance roundrobin
8843 stick match src table pop
8844 server s1 192.168.1.1:25
8845 server s2 192.168.1.1:25
8846
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008847 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02008848 about ACLs and samples fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008849
8850
8851stick on <pattern> [table <table>] [{if | unless} <condition>]
8852 Define a request pattern to associate a user to a server
8853 May be used in sections : defaults | frontend | listen | backend
8854 no | no | yes | yes
8855
8856 Note : This form is exactly equivalent to "stick match" followed by
8857 "stick store-request", all with the same arguments. Please refer
8858 to both keywords for details. It is only provided as a convenience
8859 for writing more maintainable configurations.
8860
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008861 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8862 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008863 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008864
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008865 Examples :
8866 # The following form ...
Willy Tarreauec579d82010-02-26 19:15:04 +01008867 stick on src table pop if !localhost
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008868
8869 # ...is strictly equivalent to this one :
8870 stick match src table pop if !localhost
8871 stick store-request src table pop if !localhost
8872
8873
8874 # Use cookie persistence for HTTP, and stick on source address for HTTPS as
8875 # well as HTTP without cookie. Share the same table between both accesses.
8876 backend http
8877 mode http
8878 balance roundrobin
8879 stick on src table https
8880 cookie SRV insert indirect nocache
8881 server s1 192.168.1.1:80 cookie s1
8882 server s2 192.168.1.1:80 cookie s2
8883
8884 backend https
8885 mode tcp
8886 balance roundrobin
8887 stick-table type ip size 200k expire 30m
8888 stick on src
8889 server s1 192.168.1.1:443
8890 server s2 192.168.1.1:443
8891
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008892 See also : "stick match", "stick store-request", "nbproc" and "bind-process".
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008893
8894
8895stick store-request <pattern> [table <table>] [{if | unless} <condition>]
8896 Define a request pattern used to create an entry in a stickiness table
8897 May be used in sections : defaults | frontend | listen | backend
8898 no | no | yes | yes
8899
8900 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02008901 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008902 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01008903 will be analyzed, extracted and stored in the table once a
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008904 server is selected.
8905
8906 <table> is an optional stickiness table name. If unspecified, the same
8907 backend's table is used. A stickiness table is declared using
8908 the "stick-table" statement.
8909
8910 <cond> is an optional storage condition. It makes it possible to store
8911 certain criteria only when some conditions are met (or not met).
8912 For instance, it could be used to store the source IP address
8913 except when the request passes through a known proxy, in which
8914 case we'd store a converted form of a header containing that IP
8915 address.
8916
8917 Some protocols or applications require complex stickiness rules and cannot
8918 always simply rely on cookies nor hashing. The "stick store-request" statement
8919 describes a rule to decide what to extract from the request and when to do
8920 it, in order to store it into a stickiness table for further requests to
8921 match it using the "stick match" statement. Obviously the extracted part must
8922 make sense and have a chance to be matched in a further request. Storing a
8923 client's IP address for instance often makes sense. Storing an ID found in a
8924 URL parameter also makes sense. Storing a source port will almost never make
8925 any sense because it will be randomly matched. See section 7 for a complete
8926 list of possible patterns and transformation rules.
8927
8928 The table has to be declared using the "stick-table" statement. It must be of
8929 a type compatible with the pattern. By default it is the one which is present
8930 in the same backend. It is possible to share a table with other backends by
8931 referencing it using the "table" keyword. If another table is referenced,
8932 the server's ID inside the backends are used. By default, all server IDs
8933 start at 1 in each backend, so the server ordering is enough. But in case of
8934 doubt, it is highly recommended to force server IDs using their "id" setting.
8935
8936 It is possible to restrict the conditions where a "stick store-request"
8937 statement will apply, using "if" or "unless" followed by a condition. This
8938 condition will be evaluated while parsing the request, so any criteria can be
8939 used. See section 7 for ACL based conditions.
8940
8941 There is no limit on the number of "stick store-request" statements, but
8942 there is a limit of 8 simultaneous stores per request or response. This
8943 makes it possible to store up to 8 criteria, all extracted from either the
8944 request or the response, regardless of the number of rules. Only the 8 first
8945 ones which match will be kept. Using this, it is possible to feed multiple
8946 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01008947 another protocol or access method. Using multiple store-request rules with
8948 the same table is possible and may be used to find the best criterion to rely
8949 on, by arranging the rules by decreasing preference order. Only the first
8950 extracted criterion for a given table will be stored. All subsequent store-
8951 request rules referencing the same table will be skipped and their ACLs will
8952 not be evaluated.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008953
8954 The "store-request" rules are evaluated once the server connection has been
8955 established, so that the table will contain the real server that processed
8956 the request.
8957
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008958 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8959 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008960 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008961
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008962 Example :
8963 # forward SMTP users to the same server they just used for POP in the
8964 # last 30 minutes
8965 backend pop
8966 mode tcp
8967 balance roundrobin
8968 stick store-request src
8969 stick-table type ip size 200k expire 30m
8970 server s1 192.168.1.1:110
8971 server s2 192.168.1.1:110
8972
8973 backend smtp
8974 mode tcp
8975 balance roundrobin
8976 stick match src table pop
8977 server s1 192.168.1.1:25
8978 server s2 192.168.1.1:25
8979
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008980 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02008981 about ACLs and sample fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008982
8983
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008984stick-table type {ip | integer | string [len <length>] | binary [len <length>]}
Emeric Brunf099e792010-09-27 12:05:28 +02008985 size <size> [expire <expire>] [nopurge] [peers <peersect>]
8986 [store <data_type>]*
Godbach64cef792013-12-04 16:08:22 +08008987 Configure the stickiness table for the current section
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008988 May be used in sections : defaults | frontend | listen | backend
Willy Tarreauc00cdc22010-06-06 16:48:26 +02008989 no | yes | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008990
8991 Arguments :
8992 ip a table declared with "type ip" will only store IPv4 addresses.
8993 This form is very compact (about 50 bytes per entry) and allows
8994 very fast entry lookup and stores with almost no overhead. This
8995 is mainly used to store client source IP addresses.
8996
David du Colombier9a6d3c92011-03-17 10:40:24 +01008997 ipv6 a table declared with "type ipv6" will only store IPv6 addresses.
8998 This form is very compact (about 60 bytes per entry) and allows
8999 very fast entry lookup and stores with almost no overhead. This
9000 is mainly used to store client source IP addresses.
9001
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009002 integer a table declared with "type integer" will store 32bit integers
9003 which can represent a client identifier found in a request for
9004 instance.
9005
9006 string a table declared with "type string" will store substrings of up
9007 to <len> characters. If the string provided by the pattern
9008 extractor is larger than <len>, it will be truncated before
9009 being stored. During matching, at most <len> characters will be
9010 compared between the string in the table and the extracted
9011 pattern. When not specified, the string is automatically limited
Emeric Brun7c6b82e2010-09-24 16:34:28 +02009012 to 32 characters.
9013
9014 binary a table declared with "type binary" will store binary blocks
9015 of <len> bytes. If the block provided by the pattern
9016 extractor is larger than <len>, it will be truncated before
Willy Tarreaube722a22014-06-13 16:31:59 +02009017 being stored. If the block provided by the sample expression
Emeric Brun7c6b82e2010-09-24 16:34:28 +02009018 is shorter than <len>, it will be padded by 0. When not
9019 specified, the block is automatically limited to 32 bytes.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009020
9021 <length> is the maximum number of characters that will be stored in a
Emeric Brun7c6b82e2010-09-24 16:34:28 +02009022 "string" type table (See type "string" above). Or the number
9023 of bytes of the block in "binary" type table. Be careful when
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009024 changing this parameter as memory usage will proportionally
9025 increase.
9026
9027 <size> is the maximum number of entries that can fit in the table. This
Cyril Bonté78caf842010-03-10 22:41:43 +01009028 value directly impacts memory usage. Count approximately
9029 50 bytes per entry, plus the size of a string if any. The size
9030 supports suffixes "k", "m", "g" for 2^10, 2^20 and 2^30 factors.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009031
9032 [nopurge] indicates that we refuse to purge older entries when the table
9033 is full. When not specified and the table is full when haproxy
9034 wants to store an entry in it, it will flush a few of the oldest
9035 entries in order to release some space for the new ones. This is
Davor Ocelice9ed2812017-12-25 17:49:28 +01009036 most often the desired behavior. In some specific cases, it
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009037 be desirable to refuse new entries instead of purging the older
9038 ones. That may be the case when the amount of data to store is
9039 far above the hardware limits and we prefer not to offer access
9040 to new clients than to reject the ones already connected. When
9041 using this parameter, be sure to properly set the "expire"
9042 parameter (see below).
9043
Emeric Brunf099e792010-09-27 12:05:28 +02009044 <peersect> is the name of the peers section to use for replication. Entries
9045 which associate keys to server IDs are kept synchronized with
9046 the remote peers declared in this section. All entries are also
9047 automatically learned from the local peer (old process) during a
9048 soft restart.
9049
Willy Tarreau1abc6732015-05-01 19:21:02 +02009050 NOTE : each peers section may be referenced only by tables
9051 belonging to the same unique process.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009052
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009053 <expire> defines the maximum duration of an entry in the table since it
9054 was last created, refreshed or matched. The expiration delay is
9055 defined using the standard time format, similarly as the various
9056 timeouts. The maximum duration is slightly above 24 days. See
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +03009057 section 2.4 for more information. If this delay is not specified,
Cyril Bontédc4d9032012-04-08 21:57:39 +02009058 the session won't automatically expire, but older entries will
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009059 be removed once full. Be sure not to use the "nopurge" parameter
9060 if not expiration delay is specified.
9061
Willy Tarreau08d5f982010-06-06 13:34:54 +02009062 <data_type> is used to store additional information in the stick-table. This
9063 may be used by ACLs in order to control various criteria related
9064 to the activity of the client matching the stick-table. For each
9065 item specified here, the size of each entry will be inflated so
Willy Tarreauc9705a12010-07-27 20:05:50 +02009066 that the additional data can fit. Several data types may be
9067 stored with an entry. Multiple data types may be specified after
9068 the "store" keyword, as a comma-separated list. Alternatively,
9069 it is possible to repeat the "store" keyword followed by one or
9070 several data types. Except for the "server_id" type which is
9071 automatically detected and enabled, all data types must be
9072 explicitly declared to be stored. If an ACL references a data
9073 type which is not stored, the ACL will simply not match. Some
9074 data types require an argument which must be passed just after
9075 the type between parenthesis. See below for the supported data
9076 types and their arguments.
9077
9078 The data types that can be stored with an entry are the following :
9079 - server_id : this is an integer which holds the numeric ID of the server a
9080 request was assigned to. It is used by the "stick match", "stick store",
9081 and "stick on" rules. It is automatically enabled when referenced.
9082
9083 - gpc0 : first General Purpose Counter. It is a positive 32-bit integer
9084 integer which may be used for anything. Most of the time it will be used
9085 to put a special tag on some entries, for instance to note that a
Davor Ocelice9ed2812017-12-25 17:49:28 +01009086 specific behavior was detected and must be known for future matches.
Willy Tarreauc9705a12010-07-27 20:05:50 +02009087
Willy Tarreauba2ffd12013-05-29 15:54:14 +02009088 - gpc0_rate(<period>) : increment rate of the first General Purpose Counter
9089 over a period. It is a positive 32-bit integer integer which may be used
9090 for anything. Just like <gpc0>, it counts events, but instead of keeping
Davor Ocelice9ed2812017-12-25 17:49:28 +01009091 a cumulative number, it maintains the rate at which the counter is
Willy Tarreauba2ffd12013-05-29 15:54:14 +02009092 incremented. Most of the time it will be used to measure the frequency of
Davor Ocelice9ed2812017-12-25 17:49:28 +01009093 occurrence of certain events (e.g. requests to a specific URL).
Willy Tarreauba2ffd12013-05-29 15:54:14 +02009094
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009095 - gpc1 : second General Purpose Counter. It is a positive 32-bit integer
9096 integer which may be used for anything. Most of the time it will be used
9097 to put a special tag on some entries, for instance to note that a
9098 specific behavior was detected and must be known for future matches.
9099
9100 - gpc1_rate(<period>) : increment rate of the second General Purpose Counter
9101 over a period. It is a positive 32-bit integer integer which may be used
9102 for anything. Just like <gpc1>, it counts events, but instead of keeping
9103 a cumulative number, it maintains the rate at which the counter is
9104 incremented. Most of the time it will be used to measure the frequency of
9105 occurrence of certain events (e.g. requests to a specific URL).
9106
Willy Tarreauc9705a12010-07-27 20:05:50 +02009107 - conn_cnt : Connection Count. It is a positive 32-bit integer which counts
9108 the absolute number of connections received from clients which matched
9109 this entry. It does not mean the connections were accepted, just that
9110 they were received.
9111
9112 - conn_cur : Current Connections. It is a positive 32-bit integer which
9113 stores the concurrent connection counts for the entry. It is incremented
9114 once an incoming connection matches the entry, and decremented once the
9115 connection leaves. That way it is possible to know at any time the exact
9116 number of concurrent connections for an entry.
9117
9118 - conn_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9119 integer parameter <period> which indicates in milliseconds the length
9120 of the period over which the average is measured. It reports the average
9121 incoming connection rate over that period, in connections per period. The
9122 result is an integer which can be matched using ACLs.
9123
9124 - sess_cnt : Session Count. It is a positive 32-bit integer which counts
9125 the absolute number of sessions received from clients which matched this
9126 entry. A session is a connection that was accepted by the layer 4 rules.
9127
9128 - sess_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9129 integer parameter <period> which indicates in milliseconds the length
9130 of the period over which the average is measured. It reports the average
9131 incoming session rate over that period, in sessions per period. The
9132 result is an integer which can be matched using ACLs.
9133
9134 - http_req_cnt : HTTP request Count. It is a positive 32-bit integer which
9135 counts the absolute number of HTTP requests received from clients which
9136 matched this entry. It does not matter whether they are valid requests or
9137 not. Note that this is different from sessions when keep-alive is used on
9138 the client side.
9139
9140 - http_req_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9141 integer parameter <period> which indicates in milliseconds the length
9142 of the period over which the average is measured. It reports the average
9143 HTTP request rate over that period, in requests per period. The result is
9144 an integer which can be matched using ACLs. It does not matter whether
9145 they are valid requests or not. Note that this is different from sessions
9146 when keep-alive is used on the client side.
9147
9148 - http_err_cnt : HTTP Error Count. It is a positive 32-bit integer which
9149 counts the absolute number of HTTP requests errors induced by clients
9150 which matched this entry. Errors are counted on invalid and truncated
9151 requests, as well as on denied or tarpitted requests, and on failed
9152 authentications. If the server responds with 4xx, then the request is
9153 also counted as an error since it's an error triggered by the client
Davor Ocelice9ed2812017-12-25 17:49:28 +01009154 (e.g. vulnerability scan).
Willy Tarreauc9705a12010-07-27 20:05:50 +02009155
9156 - http_err_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9157 integer parameter <period> which indicates in milliseconds the length
9158 of the period over which the average is measured. It reports the average
9159 HTTP request error rate over that period, in requests per period (see
9160 http_err_cnt above for what is accounted as an error). The result is an
9161 integer which can be matched using ACLs.
9162
9163 - bytes_in_cnt : client to server byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +01009164 integer which counts the cumulative number of bytes received from clients
Willy Tarreauc9705a12010-07-27 20:05:50 +02009165 which matched this entry. Headers are included in the count. This may be
9166 used to limit abuse of upload features on photo or video servers.
9167
9168 - bytes_in_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9169 integer parameter <period> which indicates in milliseconds the length
9170 of the period over which the average is measured. It reports the average
9171 incoming bytes rate over that period, in bytes per period. It may be used
9172 to detect users which upload too much and too fast. Warning: with large
9173 uploads, it is possible that the amount of uploaded data will be counted
9174 once upon termination, thus causing spikes in the average transfer speed
9175 instead of having a smooth one. This may partially be smoothed with
9176 "option contstats" though this is not perfect yet. Use of byte_in_cnt is
9177 recommended for better fairness.
9178
9179 - bytes_out_cnt : server to client byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +01009180 integer which counts the cumulative number of bytes sent to clients which
Willy Tarreauc9705a12010-07-27 20:05:50 +02009181 matched this entry. Headers are included in the count. This may be used
9182 to limit abuse of bots sucking the whole site.
9183
9184 - bytes_out_rate(<period>) : frequency counter (takes 12 bytes). It takes
9185 an integer parameter <period> which indicates in milliseconds the length
9186 of the period over which the average is measured. It reports the average
9187 outgoing bytes rate over that period, in bytes per period. It may be used
9188 to detect users which download too much and too fast. Warning: with large
9189 transfers, it is possible that the amount of transferred data will be
9190 counted once upon termination, thus causing spikes in the average
9191 transfer speed instead of having a smooth one. This may partially be
9192 smoothed with "option contstats" though this is not perfect yet. Use of
9193 byte_out_cnt is recommended for better fairness.
Willy Tarreau08d5f982010-06-06 13:34:54 +02009194
Willy Tarreauc00cdc22010-06-06 16:48:26 +02009195 There is only one stick-table per proxy. At the moment of writing this doc,
9196 it does not seem useful to have multiple tables per proxy. If this happens
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009197 to be required, simply create a dummy backend with a stick-table in it and
9198 reference it.
9199
9200 It is important to understand that stickiness based on learning information
9201 has some limitations, including the fact that all learned associations are
Baptiste Assmann123ff042016-03-06 23:29:28 +01009202 lost upon restart unless peers are properly configured to transfer such
9203 information upon restart (recommended). In general it can be good as a
9204 complement but not always as an exclusive stickiness.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009205
Willy Tarreauc9705a12010-07-27 20:05:50 +02009206 Last, memory requirements may be important when storing many data types.
9207 Indeed, storing all indicators above at once in each entry requires 116 bytes
9208 per entry, or 116 MB for a 1-million entries table. This is definitely not
9209 something that can be ignored.
9210
9211 Example:
9212 # Keep track of counters of up to 1 million IP addresses over 5 minutes
9213 # and store a general purpose counter and the average connection rate
9214 # computed over a sliding window of 30 seconds.
9215 stick-table type ip size 1m expire 5m store gpc0,conn_rate(30s)
9216
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +03009217 See also : "stick match", "stick on", "stick store-request", section 2.4
David du Colombiera13d1b92011-03-17 10:40:22 +01009218 about time format and section 7 about ACLs.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009219
9220
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009221stick store-response <pattern> [table <table>] [{if | unless} <condition>]
Baptiste Assmann2f2d2ec2016-03-06 23:27:24 +01009222 Define a response pattern used to create an entry in a stickiness table
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009223 May be used in sections : defaults | frontend | listen | backend
9224 no | no | yes | yes
9225
9226 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02009227 <pattern> is a sample expression rule as described in section 7.3. It
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009228 describes what elements of the response or connection will
Davor Ocelice9ed2812017-12-25 17:49:28 +01009229 be analyzed, extracted and stored in the table once a
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009230 server is selected.
9231
9232 <table> is an optional stickiness table name. If unspecified, the same
9233 backend's table is used. A stickiness table is declared using
9234 the "stick-table" statement.
9235
9236 <cond> is an optional storage condition. It makes it possible to store
9237 certain criteria only when some conditions are met (or not met).
9238 For instance, it could be used to store the SSL session ID only
9239 when the response is a SSL server hello.
9240
9241 Some protocols or applications require complex stickiness rules and cannot
9242 always simply rely on cookies nor hashing. The "stick store-response"
9243 statement describes a rule to decide what to extract from the response and
9244 when to do it, in order to store it into a stickiness table for further
9245 requests to match it using the "stick match" statement. Obviously the
9246 extracted part must make sense and have a chance to be matched in a further
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009247 request. Storing an ID found in a header of a response makes sense.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009248 See section 7 for a complete list of possible patterns and transformation
9249 rules.
9250
9251 The table has to be declared using the "stick-table" statement. It must be of
9252 a type compatible with the pattern. By default it is the one which is present
9253 in the same backend. It is possible to share a table with other backends by
9254 referencing it using the "table" keyword. If another table is referenced,
9255 the server's ID inside the backends are used. By default, all server IDs
9256 start at 1 in each backend, so the server ordering is enough. But in case of
9257 doubt, it is highly recommended to force server IDs using their "id" setting.
9258
9259 It is possible to restrict the conditions where a "stick store-response"
9260 statement will apply, using "if" or "unless" followed by a condition. This
9261 condition will be evaluated while parsing the response, so any criteria can
9262 be used. See section 7 for ACL based conditions.
9263
9264 There is no limit on the number of "stick store-response" statements, but
9265 there is a limit of 8 simultaneous stores per request or response. This
9266 makes it possible to store up to 8 criteria, all extracted from either the
9267 request or the response, regardless of the number of rules. Only the 8 first
9268 ones which match will be kept. Using this, it is possible to feed multiple
9269 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01009270 another protocol or access method. Using multiple store-response rules with
9271 the same table is possible and may be used to find the best criterion to rely
9272 on, by arranging the rules by decreasing preference order. Only the first
9273 extracted criterion for a given table will be stored. All subsequent store-
9274 response rules referencing the same table will be skipped and their ACLs will
9275 not be evaluated. However, even if a store-request rule references a table, a
9276 store-response rule may also use the same table. This means that each table
9277 may learn exactly one element from the request and one element from the
9278 response at once.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009279
9280 The table will contain the real server that processed the request.
9281
9282 Example :
9283 # Learn SSL session ID from both request and response and create affinity.
9284 backend https
9285 mode tcp
9286 balance roundrobin
Cyril Bontédc4d9032012-04-08 21:57:39 +02009287 # maximum SSL session ID length is 32 bytes.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009288 stick-table type binary len 32 size 30k expire 30m
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009289
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009290 acl clienthello req_ssl_hello_type 1
9291 acl serverhello rep_ssl_hello_type 2
9292
9293 # use tcp content accepts to detects ssl client and server hello.
9294 tcp-request inspect-delay 5s
9295 tcp-request content accept if clienthello
9296
9297 # no timeout on response inspect delay by default.
9298 tcp-response content accept if serverhello
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009299
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009300 # SSL session ID (SSLID) may be present on a client or server hello.
9301 # Its length is coded on 1 byte at offset 43 and its value starts
9302 # at offset 44.
9303
9304 # Match and learn on request if client hello.
9305 stick on payload_lv(43,1) if clienthello
9306
9307 # Learn on response if server hello.
9308 stick store-response payload_lv(43,1) if serverhello
Cyril Bontédc4d9032012-04-08 21:57:39 +02009309
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009310 server s1 192.168.1.1:443
9311 server s2 192.168.1.1:443
9312
9313 See also : "stick-table", "stick on", and section 7 about ACLs and pattern
9314 extraction.
9315
9316
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009317tcp-check connect [params*]
9318 Opens a new connection
9319 May be used in sections: defaults | frontend | listen | backend
9320 no | no | yes | yes
9321
9322 When an application lies on more than a single TCP port or when HAProxy
9323 load-balance many services in a single backend, it makes sense to probe all
9324 the services individually before considering a server as operational.
9325
9326 When there are no TCP port configured on the server line neither server port
9327 directive, then the 'tcp-check connect port <port>' must be the first step
9328 of the sequence.
9329
9330 In a tcp-check ruleset a 'connect' is required, it is also mandatory to start
9331 the ruleset with a 'connect' rule. Purpose is to ensure admin know what they
9332 do.
9333
9334 Parameters :
9335 They are optional and can be used to describe how HAProxy should open and
9336 use the TCP connection.
9337
9338 port if not set, check port or server port is used.
9339 It tells HAProxy where to open the connection to.
9340 <port> must be a valid TCP port source integer, from 1 to 65535.
9341
9342 send-proxy send a PROXY protocol string
9343
9344 ssl opens a ciphered connection
9345
9346 Examples:
9347 # check HTTP and HTTPs services on a server.
9348 # first open port 80 thanks to server line port directive, then
9349 # tcp-check opens port 443, ciphered and run a request on it:
9350 option tcp-check
9351 tcp-check connect
9352 tcp-check send GET\ /\ HTTP/1.0\r\n
9353 tcp-check send Host:\ haproxy.1wt.eu\r\n
9354 tcp-check send \r\n
9355 tcp-check expect rstring (2..|3..)
9356 tcp-check connect port 443 ssl
9357 tcp-check send GET\ /\ HTTP/1.0\r\n
9358 tcp-check send Host:\ haproxy.1wt.eu\r\n
9359 tcp-check send \r\n
9360 tcp-check expect rstring (2..|3..)
9361 server www 10.0.0.1 check port 80
9362
9363 # check both POP and IMAP from a single server:
9364 option tcp-check
9365 tcp-check connect port 110
9366 tcp-check expect string +OK\ POP3\ ready
9367 tcp-check connect port 143
9368 tcp-check expect string *\ OK\ IMAP4\ ready
9369 server mail 10.0.0.1 check
9370
9371 See also : "option tcp-check", "tcp-check send", "tcp-check expect"
9372
9373
9374tcp-check expect [!] <match> <pattern>
Davor Ocelice9ed2812017-12-25 17:49:28 +01009375 Specify data to be collected and analyzed during a generic health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009376 May be used in sections: defaults | frontend | listen | backend
9377 no | no | yes | yes
9378
9379 Arguments :
9380 <match> is a keyword indicating how to look for a specific pattern in the
9381 response. The keyword may be one of "string", "rstring" or
9382 binary.
9383 The keyword may be preceded by an exclamation mark ("!") to negate
9384 the match. Spaces are allowed between the exclamation mark and the
9385 keyword. See below for more details on the supported keywords.
9386
9387 <pattern> is the pattern to look for. It may be a string or a regular
9388 expression. If the pattern contains spaces, they must be escaped
9389 with the usual backslash ('\').
9390 If the match is set to binary, then the pattern must be passed as
Davor Ocelice9ed2812017-12-25 17:49:28 +01009391 a series of hexadecimal digits in an even number. Each sequence of
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009392 two digits will represent a byte. The hexadecimal digits may be
9393 used upper or lower case.
9394
9395
9396 The available matches are intentionally similar to their http-check cousins :
9397
9398 string <string> : test the exact string matches in the response buffer.
9399 A health check response will be considered valid if the
9400 response's buffer contains this exact string. If the
9401 "string" keyword is prefixed with "!", then the response
9402 will be considered invalid if the body contains this
9403 string. This can be used to look for a mandatory pattern
9404 in a protocol response, or to detect a failure when a
9405 specific error appears in a protocol banner.
9406
9407 rstring <regex> : test a regular expression on the response buffer.
9408 A health check response will be considered valid if the
9409 response's buffer matches this expression. If the
9410 "rstring" keyword is prefixed with "!", then the response
9411 will be considered invalid if the body matches the
9412 expression.
9413
9414 binary <hexstring> : test the exact string in its hexadecimal form matches
9415 in the response buffer. A health check response will
9416 be considered valid if the response's buffer contains
9417 this exact hexadecimal string.
9418 Purpose is to match data on binary protocols.
9419
9420 It is important to note that the responses will be limited to a certain size
9421 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
9422 Thus, too large responses may not contain the mandatory pattern when using
9423 "string", "rstring" or binary. If a large response is absolutely required, it
9424 is possible to change the default max size by setting the global variable.
9425 However, it is worth keeping in mind that parsing very large responses can
9426 waste some CPU cycles, especially when regular expressions are used, and that
9427 it is always better to focus the checks on smaller resources. Also, in its
9428 current state, the check will not find any string nor regex past a null
9429 character in the response. Similarly it is not possible to request matching
9430 the null character.
9431
9432 Examples :
9433 # perform a POP check
9434 option tcp-check
9435 tcp-check expect string +OK\ POP3\ ready
9436
9437 # perform an IMAP check
9438 option tcp-check
9439 tcp-check expect string *\ OK\ IMAP4\ ready
9440
9441 # look for the redis master server
9442 option tcp-check
9443 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02009444 tcp-check expect string +PONG
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009445 tcp-check send info\ replication\r\n
9446 tcp-check expect string role:master
9447 tcp-check send QUIT\r\n
9448 tcp-check expect string +OK
9449
9450
9451 See also : "option tcp-check", "tcp-check connect", "tcp-check send",
9452 "tcp-check send-binary", "http-check expect", tune.chksize
9453
9454
9455tcp-check send <data>
9456 Specify a string to be sent as a question during a generic health check
9457 May be used in sections: defaults | frontend | listen | backend
9458 no | no | yes | yes
9459
9460 <data> : the data to be sent as a question during a generic health check
9461 session. For now, <data> must be a string.
9462
9463 Examples :
9464 # look for the redis master server
9465 option tcp-check
9466 tcp-check send info\ replication\r\n
9467 tcp-check expect string role:master
9468
9469 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
9470 "tcp-check send-binary", tune.chksize
9471
9472
Davor Ocelice9ed2812017-12-25 17:49:28 +01009473tcp-check send-binary <hexstring>
9474 Specify a hex digits string to be sent as a binary question during a raw
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009475 tcp health check
9476 May be used in sections: defaults | frontend | listen | backend
9477 no | no | yes | yes
9478
9479 <data> : the data to be sent as a question during a generic health check
9480 session. For now, <data> must be a string.
Davor Ocelice9ed2812017-12-25 17:49:28 +01009481 <hexstring> : test the exact string in its hexadecimal form matches in the
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009482 response buffer. A health check response will be considered
9483 valid if the response's buffer contains this exact
9484 hexadecimal string.
9485 Purpose is to send binary data to ask on binary protocols.
9486
9487 Examples :
9488 # redis check in binary
9489 option tcp-check
9490 tcp-check send-binary 50494e470d0a # PING\r\n
9491 tcp-check expect binary 2b504F4e47 # +PONG
9492
9493
9494 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
9495 "tcp-check send", tune.chksize
9496
9497
Willy Tarreaue9656522010-08-17 15:40:09 +02009498tcp-request connection <action> [{if | unless} <condition>]
9499 Perform an action on an incoming connection depending on a layer 4 condition
Willy Tarreau1a687942010-05-23 22:40:30 +02009500 May be used in sections : defaults | frontend | listen | backend
9501 no | yes | yes | no
Willy Tarreaue9656522010-08-17 15:40:09 +02009502 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02009503 <action> defines the action to perform if the condition applies. See
9504 below.
Willy Tarreau1a687942010-05-23 22:40:30 +02009505
Willy Tarreaue9656522010-08-17 15:40:09 +02009506 <condition> is a standard layer4-only ACL-based condition (see section 7).
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009507
9508 Immediately after acceptance of a new incoming connection, it is possible to
9509 evaluate some conditions to decide whether this connection must be accepted
Willy Tarreaue9656522010-08-17 15:40:09 +02009510 or dropped or have its counters tracked. Those conditions cannot make use of
9511 any data contents because the connection has not been read from yet, and the
9512 buffers are not yet allocated. This is used to selectively and very quickly
9513 accept or drop connections from various sources with a very low overhead. If
9514 some contents need to be inspected in order to take the decision, the
9515 "tcp-request content" statements must be used instead.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009516
Willy Tarreaue9656522010-08-17 15:40:09 +02009517 The "tcp-request connection" rules are evaluated in their exact declaration
9518 order. If no rule matches or if there is no rule, the default action is to
9519 accept the incoming connection. There is no specific limit to the number of
9520 rules which may be inserted.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009521
Willy Tarreaua9083d02015-05-08 15:27:59 +02009522 Four types of actions are supported :
Willy Tarreaue9656522010-08-17 15:40:09 +02009523 - accept :
9524 accepts the connection if the condition is true (when used with "if")
9525 or false (when used with "unless"). The first such rule executed ends
9526 the rules evaluation.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009527
Willy Tarreaue9656522010-08-17 15:40:09 +02009528 - reject :
9529 rejects the connection if the condition is true (when used with "if")
9530 or false (when used with "unless"). The first such rule executed ends
9531 the rules evaluation. Rejected connections do not even become a
9532 session, which is why they are accounted separately for in the stats,
9533 as "denied connections". They are not considered for the session
9534 rate-limit and are not logged either. The reason is that these rules
9535 should only be used to filter extremely high connection rates such as
9536 the ones encountered during a massive DDoS attack. Under these extreme
9537 conditions, the simple action of logging each event would make the
9538 system collapse and would considerably lower the filtering capacity. If
9539 logging is absolutely desired, then "tcp-request content" rules should
Willy Tarreau4f614292016-10-21 17:49:36 +02009540 be used instead, as "tcp-request session" rules will not log either.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009541
Willy Tarreau4f0d9192013-06-11 20:40:55 +02009542 - expect-proxy layer4 :
9543 configures the client-facing connection to receive a PROXY protocol
9544 header before any byte is read from the socket. This is equivalent to
9545 having the "accept-proxy" keyword on the "bind" line, except that using
9546 the TCP rule allows the PROXY protocol to be accepted only for certain
9547 IP address ranges using an ACL. This is convenient when multiple layers
9548 of load balancers are passed through by traffic coming from public
9549 hosts.
9550
Bertrand Jacquin90759682016-06-06 15:35:39 +01009551 - expect-netscaler-cip layer4 :
9552 configures the client-facing connection to receive a NetScaler Client
9553 IP insertion protocol header before any byte is read from the socket.
9554 This is equivalent to having the "accept-netscaler-cip" keyword on the
9555 "bind" line, except that using the TCP rule allows the PROXY protocol
9556 to be accepted only for certain IP address ranges using an ACL. This
9557 is convenient when multiple layers of load balancers are passed
9558 through by traffic coming from public hosts.
9559
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009560 - capture <sample> len <length> :
9561 This only applies to "tcp-request content" rules. It captures sample
9562 expression <sample> from the request buffer, and converts it to a
9563 string of at most <len> characters. The resulting string is stored into
9564 the next request "capture" slot, so it will possibly appear next to
9565 some captured HTTP headers. It will then automatically appear in the
9566 logs, and it will be possible to extract it using sample fetch rules to
9567 feed it into headers or anything. The length should be limited given
9568 that this size will be allocated for each capture during the whole
Willy Tarreaua9083d02015-05-08 15:27:59 +02009569 session life. Please check section 7.3 (Fetching samples) and "capture
9570 request header" for more information.
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009571
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009572 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] :
Willy Tarreaue9656522010-08-17 15:40:09 +02009573 enables tracking of sticky counters from current connection. These
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +02009574 rules do not stop evaluation and do not change default action. The
9575 number of counters that may be simultaneously tracked by the same
9576 connection is set in MAX_SESS_STKCTR at build time (reported in
John Roeslerfb2fce12019-07-10 15:45:51 -05009577 haproxy -vv) which defaults to 3, so the track-sc number is between 0
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +02009578 and (MAX_SESS_STCKTR-1). The first "track-sc0" rule executed enables
9579 tracking of the counters of the specified table as the first set. The
9580 first "track-sc1" rule executed enables tracking of the counters of the
9581 specified table as the second set. The first "track-sc2" rule executed
9582 enables tracking of the counters of the specified table as the third
9583 set. It is a recommended practice to use the first set of counters for
9584 the per-frontend counters and the second set for the per-backend ones.
9585 But this is just a guideline, all may be used everywhere.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009586
Willy Tarreaue9656522010-08-17 15:40:09 +02009587 These actions take one or two arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02009588 <key> is mandatory, and is a sample expression rule as described
Willy Tarreau74ca5042013-06-11 23:12:07 +02009589 in section 7.3. It describes what elements of the incoming
Davor Ocelice9ed2812017-12-25 17:49:28 +01009590 request or connection will be analyzed, extracted, combined,
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009591 and used to select which table entry to update the counters.
9592 Note that "tcp-request connection" cannot use content-based
9593 fetches.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009594
Willy Tarreaue9656522010-08-17 15:40:09 +02009595 <table> is an optional table to be used instead of the default one,
9596 which is the stick-table declared in the current proxy. All
9597 the counters for the matches and updates for the key will
9598 then be performed in that table until the session ends.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009599
Willy Tarreaue9656522010-08-17 15:40:09 +02009600 Once a "track-sc*" rule is executed, the key is looked up in the table
9601 and if it is not found, an entry is allocated for it. Then a pointer to
9602 that entry is kept during all the session's life, and this entry's
9603 counters are updated as often as possible, every time the session's
9604 counters are updated, and also systematically when the session ends.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009605 Counters are only updated for events that happen after the tracking has
9606 been started. For example, connection counters will not be updated when
9607 tracking layer 7 information, since the connection event happens before
9608 layer7 information is extracted.
9609
Willy Tarreaue9656522010-08-17 15:40:09 +02009610 If the entry tracks concurrent connection counters, one connection is
9611 counted for as long as the entry is tracked, and the entry will not
9612 expire during that time. Tracking counters also provides a performance
9613 advantage over just checking the keys, because only one table lookup is
9614 performed for all ACL checks that make use of it.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009615
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02009616 - sc-inc-gpc0(<sc-id>):
9617 The "sc-inc-gpc0" increments the GPC0 counter according to the sticky
9618 counter designated by <sc-id>. If an error occurs, this action silently
9619 fails and the actions evaluation continues.
9620
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009621 - sc-inc-gpc1(<sc-id>):
9622 The "sc-inc-gpc1" increments the GPC1 counter according to the sticky
9623 counter designated by <sc-id>. If an error occurs, this action silently
9624 fails and the actions evaluation continues.
9625
Cédric Dufour0d7712d2019-11-06 18:38:53 +01009626 - sc-set-gpt0(<sc-id>) { <int> | <expr> }:
9627 This action sets the 32-bit unsigned GPT0 tag according to the sticky
9628 counter designated by <sc-id> and the value of <int>/<expr>. The
9629 expected result is a boolean. If an error occurs, this action silently
9630 fails and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009631
William Lallemand2e785f22016-05-25 01:48:42 +02009632 - set-src <expr> :
9633 Is used to set the source IP address to the value of specified
9634 expression. Useful if you want to mask source IP for privacy.
9635 If you want to provide an IP from a HTTP header use "http-request
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02009636 set-src".
William Lallemand2e785f22016-05-25 01:48:42 +02009637
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02009638 Arguments:
9639 <expr> Is a standard HAProxy expression formed by a sample-fetch
9640 followed by some converters.
William Lallemand2e785f22016-05-25 01:48:42 +02009641
9642 Example:
William Lallemand2e785f22016-05-25 01:48:42 +02009643 tcp-request connection set-src src,ipmask(24)
9644
Willy Tarreau0c630532016-10-21 17:52:58 +02009645 When possible, set-src preserves the original source port as long as the
9646 address family allows it, otherwise the source port is set to 0.
William Lallemand2e785f22016-05-25 01:48:42 +02009647
William Lallemand44be6402016-05-25 01:51:35 +02009648 - set-src-port <expr> :
9649 Is used to set the source port address to the value of specified
9650 expression.
9651
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02009652 Arguments:
9653 <expr> Is a standard HAProxy expression formed by a sample-fetch
9654 followed by some converters.
William Lallemand44be6402016-05-25 01:51:35 +02009655
9656 Example:
William Lallemand44be6402016-05-25 01:51:35 +02009657 tcp-request connection set-src-port int(4000)
9658
Willy Tarreau0c630532016-10-21 17:52:58 +02009659 When possible, set-src-port preserves the original source address as long
9660 as the address family supports a port, otherwise it forces the source
9661 address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +02009662
William Lallemand13e9b0c2016-05-25 02:34:07 +02009663 - set-dst <expr> :
9664 Is used to set the destination IP address to the value of specified
9665 expression. Useful if you want to mask IP for privacy in log.
9666 If you want to provide an IP from a HTTP header use "http-request
9667 set-dst". If you want to connect to the new address/port, use
9668 '0.0.0.0:0' as a server address in the backend.
9669
9670 <expr> Is a standard HAProxy expression formed by a sample-fetch
9671 followed by some converters.
9672
9673 Example:
9674
9675 tcp-request connection set-dst dst,ipmask(24)
9676 tcp-request connection set-dst ipv4(10.0.0.1)
9677
Willy Tarreau0c630532016-10-21 17:52:58 +02009678 When possible, set-dst preserves the original destination port as long as
9679 the address family allows it, otherwise the destination port is set to 0.
9680
William Lallemand13e9b0c2016-05-25 02:34:07 +02009681 - set-dst-port <expr> :
9682 Is used to set the destination port address to the value of specified
9683 expression. If you want to connect to the new address/port, use
9684 '0.0.0.0:0' as a server address in the backend.
9685
9686
9687 <expr> Is a standard HAProxy expression formed by a sample-fetch
9688 followed by some converters.
9689
9690 Example:
9691
9692 tcp-request connection set-dst-port int(4000)
9693
Willy Tarreau0c630532016-10-21 17:52:58 +02009694 When possible, set-dst-port preserves the original destination address as
9695 long as the address family supports a port, otherwise it forces the
9696 destination address to IPv4 "0.0.0.0" before rewriting the port.
9697
Willy Tarreau2d392c22015-08-24 01:43:45 +02009698 - "silent-drop" :
9699 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +01009700 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +02009701 to prevent the client from being notified. The effect it then that the
9702 client still sees an established connection while there's none on
9703 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
9704 except that it doesn't use any local resource at all on the machine
9705 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +01009706 slow down stronger attackers. It is important to understand the impact
9707 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009708 client and HAProxy (firewalls, proxies, load balancers) will also keep
9709 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +01009710 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009711 TCP_REPAIR socket option is used to block the emission of a TCP
9712 reset. On other systems, the socket's TTL is reduced to 1 so that the
9713 TCP reset doesn't pass the first router, though it's still delivered to
9714 local networks. Do not use it unless you fully understand how it works.
9715
Willy Tarreaue9656522010-08-17 15:40:09 +02009716 Note that the "if/unless" condition is optional. If no condition is set on
9717 the action, it is simply performed unconditionally. That can be useful for
9718 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009719
Willy Tarreaue9656522010-08-17 15:40:09 +02009720 Example: accept all connections from white-listed hosts, reject too fast
9721 connection without counting them, and track accepted connections.
9722 This results in connection rate being capped from abusive sources.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009723
Willy Tarreaue9656522010-08-17 15:40:09 +02009724 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009725 tcp-request connection reject if { src_conn_rate gt 10 }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009726 tcp-request connection track-sc0 src
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009727
Willy Tarreaue9656522010-08-17 15:40:09 +02009728 Example: accept all connections from white-listed hosts, count all other
9729 connections and reject too fast ones. This results in abusive ones
9730 being blocked as long as they don't slow down.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009731
Willy Tarreaue9656522010-08-17 15:40:09 +02009732 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009733 tcp-request connection track-sc0 src
9734 tcp-request connection reject if { sc0_conn_rate gt 10 }
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009735
Willy Tarreau4f0d9192013-06-11 20:40:55 +02009736 Example: enable the PROXY protocol for traffic coming from all known proxies.
9737
9738 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
9739
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009740 See section 7 about ACL usage.
9741
Willy Tarreau4f614292016-10-21 17:49:36 +02009742 See also : "tcp-request session", "tcp-request content", "stick-table"
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009743
9744
Willy Tarreaue9656522010-08-17 15:40:09 +02009745tcp-request content <action> [{if | unless} <condition>]
9746 Perform an action on a new session depending on a layer 4-7 condition
Willy Tarreau62644772008-07-16 18:36:06 +02009747 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +02009748 no | yes | yes | yes
Willy Tarreaue9656522010-08-17 15:40:09 +02009749 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02009750 <action> defines the action to perform if the condition applies. See
9751 below.
Willy Tarreau62644772008-07-16 18:36:06 +02009752
Willy Tarreaue9656522010-08-17 15:40:09 +02009753 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
Willy Tarreau62644772008-07-16 18:36:06 +02009754
Davor Ocelice9ed2812017-12-25 17:49:28 +01009755 A request's contents can be analyzed at an early stage of request processing
Willy Tarreaue9656522010-08-17 15:40:09 +02009756 called "TCP content inspection". During this stage, ACL-based rules are
9757 evaluated every time the request contents are updated, until either an
9758 "accept" or a "reject" rule matches, or the TCP request inspection delay
9759 expires with no matching rule.
Willy Tarreau62644772008-07-16 18:36:06 +02009760
Willy Tarreaue9656522010-08-17 15:40:09 +02009761 The first difference between these rules and "tcp-request connection" rules
9762 is that "tcp-request content" rules can make use of contents to take a
9763 decision. Most often, these decisions will consider a protocol recognition or
9764 validity. The second difference is that content-based rules can be used in
Willy Tarreauf3338342014-01-28 21:40:28 +01009765 both frontends and backends. In case of HTTP keep-alive with the client, all
9766 tcp-request content rules are evaluated again, so haproxy keeps a record of
9767 what sticky counters were assigned by a "tcp-request connection" versus a
9768 "tcp-request content" rule, and flushes all the content-related ones after
9769 processing an HTTP request, so that they may be evaluated again by the rules
9770 being evaluated again for the next request. This is of particular importance
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009771 when the rule tracks some L7 information or when it is conditioned by an
Willy Tarreauf3338342014-01-28 21:40:28 +01009772 L7-based ACL, since tracking may change between requests.
Willy Tarreau62644772008-07-16 18:36:06 +02009773
Willy Tarreaue9656522010-08-17 15:40:09 +02009774 Content-based rules are evaluated in their exact declaration order. If no
9775 rule matches or if there is no rule, the default action is to accept the
9776 contents. There is no specific limit to the number of rules which may be
9777 inserted.
Willy Tarreau62644772008-07-16 18:36:06 +02009778
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009779 Several types of actions are supported :
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009780 - accept : the request is accepted
Baptiste Assmann333939c2019-01-21 08:34:50 +01009781 - do-resolve: perform a DNS resolution
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009782 - reject : the request is rejected and the connection is closed
9783 - capture : the specified sample expression is captured
Patrick Hemmer268a7072018-05-11 12:52:31 -04009784 - set-priority-class <expr> | set-priority-offset <expr>
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009785 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02009786 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009787 - sc-inc-gpc1(<sc-id>)
Cédric Dufour0d7712d2019-11-06 18:38:53 +01009788 - sc-set-gpt0(<sc-id>) { <int> | <expr> }
Baptiste Assmanne1afd4f2019-04-18 16:21:13 +02009789 - set-dst <expr>
9790 - set-dst-port <expr>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009791 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +01009792 - unset-var(<var-name>)
Willy Tarreau2d392c22015-08-24 01:43:45 +02009793 - silent-drop
Davor Ocelice9ed2812017-12-25 17:49:28 +01009794 - send-spoe-group <engine-name> <group-name>
Christopher Faulet579d83b2019-11-22 15:34:17 +01009795 - use-service <service-name>
Willy Tarreau62644772008-07-16 18:36:06 +02009796
Willy Tarreaue9656522010-08-17 15:40:09 +02009797 They have the same meaning as their counter-parts in "tcp-request connection"
9798 so please refer to that section for a complete description.
Baptiste Assmann333939c2019-01-21 08:34:50 +01009799 For "do-resolve" action, please check the "http-request do-resolve"
9800 configuration section.
Willy Tarreau62644772008-07-16 18:36:06 +02009801
Willy Tarreauf3338342014-01-28 21:40:28 +01009802 While there is nothing mandatory about it, it is recommended to use the
9803 track-sc0 in "tcp-request connection" rules, track-sc1 for "tcp-request
9804 content" rules in the frontend, and track-sc2 for "tcp-request content"
9805 rules in the backend, because that makes the configuration more readable
9806 and easier to troubleshoot, but this is just a guideline and all counters
9807 may be used everywhere.
Willy Tarreau62644772008-07-16 18:36:06 +02009808
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009809 Note that the "if/unless" condition is optional. If no condition is set on
Willy Tarreaue9656522010-08-17 15:40:09 +02009810 the action, it is simply performed unconditionally. That can be useful for
9811 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau62644772008-07-16 18:36:06 +02009812
Willy Tarreaue9656522010-08-17 15:40:09 +02009813 It is perfectly possible to match layer 7 contents with "tcp-request content"
Willy Tarreauc0239e02012-04-16 14:42:55 +02009814 rules, since HTTP-specific ACL matches are able to preliminarily parse the
9815 contents of a buffer before extracting the required data. If the buffered
9816 contents do not parse as a valid HTTP message, then the ACL does not match.
9817 The parser which is involved there is exactly the same as for all other HTTP
Willy Tarreauf3338342014-01-28 21:40:28 +01009818 processing, so there is no risk of parsing something differently. In an HTTP
9819 backend connected to from an HTTP frontend, it is guaranteed that HTTP
9820 contents will always be immediately present when the rule is evaluated first.
Willy Tarreau62644772008-07-16 18:36:06 +02009821
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009822 Tracking layer7 information is also possible provided that the information
Willy Tarreau4d54c7c2014-09-16 15:48:15 +02009823 are present when the rule is processed. The rule processing engine is able to
9824 wait until the inspect delay expires when the data to be tracked is not yet
9825 available.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009826
Baptiste Assmanne1afd4f2019-04-18 16:21:13 +02009827 The "set-dst" and "set-dst-port" are used to set respectively the destination
9828 IP and port. More information on how to use it at "http-request set-dst".
9829
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009830 The "set-var" is used to set the content of a variable. The variable is
Willy Tarreau4f614292016-10-21 17:49:36 +02009831 declared inline. For "tcp-request session" rules, only session-level
9832 variables can be used, without any layer7 contents.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009833
Daniel Schneller0b547052016-03-21 20:46:57 +01009834 <var-name> The name of the variable starts with an indication about
9835 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +01009836 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +01009837 "sess" : the variable is shared with the whole session
9838 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009839 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +01009840 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009841 processing
Daniel Schneller0b547052016-03-21 20:46:57 +01009842 "res" : the variable is shared only during response
9843 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009844 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +01009845 The name may only contain characters 'a-z', 'A-Z', '0-9',
9846 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009847
9848 <expr> Is a standard HAProxy expression formed by a sample-fetch
9849 followed by some converters.
9850
Christopher Faulet85d79c92016-11-09 16:54:56 +01009851 The "unset-var" is used to unset a variable. See above for details about
9852 <var-name>.
9853
Patrick Hemmer268a7072018-05-11 12:52:31 -04009854 The "set-priority-class" is used to set the queue priority class of the
9855 current request. The value must be a sample expression which converts to an
9856 integer in the range -2047..2047. Results outside this range will be
9857 truncated. The priority class determines the order in which queued requests
9858 are processed. Lower values have higher priority.
9859
9860 The "set-priority-offset" is used to set the queue priority timestamp offset
9861 of the current request. The value must be a sample expression which converts
9862 to an integer in the range -524287..524287. Results outside this range will be
9863 truncated. When a request is queued, it is ordered first by the priority
9864 class, then by the current timestamp adjusted by the given offset in
9865 milliseconds. Lower values have higher priority.
9866 Note that the resulting timestamp is is only tracked with enough precision for
9867 524,287ms (8m44s287ms). If the request is queued long enough to where the
9868 adjusted timestamp exceeds this value, it will be misidentified as highest
9869 priority. Thus it is important to set "timeout queue" to a value, where when
9870 combined with the offset, does not exceed this limit.
9871
Christopher Faulet76c09ef2017-09-21 11:03:52 +02009872 The "send-spoe-group" is used to trigger sending of a group of SPOE
9873 messages. To do so, the SPOE engine used to send messages must be defined, as
9874 well as the SPOE group to send. Of course, the SPOE engine must refer to an
9875 existing SPOE filter. If not engine name is provided on the SPOE filter line,
9876 the SPOE agent name must be used.
9877
9878 <engine-name> The SPOE engine name.
9879
9880 <group-name> The SPOE group name as specified in the engine configuration.
9881
Christopher Faulet579d83b2019-11-22 15:34:17 +01009882 The "use-service" is used to executes a TCP service which will reply to the
9883 request and stop the evaluation of the rules. This service may choose to
9884 reply by sending any valid response or it may immediately close the
9885 connection without sending anything. Outside natives services, it is possible
9886 to write your own services in Lua. No further "tcp-request" rules are
9887 evaluated.
9888
9889 Example:
9890 tcp-request content use-service lua.deny { src -f /etc/haproxy/blacklist.lst }
9891
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009892 Example:
9893
9894 tcp-request content set-var(sess.my_var) src
Christopher Faulet85d79c92016-11-09 16:54:56 +01009895 tcp-request content unset-var(sess.my_var2)
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009896
Willy Tarreau62644772008-07-16 18:36:06 +02009897 Example:
Willy Tarreaue9656522010-08-17 15:40:09 +02009898 # Accept HTTP requests containing a Host header saying "example.com"
9899 # and reject everything else.
9900 acl is_host_com hdr(Host) -i example.com
9901 tcp-request inspect-delay 30s
Willy Tarreauc0239e02012-04-16 14:42:55 +02009902 tcp-request content accept if is_host_com
Willy Tarreaue9656522010-08-17 15:40:09 +02009903 tcp-request content reject
9904
9905 Example:
Willy Tarreau62644772008-07-16 18:36:06 +02009906 # reject SMTP connection if client speaks first
9907 tcp-request inspect-delay 30s
9908 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009909 tcp-request content reject if content_present
Willy Tarreau62644772008-07-16 18:36:06 +02009910
9911 # Forward HTTPS connection only if client speaks
9912 tcp-request inspect-delay 30s
9913 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009914 tcp-request content accept if content_present
Willy Tarreaue9656522010-08-17 15:40:09 +02009915 tcp-request content reject
9916
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009917 Example:
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009918 # Track the last IP(stick-table type string) from X-Forwarded-For
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009919 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +02009920 tcp-request content track-sc0 hdr(x-forwarded-for,-1)
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009921 # Or track the last IP(stick-table type ip|ipv6) from X-Forwarded-For
9922 tcp-request content track-sc0 req.hdr_ip(x-forwarded-for,-1)
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009923
9924 Example:
9925 # track request counts per "base" (concatenation of Host+URL)
9926 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +02009927 tcp-request content track-sc0 base table req-rate
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009928
Willy Tarreaue9656522010-08-17 15:40:09 +02009929 Example: track per-frontend and per-backend counters, block abusers at the
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009930 frontend when the backend detects abuse(and marks gpc0).
Willy Tarreaue9656522010-08-17 15:40:09 +02009931
9932 frontend http
Davor Ocelice9ed2812017-12-25 17:49:28 +01009933 # Use General Purpose Counter 0 in SC0 as a global abuse counter
Willy Tarreaue9656522010-08-17 15:40:09 +02009934 # protecting all our sites
9935 stick-table type ip size 1m expire 5m store gpc0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009936 tcp-request connection track-sc0 src
9937 tcp-request connection reject if { sc0_get_gpc0 gt 0 }
Willy Tarreaue9656522010-08-17 15:40:09 +02009938 ...
9939 use_backend http_dynamic if { path_end .php }
9940
9941 backend http_dynamic
9942 # if a source makes too fast requests to this dynamic site (tracked
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009943 # by SC1), block it globally in the frontend.
Willy Tarreaue9656522010-08-17 15:40:09 +02009944 stick-table type ip size 1m expire 5m store http_req_rate(10s)
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009945 acl click_too_fast sc1_http_req_rate gt 10
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009946 acl mark_as_abuser sc0_inc_gpc0(http) gt 0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009947 tcp-request content track-sc1 src
Willy Tarreaue9656522010-08-17 15:40:09 +02009948 tcp-request content reject if click_too_fast mark_as_abuser
Willy Tarreau62644772008-07-16 18:36:06 +02009949
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009950 See section 7 about ACL usage.
Willy Tarreau62644772008-07-16 18:36:06 +02009951
Jarno Huuskonen95b012b2017-04-06 13:59:14 +03009952 See also : "tcp-request connection", "tcp-request session",
9953 "tcp-request inspect-delay", and "http-request".
Willy Tarreau62644772008-07-16 18:36:06 +02009954
9955
9956tcp-request inspect-delay <timeout>
9957 Set the maximum allowed time to wait for data during content inspection
9958 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +02009959 no | yes | yes | yes
Willy Tarreau62644772008-07-16 18:36:06 +02009960 Arguments :
9961 <timeout> is the timeout value specified in milliseconds by default, but
9962 can be in any other unit if the number is suffixed by the unit,
9963 as explained at the top of this document.
9964
9965 People using haproxy primarily as a TCP relay are often worried about the
9966 risk of passing any type of protocol to a server without any analysis. In
9967 order to be able to analyze the request contents, we must first withhold
9968 the data then analyze them. This statement simply enables withholding of
9969 data for at most the specified amount of time.
9970
Willy Tarreaufb356202010-08-03 14:02:05 +02009971 TCP content inspection applies very early when a connection reaches a
9972 frontend, then very early when the connection is forwarded to a backend. This
9973 means that a connection may experience a first delay in the frontend and a
9974 second delay in the backend if both have tcp-request rules.
9975
Willy Tarreau62644772008-07-16 18:36:06 +02009976 Note that when performing content inspection, haproxy will evaluate the whole
9977 rules for every new chunk which gets in, taking into account the fact that
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009978 those data are partial. If no rule matches before the aforementioned delay,
Willy Tarreau62644772008-07-16 18:36:06 +02009979 a last check is performed upon expiration, this time considering that the
Willy Tarreaud869b242009-03-15 14:43:58 +01009980 contents are definitive. If no delay is set, haproxy will not wait at all
9981 and will immediately apply a verdict based on the available information.
9982 Obviously this is unlikely to be very useful and might even be racy, so such
9983 setups are not recommended.
Willy Tarreau62644772008-07-16 18:36:06 +02009984
9985 As soon as a rule matches, the request is released and continues as usual. If
9986 the timeout is reached and no rule matches, the default policy will be to let
9987 it pass through unaffected.
9988
9989 For most protocols, it is enough to set it to a few seconds, as most clients
9990 send the full request immediately upon connection. Add 3 or more seconds to
9991 cover TCP retransmits but that's all. For some protocols, it may make sense
Willy Tarreaud72758d2010-01-12 10:42:19 +01009992 to use large values, for instance to ensure that the client never talks
Davor Ocelice9ed2812017-12-25 17:49:28 +01009993 before the server (e.g. SMTP), or to wait for a client to talk before passing
9994 data to the server (e.g. SSL). Note that the client timeout must cover at
Willy Tarreaub824b002010-09-29 16:36:16 +02009995 least the inspection delay, otherwise it will expire first. If the client
9996 closes the connection or if the buffer is full, the delay immediately expires
9997 since the contents will not be able to change anymore.
Willy Tarreau62644772008-07-16 18:36:06 +02009998
Willy Tarreau55165fe2009-05-10 12:02:55 +02009999 See also : "tcp-request content accept", "tcp-request content reject",
Willy Tarreau62644772008-07-16 18:36:06 +020010000 "timeout client".
10001
10002
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010003tcp-response content <action> [{if | unless} <condition>]
10004 Perform an action on a session response depending on a layer 4-7 condition
10005 May be used in sections : defaults | frontend | listen | backend
10006 no | no | yes | yes
10007 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +020010008 <action> defines the action to perform if the condition applies. See
10009 below.
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010010
10011 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
10012
Davor Ocelice9ed2812017-12-25 17:49:28 +010010013 Response contents can be analyzed at an early stage of response processing
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010014 called "TCP content inspection". During this stage, ACL-based rules are
10015 evaluated every time the response contents are updated, until either an
Willy Tarreaucc1e04b2013-09-11 23:20:29 +020010016 "accept", "close" or a "reject" rule matches, or a TCP response inspection
10017 delay is set and expires with no matching rule.
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010018
10019 Most often, these decisions will consider a protocol recognition or validity.
10020
10021 Content-based rules are evaluated in their exact declaration order. If no
10022 rule matches or if there is no rule, the default action is to accept the
10023 contents. There is no specific limit to the number of rules which may be
10024 inserted.
10025
Thierry FOURNIER236657b2015-08-19 08:25:14 +020010026 Several types of actions are supported :
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010027 - accept :
10028 accepts the response if the condition is true (when used with "if")
10029 or false (when used with "unless"). The first such rule executed ends
10030 the rules evaluation.
10031
Willy Tarreaucc1e04b2013-09-11 23:20:29 +020010032 - close :
10033 immediately closes the connection with the server if the condition is
10034 true (when used with "if"), or false (when used with "unless"). The
10035 first such rule executed ends the rules evaluation. The main purpose of
10036 this action is to force a connection to be finished between a client
10037 and a server after an exchange when the application protocol expects
10038 some long time outs to elapse first. The goal is to eliminate idle
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010039 connections which take significant resources on servers with certain
Willy Tarreaucc1e04b2013-09-11 23:20:29 +020010040 protocols.
10041
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010042 - reject :
10043 rejects the response if the condition is true (when used with "if")
10044 or false (when used with "unless"). The first such rule executed ends
Jamie Gloudonaaa21002012-08-25 00:18:33 -040010045 the rules evaluation. Rejected session are immediately closed.
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010046
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010047 - set-var(<var-name>) <expr>
10048 Sets a variable.
10049
Christopher Faulet85d79c92016-11-09 16:54:56 +010010050 - unset-var(<var-name>)
10051 Unsets a variable.
10052
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +020010053 - sc-inc-gpc0(<sc-id>):
10054 This action increments the GPC0 counter according to the sticky
10055 counter designated by <sc-id>. If an error occurs, this action fails
10056 silently and the actions evaluation continues.
10057
Frédéric Lécaille6778b272018-01-29 15:22:53 +010010058 - sc-inc-gpc1(<sc-id>):
10059 This action increments the GPC1 counter according to the sticky
10060 counter designated by <sc-id>. If an error occurs, this action fails
10061 silently and the actions evaluation continues.
10062
Cédric Dufour0d7712d2019-11-06 18:38:53 +010010063 - sc-set-gpt0(<sc-id>) { <int> | <expr> }
10064 This action sets the 32-bit unsigned GPT0 tag according to the sticky
10065 counter designated by <sc-id> and the value of <int>/<expr>. The
10066 expected result is a boolean. If an error occurs, this action silently
10067 fails and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +020010068
Willy Tarreau2d392c22015-08-24 01:43:45 +020010069 - "silent-drop" :
10070 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +010010071 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +020010072 to prevent the client from being notified. The effect it then that the
10073 client still sees an established connection while there's none on
10074 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
10075 except that it doesn't use any local resource at all on the machine
10076 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +010010077 slow down stronger attackers. It is important to understand the impact
10078 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +020010079 client and HAProxy (firewalls, proxies, load balancers) will also keep
10080 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +010010081 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +020010082 TCP_REPAIR socket option is used to block the emission of a TCP
10083 reset. On other systems, the socket's TTL is reduced to 1 so that the
10084 TCP reset doesn't pass the first router, though it's still delivered to
10085 local networks. Do not use it unless you fully understand how it works.
10086
Christopher Faulet76c09ef2017-09-21 11:03:52 +020010087 - send-spoe-group <engine-name> <group-name>
10088 Send a group of SPOE messages.
10089
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010090 Note that the "if/unless" condition is optional. If no condition is set on
10091 the action, it is simply performed unconditionally. That can be useful for
10092 for changing the default action to a reject.
10093
Jamie Gloudonaaa21002012-08-25 00:18:33 -040010094 It is perfectly possible to match layer 7 contents with "tcp-response
10095 content" rules, but then it is important to ensure that a full response has
10096 been buffered, otherwise no contents will match. In order to achieve this,
10097 the best solution involves detecting the HTTP protocol during the inspection
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010098 period.
10099
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010100 The "set-var" is used to set the content of a variable. The variable is
10101 declared inline.
10102
Daniel Schneller0b547052016-03-21 20:46:57 +010010103 <var-name> The name of the variable starts with an indication about
10104 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010010105 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010010106 "sess" : the variable is shared with the whole session
10107 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010108 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +010010109 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010110 processing
Daniel Schneller0b547052016-03-21 20:46:57 +010010111 "res" : the variable is shared only during response
10112 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010113 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +010010114 The name may only contain characters 'a-z', 'A-Z', '0-9',
10115 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010116
10117 <expr> Is a standard HAProxy expression formed by a sample-fetch
10118 followed by some converters.
10119
10120 Example:
10121
10122 tcp-request content set-var(sess.my_var) src
10123
Christopher Faulet85d79c92016-11-09 16:54:56 +010010124 The "unset-var" is used to unset a variable. See above for details about
10125 <var-name>.
10126
10127 Example:
10128
10129 tcp-request content unset-var(sess.my_var)
10130
Christopher Faulet76c09ef2017-09-21 11:03:52 +020010131 The "send-spoe-group" is used to trigger sending of a group of SPOE
10132 messages. To do so, the SPOE engine used to send messages must be defined, as
10133 well as the SPOE group to send. Of course, the SPOE engine must refer to an
10134 existing SPOE filter. If not engine name is provided on the SPOE filter line,
10135 the SPOE agent name must be used.
10136
10137 <engine-name> The SPOE engine name.
10138
10139 <group-name> The SPOE group name as specified in the engine configuration.
10140
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010141 See section 7 about ACL usage.
10142
10143 See also : "tcp-request content", "tcp-response inspect-delay"
10144
10145
Willy Tarreau4f614292016-10-21 17:49:36 +020010146tcp-request session <action> [{if | unless} <condition>]
10147 Perform an action on a validated session depending on a layer 5 condition
10148 May be used in sections : defaults | frontend | listen | backend
10149 no | yes | yes | no
10150 Arguments :
10151 <action> defines the action to perform if the condition applies. See
10152 below.
10153
10154 <condition> is a standard layer5-only ACL-based condition (see section 7).
10155
Davor Ocelice9ed2812017-12-25 17:49:28 +010010156 Once a session is validated, (i.e. after all handshakes have been completed),
Willy Tarreau4f614292016-10-21 17:49:36 +020010157 it is possible to evaluate some conditions to decide whether this session
10158 must be accepted or dropped or have its counters tracked. Those conditions
10159 cannot make use of any data contents because no buffers are allocated yet and
10160 the processing cannot wait at this stage. The main use case it to copy some
10161 early information into variables (since variables are accessible in the
10162 session), or to keep track of some information collected after the handshake,
10163 such as SSL-level elements (SNI, ciphers, client cert's CN) or information
Davor Ocelice9ed2812017-12-25 17:49:28 +010010164 from the PROXY protocol header (e.g. track a source forwarded this way). The
Willy Tarreau4f614292016-10-21 17:49:36 +020010165 extracted information can thus be copied to a variable or tracked using
10166 "track-sc" rules. Of course it is also possible to decide to accept/reject as
10167 with other rulesets. Most operations performed here could also be performed
10168 in "tcp-request content" rules, except that in HTTP these rules are evaluated
10169 for each new request, and that might not always be acceptable. For example a
10170 rule might increment a counter on each evaluation. It would also be possible
10171 that a country is resolved by geolocation from the source IP address,
10172 assigned to a session-wide variable, then the source address rewritten from
10173 an HTTP header for all requests. If some contents need to be inspected in
10174 order to take the decision, the "tcp-request content" statements must be used
10175 instead.
10176
10177 The "tcp-request session" rules are evaluated in their exact declaration
10178 order. If no rule matches or if there is no rule, the default action is to
10179 accept the incoming session. There is no specific limit to the number of
10180 rules which may be inserted.
10181
10182 Several types of actions are supported :
10183 - accept : the request is accepted
10184 - reject : the request is rejected and the connection is closed
10185 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
10186 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +010010187 - sc-inc-gpc1(<sc-id>)
Cédric Dufour0d7712d2019-11-06 18:38:53 +010010188 - sc-set-gpt0(<sc-id>) { <int> | <expr> }
Willy Tarreau4f614292016-10-21 17:49:36 +020010189 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +010010190 - unset-var(<var-name>)
Willy Tarreau4f614292016-10-21 17:49:36 +020010191 - silent-drop
10192
10193 These actions have the same meaning as their respective counter-parts in
10194 "tcp-request connection" and "tcp-request content", so please refer to these
10195 sections for a complete description.
10196
10197 Note that the "if/unless" condition is optional. If no condition is set on
10198 the action, it is simply performed unconditionally. That can be useful for
10199 "track-sc*" actions as well as for changing the default action to a reject.
10200
10201 Example: track the original source address by default, or the one advertised
10202 in the PROXY protocol header for connection coming from the local
10203 proxies. The first connection-level rule enables receipt of the
10204 PROXY protocol for these ones, the second rule tracks whatever
10205 address we decide to keep after optional decoding.
10206
10207 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
10208 tcp-request session track-sc0 src
10209
10210 Example: accept all sessions from white-listed hosts, reject too fast
10211 sessions without counting them, and track accepted sessions.
10212 This results in session rate being capped from abusive sources.
10213
10214 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
10215 tcp-request session reject if { src_sess_rate gt 10 }
10216 tcp-request session track-sc0 src
10217
10218 Example: accept all sessions from white-listed hosts, count all other
10219 sessions and reject too fast ones. This results in abusive ones
10220 being blocked as long as they don't slow down.
10221
10222 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
10223 tcp-request session track-sc0 src
10224 tcp-request session reject if { sc0_sess_rate gt 10 }
10225
10226 See section 7 about ACL usage.
10227
10228 See also : "tcp-request connection", "tcp-request content", "stick-table"
10229
10230
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010231tcp-response inspect-delay <timeout>
10232 Set the maximum allowed time to wait for a response during content inspection
10233 May be used in sections : defaults | frontend | listen | backend
10234 no | no | yes | yes
10235 Arguments :
10236 <timeout> is the timeout value specified in milliseconds by default, but
10237 can be in any other unit if the number is suffixed by the unit,
10238 as explained at the top of this document.
10239
10240 See also : "tcp-response content", "tcp-request inspect-delay".
10241
10242
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010243timeout check <timeout>
10244 Set additional check timeout, but only after a connection has been already
10245 established.
10246
10247 May be used in sections: defaults | frontend | listen | backend
10248 yes | no | yes | yes
10249 Arguments:
10250 <timeout> is the timeout value specified in milliseconds by default, but
10251 can be in any other unit if the number is suffixed by the unit,
10252 as explained at the top of this document.
10253
10254 If set, haproxy uses min("timeout connect", "inter") as a connect timeout
10255 for check and "timeout check" as an additional read timeout. The "min" is
Davor Ocelice9ed2812017-12-25 17:49:28 +010010256 used so that people running with *very* long "timeout connect" (e.g. those
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010257 who needed this due to the queue or tarpit) do not slow down their checks.
Willy Tarreaud7550a22010-02-10 05:10:19 +010010258 (Please also note that there is no valid reason to have such long connect
10259 timeouts, because "timeout queue" and "timeout tarpit" can always be used to
10260 avoid that).
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010261
10262 If "timeout check" is not set haproxy uses "inter" for complete check
10263 timeout (connect + read) exactly like all <1.3.15 version.
10264
10265 In most cases check request is much simpler and faster to handle than normal
10266 requests and people may want to kick out laggy servers so this timeout should
Willy Tarreau41a340d2008-01-22 12:25:31 +010010267 be smaller than "timeout server".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010268
10269 This parameter is specific to backends, but can be specified once for all in
10270 "defaults" sections. This is in fact one of the easiest solutions not to
10271 forget about it.
10272
Willy Tarreau41a340d2008-01-22 12:25:31 +010010273 See also: "timeout connect", "timeout queue", "timeout server",
10274 "timeout tarpit".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010275
10276
Willy Tarreau0ba27502007-12-24 16:55:16 +010010277timeout client <timeout>
Willy Tarreau0ba27502007-12-24 16:55:16 +010010278 Set the maximum inactivity time on the client side.
10279 May be used in sections : defaults | frontend | listen | backend
10280 yes | yes | yes | no
10281 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010282 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010010283 can be in any other unit if the number is suffixed by the unit,
10284 as explained at the top of this document.
10285
10286 The inactivity timeout applies when the client is expected to acknowledge or
10287 send data. In HTTP mode, this timeout is particularly important to consider
10288 during the first phase, when the client sends the request, and during the
Baptiste Assmann2e1941e2016-03-06 23:24:12 +010010289 response while it is reading data sent by the server. That said, for the
10290 first phase, it is preferable to set the "timeout http-request" to better
10291 protect HAProxy from Slowloris like attacks. The value is specified in
10292 milliseconds by default, but can be in any other unit if the number is
Willy Tarreau0ba27502007-12-24 16:55:16 +010010293 suffixed by the unit, as specified at the top of this document. In TCP mode
10294 (and to a lesser extent, in HTTP mode), it is highly recommended that the
10295 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010296 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +010010297 losses by specifying timeouts that are slightly above multiples of 3 seconds
Davor Ocelice9ed2812017-12-25 17:49:28 +010010298 (e.g. 4 or 5 seconds). If some long-lived sessions are mixed with short-lived
10299 sessions (e.g. WebSocket and HTTP), it's worth considering "timeout tunnel",
Willy Tarreau05cdd962014-05-10 14:30:07 +020010300 which overrides "timeout client" and "timeout server" for tunnels, as well as
10301 "timeout client-fin" for half-closed connections.
Willy Tarreau0ba27502007-12-24 16:55:16 +010010302
10303 This parameter is specific to frontends, but can be specified once for all in
10304 "defaults" sections. This is in fact one of the easiest solutions not to
10305 forget about it. An unspecified timeout results in an infinite timeout, which
10306 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050010307 during startup because it may result in accumulation of expired sessions in
Willy Tarreau0ba27502007-12-24 16:55:16 +010010308 the system if the system's timeouts are not configured either.
10309
Willy Tarreau95c4e142017-11-26 12:18:55 +010010310 This also applies to HTTP/2 connections, which will be closed with GOAWAY.
Lukas Tribus75df9d72017-11-24 19:05:12 +010010311
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020010312 See also : "timeout server", "timeout tunnel", "timeout http-request".
Willy Tarreau0ba27502007-12-24 16:55:16 +010010313
Willy Tarreau0ba27502007-12-24 16:55:16 +010010314
Willy Tarreau05cdd962014-05-10 14:30:07 +020010315timeout client-fin <timeout>
10316 Set the inactivity timeout on the client side for half-closed connections.
10317 May be used in sections : defaults | frontend | listen | backend
10318 yes | yes | yes | no
10319 Arguments :
10320 <timeout> is the timeout value specified in milliseconds by default, but
10321 can be in any other unit if the number is suffixed by the unit,
10322 as explained at the top of this document.
10323
10324 The inactivity timeout applies when the client is expected to acknowledge or
10325 send data while one direction is already shut down. This timeout is different
10326 from "timeout client" in that it only applies to connections which are closed
10327 in one direction. This is particularly useful to avoid keeping connections in
10328 FIN_WAIT state for too long when clients do not disconnect cleanly. This
10329 problem is particularly common long connections such as RDP or WebSocket.
10330 Note that this timeout can override "timeout tunnel" when a connection shuts
Willy Tarreau599391a2017-11-24 10:16:00 +010010331 down in one direction. It is applied to idle HTTP/2 connections once a GOAWAY
10332 frame was sent, often indicating an expectation that the connection quickly
10333 ends.
Willy Tarreau05cdd962014-05-10 14:30:07 +020010334
10335 This parameter is specific to frontends, but can be specified once for all in
10336 "defaults" sections. By default it is not set, so half-closed connections
10337 will use the other timeouts (timeout.client or timeout.tunnel).
10338
10339 See also : "timeout client", "timeout server-fin", and "timeout tunnel".
10340
10341
Willy Tarreau0ba27502007-12-24 16:55:16 +010010342timeout connect <timeout>
Willy Tarreau0ba27502007-12-24 16:55:16 +010010343 Set the maximum time to wait for a connection attempt to a server to succeed.
10344 May be used in sections : defaults | frontend | listen | backend
10345 yes | no | yes | yes
10346 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010347 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010010348 can be in any other unit if the number is suffixed by the unit,
10349 as explained at the top of this document.
10350
10351 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010352 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +010010353 cover one or several TCP packet losses by specifying timeouts that are
Davor Ocelice9ed2812017-12-25 17:49:28 +010010354 slightly above multiples of 3 seconds (e.g. 4 or 5 seconds). By default, the
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010355 connect timeout also presets both queue and tarpit timeouts to the same value
10356 if these have not been specified.
Willy Tarreau0ba27502007-12-24 16:55:16 +010010357
10358 This parameter is specific to backends, but can be specified once for all in
10359 "defaults" sections. This is in fact one of the easiest solutions not to
10360 forget about it. An unspecified timeout results in an infinite timeout, which
10361 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050010362 during startup because it may result in accumulation of failed sessions in
Willy Tarreau0ba27502007-12-24 16:55:16 +010010363 the system if the system's timeouts are not configured either.
10364
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020010365 See also: "timeout check", "timeout queue", "timeout server", "timeout tarpit".
Willy Tarreau0ba27502007-12-24 16:55:16 +010010366
Willy Tarreau0ba27502007-12-24 16:55:16 +010010367
Willy Tarreaub16a5742010-01-10 14:46:16 +010010368timeout http-keep-alive <timeout>
10369 Set the maximum allowed time to wait for a new HTTP request to appear
10370 May be used in sections : defaults | frontend | listen | backend
10371 yes | yes | yes | yes
10372 Arguments :
10373 <timeout> is the timeout value specified in milliseconds by default, but
10374 can be in any other unit if the number is suffixed by the unit,
10375 as explained at the top of this document.
10376
10377 By default, the time to wait for a new request in case of keep-alive is set
10378 by "timeout http-request". However this is not always convenient because some
10379 people want very short keep-alive timeouts in order to release connections
10380 faster, and others prefer to have larger ones but still have short timeouts
10381 once the request has started to present itself.
10382
10383 The "http-keep-alive" timeout covers these needs. It will define how long to
10384 wait for a new HTTP request to start coming after a response was sent. Once
10385 the first byte of request has been seen, the "http-request" timeout is used
10386 to wait for the complete request to come. Note that empty lines prior to a
10387 new request do not refresh the timeout and are not counted as a new request.
10388
10389 There is also another difference between the two timeouts : when a connection
10390 expires during timeout http-keep-alive, no error is returned, the connection
10391 just closes. If the connection expires in "http-request" while waiting for a
10392 connection to complete, a HTTP 408 error is returned.
10393
10394 In general it is optimal to set this value to a few tens to hundreds of
10395 milliseconds, to allow users to fetch all objects of a page at once but
Davor Ocelice9ed2812017-12-25 17:49:28 +010010396 without waiting for further clicks. Also, if set to a very small value (e.g.
Willy Tarreaub16a5742010-01-10 14:46:16 +010010397 1 millisecond) it will probably only accept pipelined requests but not the
10398 non-pipelined ones. It may be a nice trade-off for very large sites running
Patrick Mézard2382ad62010-05-09 10:43:32 +020010399 with tens to hundreds of thousands of clients.
Willy Tarreaub16a5742010-01-10 14:46:16 +010010400
10401 If this parameter is not set, the "http-request" timeout applies, and if both
10402 are not set, "timeout client" still applies at the lower level. It should be
10403 set in the frontend to take effect, unless the frontend is in TCP mode, in
10404 which case the HTTP backend's timeout will be used.
10405
Willy Tarreau95c4e142017-11-26 12:18:55 +010010406 When using HTTP/2 "timeout client" is applied instead. This is so we can keep
10407 using short keep-alive timeouts in HTTP/1.1 while using longer ones in HTTP/2
Lukas Tribus75df9d72017-11-24 19:05:12 +010010408 (where we only have one connection per client and a connection setup).
10409
Willy Tarreaub16a5742010-01-10 14:46:16 +010010410 See also : "timeout http-request", "timeout client".
10411
10412
Willy Tarreau036fae02008-01-06 13:24:40 +010010413timeout http-request <timeout>
10414 Set the maximum allowed time to wait for a complete HTTP request
10415 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaucd7afc02009-07-12 10:03:17 +020010416 yes | yes | yes | yes
Willy Tarreau036fae02008-01-06 13:24:40 +010010417 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010418 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau036fae02008-01-06 13:24:40 +010010419 can be in any other unit if the number is suffixed by the unit,
10420 as explained at the top of this document.
10421
10422 In order to offer DoS protection, it may be required to lower the maximum
10423 accepted time to receive a complete HTTP request without affecting the client
10424 timeout. This helps protecting against established connections on which
10425 nothing is sent. The client timeout cannot offer a good protection against
10426 this abuse because it is an inactivity timeout, which means that if the
10427 attacker sends one character every now and then, the timeout will not
10428 trigger. With the HTTP request timeout, no matter what speed the client
Willy Tarreau2705a612014-05-23 17:38:34 +020010429 types, the request will be aborted if it does not complete in time. When the
10430 timeout expires, an HTTP 408 response is sent to the client to inform it
10431 about the problem, and the connection is closed. The logs will report
10432 termination codes "cR". Some recent browsers are having problems with this
Davor Ocelice9ed2812017-12-25 17:49:28 +010010433 standard, well-documented behavior, so it might be needed to hide the 408
Willy Tarreau0f228a02015-05-01 15:37:53 +020010434 code using "option http-ignore-probes" or "errorfile 408 /dev/null". See
10435 more details in the explanations of the "cR" termination code in section 8.5.
Willy Tarreau036fae02008-01-06 13:24:40 +010010436
Baptiste Assmanneccdf432015-10-28 13:49:01 +010010437 By default, this timeout only applies to the header part of the request,
10438 and not to any data. As soon as the empty line is received, this timeout is
10439 not used anymore. When combined with "option http-buffer-request", this
10440 timeout also applies to the body of the request..
10441 It is used again on keep-alive connections to wait for a second
Willy Tarreaub16a5742010-01-10 14:46:16 +010010442 request if "timeout http-keep-alive" is not set.
Willy Tarreau036fae02008-01-06 13:24:40 +010010443
10444 Generally it is enough to set it to a few seconds, as most clients send the
10445 full request immediately upon connection. Add 3 or more seconds to cover TCP
Davor Ocelice9ed2812017-12-25 17:49:28 +010010446 retransmits but that's all. Setting it to very low values (e.g. 50 ms) will
Willy Tarreau036fae02008-01-06 13:24:40 +010010447 generally work on local networks as long as there are no packet losses. This
10448 will prevent people from sending bare HTTP requests using telnet.
10449
10450 If this parameter is not set, the client timeout still applies between each
Willy Tarreaucd7afc02009-07-12 10:03:17 +020010451 chunk of the incoming request. It should be set in the frontend to take
10452 effect, unless the frontend is in TCP mode, in which case the HTTP backend's
10453 timeout will be used.
Willy Tarreau036fae02008-01-06 13:24:40 +010010454
Willy Tarreau0f228a02015-05-01 15:37:53 +020010455 See also : "errorfile", "http-ignore-probes", "timeout http-keep-alive", and
Baptiste Assmanneccdf432015-10-28 13:49:01 +010010456 "timeout client", "option http-buffer-request".
Willy Tarreau036fae02008-01-06 13:24:40 +010010457
Willy Tarreau844e3c52008-01-11 16:28:18 +010010458
10459timeout queue <timeout>
10460 Set the maximum time to wait in the queue for a connection slot to be free
10461 May be used in sections : defaults | frontend | listen | backend
10462 yes | no | yes | yes
10463 Arguments :
10464 <timeout> is the timeout value specified in milliseconds by default, but
10465 can be in any other unit if the number is suffixed by the unit,
10466 as explained at the top of this document.
10467
10468 When a server's maxconn is reached, connections are left pending in a queue
10469 which may be server-specific or global to the backend. In order not to wait
10470 indefinitely, a timeout is applied to requests pending in the queue. If the
10471 timeout is reached, it is considered that the request will almost never be
10472 served, so it is dropped and a 503 error is returned to the client.
10473
10474 The "timeout queue" statement allows to fix the maximum time for a request to
10475 be left pending in a queue. If unspecified, the same value as the backend's
10476 connection timeout ("timeout connect") is used, for backwards compatibility
10477 with older versions with no "timeout queue" parameter.
10478
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020010479 See also : "timeout connect".
Willy Tarreau844e3c52008-01-11 16:28:18 +010010480
10481
10482timeout server <timeout>
Willy Tarreau844e3c52008-01-11 16:28:18 +010010483 Set the maximum inactivity time on the server side.
10484 May be used in sections : defaults | frontend | listen | backend
10485 yes | no | yes | yes
10486 Arguments :
10487 <timeout> is the timeout value specified in milliseconds by default, but
10488 can be in any other unit if the number is suffixed by the unit,
10489 as explained at the top of this document.
10490
10491 The inactivity timeout applies when the server is expected to acknowledge or
10492 send data. In HTTP mode, this timeout is particularly important to consider
10493 during the first phase of the server's response, when it has to send the
10494 headers, as it directly represents the server's processing time for the
10495 request. To find out what value to put there, it's often good to start with
10496 what would be considered as unacceptable response times, then check the logs
10497 to observe the response time distribution, and adjust the value accordingly.
10498
10499 The value is specified in milliseconds by default, but can be in any other
10500 unit if the number is suffixed by the unit, as specified at the top of this
10501 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
10502 recommended that the client timeout remains equal to the server timeout in
10503 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010504 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +010010505 packet losses by specifying timeouts that are slightly above multiples of 3
Davor Ocelice9ed2812017-12-25 17:49:28 +010010506 seconds (e.g. 4 or 5 seconds minimum). If some long-lived sessions are mixed
10507 with short-lived sessions (e.g. WebSocket and HTTP), it's worth considering
Willy Tarreauce887fd2012-05-12 12:50:00 +020010508 "timeout tunnel", which overrides "timeout client" and "timeout server" for
10509 tunnels.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010510
10511 This parameter is specific to backends, but can be specified once for all in
10512 "defaults" sections. This is in fact one of the easiest solutions not to
10513 forget about it. An unspecified timeout results in an infinite timeout, which
10514 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050010515 during startup because it may result in accumulation of expired sessions in
Willy Tarreau844e3c52008-01-11 16:28:18 +010010516 the system if the system's timeouts are not configured either.
10517
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020010518 See also : "timeout client" and "timeout tunnel".
Willy Tarreau844e3c52008-01-11 16:28:18 +010010519
Willy Tarreau05cdd962014-05-10 14:30:07 +020010520
10521timeout server-fin <timeout>
10522 Set the inactivity timeout on the server side for half-closed connections.
10523 May be used in sections : defaults | frontend | listen | backend
10524 yes | no | yes | yes
10525 Arguments :
10526 <timeout> is the timeout value specified in milliseconds by default, but
10527 can be in any other unit if the number is suffixed by the unit,
10528 as explained at the top of this document.
10529
10530 The inactivity timeout applies when the server is expected to acknowledge or
10531 send data while one direction is already shut down. This timeout is different
10532 from "timeout server" in that it only applies to connections which are closed
10533 in one direction. This is particularly useful to avoid keeping connections in
10534 FIN_WAIT state for too long when a remote server does not disconnect cleanly.
10535 This problem is particularly common long connections such as RDP or WebSocket.
10536 Note that this timeout can override "timeout tunnel" when a connection shuts
10537 down in one direction. This setting was provided for completeness, but in most
10538 situations, it should not be needed.
10539
10540 This parameter is specific to backends, but can be specified once for all in
10541 "defaults" sections. By default it is not set, so half-closed connections
10542 will use the other timeouts (timeout.server or timeout.tunnel).
10543
10544 See also : "timeout client-fin", "timeout server", and "timeout tunnel".
10545
Willy Tarreau844e3c52008-01-11 16:28:18 +010010546
10547timeout tarpit <timeout>
Cyril Bonté78caf842010-03-10 22:41:43 +010010548 Set the duration for which tarpitted connections will be maintained
Willy Tarreau844e3c52008-01-11 16:28:18 +010010549 May be used in sections : defaults | frontend | listen | backend
10550 yes | yes | yes | yes
10551 Arguments :
10552 <timeout> is the tarpit duration specified in milliseconds by default, but
10553 can be in any other unit if the number is suffixed by the unit,
10554 as explained at the top of this document.
10555
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020010556 When a connection is tarpitted using "http-request tarpit", it is maintained
10557 open with no activity for a certain amount of time, then closed. "timeout
10558 tarpit" defines how long it will be maintained open.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010559
10560 The value is specified in milliseconds by default, but can be in any other
10561 unit if the number is suffixed by the unit, as specified at the top of this
10562 document. If unspecified, the same value as the backend's connection timeout
10563 ("timeout connect") is used, for backwards compatibility with older versions
Cyril Bonté78caf842010-03-10 22:41:43 +010010564 with no "timeout tarpit" parameter.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010565
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020010566 See also : "timeout connect".
Willy Tarreau844e3c52008-01-11 16:28:18 +010010567
10568
Willy Tarreauce887fd2012-05-12 12:50:00 +020010569timeout tunnel <timeout>
10570 Set the maximum inactivity time on the client and server side for tunnels.
10571 May be used in sections : defaults | frontend | listen | backend
10572 yes | no | yes | yes
10573 Arguments :
10574 <timeout> is the timeout value specified in milliseconds by default, but
10575 can be in any other unit if the number is suffixed by the unit,
10576 as explained at the top of this document.
10577
Jamie Gloudonaaa21002012-08-25 00:18:33 -040010578 The tunnel timeout applies when a bidirectional connection is established
Willy Tarreauce887fd2012-05-12 12:50:00 +020010579 between a client and a server, and the connection remains inactive in both
10580 directions. This timeout supersedes both the client and server timeouts once
10581 the connection becomes a tunnel. In TCP, this timeout is used as soon as no
Davor Ocelice9ed2812017-12-25 17:49:28 +010010582 analyzer remains attached to either connection (e.g. tcp content rules are
10583 accepted). In HTTP, this timeout is used when a connection is upgraded (e.g.
Willy Tarreauce887fd2012-05-12 12:50:00 +020010584 when switching to the WebSocket protocol, or forwarding a CONNECT request
10585 to a proxy), or after the first response when no keepalive/close option is
10586 specified.
10587
Willy Tarreau05cdd962014-05-10 14:30:07 +020010588 Since this timeout is usually used in conjunction with long-lived connections,
10589 it usually is a good idea to also set "timeout client-fin" to handle the
10590 situation where a client suddenly disappears from the net and does not
10591 acknowledge a close, or sends a shutdown and does not acknowledge pending
10592 data anymore. This can happen in lossy networks where firewalls are present,
10593 and is detected by the presence of large amounts of sessions in a FIN_WAIT
10594 state.
10595
Willy Tarreauce887fd2012-05-12 12:50:00 +020010596 The value is specified in milliseconds by default, but can be in any other
10597 unit if the number is suffixed by the unit, as specified at the top of this
10598 document. Whatever the expected normal idle time, it is a good practice to
10599 cover at least one or several TCP packet losses by specifying timeouts that
Davor Ocelice9ed2812017-12-25 17:49:28 +010010600 are slightly above multiples of 3 seconds (e.g. 4 or 5 seconds minimum).
Willy Tarreauce887fd2012-05-12 12:50:00 +020010601
10602 This parameter is specific to backends, but can be specified once for all in
10603 "defaults" sections. This is in fact one of the easiest solutions not to
10604 forget about it.
10605
10606 Example :
10607 defaults http
10608 option http-server-close
10609 timeout connect 5s
10610 timeout client 30s
Willy Tarreau05cdd962014-05-10 14:30:07 +020010611 timeout client-fin 30s
Willy Tarreauce887fd2012-05-12 12:50:00 +020010612 timeout server 30s
10613 timeout tunnel 1h # timeout to use with WebSocket and CONNECT
10614
Willy Tarreau05cdd962014-05-10 14:30:07 +020010615 See also : "timeout client", "timeout client-fin", "timeout server".
Willy Tarreauce887fd2012-05-12 12:50:00 +020010616
10617
Willy Tarreau844e3c52008-01-11 16:28:18 +010010618transparent (deprecated)
10619 Enable client-side transparent proxying
10620 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +010010621 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +010010622 Arguments : none
10623
10624 This keyword was introduced in order to provide layer 7 persistence to layer
10625 3 load balancers. The idea is to use the OS's ability to redirect an incoming
10626 connection for a remote address to a local process (here HAProxy), and let
10627 this process know what address was initially requested. When this option is
10628 used, sessions without cookies will be forwarded to the original destination
10629 IP address of the incoming request (which should match that of another
10630 equipment), while requests with cookies will still be forwarded to the
10631 appropriate server.
10632
10633 The "transparent" keyword is deprecated, use "option transparent" instead.
10634
10635 Note that contrary to a common belief, this option does NOT make HAProxy
10636 present the client's IP to the server when establishing the connection.
10637
Willy Tarreau844e3c52008-01-11 16:28:18 +010010638 See also: "option transparent"
10639
William Lallemanda73203e2012-03-12 12:48:57 +010010640unique-id-format <string>
10641 Generate a unique ID for each request.
10642 May be used in sections : defaults | frontend | listen | backend
10643 yes | yes | yes | no
10644 Arguments :
10645 <string> is a log-format string.
10646
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010647 This keyword creates a ID for each request using the custom log format. A
10648 unique ID is useful to trace a request passing through many components of
10649 a complex infrastructure. The newly created ID may also be logged using the
10650 %ID tag the log-format string.
William Lallemanda73203e2012-03-12 12:48:57 +010010651
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010652 The format should be composed from elements that are guaranteed to be
10653 unique when combined together. For instance, if multiple haproxy instances
10654 are involved, it might be important to include the node name. It is often
10655 needed to log the incoming connection's source and destination addresses
10656 and ports. Note that since multiple requests may be performed over the same
10657 connection, including a request counter may help differentiate them.
10658 Similarly, a timestamp may protect against a rollover of the counter.
10659 Logging the process ID will avoid collisions after a service restart.
William Lallemanda73203e2012-03-12 12:48:57 +010010660
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010661 It is recommended to use hexadecimal notation for many fields since it
10662 makes them more compact and saves space in logs.
William Lallemanda73203e2012-03-12 12:48:57 +010010663
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010664 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010010665
Julien Vehentf21be322014-03-07 08:27:34 -050010666 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010010667
10668 will generate:
10669
10670 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
10671
10672 See also: "unique-id-header"
10673
10674unique-id-header <name>
10675 Add a unique ID header in the HTTP request.
10676 May be used in sections : defaults | frontend | listen | backend
10677 yes | yes | yes | no
10678 Arguments :
10679 <name> is the name of the header.
10680
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010681 Add a unique-id header in the HTTP request sent to the server, using the
10682 unique-id-format. It can't work if the unique-id-format doesn't exist.
William Lallemanda73203e2012-03-12 12:48:57 +010010683
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010684 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010010685
Julien Vehentf21be322014-03-07 08:27:34 -050010686 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010010687 unique-id-header X-Unique-ID
10688
10689 will generate:
10690
10691 X-Unique-ID: 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
10692
10693 See also: "unique-id-format"
Willy Tarreau844e3c52008-01-11 16:28:18 +010010694
Willy Tarreauf51658d2014-04-23 01:21:56 +020010695use_backend <backend> [{if | unless} <condition>]
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020010696 Switch to a specific backend if/unless an ACL-based condition is matched.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010697 May be used in sections : defaults | frontend | listen | backend
10698 no | yes | yes | no
10699 Arguments :
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010010700 <backend> is the name of a valid backend or "listen" section, or a
10701 "log-format" string resolving to a backend name.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010702
Willy Tarreauf51658d2014-04-23 01:21:56 +020010703 <condition> is a condition composed of ACLs, as described in section 7. If
10704 it is omitted, the rule is unconditionally applied.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010705
10706 When doing content-switching, connections arrive on a frontend and are then
10707 dispatched to various backends depending on a number of conditions. The
10708 relation between the conditions and the backends is described with the
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020010709 "use_backend" keyword. While it is normally used with HTTP processing, it can
Davor Ocelice9ed2812017-12-25 17:49:28 +010010710 also be used in pure TCP, either without content using stateless ACLs (e.g.
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020010711 source address validation) or combined with a "tcp-request" rule to wait for
10712 some payload.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010713
10714 There may be as many "use_backend" rules as desired. All of these rules are
10715 evaluated in their declaration order, and the first one which matches will
10716 assign the backend.
10717
10718 In the first form, the backend will be used if the condition is met. In the
10719 second form, the backend will be used if the condition is not met. If no
10720 condition is valid, the backend defined with "default_backend" will be used.
10721 If no default backend is defined, either the servers in the same section are
10722 used (in case of a "listen" section) or, in case of a frontend, no server is
10723 used and a 503 service unavailable response is returned.
10724
Willy Tarreau51aecc72009-07-12 09:47:04 +020010725 Note that it is possible to switch from a TCP frontend to an HTTP backend. In
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010010726 this case, either the frontend has already checked that the protocol is HTTP,
Willy Tarreau51aecc72009-07-12 09:47:04 +020010727 and backend processing will immediately follow, or the backend will wait for
10728 a complete HTTP request to get in. This feature is useful when a frontend
10729 must decode several protocols on a unique port, one of them being HTTP.
10730
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010010731 When <backend> is a simple name, it is resolved at configuration time, and an
10732 error is reported if the specified backend does not exist. If <backend> is
10733 a log-format string instead, no check may be done at configuration time, so
10734 the backend name is resolved dynamically at run time. If the resulting
10735 backend name does not correspond to any valid backend, no other rule is
10736 evaluated, and the default_backend directive is applied instead. Note that
10737 when using dynamic backend names, it is highly recommended to use a prefix
10738 that no other backend uses in order to ensure that an unauthorized backend
10739 cannot be forced from the request.
10740
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010741 It is worth mentioning that "use_backend" rules with an explicit name are
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010010742 used to detect the association between frontends and backends to compute the
10743 backend's "fullconn" setting. This cannot be done for dynamic names.
10744
10745 See also: "default_backend", "tcp-request", "fullconn", "log-format", and
10746 section 7 about ACLs.
Willy Tarreaud72758d2010-01-12 10:42:19 +010010747
Christopher Fauletb30b3102019-09-12 23:03:09 +020010748use-fcgi-app <name>
10749 Defines the FastCGI application to use for the backend.
10750 May be used in sections : defaults | frontend | listen | backend
10751 no | no | yes | yes
10752 Arguments :
10753 <name> is the name of the FastCGI application to use.
10754
10755 See section 10.1 about FastCGI application setup for details.
Willy Tarreau036fae02008-01-06 13:24:40 +010010756
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010757use-server <server> if <condition>
10758use-server <server> unless <condition>
10759 Only use a specific server if/unless an ACL-based condition is matched.
10760 May be used in sections : defaults | frontend | listen | backend
10761 no | no | yes | yes
10762 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010763 <server> is the name of a valid server in the same backend section.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010764
10765 <condition> is a condition composed of ACLs, as described in section 7.
10766
10767 By default, connections which arrive to a backend are load-balanced across
10768 the available servers according to the configured algorithm, unless a
10769 persistence mechanism such as a cookie is used and found in the request.
10770
10771 Sometimes it is desirable to forward a particular request to a specific
10772 server without having to declare a dedicated backend for this server. This
10773 can be achieved using the "use-server" rules. These rules are evaluated after
10774 the "redirect" rules and before evaluating cookies, and they have precedence
10775 on them. There may be as many "use-server" rules as desired. All of these
10776 rules are evaluated in their declaration order, and the first one which
10777 matches will assign the server.
10778
10779 If a rule designates a server which is down, and "option persist" is not used
10780 and no force-persist rule was validated, it is ignored and evaluation goes on
10781 with the next rules until one matches.
10782
10783 In the first form, the server will be used if the condition is met. In the
10784 second form, the server will be used if the condition is not met. If no
10785 condition is valid, the processing continues and the server will be assigned
10786 according to other persistence mechanisms.
10787
10788 Note that even if a rule is matched, cookie processing is still performed but
10789 does not assign the server. This allows prefixed cookies to have their prefix
10790 stripped.
10791
10792 The "use-server" statement works both in HTTP and TCP mode. This makes it
10793 suitable for use with content-based inspection. For instance, a server could
10794 be selected in a farm according to the TLS SNI field. And if these servers
10795 have their weight set to zero, they will not be used for other traffic.
10796
10797 Example :
10798 # intercept incoming TLS requests based on the SNI field
10799 use-server www if { req_ssl_sni -i www.example.com }
10800 server www 192.168.0.1:443 weight 0
10801 use-server mail if { req_ssl_sni -i mail.example.com }
10802 server mail 192.168.0.1:587 weight 0
10803 use-server imap if { req_ssl_sni -i imap.example.com }
Lukas Tribus98a3e3f2017-03-26 12:55:35 +000010804 server imap 192.168.0.1:993 weight 0
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010805 # all the rest is forwarded to this server
10806 server default 192.168.0.2:443 check
10807
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010808 See also: "use_backend", section 5 about server and section 7 about ACLs.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010809
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010810
Davor Ocelice9ed2812017-12-25 17:49:28 +0100108115. Bind and server options
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010812--------------------------
10813
10814The "bind", "server" and "default-server" keywords support a number of settings
10815depending on some build options and on the system HAProxy was built on. These
10816settings generally each consist in one word sometimes followed by a value,
10817written on the same line as the "bind" or "server" line. All these options are
10818described in this section.
10819
10820
108215.1. Bind options
10822-----------------
10823
10824The "bind" keyword supports a certain number of settings which are all passed
10825as arguments on the same line. The order in which those arguments appear makes
10826no importance, provided that they appear after the bind address. All of these
10827parameters are optional. Some of them consist in a single words (booleans),
10828while other ones expect a value after them. In this case, the value must be
10829provided immediately after the setting name.
10830
10831The currently supported settings are the following ones.
10832
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010010833accept-netscaler-cip <magic number>
10834 Enforces the use of the NetScaler Client IP insertion protocol over any
10835 connection accepted by any of the TCP sockets declared on the same line. The
10836 NetScaler Client IP insertion protocol dictates the layer 3/4 addresses of
10837 the incoming connection to be used everywhere an address is used, with the
10838 only exception of "tcp-request connection" rules which will only see the
10839 real connection address. Logs will reflect the addresses indicated in the
10840 protocol, unless it is violated, in which case the real address will still
10841 be used. This keyword combined with support from external components can be
10842 used as an efficient and reliable alternative to the X-Forwarded-For
Bertrand Jacquin90759682016-06-06 15:35:39 +010010843 mechanism which is not always reliable and not even always usable. See also
10844 "tcp-request connection expect-netscaler-cip" for a finer-grained setting of
10845 which client is allowed to use the protocol.
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010010846
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010847accept-proxy
10848 Enforces the use of the PROXY protocol over any connection accepted by any of
Willy Tarreau77992672014-06-14 11:06:17 +020010849 the sockets declared on the same line. Versions 1 and 2 of the PROXY protocol
10850 are supported and correctly detected. The PROXY protocol dictates the layer
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010851 3/4 addresses of the incoming connection to be used everywhere an address is
10852 used, with the only exception of "tcp-request connection" rules which will
10853 only see the real connection address. Logs will reflect the addresses
10854 indicated in the protocol, unless it is violated, in which case the real
Davor Ocelice9ed2812017-12-25 17:49:28 +010010855 address will still be used. This keyword combined with support from external
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010856 components can be used as an efficient and reliable alternative to the
10857 X-Forwarded-For mechanism which is not always reliable and not even always
Willy Tarreau4f0d9192013-06-11 20:40:55 +020010858 usable. See also "tcp-request connection expect-proxy" for a finer-grained
10859 setting of which client is allowed to use the protocol.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010860
Olivier Houchardc2aae742017-09-22 18:26:28 +020010861allow-0rtt
Bertrand Jacquina25282b2018-08-14 00:56:13 +010010862 Allow receiving early data when using TLSv1.3. This is disabled by default,
Olivier Houchard69752962019-01-08 15:35:32 +010010863 due to security considerations. Because it is vulnerable to replay attacks,
John Roeslerfb2fce12019-07-10 15:45:51 -050010864 you should only allow if for requests that are safe to replay, i.e. requests
Olivier Houchard69752962019-01-08 15:35:32 +010010865 that are idempotent. You can use the "wait-for-handshake" action for any
10866 request that wouldn't be safe with early data.
Olivier Houchardc2aae742017-09-22 18:26:28 +020010867
Willy Tarreauab861d32013-04-02 02:30:41 +020010868alpn <protocols>
10869 This enables the TLS ALPN extension and advertises the specified protocol
10870 list as supported on top of ALPN. The protocol list consists in a comma-
10871 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
John Roeslerfb2fce12019-07-10 15:45:51 -050010872 quotes). This requires that the SSL library is built with support for TLS
Willy Tarreauab861d32013-04-02 02:30:41 +020010873 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
Willy Tarreau95c4e142017-11-26 12:18:55 +010010874 initial NPN extension. ALPN is required to enable HTTP/2 on an HTTP frontend.
10875 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
10876 now obsolete NPN extension. At the time of writing this, most browsers still
10877 support both ALPN and NPN for HTTP/2 so a fallback to NPN may still work for
10878 a while. But ALPN must be used whenever possible. If both HTTP/2 and HTTP/1.1
10879 are expected to be supported, both versions can be advertised, in order of
10880 preference, like below :
10881
10882 bind :443 ssl crt pub.pem alpn h2,http/1.1
Willy Tarreauab861d32013-04-02 02:30:41 +020010883
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010884backlog <backlog>
Willy Tarreaue2711c72019-02-27 15:39:41 +010010885 Sets the socket's backlog to this value. If unspecified or 0, the frontend's
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010886 backlog is used instead, which generally defaults to the maxconn value.
10887
Emmanuel Hocdete7f2b732017-01-09 16:15:54 +010010888curves <curves>
10889 This setting is only available when support for OpenSSL was built in. It sets
10890 the string describing the list of elliptic curves algorithms ("curve suite")
10891 that are negotiated during the SSL/TLS handshake with ECDHE. The format of the
10892 string is a colon-delimited list of curve name.
10893 Example: "X25519:P-256" (without quote)
10894 When "curves" is set, "ecdhe" parameter is ignored.
10895
Emeric Brun7fb34422012-09-28 15:26:15 +020010896ecdhe <named curve>
10897 This setting is only available when support for OpenSSL was built in. It sets
Emeric Brun6924ef82013-03-06 14:08:53 +010010898 the named curve (RFC 4492) used to generate ECDH ephemeral keys. By default,
10899 used named curve is prime256v1.
Emeric Brun7fb34422012-09-28 15:26:15 +020010900
Emeric Brunfd33a262012-10-11 16:28:27 +020010901ca-file <cafile>
Emeric Brun1a073b42012-09-28 17:07:34 +020010902 This setting is only available when support for OpenSSL was built in. It
10903 designates a PEM file from which to load CA certificates used to verify
10904 client's certificate.
10905
Emeric Brunb6dc9342012-09-28 17:55:37 +020010906ca-ignore-err [all|<errorID>,...]
10907 This setting is only available when support for OpenSSL was built in.
10908 Sets a comma separated list of errorIDs to ignore during verify at depth > 0.
10909 If set to 'all', all errors are ignored. SSL handshake is not aborted if an
10910 error is ignored.
10911
Christopher Faulet31af49d2015-06-09 17:29:50 +020010912ca-sign-file <cafile>
10913 This setting is only available when support for OpenSSL was built in. It
10914 designates a PEM file containing both the CA certificate and the CA private
10915 key used to create and sign server's certificates. This is a mandatory
10916 setting when the dynamic generation of certificates is enabled. See
10917 'generate-certificates' for details.
10918
Bertrand Jacquind4d0a232016-11-13 16:37:12 +000010919ca-sign-pass <passphrase>
Christopher Faulet31af49d2015-06-09 17:29:50 +020010920 This setting is only available when support for OpenSSL was built in. It is
10921 the CA private key passphrase. This setting is optional and used only when
10922 the dynamic generation of certificates is enabled. See
10923 'generate-certificates' for details.
10924
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010925ciphers <ciphers>
10926 This setting is only available when support for OpenSSL was built in. It sets
10927 the string describing the list of cipher algorithms ("cipher suite") that are
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +000010928 negotiated during the SSL/TLS handshake up to TLSv1.2. The format of the
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000010929 string is defined in "man 1 ciphers" from OpenSSL man pages. For background
Dirkjan Bussink415150f2018-09-14 11:14:21 +020010930 information and recommendations see e.g.
10931 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
10932 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
10933 cipher configuration, please check the "ciphersuites" keyword.
10934
10935ciphersuites <ciphersuites>
10936 This setting is only available when support for OpenSSL was built in and
10937 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the string describing
10938 the list of cipher algorithms ("cipher suite") that are negotiated during the
10939 TLSv1.3 handshake. The format of the string is defined in "man 1 ciphers" from
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000010940 OpenSSL man pages under the "ciphersuites" section. For cipher configuration
10941 for TLSv1.2 and earlier, please check the "ciphers" keyword.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010942
Emeric Brunfd33a262012-10-11 16:28:27 +020010943crl-file <crlfile>
Emeric Brun1a073b42012-09-28 17:07:34 +020010944 This setting is only available when support for OpenSSL was built in. It
10945 designates a PEM file from which to load certificate revocation list used
10946 to verify client's certificate.
10947
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010948crt <cert>
Alex Davies0fbf0162013-03-02 16:04:50 +000010949 This setting is only available when support for OpenSSL was built in. It
10950 designates a PEM file containing both the required certificates and any
10951 associated private keys. This file can be built by concatenating multiple
10952 PEM files into one (e.g. cat cert.pem key.pem > combined.pem). If your CA
10953 requires an intermediate certificate, this can also be concatenated into this
10954 file.
10955
10956 If the OpenSSL used supports Diffie-Hellman, parameters present in this file
10957 are loaded.
10958
10959 If a directory name is used instead of a PEM file, then all files found in
Cyril Bonté3180f7b2015-01-25 00:16:08 +010010960 that directory will be loaded in alphabetic order unless their name ends with
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010010961 '.issuer', '.ocsp' or '.sctl' (reserved extensions). This directive may be
10962 specified multiple times in order to load certificates from multiple files or
10963 directories. The certificates will be presented to clients who provide a
10964 valid TLS Server Name Indication field matching one of their CN or alt
Davor Ocelice9ed2812017-12-25 17:49:28 +010010965 subjects. Wildcards are supported, where a wildcard character '*' is used
10966 instead of the first hostname component (e.g. *.example.org matches
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010010967 www.example.org but not www.sub.example.org).
Alex Davies0fbf0162013-03-02 16:04:50 +000010968
10969 If no SNI is provided by the client or if the SSL library does not support
10970 TLS extensions, or if the client provides an SNI hostname which does not
10971 match any certificate, then the first loaded certificate will be presented.
10972 This means that when loading certificates from a directory, it is highly
Cyril Bonté3180f7b2015-01-25 00:16:08 +010010973 recommended to load the default one first as a file or to ensure that it will
10974 always be the first one in the directory.
Alex Davies0fbf0162013-03-02 16:04:50 +000010975
Emeric Brune032bfa2012-09-28 13:01:45 +020010976 Note that the same cert may be loaded multiple times without side effects.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010977
Davor Ocelice9ed2812017-12-25 17:49:28 +010010978 Some CAs (such as GoDaddy) offer a drop down list of server types that do not
Alex Davies0fbf0162013-03-02 16:04:50 +000010979 include HAProxy when obtaining a certificate. If this happens be sure to
Davor Ocelice9ed2812017-12-25 17:49:28 +010010980 choose a web server that the CA believes requires an intermediate CA (for
10981 GoDaddy, selection Apache Tomcat will get the correct bundle, but many
Alex Davies0fbf0162013-03-02 16:04:50 +000010982 others, e.g. nginx, result in a wrong bundle that will not work for some
10983 clients).
10984
Emeric Brun4147b2e2014-06-16 18:36:30 +020010985 For each PEM file, haproxy checks for the presence of file at the same path
10986 suffixed by ".ocsp". If such file is found, support for the TLS Certificate
10987 Status Request extension (also known as "OCSP stapling") is automatically
10988 enabled. The content of this file is optional. If not empty, it must contain
10989 a valid OCSP Response in DER format. In order to be valid an OCSP Response
10990 must comply with the following rules: it has to indicate a good status,
10991 it has to be a single response for the certificate of the PEM file, and it
10992 has to be valid at the moment of addition. If these rules are not respected
10993 the OCSP Response is ignored and a warning is emitted. In order to identify
10994 which certificate an OCSP Response applies to, the issuer's certificate is
10995 necessary. If the issuer's certificate is not found in the PEM file, it will
10996 be loaded from a file at the same path as the PEM file suffixed by ".issuer"
10997 if it exists otherwise it will fail with an error.
10998
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010010999 For each PEM file, haproxy also checks for the presence of file at the same
11000 path suffixed by ".sctl". If such file is found, support for Certificate
11001 Transparency (RFC6962) TLS extension is enabled. The file must contain a
11002 valid Signed Certificate Timestamp List, as described in RFC. File is parsed
11003 to check basic syntax, but no signatures are verified.
11004
yanbzhu6c25e9e2016-01-05 12:52:02 -050011005 There are cases where it is desirable to support multiple key types, e.g. RSA
11006 and ECDSA in the cipher suites offered to the clients. This allows clients
11007 that support EC certificates to be able to use EC ciphers, while
11008 simultaneously supporting older, RSA only clients.
yanbzhud19630c2015-12-14 15:10:25 -050011009
11010 In order to provide this functionality, multiple PEM files, each with a
11011 different key type, are required. To associate these PEM files into a
11012 "cert bundle" that is recognized by haproxy, they must be named in the
11013 following way: All PEM files that are to be bundled must have the same base
11014 name, with a suffix indicating the key type. Currently, three suffixes are
11015 supported: rsa, dsa and ecdsa. For example, if www.example.com has two PEM
11016 files, an RSA file and an ECDSA file, they must be named: "example.pem.rsa"
11017 and "example.pem.ecdsa". The first part of the filename is arbitrary; only the
11018 suffix matters. To load this bundle into haproxy, specify the base name only:
11019
11020 Example : bind :8443 ssl crt example.pem
11021
yanbzhu6c25e9e2016-01-05 12:52:02 -050011022 Note that the suffix is not given to haproxy; this tells haproxy to look for
yanbzhud19630c2015-12-14 15:10:25 -050011023 a cert bundle.
11024
Davor Ocelice9ed2812017-12-25 17:49:28 +010011025 HAProxy will load all PEM files in the bundle at the same time to try to
yanbzhud19630c2015-12-14 15:10:25 -050011026 support multiple key types. PEM files are combined based on Common Name
11027 (CN) and Subject Alternative Name (SAN) to support SNI lookups. This means
11028 that even if you give haproxy a cert bundle, if there are no shared CN/SAN
11029 entries in the certificates in that bundle, haproxy will not be able to
11030 provide multi-cert support.
11031
11032 Assuming bundle in the example above contained the following:
11033
11034 Filename | CN | SAN
11035 -------------------+-----------------+-------------------
11036 example.pem.rsa | www.example.com | rsa.example.com
yanbzhu6c25e9e2016-01-05 12:52:02 -050011037 -------------------+-----------------+-------------------
yanbzhud19630c2015-12-14 15:10:25 -050011038 example.pem.ecdsa | www.example.com | ecdsa.example.com
11039 -------------------+-----------------+-------------------
11040
11041 Users connecting with an SNI of "www.example.com" will be able
11042 to use both RSA and ECDSA cipher suites. Users connecting with an SNI of
11043 "rsa.example.com" will only be able to use RSA cipher suites, and users
11044 connecting with "ecdsa.example.com" will only be able to use ECDSA cipher
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020011045 suites. With BoringSSL and Openssl >= 1.1.1 multi-cert is natively supported,
11046 no need to bundle certificates. ECDSA certificate will be preferred if client
11047 support it.
yanbzhud19630c2015-12-14 15:10:25 -050011048
11049 If a directory name is given as the <cert> argument, haproxy will
11050 automatically search and load bundled files in that directory.
11051
11052 OSCP files (.ocsp) and issuer files (.issuer) are supported with multi-cert
11053 bundling. Each certificate can have its own .ocsp and .issuer file. At this
11054 time, sctl is not supported in multi-certificate bundling.
11055
Emeric Brunb6dc9342012-09-28 17:55:37 +020011056crt-ignore-err <errors>
Alex Davies0fbf0162013-03-02 16:04:50 +000011057 This setting is only available when support for OpenSSL was built in. Sets a
Davor Ocelice9ed2812017-12-25 17:49:28 +010011058 comma separated list of errorIDs to ignore during verify at depth == 0. If
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011059 set to 'all', all errors are ignored. SSL handshake is not aborted if an error
Alex Davies0fbf0162013-03-02 16:04:50 +000011060 is ignored.
Emeric Brunb6dc9342012-09-28 17:55:37 +020011061
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010011062crt-list <file>
11063 This setting is only available when support for OpenSSL was built in. It
Emmanuel Hocdet98263292016-12-29 18:26:15 +010011064 designates a list of PEM file with an optional ssl configuration and a SNI
11065 filter per certificate, with the following format for each line :
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010011066
Emmanuel Hocdet98263292016-12-29 18:26:15 +010011067 <crtfile> [\[<sslbindconf> ...\]] [[!]<snifilter> ...]
11068
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020011069 sslbindconf support "npn", "alpn", "verify", "ca-file", "no-ca-names",
11070 crl-file", "ecdhe", "curves", "ciphers" configuration. With BoringSSL
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020011071 and Openssl >= 1.1.1 "ssl-min-ver" and "ssl-max-ver" are also supported.
Emmanuel Hocdet98263292016-12-29 18:26:15 +010011072 It override the configuration set in bind line for the certificate.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010011073
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +020011074 Wildcards are supported in the SNI filter. Negative filter are also supported,
11075 only useful in combination with a wildcard filter to exclude a particular SNI.
11076 The certificates will be presented to clients who provide a valid TLS Server
11077 Name Indication field matching one of the SNI filters. If no SNI filter is
11078 specified, the CN and alt subjects are used. This directive may be specified
11079 multiple times. See the "crt" option for more information. The default
11080 certificate is still needed to meet OpenSSL expectations. If it is not used,
11081 the 'strict-sni' option may be used.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010011082
yanbzhu6c25e9e2016-01-05 12:52:02 -050011083 Multi-cert bundling (see "crt") is supported with crt-list, as long as only
Emmanuel Hocdetd294aea2016-05-13 11:14:06 +020011084 the base name is given in the crt-list. SNI filter will do the same work on
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020011085 all bundled certificates. With BoringSSL and Openssl >= 1.1.1 multi-cert is
11086 natively supported, avoid multi-cert bundling. RSA and ECDSA certificates can
11087 be declared in a row, and set different ssl and filter parameter.
yanbzhud19630c2015-12-14 15:10:25 -050011088
Emmanuel Hocdet98263292016-12-29 18:26:15 +010011089 crt-list file example:
11090 cert1.pem
Emmanuel Hocdet05942112017-02-20 16:11:50 +010011091 cert2.pem [alpn h2,http/1.1]
Emmanuel Hocdet98263292016-12-29 18:26:15 +010011092 certW.pem *.domain.tld !secure.domain.tld
Emmanuel Hocdet05942112017-02-20 16:11:50 +010011093 certS.pem [curves X25519:P-256 ciphers ECDHE-ECDSA-AES256-GCM-SHA384] secure.domain.tld
Emmanuel Hocdet98263292016-12-29 18:26:15 +010011094
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011095defer-accept
11096 Is an optional keyword which is supported only on certain Linux kernels. It
11097 states that a connection will only be accepted once some data arrive on it,
11098 or at worst after the first retransmit. This should be used only on protocols
Davor Ocelice9ed2812017-12-25 17:49:28 +010011099 for which the client talks first (e.g. HTTP). It can slightly improve
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011100 performance by ensuring that most of the request is already available when
11101 the connection is accepted. On the other hand, it will not be able to detect
11102 connections which don't talk. It is important to note that this option is
11103 broken in all kernels up to 2.6.31, as the connection is never accepted until
11104 the client talks. This can cause issues with front firewalls which would see
11105 an established connection while the proxy will only see it in SYN_RECV. This
11106 option is only supported on TCPv4/TCPv6 sockets and ignored by other ones.
11107
William Lallemandf6975e92017-05-26 17:42:10 +020011108expose-fd listeners
11109 This option is only usable with the stats socket. It gives your stats socket
11110 the capability to pass listeners FD to another HAProxy process.
William Lallemande202b1e2017-06-01 17:38:56 +020011111 During a reload with the master-worker mode, the process is automatically
11112 reexecuted adding -x and one of the stats socket with this option.
Davor Ocelice9ed2812017-12-25 17:49:28 +010011113 See also "-x" in the management guide.
William Lallemandf6975e92017-05-26 17:42:10 +020011114
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011115force-sslv3
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011116 This option enforces use of SSLv3 only on SSL connections instantiated from
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011117 this listener. SSLv3 is generally less expensive than the TLS counterparts
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011118 for high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011119 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011120
11121force-tlsv10
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011122 This option enforces use of TLSv1.0 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011123 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011124 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011125
11126force-tlsv11
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011127 This option enforces use of TLSv1.1 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011128 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011129 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011130
11131force-tlsv12
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011132 This option enforces use of TLSv1.2 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011133 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011134 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011135
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011136force-tlsv13
11137 This option enforces use of TLSv1.3 only on SSL connections instantiated from
11138 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011139 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011140
Christopher Faulet31af49d2015-06-09 17:29:50 +020011141generate-certificates
11142 This setting is only available when support for OpenSSL was built in. It
11143 enables the dynamic SSL certificates generation. A CA certificate and its
11144 private key are necessary (see 'ca-sign-file'). When HAProxy is configured as
11145 a transparent forward proxy, SSL requests generate errors because of a common
11146 name mismatch on the certificate presented to the client. With this option
11147 enabled, HAProxy will try to forge a certificate using the SNI hostname
11148 indicated by the client. This is done only if no certificate matches the SNI
11149 hostname (see 'crt-list'). If an error occurs, the default certificate is
11150 used, else the 'strict-sni' option is set.
11151 It can also be used when HAProxy is configured as a reverse proxy to ease the
11152 deployment of an architecture with many backends.
11153
11154 Creating a SSL certificate is an expensive operation, so a LRU cache is used
11155 to store forged certificates (see 'tune.ssl.ssl-ctx-cache-size'). It
Davor Ocelice9ed2812017-12-25 17:49:28 +010011156 increases the HAProxy's memory footprint to reduce latency when the same
Christopher Faulet31af49d2015-06-09 17:29:50 +020011157 certificate is used many times.
11158
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011159gid <gid>
11160 Sets the group of the UNIX sockets to the designated system gid. It can also
11161 be set by default in the global section's "unix-bind" statement. Note that
11162 some platforms simply ignore this. This setting is equivalent to the "group"
11163 setting except that the group ID is used instead of its name. This setting is
11164 ignored by non UNIX sockets.
11165
11166group <group>
11167 Sets the group of the UNIX sockets to the designated system group. It can
11168 also be set by default in the global section's "unix-bind" statement. Note
11169 that some platforms simply ignore this. This setting is equivalent to the
11170 "gid" setting except that the group name is used instead of its gid. This
11171 setting is ignored by non UNIX sockets.
11172
11173id <id>
11174 Fixes the socket ID. By default, socket IDs are automatically assigned, but
11175 sometimes it is more convenient to fix them to ease monitoring. This value
11176 must be strictly positive and unique within the listener/frontend. This
11177 option can only be used when defining only a single socket.
11178
11179interface <interface>
Lukas Tribusfce2e962013-02-12 22:13:19 +010011180 Restricts the socket to a specific interface. When specified, only packets
11181 received from that particular interface are processed by the socket. This is
11182 currently only supported on Linux. The interface must be a primary system
11183 interface, not an aliased interface. It is also possible to bind multiple
11184 frontends to the same address if they are bound to different interfaces. Note
11185 that binding to a network interface requires root privileges. This parameter
Jérôme Magnin61275192018-02-07 11:39:58 +010011186 is only compatible with TCPv4/TCPv6 sockets. When specified, return traffic
11187 uses the same interface as inbound traffic, and its associated routing table,
11188 even if there are explicit routes through different interfaces configured.
11189 This can prove useful to address asymmetric routing issues when the same
11190 client IP addresses need to be able to reach frontends hosted on different
11191 interfaces.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011192
Willy Tarreauabb175f2012-09-24 12:43:26 +020011193level <level>
11194 This setting is used with the stats sockets only to restrict the nature of
11195 the commands that can be issued on the socket. It is ignored by other
11196 sockets. <level> can be one of :
Davor Ocelice9ed2812017-12-25 17:49:28 +010011197 - "user" is the least privileged level; only non-sensitive stats can be
Willy Tarreauabb175f2012-09-24 12:43:26 +020011198 read, and no change is allowed. It would make sense on systems where it
11199 is not easy to restrict access to the socket.
11200 - "operator" is the default level and fits most common uses. All data can
Davor Ocelice9ed2812017-12-25 17:49:28 +010011201 be read, and only non-sensitive changes are permitted (e.g. clear max
Willy Tarreauabb175f2012-09-24 12:43:26 +020011202 counters).
Davor Ocelice9ed2812017-12-25 17:49:28 +010011203 - "admin" should be used with care, as everything is permitted (e.g. clear
Willy Tarreauabb175f2012-09-24 12:43:26 +020011204 all counters).
11205
Andjelko Iharosc4df59e2017-07-20 11:59:48 +020011206severity-output <format>
11207 This setting is used with the stats sockets only to configure severity
11208 level output prepended to informational feedback messages. Severity
11209 level of messages can range between 0 and 7, conforming to syslog
11210 rfc5424. Valid and successful socket commands requesting data
11211 (i.e. "show map", "get acl foo" etc.) will never have a severity level
11212 prepended. It is ignored by other sockets. <format> can be one of :
11213 - "none" (default) no severity level is prepended to feedback messages.
11214 - "number" severity level is prepended as a number.
11215 - "string" severity level is prepended as a string following the
11216 rfc5424 convention.
11217
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011218maxconn <maxconn>
11219 Limits the sockets to this number of concurrent connections. Extraneous
11220 connections will remain in the system's backlog until a connection is
11221 released. If unspecified, the limit will be the same as the frontend's
11222 maxconn. Note that in case of port ranges or multiple addresses, the same
11223 value will be applied to each socket. This setting enables different
11224 limitations on expensive sockets, for instance SSL entries which may easily
11225 eat all memory.
11226
11227mode <mode>
11228 Sets the octal mode used to define access permissions on the UNIX socket. It
11229 can also be set by default in the global section's "unix-bind" statement.
11230 Note that some platforms simply ignore this. This setting is ignored by non
11231 UNIX sockets.
11232
11233mss <maxseg>
11234 Sets the TCP Maximum Segment Size (MSS) value to be advertised on incoming
11235 connections. This can be used to force a lower MSS for certain specific
11236 ports, for instance for connections passing through a VPN. Note that this
11237 relies on a kernel feature which is theoretically supported under Linux but
11238 was buggy in all versions prior to 2.6.28. It may or may not work on other
11239 operating systems. It may also not change the advertised value but change the
11240 effective size of outgoing segments. The commonly advertised value for TCPv4
11241 over Ethernet networks is 1460 = 1500(MTU) - 40(IP+TCP). If this value is
11242 positive, it will be used as the advertised MSS. If it is negative, it will
11243 indicate by how much to reduce the incoming connection's advertised MSS for
11244 outgoing segments. This parameter is only compatible with TCP v4/v6 sockets.
11245
11246name <name>
11247 Sets an optional name for these sockets, which will be reported on the stats
11248 page.
11249
Willy Tarreaud72f0f32015-10-13 14:50:22 +020011250namespace <name>
11251 On Linux, it is possible to specify which network namespace a socket will
11252 belong to. This directive makes it possible to explicitly bind a listener to
11253 a namespace different from the default one. Please refer to your operating
11254 system's documentation to find more details about network namespaces.
11255
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011256nice <nice>
11257 Sets the 'niceness' of connections initiated from the socket. Value must be
11258 in the range -1024..1024 inclusive, and defaults to zero. Positive values
11259 means that such connections are more friendly to others and easily offer
11260 their place in the scheduler. On the opposite, negative values mean that
11261 connections want to run with a higher priority than others. The difference
11262 only happens under high loads when the system is close to saturation.
11263 Negative values are appropriate for low-latency or administration services,
11264 and high values are generally recommended for CPU intensive tasks such as SSL
11265 processing or bulk transfers which are less sensible to latency. For example,
11266 it may make sense to use a positive value for an SMTP socket and a negative
11267 one for an RDP socket.
11268
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020011269no-ca-names
11270 This setting is only available when support for OpenSSL was built in. It
11271 prevents from send CA names in server hello message when ca-file is used.
11272
Emeric Brun9b3009b2012-10-05 11:55:06 +020011273no-sslv3
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011274 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011275 disables support for SSLv3 on any sockets instantiated from the listener when
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011276 SSL is supported. Note that SSLv2 is forced disabled in the code and cannot
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011277 be enabled using any configuration option. This option is also available on
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011278 global statement "ssl-default-bind-options". Use "ssl-min-ver" and
11279 "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011280
Emeric Brun90ad8722012-10-02 14:00:59 +020011281no-tls-tickets
11282 This setting is only available when support for OpenSSL was built in. It
11283 disables the stateless session resumption (RFC 5077 TLS Ticket
11284 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011285 session resumption is more expensive in CPU usage. This option is also
11286 available on global statement "ssl-default-bind-options".
Emeric Brun90ad8722012-10-02 14:00:59 +020011287
Emeric Brun9b3009b2012-10-05 11:55:06 +020011288no-tlsv10
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011289 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011290 disables support for TLSv1.0 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011291 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011292 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011293 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11294 and "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011295
Emeric Brun9b3009b2012-10-05 11:55:06 +020011296no-tlsv11
Emeric Brunf5da4932012-09-28 19:42:54 +020011297 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011298 disables support for TLSv1.1 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011299 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011300 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011301 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11302 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020011303
Emeric Brun9b3009b2012-10-05 11:55:06 +020011304no-tlsv12
Emeric Brunf5da4932012-09-28 19:42:54 +020011305 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011306 disables support for TLSv1.2 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011307 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011308 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011309 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11310 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020011311
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011312no-tlsv13
11313 This setting is only available when support for OpenSSL was built in. It
11314 disables support for TLSv1.3 on any sockets instantiated from the listener
11315 when SSL is supported. Note that SSLv2 is forced disabled in the code and
11316 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011317 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11318 and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011319
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020011320npn <protocols>
11321 This enables the NPN TLS extension and advertises the specified protocol list
11322 as supported on top of NPN. The protocol list consists in a comma-delimited
11323 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
John Roeslerfb2fce12019-07-10 15:45:51 -050011324 This requires that the SSL library is built with support for TLS extensions
Willy Tarreauab861d32013-04-02 02:30:41 +020011325 enabled (check with haproxy -vv). Note that the NPN extension has been
Willy Tarreau95c4e142017-11-26 12:18:55 +010011326 replaced with the ALPN extension (see the "alpn" keyword), though this one is
11327 only available starting with OpenSSL 1.0.2. If HTTP/2 is desired on an older
11328 version of OpenSSL, NPN might still be used as most clients still support it
11329 at the time of writing this. It is possible to enable both NPN and ALPN
11330 though it probably doesn't make any sense out of testing.
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020011331
Lukas Tribus53ae85c2017-05-04 15:45:40 +000011332prefer-client-ciphers
11333 Use the client's preference when selecting the cipher suite, by default
11334 the server's preference is enforced. This option is also available on
11335 global statement "ssl-default-bind-options".
Lukas Tribus926594f2018-05-18 17:55:57 +020011336 Note that with OpenSSL >= 1.1.1 ChaCha20-Poly1305 is reprioritized anyway
11337 (without setting this option), if a ChaCha20-Poly1305 cipher is at the top of
11338 the client cipher list.
Lukas Tribus53ae85c2017-05-04 15:45:40 +000011339
Christopher Fauletc644fa92017-11-23 22:44:11 +010011340process <process-set>[/<thread-set>]
Willy Tarreaua36b3242019-02-02 13:14:34 +010011341 This restricts the list of processes or threads on which this listener is
Christopher Fauletc644fa92017-11-23 22:44:11 +010011342 allowed to run. It does not enforce any process but eliminates those which do
Davor Ocelice9ed2812017-12-25 17:49:28 +010011343 not match. If the frontend uses a "bind-process" setting, the intersection
Christopher Fauletc644fa92017-11-23 22:44:11 +010011344 between the two is applied. If in the end the listener is not allowed to run
11345 on any remaining process, a warning is emitted, and the listener will either
11346 run on the first process of the listener if a single process was specified,
11347 or on all of its processes if multiple processes were specified. If a thread
Davor Ocelice9ed2812017-12-25 17:49:28 +010011348 set is specified, it limits the threads allowed to process incoming
Willy Tarreaua36b3242019-02-02 13:14:34 +010011349 connections for this listener, for the the process set. If multiple processes
11350 and threads are configured, a warning is emitted, as it either results from a
11351 configuration error or a misunderstanding of these models. For the unlikely
11352 case where several ranges are needed, this directive may be repeated.
11353 <process-set> and <thread-set> must use the format
Christopher Fauletc644fa92017-11-23 22:44:11 +010011354
11355 all | odd | even | number[-[number]]
11356
11357 Ranges can be partially defined. The higher bound can be omitted. In such
11358 case, it is replaced by the corresponding maximum value. The main purpose of
11359 this directive is to be used with the stats sockets and have one different
11360 socket per process. The second purpose is to have multiple bind lines sharing
11361 the same IP:port but not the same process in a listener, so that the system
11362 can distribute the incoming connections into multiple queues and allow a
11363 smoother inter-process load balancing. Currently Linux 3.9 and above is known
11364 for supporting this. See also "bind-process" and "nbproc".
Willy Tarreau6ae1ba62014-05-07 19:01:58 +020011365
Christopher Fauleta717b992018-04-10 14:43:00 +020011366proto <name>
11367 Forces the multiplexer's protocol to use for the incoming connections. It
11368 must be compatible with the mode of the frontend (TCP or HTTP). It must also
11369 be usable on the frontend side. The list of available protocols is reported
11370 in haproxy -vv.
11371 Idea behind this optipon is to bypass the selection of the best multiplexer's
11372 protocol for all connections instantiated from this listening socket. For
Joseph Herlant71b4b152018-11-13 16:55:16 -080011373 instance, it is possible to force the http/2 on clear TCP by specifying "proto
Christopher Fauleta717b992018-04-10 14:43:00 +020011374 h2" on the bind line.
11375
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011376ssl
11377 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011378 enables SSL deciphering on connections instantiated from this listener. A
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011379 certificate is necessary (see "crt" above). All contents in the buffers will
11380 appear in clear text, so that ACLs and HTTP processing will only have access
Emmanuel Hocdetbd695fe2017-05-15 15:53:41 +020011381 to deciphered contents. SSLv3 is disabled per default, use "ssl-min-ver SSLv3"
11382 to enable it.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011383
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011384ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
11385 This option enforces use of <version> or lower on SSL connections instantiated
11386 from this listener. This option is also available on global statement
11387 "ssl-default-bind-options". See also "ssl-min-ver".
11388
11389ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
11390 This option enforces use of <version> or upper on SSL connections instantiated
11391 from this listener. This option is also available on global statement
11392 "ssl-default-bind-options". See also "ssl-max-ver".
11393
Emmanuel Hocdet65623372013-01-24 17:17:15 +010011394strict-sni
11395 This setting is only available when support for OpenSSL was built in. The
11396 SSL/TLS negotiation is allow only if the client provided an SNI which match
11397 a certificate. The default certificate is not used.
11398 See the "crt" option for more information.
11399
Willy Tarreau2af207a2015-02-04 00:45:58 +010011400tcp-ut <delay>
Tim Düsterhus4896c442016-11-29 02:15:19 +010011401 Sets the TCP User Timeout for all incoming connections instantiated from this
Willy Tarreau2af207a2015-02-04 00:45:58 +010011402 listening socket. This option is available on Linux since version 2.6.37. It
11403 allows haproxy to configure a timeout for sockets which contain data not
Davor Ocelice9ed2812017-12-25 17:49:28 +010011404 receiving an acknowledgment for the configured delay. This is especially
Willy Tarreau2af207a2015-02-04 00:45:58 +010011405 useful on long-lived connections experiencing long idle periods such as
11406 remote terminals or database connection pools, where the client and server
11407 timeouts must remain high to allow a long period of idle, but where it is
11408 important to detect that the client has disappeared in order to release all
11409 resources associated with its connection (and the server's session). The
11410 argument is a delay expressed in milliseconds by default. This only works
11411 for regular TCP connections, and is ignored for other protocols.
11412
Willy Tarreau1c862c52012-10-05 16:21:00 +020011413tfo
Lukas Tribus0defb902013-02-13 23:35:39 +010011414 Is an optional keyword which is supported only on Linux kernels >= 3.7. It
Willy Tarreau1c862c52012-10-05 16:21:00 +020011415 enables TCP Fast Open on the listening socket, which means that clients which
11416 support this feature will be able to send a request and receive a response
11417 during the 3-way handshake starting from second connection, thus saving one
11418 round-trip after the first connection. This only makes sense with protocols
11419 that use high connection rates and where each round trip matters. This can
11420 possibly cause issues with many firewalls which do not accept data on SYN
11421 packets, so this option should only be enabled once well tested. This option
Lukas Tribus0999f762013-04-02 16:43:24 +020011422 is only supported on TCPv4/TCPv6 sockets and ignored by other ones. You may
11423 need to build HAProxy with USE_TFO=1 if your libc doesn't define
11424 TCP_FASTOPEN.
Willy Tarreau1c862c52012-10-05 16:21:00 +020011425
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010011426tls-ticket-keys <keyfile>
11427 Sets the TLS ticket keys file to load the keys from. The keys need to be 48
Emeric Brun9e754772019-01-10 17:51:55 +010011428 or 80 bytes long, depending if aes128 or aes256 is used, encoded with base64
11429 with one line per key (ex. openssl rand 80 | openssl base64 -A | xargs echo).
11430 The first key determines the key length used for next keys: you can't mix
11431 aes128 and aes256 keys. Number of keys is specified by the TLS_TICKETS_NO
11432 build option (default 3) and at least as many keys need to be present in
11433 the file. Last TLS_TICKETS_NO keys will be used for decryption and the
11434 penultimate one for encryption. This enables easy key rotation by just
11435 appending new key to the file and reloading the process. Keys must be
11436 periodically rotated (ex. every 12h) or Perfect Forward Secrecy is
11437 compromised. It is also a good idea to keep the keys off any permanent
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010011438 storage such as hard drives (hint: use tmpfs and don't swap those files).
11439 Lifetime hint can be changed using tune.ssl.timeout.
11440
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011441transparent
11442 Is an optional keyword which is supported only on certain Linux kernels. It
11443 indicates that the addresses will be bound even if they do not belong to the
11444 local machine, and that packets targeting any of these addresses will be
11445 intercepted just as if the addresses were locally configured. This normally
11446 requires that IP forwarding is enabled. Caution! do not use this with the
11447 default address '*', as it would redirect any traffic for the specified port.
11448 This keyword is available only when HAProxy is built with USE_LINUX_TPROXY=1.
11449 This parameter is only compatible with TCPv4 and TCPv6 sockets, depending on
11450 kernel version. Some distribution kernels include backports of the feature,
11451 so check for support with your vendor.
11452
Willy Tarreau77e3af92012-11-24 15:07:23 +010011453v4v6
11454 Is an optional keyword which is supported only on most recent systems
11455 including Linux kernels >= 2.4.21. It is used to bind a socket to both IPv4
11456 and IPv6 when it uses the default address. Doing so is sometimes necessary
11457 on systems which bind to IPv6 only by default. It has no effect on non-IPv6
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011458 sockets, and is overridden by the "v6only" option.
Willy Tarreau77e3af92012-11-24 15:07:23 +010011459
Willy Tarreau9b6700f2012-11-24 11:55:28 +010011460v6only
11461 Is an optional keyword which is supported only on most recent systems
11462 including Linux kernels >= 2.4.21. It is used to bind a socket to IPv6 only
11463 when it uses the default address. Doing so is sometimes preferred to doing it
Willy Tarreau77e3af92012-11-24 15:07:23 +010011464 system-wide as it is per-listener. It has no effect on non-IPv6 sockets and
11465 has precedence over the "v4v6" option.
Willy Tarreau9b6700f2012-11-24 11:55:28 +010011466
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011467uid <uid>
11468 Sets the owner of the UNIX sockets to the designated system uid. It can also
11469 be set by default in the global section's "unix-bind" statement. Note that
11470 some platforms simply ignore this. This setting is equivalent to the "user"
11471 setting except that the user numeric ID is used instead of its name. This
11472 setting is ignored by non UNIX sockets.
11473
11474user <user>
11475 Sets the owner of the UNIX sockets to the designated system user. It can also
11476 be set by default in the global section's "unix-bind" statement. Note that
11477 some platforms simply ignore this. This setting is equivalent to the "uid"
11478 setting except that the user name is used instead of its uid. This setting is
11479 ignored by non UNIX sockets.
11480
Emeric Brun1a073b42012-09-28 17:07:34 +020011481verify [none|optional|required]
11482 This setting is only available when support for OpenSSL was built in. If set
11483 to 'none', client certificate is not requested. This is the default. In other
11484 cases, a client certificate is requested. If the client does not provide a
11485 certificate after the request and if 'verify' is set to 'required', then the
11486 handshake is aborted, while it would have succeeded if set to 'optional'. The
Emeric Brunfd33a262012-10-11 16:28:27 +020011487 certificate provided by the client is always verified using CAs from
11488 'ca-file' and optional CRLs from 'crl-file'. On verify failure the handshake
11489 is aborted, regardless of the 'verify' option, unless the error code exactly
11490 matches one of those listed with 'ca-ignore-err' or 'crt-ignore-err'.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020011491
Willy Tarreaub6205fd2012-09-24 12:27:33 +0200114925.2. Server and default-server options
Cyril Bontéf0c60612010-02-06 14:44:47 +010011493------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020011494
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010011495The "server" and "default-server" keywords support a certain number of settings
11496which are all passed as arguments on the server line. The order in which those
11497arguments appear does not count, and they are all optional. Some of those
11498settings are single words (booleans) while others expect one or several values
11499after them. In this case, the values must immediately follow the setting name.
11500Except default-server, all those settings must be specified after the server's
11501address if they are used:
Willy Tarreau6a06a402007-07-15 20:15:28 +020011502
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011503 server <name> <address>[:port] [settings ...]
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010011504 default-server [settings ...]
Willy Tarreau6a06a402007-07-15 20:15:28 +020011505
Frédéric Lécailled2376272017-03-21 18:52:12 +010011506Note that all these settings are supported both by "server" and "default-server"
11507keywords, except "id" which is only supported by "server".
11508
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011509The currently supported settings are the following ones.
Willy Tarreau0ba27502007-12-24 16:55:16 +010011510
Willy Tarreauceb4ac92012-04-28 00:41:46 +020011511addr <ipv4|ipv6>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011512 Using the "addr" parameter, it becomes possible to use a different IP address
Baptiste Assmann13f83532016-03-06 23:14:36 +010011513 to send health-checks or to probe the agent-check. On some servers, it may be
11514 desirable to dedicate an IP address to specific component able to perform
11515 complex tests which are more suitable to health-checks than the application.
11516 This parameter is ignored if the "check" parameter is not set. See also the
11517 "port" parameter.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011518
Simon Hormand60d6912013-11-25 10:46:36 +090011519agent-check
11520 Enable an auxiliary agent check which is run independently of a regular
Willy Tarreau81f5d942013-12-09 20:51:51 +010011521 health check. An agent health check is performed by making a TCP connection
Willy Tarreau7a0139e2018-12-16 08:42:56 +010011522 to the port set by the "agent-port" parameter and reading an ASCII string
11523 terminated by the first '\r' or '\n' met. The string is made of a series of
11524 words delimited by spaces, tabs or commas in any order, each consisting of :
Simon Hormand60d6912013-11-25 10:46:36 +090011525
Willy Tarreau81f5d942013-12-09 20:51:51 +010011526 - An ASCII representation of a positive integer percentage, e.g. "75%".
Simon Hormand60d6912013-11-25 10:46:36 +090011527 Values in this format will set the weight proportional to the initial
Willy Tarreauc5af3a62014-10-07 15:27:33 +020011528 weight of a server as configured when haproxy starts. Note that a zero
11529 weight is reported on the stats page as "DRAIN" since it has the same
11530 effect on the server (it's removed from the LB farm).
Simon Hormand60d6912013-11-25 10:46:36 +090011531
Davor Ocelice9ed2812017-12-25 17:49:28 +010011532 - The string "maxconn:" followed by an integer (no space between). Values
11533 in this format will set the maxconn of a server. The maximum number of
11534 connections advertised needs to be multiplied by the number of load
11535 balancers and different backends that use this health check to get the
11536 total number of connections the server might receive. Example: maxconn:30
Nenad Merdanovic174dd372016-04-24 23:10:06 +020011537
Willy Tarreau81f5d942013-12-09 20:51:51 +010011538 - The word "ready". This will turn the server's administrative state to the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011539 READY mode, thus canceling any DRAIN or MAINT state
Simon Hormand60d6912013-11-25 10:46:36 +090011540
Willy Tarreau81f5d942013-12-09 20:51:51 +010011541 - The word "drain". This will turn the server's administrative state to the
11542 DRAIN mode, thus it will not accept any new connections other than those
11543 that are accepted via persistence.
Simon Hormand60d6912013-11-25 10:46:36 +090011544
Willy Tarreau81f5d942013-12-09 20:51:51 +010011545 - The word "maint". This will turn the server's administrative state to the
11546 MAINT mode, thus it will not accept any new connections at all, and health
11547 checks will be stopped.
Simon Hormand60d6912013-11-25 10:46:36 +090011548
Willy Tarreau81f5d942013-12-09 20:51:51 +010011549 - The words "down", "failed", or "stopped", optionally followed by a
11550 description string after a sharp ('#'). All of these mark the server's
11551 operating state as DOWN, but since the word itself is reported on the stats
11552 page, the difference allows an administrator to know if the situation was
11553 expected or not : the service may intentionally be stopped, may appear up
Davor Ocelice9ed2812017-12-25 17:49:28 +010011554 but fail some validity tests, or may be seen as down (e.g. missing process,
Willy Tarreau81f5d942013-12-09 20:51:51 +010011555 or port not responding).
Simon Hormand60d6912013-11-25 10:46:36 +090011556
Willy Tarreau81f5d942013-12-09 20:51:51 +010011557 - The word "up" sets back the server's operating state as UP if health checks
11558 also report that the service is accessible.
Simon Hormand60d6912013-11-25 10:46:36 +090011559
Willy Tarreau81f5d942013-12-09 20:51:51 +010011560 Parameters which are not advertised by the agent are not changed. For
11561 example, an agent might be designed to monitor CPU usage and only report a
11562 relative weight and never interact with the operating status. Similarly, an
11563 agent could be designed as an end-user interface with 3 radio buttons
11564 allowing an administrator to change only the administrative state. However,
11565 it is important to consider that only the agent may revert its own actions,
11566 so if a server is set to DRAIN mode or to DOWN state using the agent, the
11567 agent must implement the other equivalent actions to bring the service into
11568 operations again.
Simon Hormand60d6912013-11-25 10:46:36 +090011569
Simon Horman2f1f9552013-11-25 10:46:37 +090011570 Failure to connect to the agent is not considered an error as connectivity
11571 is tested by the regular health check which is enabled by the "check"
Willy Tarreau81f5d942013-12-09 20:51:51 +010011572 parameter. Warning though, it is not a good idea to stop an agent after it
11573 reports "down", since only an agent reporting "up" will be able to turn the
11574 server up again. Note that the CLI on the Unix stats socket is also able to
Willy Tarreau989222a2016-01-15 10:26:26 +010011575 force an agent's result in order to work around a bogus agent if needed.
Simon Horman2f1f9552013-11-25 10:46:37 +090011576
Willy Tarreau81f5d942013-12-09 20:51:51 +010011577 Requires the "agent-port" parameter to be set. See also the "agent-inter"
Frédéric Lécailled2376272017-03-21 18:52:12 +010011578 and "no-agent-check" parameters.
Simon Hormand60d6912013-11-25 10:46:36 +090011579
James Brown55f9ff12015-10-21 18:19:05 -070011580agent-send <string>
11581 If this option is specified, haproxy will send the given string (verbatim)
11582 to the agent server upon connection. You could, for example, encode
11583 the backend name into this string, which would enable your agent to send
11584 different responses based on the backend. Make sure to include a '\n' if
11585 you want to terminate your request with a newline.
11586
Simon Hormand60d6912013-11-25 10:46:36 +090011587agent-inter <delay>
11588 The "agent-inter" parameter sets the interval between two agent checks
11589 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
11590
11591 Just as with every other time-based parameter, it may be entered in any
11592 other explicit unit among { us, ms, s, m, h, d }. The "agent-inter"
11593 parameter also serves as a timeout for agent checks "timeout check" is
11594 not set. In order to reduce "resonance" effects when multiple servers are
11595 hosted on the same hardware, the agent and health checks of all servers
11596 are started with a small time offset between them. It is also possible to
11597 add some random noise in the agent and health checks interval using the
11598 global "spread-checks" keyword. This makes sense for instance when a lot
11599 of backends use the same servers.
11600
11601 See also the "agent-check" and "agent-port" parameters.
11602
Misiek768d8602017-01-09 09:52:43 +010011603agent-addr <addr>
11604 The "agent-addr" parameter sets address for agent check.
11605
11606 You can offload agent-check to another target, so you can make single place
11607 managing status and weights of servers defined in haproxy in case you can't
11608 make self-aware and self-managing services. You can specify both IP or
11609 hostname, it will be resolved.
11610
Simon Hormand60d6912013-11-25 10:46:36 +090011611agent-port <port>
11612 The "agent-port" parameter sets the TCP port used for agent checks.
11613
11614 See also the "agent-check" and "agent-inter" parameters.
11615
Olivier Houchard8cb2d2e2019-05-06 18:58:48 +020011616allow-0rtt
11617 Allow sending early data to the server when using TLS 1.3.
Olivier Houchard22c9b442019-05-06 19:01:04 +020011618 Note that early data will be sent only if the client used early data, or
11619 if the backend uses "retry-on" with the "0rtt-rejected" keyword.
Olivier Houchard8cb2d2e2019-05-06 18:58:48 +020011620
Olivier Houchardc7566002018-11-20 23:33:50 +010011621alpn <protocols>
11622 This enables the TLS ALPN extension and advertises the specified protocol
11623 list as supported on top of ALPN. The protocol list consists in a comma-
11624 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
John Roeslerfb2fce12019-07-10 15:45:51 -050011625 quotes). This requires that the SSL library is built with support for TLS
Olivier Houchardc7566002018-11-20 23:33:50 +010011626 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
11627 initial NPN extension. ALPN is required to connect to HTTP/2 servers.
11628 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
11629 now obsolete NPN extension.
11630 If both HTTP/2 and HTTP/1.1 are expected to be supported, both versions can
11631 be advertised, in order of preference, like below :
11632
11633 server 127.0.0.1:443 ssl crt pub.pem alpn h2,http/1.1
11634
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011635backup
11636 When "backup" is present on a server line, the server is only used in load
11637 balancing when all other non-backup servers are unavailable. Requests coming
11638 with a persistence cookie referencing the server will always be served
11639 though. By default, only the first operational backup server is used, unless
Frédéric Lécailled2376272017-03-21 18:52:12 +010011640 the "allbackups" option is set in the backend. See also the "no-backup" and
11641 "allbackups" options.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011642
Emeric Brunef42d922012-10-11 16:11:36 +020011643ca-file <cafile>
11644 This setting is only available when support for OpenSSL was built in. It
11645 designates a PEM file from which to load CA certificates used to verify
11646 server's certificate.
11647
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011648check
11649 This option enables health checks on the server. By default, a server is
Patrick Mézardb7aeec62012-01-22 16:01:22 +010011650 always considered available. If "check" is set, the server is available when
11651 accepting periodic TCP connections, to ensure that it is really able to serve
11652 requests. The default address and port to send the tests to are those of the
11653 server, and the default source is the same as the one defined in the
11654 backend. It is possible to change the address using the "addr" parameter, the
11655 port using the "port" parameter, the source address using the "source"
11656 address, and the interval and timers using the "inter", "rise" and "fall"
Simon Hormanafc47ee2013-11-25 10:46:35 +090011657 parameters. The request method is define in the backend using the "httpchk",
11658 "smtpchk", "mysql-check", "pgsql-check" and "ssl-hello-chk" options. Please
Frédéric Lécailled2376272017-03-21 18:52:12 +010011659 refer to those options and parameters for more information. See also
11660 "no-check" option.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011661
Willy Tarreau6c16adc2012-10-05 00:04:16 +020011662check-send-proxy
11663 This option forces emission of a PROXY protocol line with outgoing health
11664 checks, regardless of whether the server uses send-proxy or not for the
11665 normal traffic. By default, the PROXY protocol is enabled for health checks
11666 if it is already enabled for normal traffic and if no "port" nor "addr"
11667 directive is present. However, if such a directive is present, the
11668 "check-send-proxy" option needs to be used to force the use of the
11669 protocol. See also the "send-proxy" option for more information.
11670
Olivier Houchard92150142018-12-21 19:47:01 +010011671check-alpn <protocols>
11672 Defines which protocols to advertise with ALPN. The protocol list consists in
11673 a comma-delimited list of protocol names, for instance: "http/1.1,http/1.0"
11674 (without quotes). If it is not set, the server ALPN is used.
11675
Jérôme Magninae9bb762018-12-09 16:08:26 +010011676check-sni <sni>
Olivier Houchard9130a962017-10-17 17:33:43 +020011677 This option allows you to specify the SNI to be used when doing health checks
Jérôme Magninae9bb762018-12-09 16:08:26 +010011678 over SSL. It is only possible to use a string to set <sni>. If you want to
11679 set a SNI for proxied traffic, see "sni".
Olivier Houchard9130a962017-10-17 17:33:43 +020011680
Willy Tarreau763a95b2012-10-04 23:15:39 +020011681check-ssl
11682 This option forces encryption of all health checks over SSL, regardless of
11683 whether the server uses SSL or not for the normal traffic. This is generally
11684 used when an explicit "port" or "addr" directive is specified and SSL health
11685 checks are not inherited. It is important to understand that this option
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011686 inserts an SSL transport layer below the checks, so that a simple TCP connect
Willy Tarreau763a95b2012-10-04 23:15:39 +020011687 check becomes an SSL connect, which replaces the old ssl-hello-chk. The most
11688 common use is to send HTTPS checks by combining "httpchk" with SSL checks.
Davor Ocelice9ed2812017-12-25 17:49:28 +010011689 All SSL settings are common to health checks and traffic (e.g. ciphers).
Frédéric Lécailled2376272017-03-21 18:52:12 +010011690 See the "ssl" option for more information and "no-check-ssl" to disable
11691 this option.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011692
Alexander Liu2a54bb72019-05-22 19:44:48 +080011693check-via-socks4
John Roeslerfb2fce12019-07-10 15:45:51 -050011694 This option enables outgoing health checks using upstream socks4 proxy. By
Alexander Liu2a54bb72019-05-22 19:44:48 +080011695 default, the health checks won't go through socks tunnel even it was enabled
11696 for normal traffic.
11697
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011698ciphers <ciphers>
Dirkjan Bussink415150f2018-09-14 11:14:21 +020011699 This setting is only available when support for OpenSSL was built in. This
11700 option sets the string describing the list of cipher algorithms that is
11701 negotiated during the SSL/TLS handshake with the server. The format of the
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000011702 string is defined in "man 1 ciphers" from OpenSSL man pages. For background
11703 information and recommendations see e.g.
11704 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
11705 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
11706 cipher configuration, please check the "ciphersuites" keyword.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011707
Dirkjan Bussink415150f2018-09-14 11:14:21 +020011708ciphersuites <ciphersuites>
11709 This setting is only available when support for OpenSSL was built in and
11710 OpenSSL 1.1.1 or later was used to build HAProxy. This option sets the string
11711 describing the list of cipher algorithms that is negotiated during the TLS
11712 1.3 handshake with the server. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000011713 "man 1 ciphers" from OpenSSL man pages under the "ciphersuites" section.
11714 For cipher configuration for TLSv1.2 and earlier, please check the "ciphers"
11715 keyword.
Dirkjan Bussink415150f2018-09-14 11:14:21 +020011716
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011717cookie <value>
11718 The "cookie" parameter sets the cookie value assigned to the server to
11719 <value>. This value will be checked in incoming requests, and the first
11720 operational server possessing the same value will be selected. In return, in
11721 cookie insertion or rewrite modes, this value will be assigned to the cookie
11722 sent to the client. There is nothing wrong in having several servers sharing
11723 the same cookie value, and it is in fact somewhat common between normal and
11724 backup servers. See also the "cookie" keyword in backend section.
11725
Emeric Brunef42d922012-10-11 16:11:36 +020011726crl-file <crlfile>
11727 This setting is only available when support for OpenSSL was built in. It
11728 designates a PEM file from which to load certificate revocation list used
11729 to verify server's certificate.
11730
Emeric Bruna7aa3092012-10-26 12:58:00 +020011731crt <cert>
11732 This setting is only available when support for OpenSSL was built in.
11733 It designates a PEM file from which to load both a certificate and the
11734 associated private key. This file can be built by concatenating both PEM
11735 files into one. This certificate will be sent if the server send a client
11736 certificate request.
11737
Willy Tarreau96839092010-03-29 10:02:24 +020011738disabled
11739 The "disabled" keyword starts the server in the "disabled" state. That means
11740 that it is marked down in maintenance mode, and no connection other than the
11741 ones allowed by persist mode will reach it. It is very well suited to setup
11742 new servers, because normal traffic will never reach them, while it is still
11743 possible to test the service by making use of the force-persist mechanism.
Frédéric Lécailled2376272017-03-21 18:52:12 +010011744 See also "enabled" setting.
Willy Tarreau96839092010-03-29 10:02:24 +020011745
Frédéric Lécailled2376272017-03-21 18:52:12 +010011746enabled
11747 This option may be used as 'server' setting to reset any 'disabled'
11748 setting which would have been inherited from 'default-server' directive as
11749 default value.
11750 It may also be used as 'default-server' setting to reset any previous
11751 'default-server' 'disabled' setting.
Willy Tarreau96839092010-03-29 10:02:24 +020011752
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011753error-limit <count>
Willy Tarreau983e01e2010-01-11 18:42:06 +010011754 If health observing is enabled, the "error-limit" parameter specifies the
11755 number of consecutive errors that triggers event selected by the "on-error"
11756 option. By default it is set to 10 consecutive errors.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011757
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011758 See also the "check", "error-limit" and "on-error".
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011759
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011760fall <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011761 The "fall" parameter states that a server will be considered as dead after
11762 <count> consecutive unsuccessful health checks. This value defaults to 3 if
11763 unspecified. See also the "check", "inter" and "rise" parameters.
11764
Emeric Brun8694b9a2012-10-05 14:39:07 +020011765force-sslv3
11766 This option enforces use of SSLv3 only when SSL is used to communicate with
11767 the server. SSLv3 is generally less expensive than the TLS counterparts for
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011768 high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011769 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011770
11771force-tlsv10
11772 This option enforces use of TLSv1.0 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011773 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011774 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011775
11776force-tlsv11
11777 This option enforces use of TLSv1.1 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011778 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011779 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011780
11781force-tlsv12
11782 This option enforces use of TLSv1.2 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011783 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011784 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011785
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011786force-tlsv13
11787 This option enforces use of TLSv1.3 only when SSL is used to communicate with
11788 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011789 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011790
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011791id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +020011792 Set a persistent ID for the server. This ID must be positive and unique for
11793 the proxy. An unused ID will automatically be assigned if unset. The first
11794 assigned value will be 1. This ID is currently only returned in statistics.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011795
Willy Tarreau6a031d12016-11-07 19:42:35 +010011796init-addr {last | libc | none | <ip>},[...]*
11797 Indicate in what order the server's address should be resolved upon startup
11798 if it uses an FQDN. Attempts are made to resolve the address by applying in
Davor Ocelice9ed2812017-12-25 17:49:28 +010011799 turn each of the methods mentioned in the comma-delimited list. The first
Willy Tarreau6a031d12016-11-07 19:42:35 +010011800 method which succeeds is used. If the end of the list is reached without
11801 finding a working method, an error is thrown. Method "last" suggests to pick
11802 the address which appears in the state file (see "server-state-file"). Method
11803 "libc" uses the libc's internal resolver (gethostbyname() or getaddrinfo()
11804 depending on the operating system and build options). Method "none"
11805 specifically indicates that the server should start without any valid IP
11806 address in a down state. It can be useful to ignore some DNS issues upon
11807 startup, waiting for the situation to get fixed later. Finally, an IP address
11808 (IPv4 or IPv6) may be provided. It can be the currently known address of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011809 server (e.g. filled by a configuration generator), or the address of a dummy
Willy Tarreau6a031d12016-11-07 19:42:35 +010011810 server used to catch old sessions and present them with a decent error
11811 message for example. When the "first" load balancing algorithm is used, this
11812 IP address could point to a fake server used to trigger the creation of new
11813 instances on the fly. This option defaults to "last,libc" indicating that the
11814 previous address found in the state file (if any) is used first, otherwise
11815 the libc's resolver is used. This ensures continued compatibility with the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011816 historic behavior.
Willy Tarreau6a031d12016-11-07 19:42:35 +010011817
11818 Example:
11819 defaults
11820 # never fail on address resolution
11821 default-server init-addr last,libc,none
11822
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011823inter <delay>
11824fastinter <delay>
11825downinter <delay>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011826 The "inter" parameter sets the interval between two consecutive health checks
11827 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
11828 It is also possible to use "fastinter" and "downinter" to optimize delays
11829 between checks depending on the server state :
11830
Pieter Baauw44fc9df2015-09-17 21:30:46 +020011831 Server state | Interval used
11832 ----------------------------------------+----------------------------------
11833 UP 100% (non-transitional) | "inter"
11834 ----------------------------------------+----------------------------------
11835 Transitionally UP (going down "fall"), | "fastinter" if set,
11836 Transitionally DOWN (going up "rise"), | "inter" otherwise.
11837 or yet unchecked. |
11838 ----------------------------------------+----------------------------------
11839 DOWN 100% (non-transitional) | "downinter" if set,
11840 | "inter" otherwise.
11841 ----------------------------------------+----------------------------------
Willy Tarreaud72758d2010-01-12 10:42:19 +010011842
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011843 Just as with every other time-based parameter, they can be entered in any
11844 other explicit unit among { us, ms, s, m, h, d }. The "inter" parameter also
11845 serves as a timeout for health checks sent to servers if "timeout check" is
11846 not set. In order to reduce "resonance" effects when multiple servers are
Simon Hormand60d6912013-11-25 10:46:36 +090011847 hosted on the same hardware, the agent and health checks of all servers
11848 are started with a small time offset between them. It is also possible to
11849 add some random noise in the agent and health checks interval using the
11850 global "spread-checks" keyword. This makes sense for instance when a lot
11851 of backends use the same servers.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011852
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011853maxconn <maxconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011854 The "maxconn" parameter specifies the maximal number of concurrent
11855 connections that will be sent to this server. If the number of incoming
Tim Duesterhuscefbbd92019-11-27 22:35:27 +010011856 concurrent connections goes higher than this value, they will be queued,
11857 waiting for a slot to be released. This parameter is very important as it can
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011858 save fragile servers from going down under extreme loads. If a "minconn"
11859 parameter is specified, the limit becomes dynamic. The default value is "0"
11860 which means unlimited. See also the "minconn" and "maxqueue" parameters, and
11861 the backend's "fullconn" keyword.
11862
Tim Duesterhuscefbbd92019-11-27 22:35:27 +010011863 In HTTP mode this parameter limits the number of concurrent requests instead
11864 of the number of connections. Multiple requests might be multiplexed over a
11865 single TCP connection to the server. As an example if you specify a maxconn
11866 of 50 you might see between 1 and 50 actual server connections, but no more
11867 than 50 concurrent requests.
11868
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011869maxqueue <maxqueue>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011870 The "maxqueue" parameter specifies the maximal number of connections which
11871 will wait in the queue for this server. If this limit is reached, next
11872 requests will be redispatched to other servers instead of indefinitely
11873 waiting to be served. This will break persistence but may allow people to
11874 quickly re-log in when the server they try to connect to is dying. The
11875 default value is "0" which means the queue is unlimited. See also the
11876 "maxconn" and "minconn" parameters.
11877
Willy Tarreau9c538e02019-01-23 10:21:49 +010011878max-reuse <count>
11879 The "max-reuse" argument indicates the HTTP connection processors that they
11880 should not reuse a server connection more than this number of times to send
11881 new requests. Permitted values are -1 (the default), which disables this
11882 limit, or any positive value. Value zero will effectively disable keep-alive.
11883 This is only used to work around certain server bugs which cause them to leak
11884 resources over time. The argument is not necessarily respected by the lower
11885 layers as there might be technical limitations making it impossible to
11886 enforce. At least HTTP/2 connections to servers will respect it.
11887
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011888minconn <minconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011889 When the "minconn" parameter is set, the maxconn limit becomes a dynamic
11890 limit following the backend's load. The server will always accept at least
11891 <minconn> connections, never more than <maxconn>, and the limit will be on
11892 the ramp between both values when the backend has less than <fullconn>
11893 concurrent connections. This makes it possible to limit the load on the
11894 server during normal loads, but push it further for important loads without
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011895 overloading the server during exceptional loads. See also the "maxconn"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011896 and "maxqueue" parameters, as well as the "fullconn" backend keyword.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011897
Willy Tarreaud72f0f32015-10-13 14:50:22 +020011898namespace <name>
11899 On Linux, it is possible to specify which network namespace a socket will
11900 belong to. This directive makes it possible to explicitly bind a server to
11901 a namespace different from the default one. Please refer to your operating
11902 system's documentation to find more details about network namespaces.
11903
Frédéric Lécailled2376272017-03-21 18:52:12 +010011904no-agent-check
11905 This option may be used as "server" setting to reset any "agent-check"
11906 setting which would have been inherited from "default-server" directive as
11907 default value.
11908 It may also be used as "default-server" setting to reset any previous
11909 "default-server" "agent-check" setting.
11910
11911no-backup
11912 This option may be used as "server" setting to reset any "backup"
11913 setting which would have been inherited from "default-server" directive as
11914 default value.
11915 It may also be used as "default-server" setting to reset any previous
11916 "default-server" "backup" setting.
11917
11918no-check
11919 This option may be used as "server" setting to reset any "check"
11920 setting which would have been inherited from "default-server" directive as
11921 default value.
11922 It may also be used as "default-server" setting to reset any previous
11923 "default-server" "check" setting.
11924
11925no-check-ssl
11926 This option may be used as "server" setting to reset any "check-ssl"
11927 setting which would have been inherited from "default-server" directive as
11928 default value.
11929 It may also be used as "default-server" setting to reset any previous
11930 "default-server" "check-ssl" setting.
11931
Frédéric Lécailled2376272017-03-21 18:52:12 +010011932no-send-proxy
11933 This option may be used as "server" setting to reset any "send-proxy"
11934 setting which would have been inherited from "default-server" directive as
11935 default value.
11936 It may also be used as "default-server" setting to reset any previous
11937 "default-server" "send-proxy" setting.
11938
11939no-send-proxy-v2
11940 This option may be used as "server" setting to reset any "send-proxy-v2"
11941 setting which would have been inherited from "default-server" directive as
11942 default value.
11943 It may also be used as "default-server" setting to reset any previous
11944 "default-server" "send-proxy-v2" setting.
11945
11946no-send-proxy-v2-ssl
11947 This option may be used as "server" setting to reset any "send-proxy-v2-ssl"
11948 setting which would have been inherited from "default-server" directive as
11949 default value.
11950 It may also be used as "default-server" setting to reset any previous
11951 "default-server" "send-proxy-v2-ssl" setting.
11952
11953no-send-proxy-v2-ssl-cn
11954 This option may be used as "server" setting to reset any "send-proxy-v2-ssl-cn"
11955 setting which would have been inherited from "default-server" directive as
11956 default value.
11957 It may also be used as "default-server" setting to reset any previous
11958 "default-server" "send-proxy-v2-ssl-cn" setting.
11959
11960no-ssl
11961 This option may be used as "server" setting to reset any "ssl"
11962 setting which would have been inherited from "default-server" directive as
11963 default value.
11964 It may also be used as "default-server" setting to reset any previous
11965 "default-server" "ssl" setting.
11966
Willy Tarreau2a3fb1c2015-02-05 16:47:07 +010011967no-ssl-reuse
11968 This option disables SSL session reuse when SSL is used to communicate with
11969 the server. It will force the server to perform a full handshake for every
11970 new connection. It's probably only useful for benchmarking, troubleshooting,
11971 and for paranoid users.
11972
Emeric Brun9b3009b2012-10-05 11:55:06 +020011973no-sslv3
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011974 This option disables support for SSLv3 when SSL is used to communicate with
11975 the server. Note that SSLv2 is disabled in the code and cannot be enabled
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011976 using any configuration option. Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011977
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011978 Supported in default-server: No
11979
Emeric Brunf9c5c472012-10-11 15:28:34 +020011980no-tls-tickets
11981 This setting is only available when support for OpenSSL was built in. It
11982 disables the stateless session resumption (RFC 5077 TLS Ticket
11983 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011984 session resumption is more expensive in CPU usage for servers. This option
11985 is also available on global statement "ssl-default-server-options".
Frédéric Lécailled2376272017-03-21 18:52:12 +010011986 See also "tls-tickets".
Emeric Brunf9c5c472012-10-11 15:28:34 +020011987
Emeric Brun9b3009b2012-10-05 11:55:06 +020011988no-tlsv10
Emeric Brun8694b9a2012-10-05 14:39:07 +020011989 This option disables support for TLSv1.0 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020011990 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11991 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011992 often makes sense to disable it when communicating with local servers. This
11993 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011994 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011995
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011996 Supported in default-server: No
11997
Emeric Brun9b3009b2012-10-05 11:55:06 +020011998no-tlsv11
Emeric Brun8694b9a2012-10-05 14:39:07 +020011999 This option disables support for TLSv1.1 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020012000 the server. Note that SSLv2 is disabled in the code and cannot be enabled
12001 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012002 often makes sense to disable it when communicating with local servers. This
12003 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012004 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020012005
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020012006 Supported in default-server: No
12007
Emeric Brun9b3009b2012-10-05 11:55:06 +020012008no-tlsv12
Emeric Brun8694b9a2012-10-05 14:39:07 +020012009 This option disables support for TLSv1.2 when SSL is used to communicate with
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012010 the server. Note that SSLv2 is disabled in the code and cannot be enabled
12011 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012012 often makes sense to disable it when communicating with local servers. This
12013 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012014 Use "ssl-min-ver" and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020012015
12016 Supported in default-server: No
12017
12018no-tlsv13
12019 This option disables support for TLSv1.3 when SSL is used to communicate with
12020 the server. Note that SSLv2 is disabled in the code and cannot be enabled
12021 using any configuration option. TLSv1 is more expensive than SSLv3 so it
12022 often makes sense to disable it when communicating with local servers. This
12023 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012024 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012025
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020012026 Supported in default-server: No
12027
Frédéric Lécailled2376272017-03-21 18:52:12 +010012028no-verifyhost
12029 This option may be used as "server" setting to reset any "verifyhost"
12030 setting which would have been inherited from "default-server" directive as
12031 default value.
12032 It may also be used as "default-server" setting to reset any previous
12033 "default-server" "verifyhost" setting.
Willy Tarreau763a95b2012-10-04 23:15:39 +020012034
Frédéric Lécaille1b9423d2019-07-04 14:19:06 +020012035no-tfo
12036 This option may be used as "server" setting to reset any "tfo"
12037 setting which would have been inherited from "default-server" directive as
12038 default value.
12039 It may also be used as "default-server" setting to reset any previous
12040 "default-server" "tfo" setting.
12041
Simon Hormanfa461682011-06-25 09:39:49 +090012042non-stick
12043 Never add connections allocated to this sever to a stick-table.
12044 This may be used in conjunction with backup to ensure that
12045 stick-table persistence is disabled for backup servers.
12046
Olivier Houchardc7566002018-11-20 23:33:50 +010012047npn <protocols>
12048 This enables the NPN TLS extension and advertises the specified protocol list
12049 as supported on top of NPN. The protocol list consists in a comma-delimited
12050 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
John Roeslerfb2fce12019-07-10 15:45:51 -050012051 This requires that the SSL library is built with support for TLS extensions
Olivier Houchardc7566002018-11-20 23:33:50 +010012052 enabled (check with haproxy -vv). Note that the NPN extension has been
12053 replaced with the ALPN extension (see the "alpn" keyword), though this one is
12054 only available starting with OpenSSL 1.0.2.
12055
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010012056observe <mode>
12057 This option enables health adjusting based on observing communication with
12058 the server. By default this functionality is disabled and enabling it also
12059 requires to enable health checks. There are two supported modes: "layer4" and
12060 "layer7". In layer4 mode, only successful/unsuccessful tcp connections are
12061 significant. In layer7, which is only allowed for http proxies, responses
12062 received from server are verified, like valid/wrong http code, unparsable
Willy Tarreau150d1462012-03-10 08:19:02 +010012063 headers, a timeout, etc. Valid status codes include 100 to 499, 501 and 505.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010012064
12065 See also the "check", "on-error" and "error-limit".
12066
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012067on-error <mode>
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010012068 Select what should happen when enough consecutive errors are detected.
12069 Currently, four modes are available:
12070 - fastinter: force fastinter
12071 - fail-check: simulate a failed check, also forces fastinter (default)
12072 - sudden-death: simulate a pre-fatal failed health check, one more failed
12073 check will mark a server down, forces fastinter
12074 - mark-down: mark the server immediately down and force fastinter
12075
12076 See also the "check", "observe" and "error-limit".
12077
Simon Hormane0d1bfb2011-06-21 14:34:58 +090012078on-marked-down <action>
12079 Modify what occurs when a server is marked down.
12080 Currently one action is available:
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070012081 - shutdown-sessions: Shutdown peer sessions. When this setting is enabled,
12082 all connections to the server are immediately terminated when the server
12083 goes down. It might be used if the health check detects more complex cases
12084 than a simple connection status, and long timeouts would cause the service
12085 to remain unresponsive for too long a time. For instance, a health check
12086 might detect that a database is stuck and that there's no chance to reuse
12087 existing connections anymore. Connections killed this way are logged with
12088 a 'D' termination code (for "Down").
Simon Hormane0d1bfb2011-06-21 14:34:58 +090012089
12090 Actions are disabled by default
12091
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070012092on-marked-up <action>
12093 Modify what occurs when a server is marked up.
12094 Currently one action is available:
12095 - shutdown-backup-sessions: Shutdown sessions on all backup servers. This is
12096 done only if the server is not in backup state and if it is not disabled
12097 (it must have an effective weight > 0). This can be used sometimes to force
12098 an active server to take all the traffic back after recovery when dealing
Davor Ocelice9ed2812017-12-25 17:49:28 +010012099 with long sessions (e.g. LDAP, SQL, ...). Doing this can cause more trouble
12100 than it tries to solve (e.g. incomplete transactions), so use this feature
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070012101 with extreme care. Sessions killed because a server comes up are logged
12102 with an 'U' termination code (for "Up").
12103
12104 Actions are disabled by default
12105
Olivier Houchard006e3102018-12-10 18:30:32 +010012106pool-max-conn <max>
12107 Set the maximum number of idling connections for a server. -1 means unlimited
12108 connections, 0 means no idle connections. The default is -1. When idle
12109 connections are enabled, orphaned idle connections which do not belong to any
12110 client session anymore are moved to a dedicated pool so that they remain
12111 usable by future clients. This only applies to connections that can be shared
12112 according to the same principles as those applying to "http-reuse".
12113
Olivier Houchardb7b3faa2018-12-14 18:15:36 +010012114pool-purge-delay <delay>
12115 Sets the delay to start purging idle connections. Each <delay> interval, half
Olivier Houcharda56eebf2019-03-19 16:44:02 +010012116 of the idle connections are closed. 0 means we don't keep any idle connection.
Willy Tarreaufb553652019-06-04 14:06:31 +020012117 The default is 5s.
Olivier Houchardb7b3faa2018-12-14 18:15:36 +010012118
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012119port <port>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012120 Using the "port" parameter, it becomes possible to use a different port to
12121 send health-checks. On some servers, it may be desirable to dedicate a port
12122 to a specific component able to perform complex tests which are more suitable
12123 to health-checks than the application. It is common to run a simple script in
12124 inetd for instance. This parameter is ignored if the "check" parameter is not
12125 set. See also the "addr" parameter.
12126
Christopher Faulet8ed0a3e2018-04-10 14:45:45 +020012127proto <name>
12128
12129 Forces the multiplexer's protocol to use for the outgoing connections to this
12130 server. It must be compatible with the mode of the backend (TCP or HTTP). It
12131 must also be usable on the backend side. The list of available protocols is
12132 reported in haproxy -vv.
12133 Idea behind this optipon is to bypass the selection of the best multiplexer's
12134 protocol for all connections established to this server.
12135
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012136redir <prefix>
12137 The "redir" parameter enables the redirection mode for all GET and HEAD
12138 requests addressing this server. This means that instead of having HAProxy
12139 forward the request to the server, it will send an "HTTP 302" response with
12140 the "Location" header composed of this prefix immediately followed by the
12141 requested URI beginning at the leading '/' of the path component. That means
12142 that no trailing slash should be used after <prefix>. All invalid requests
12143 will be rejected, and all non-GET or HEAD requests will be normally served by
12144 the server. Note that since the response is completely forged, no header
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012145 mangling nor cookie insertion is possible in the response. However, cookies in
Davor Ocelice9ed2812017-12-25 17:49:28 +010012146 requests are still analyzed, making this solution completely usable to direct
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012147 users to a remote location in case of local disaster. Main use consists in
12148 increasing bandwidth for static servers by having the clients directly
12149 connect to them. Note: never use a relative location here, it would cause a
12150 loop between the client and HAProxy!
12151
12152 Example : server srv1 192.168.1.1:80 redir http://image1.mydomain.com check
12153
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012154rise <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012155 The "rise" parameter states that a server will be considered as operational
12156 after <count> consecutive successful health checks. This value defaults to 2
12157 if unspecified. See also the "check", "inter" and "fall" parameters.
12158
Baptiste Assmann8e2d9432018-06-22 15:04:43 +020012159resolve-opts <option>,<option>,...
12160 Comma separated list of options to apply to DNS resolution linked to this
12161 server.
12162
12163 Available options:
12164
12165 * allow-dup-ip
12166 By default, HAProxy prevents IP address duplication in a backend when DNS
12167 resolution at runtime is in operation.
12168 That said, for some cases, it makes sense that two servers (in the same
12169 backend, being resolved by the same FQDN) have the same IP address.
12170 For such case, simply enable this option.
12171 This is the opposite of prevent-dup-ip.
12172
Daniel Corbettf8716912019-11-17 09:48:56 -050012173 * ignore-weight
12174 Ignore any weight that is set within an SRV record. This is useful when
12175 you would like to control the weights using an alternate method, such as
12176 using an "agent-check" or through the runtime api.
12177
Baptiste Assmann8e2d9432018-06-22 15:04:43 +020012178 * prevent-dup-ip
12179 Ensure HAProxy's default behavior is enforced on a server: prevent re-using
12180 an IP address already set to a server in the same backend and sharing the
12181 same fqdn.
12182 This is the opposite of allow-dup-ip.
12183
12184 Example:
12185 backend b_myapp
12186 default-server init-addr none resolvers dns
12187 server s1 myapp.example.com:80 check resolve-opts allow-dup-ip
12188 server s2 myapp.example.com:81 check resolve-opts allow-dup-ip
12189
12190 With the option allow-dup-ip set:
12191 * if the nameserver returns a single IP address, then both servers will use
12192 it
12193 * If the nameserver returns 2 IP addresses, then each server will pick up a
12194 different address
12195
12196 Default value: not set
12197
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012198resolve-prefer <family>
12199 When DNS resolution is enabled for a server and multiple IP addresses from
12200 different families are returned, HAProxy will prefer using an IP address
12201 from the family mentioned in the "resolve-prefer" parameter.
12202 Available families: "ipv4" and "ipv6"
12203
Baptiste Assmannc4aabae2015-08-04 22:43:06 +020012204 Default value: ipv6
12205
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012206 Example:
12207
12208 server s1 app1.domain.com:80 resolvers mydns resolve-prefer ipv6
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012209
Thierry Fournierac88cfe2016-02-17 22:05:30 +010012210resolve-net <network>[,<network[,...]]
John Roeslerfb2fce12019-07-10 15:45:51 -050012211 This option prioritizes the choice of an ip address matching a network. This is
Thierry Fournierac88cfe2016-02-17 22:05:30 +010012212 useful with clouds to prefer a local ip. In some cases, a cloud high
Tim Düsterhus4896c442016-11-29 02:15:19 +010012213 availability service can be announced with many ip addresses on many
Davor Ocelice9ed2812017-12-25 17:49:28 +010012214 different datacenters. The latency between datacenter is not negligible, so
12215 this patch permits to prefer a local datacenter. If no address matches the
Thierry Fournierac88cfe2016-02-17 22:05:30 +010012216 configured network, another address is selected.
12217
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012218 Example:
12219
12220 server s1 app1.domain.com:80 resolvers mydns resolve-net 10.0.0.0/8
Thierry Fournierac88cfe2016-02-17 22:05:30 +010012221
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012222resolvers <id>
12223 Points to an existing "resolvers" section to resolve current server's
12224 hostname.
12225
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012226 Example:
12227
12228 server s1 app1.domain.com:80 check resolvers mydns
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012229
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012230 See also section 5.3
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012231
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010012232send-proxy
12233 The "send-proxy" parameter enforces use of the PROXY protocol over any
12234 connection established to this server. The PROXY protocol informs the other
12235 end about the layer 3/4 addresses of the incoming connection, so that it can
12236 know the client's address or the public address it accessed to, whatever the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010012237 upper layer protocol. For connections accepted by an "accept-proxy" or
12238 "accept-netscaler-cip" listener, the advertised address will be used. Only
12239 TCPv4 and TCPv6 address families are supported. Other families such as
12240 Unix sockets, will report an UNKNOWN family. Servers using this option can
12241 fully be chained to another instance of haproxy listening with an
12242 "accept-proxy" setting. This setting must not be used if the server isn't
12243 aware of the protocol. When health checks are sent to the server, the PROXY
12244 protocol is automatically used when this option is set, unless there is an
12245 explicit "port" or "addr" directive, in which case an explicit
12246 "check-send-proxy" directive would also be needed to use the PROXY protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010012247 See also the "no-send-proxy" option of this section and "accept-proxy" and
12248 "accept-netscaler-cip" option of the "bind" keyword.
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010012249
David Safb76832014-05-08 23:42:08 -040012250send-proxy-v2
12251 The "send-proxy-v2" parameter enforces use of the PROXY protocol version 2
12252 over any connection established to this server. The PROXY protocol informs
12253 the other end about the layer 3/4 addresses of the incoming connection, so
12254 that it can know the client's address or the public address it accessed to,
Emmanuel Hocdet404d9782017-10-24 10:55:14 +020012255 whatever the upper layer protocol. It also send ALPN information if an alpn
12256 have been negotiated. This setting must not be used if the server isn't aware
12257 of this version of the protocol. See also the "no-send-proxy-v2" option of
12258 this section and send-proxy" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040012259
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010012260proxy-v2-options <option>[,<option>]*
12261 The "proxy-v2-options" parameter add option to send in PROXY protocol version
12262 2 when "send-proxy-v2" is used. Options available are "ssl" (see also
Emmanuel Hocdetfa8d0f12018-02-01 15:53:52 +010012263 send-proxy-v2-ssl), "cert-cn" (see also "send-proxy-v2-ssl-cn"), "ssl-cipher":
12264 name of the used cipher, "cert-sig": signature algorithm of the used
Emmanuel Hocdet253c3b72018-02-01 18:29:59 +010012265 certificate, "cert-key": key algorithm of the used certificate), "authority":
12266 host name value passed by the client (only sni from a tls connection is
Emmanuel Hocdet4399c752018-02-05 15:26:43 +010012267 supported), "crc32c": checksum of the proxy protocol v2 header.
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010012268
David Safb76832014-05-08 23:42:08 -040012269send-proxy-v2-ssl
12270 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
12271 2 over any connection established to this server. The PROXY protocol informs
12272 the other end about the layer 3/4 addresses of the incoming connection, so
12273 that it can know the client's address or the public address it accessed to,
12274 whatever the upper layer protocol. In addition, the SSL information extension
12275 of the PROXY protocol is added to the PROXY protocol header. This setting
12276 must not be used if the server isn't aware of this version of the protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010012277 See also the "no-send-proxy-v2-ssl" option of this section and the
12278 "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040012279
12280send-proxy-v2-ssl-cn
12281 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
12282 2 over any connection established to this server. The PROXY protocol informs
12283 the other end about the layer 3/4 addresses of the incoming connection, so
12284 that it can know the client's address or the public address it accessed to,
12285 whatever the upper layer protocol. In addition, the SSL information extension
12286 of the PROXY protocol, along along with the Common Name from the subject of
12287 the client certificate (if any), is added to the PROXY protocol header. This
12288 setting must not be used if the server isn't aware of this version of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012289 protocol. See also the "no-send-proxy-v2-ssl-cn" option of this section and
12290 the "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040012291
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012292slowstart <start_time_in_ms>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012293 The "slowstart" parameter for a server accepts a value in milliseconds which
12294 indicates after how long a server which has just come back up will run at
12295 full speed. Just as with every other time-based parameter, it can be entered
12296 in any other explicit unit among { us, ms, s, m, h, d }. The speed grows
12297 linearly from 0 to 100% during this time. The limitation applies to two
12298 parameters :
12299
12300 - maxconn: the number of connections accepted by the server will grow from 1
12301 to 100% of the usual dynamic limit defined by (minconn,maxconn,fullconn).
12302
12303 - weight: when the backend uses a dynamic weighted algorithm, the weight
12304 grows linearly from 1 to 100%. In this case, the weight is updated at every
12305 health-check. For this reason, it is important that the "inter" parameter
12306 is smaller than the "slowstart", in order to maximize the number of steps.
12307
12308 The slowstart never applies when haproxy starts, otherwise it would cause
12309 trouble to running servers. It only applies when a server has been previously
12310 seen as failed.
12311
Willy Tarreau732eac42015-07-09 11:40:25 +020012312sni <expression>
12313 The "sni" parameter evaluates the sample fetch expression, converts it to a
12314 string and uses the result as the host name sent in the SNI TLS extension to
12315 the server. A typical use case is to send the SNI received from the client in
12316 a bridged HTTPS scenario, using the "ssl_fc_sni" sample fetch for the
Willy Tarreau2ab88672017-07-05 18:23:03 +020012317 expression, though alternatives such as req.hdr(host) can also make sense. If
12318 "verify required" is set (which is the recommended setting), the resulting
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012319 name will also be matched against the server certificate's names. See the
Jérôme Magninb36a6d22018-12-09 16:03:40 +010012320 "verify" directive for more details. If you want to set a SNI for health
12321 checks, see the "check-sni" directive for more details.
Willy Tarreau732eac42015-07-09 11:40:25 +020012322
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020012323source <addr>[:<pl>[-<ph>]] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +020012324source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020012325source <addr>[:<pl>[-<ph>]] [interface <name>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012326 The "source" parameter sets the source address which will be used when
12327 connecting to the server. It follows the exact same parameters and principle
12328 as the backend "source" keyword, except that it only applies to the server
12329 referencing it. Please consult the "source" keyword for details.
12330
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020012331 Additionally, the "source" statement on a server line allows one to specify a
12332 source port range by indicating the lower and higher bounds delimited by a
12333 dash ('-'). Some operating systems might require a valid IP address when a
12334 source port range is specified. It is permitted to have the same IP/range for
12335 several servers. Doing so makes it possible to bypass the maximum of 64k
12336 total concurrent connections. The limit will then reach 64k connections per
12337 server.
12338
Lukas Tribus7d56c6d2016-09-13 09:51:15 +000012339 Since Linux 4.2/libc 2.23 IP_BIND_ADDRESS_NO_PORT is set for connections
12340 specifying the source address without port(s).
12341
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012342ssl
Willy Tarreau44f65392013-06-25 07:56:20 +020012343 This option enables SSL ciphering on outgoing connections to the server. It
12344 is critical to verify server certificates using "verify" when using SSL to
12345 connect to servers, otherwise the communication is prone to trivial man in
12346 the-middle attacks rendering SSL useless. When this option is used, health
12347 checks are automatically sent in SSL too unless there is a "port" or an
12348 "addr" directive indicating the check should be sent to a different location.
Frédéric Lécailled2376272017-03-21 18:52:12 +010012349 See the "no-ssl" to disable "ssl" option and "check-ssl" option to force
12350 SSL health checks.
Willy Tarreau763a95b2012-10-04 23:15:39 +020012351
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012352ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
12353 This option enforces use of <version> or lower when SSL is used to communicate
12354 with the server. This option is also available on global statement
12355 "ssl-default-server-options". See also "ssl-min-ver".
12356
12357ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
12358 This option enforces use of <version> or upper when SSL is used to communicate
12359 with the server. This option is also available on global statement
12360 "ssl-default-server-options". See also "ssl-max-ver".
12361
Frédéric Lécailled2376272017-03-21 18:52:12 +010012362ssl-reuse
12363 This option may be used as "server" setting to reset any "no-ssl-reuse"
12364 setting which would have been inherited from "default-server" directive as
12365 default value.
12366 It may also be used as "default-server" setting to reset any previous
12367 "default-server" "no-ssl-reuse" setting.
12368
12369stick
12370 This option may be used as "server" setting to reset any "non-stick"
12371 setting which would have been inherited from "default-server" directive as
12372 default value.
12373 It may also be used as "default-server" setting to reset any previous
12374 "default-server" "non-stick" setting.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012375
Alexander Liu2a54bb72019-05-22 19:44:48 +080012376socks4 <addr>:<port>
John Roeslerfb2fce12019-07-10 15:45:51 -050012377 This option enables upstream socks4 tunnel for outgoing connections to the
Alexander Liu2a54bb72019-05-22 19:44:48 +080012378 server. Using this option won't force the health check to go via socks4 by
12379 default. You will have to use the keyword "check-via-socks4" to enable it.
12380
Willy Tarreau163d4622015-10-13 16:16:41 +020012381tcp-ut <delay>
12382 Sets the TCP User Timeout for all outgoing connections to this server. This
12383 option is available on Linux since version 2.6.37. It allows haproxy to
12384 configure a timeout for sockets which contain data not receiving an
Davor Ocelice9ed2812017-12-25 17:49:28 +010012385 acknowledgment for the configured delay. This is especially useful on
Willy Tarreau163d4622015-10-13 16:16:41 +020012386 long-lived connections experiencing long idle periods such as remote
12387 terminals or database connection pools, where the client and server timeouts
12388 must remain high to allow a long period of idle, but where it is important to
12389 detect that the server has disappeared in order to release all resources
12390 associated with its connection (and the client's session). One typical use
12391 case is also to force dead server connections to die when health checks are
12392 too slow or during a soft reload since health checks are then disabled. The
12393 argument is a delay expressed in milliseconds by default. This only works for
12394 regular TCP connections, and is ignored for other protocols.
12395
Willy Tarreau034c88c2017-01-23 23:36:45 +010012396tfo
12397 This option enables using TCP fast open when connecting to servers, on
12398 systems that support it (currently only the Linux kernel >= 4.11).
12399 See the "tfo" bind option for more information about TCP fast open.
12400 Please note that when using tfo, you should also use the "conn-failure",
12401 "empty-response" and "response-timeout" keywords for "retry-on", or haproxy
Frédéric Lécaille1b9423d2019-07-04 14:19:06 +020012402 won't be able to retry the connection on failure. See also "no-tfo".
Willy Tarreau034c88c2017-01-23 23:36:45 +010012403
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012404track [<proxy>/]<server>
Willy Tarreau32091232014-05-16 13:52:00 +020012405 This option enables ability to set the current state of the server by tracking
12406 another one. It is possible to track a server which itself tracks another
12407 server, provided that at the end of the chain, a server has health checks
12408 enabled. If <proxy> is omitted the current one is used. If disable-on-404 is
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012409 used, it has to be enabled on both proxies.
12410
Frédéric Lécailled2376272017-03-21 18:52:12 +010012411tls-tickets
12412 This option may be used as "server" setting to reset any "no-tls-tickets"
12413 setting which would have been inherited from "default-server" directive as
12414 default value.
12415 It may also be used as "default-server" setting to reset any previous
12416 "default-server" "no-tlsv-tickets" setting.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012417
Emeric Brunef42d922012-10-11 16:11:36 +020012418verify [none|required]
12419 This setting is only available when support for OpenSSL was built in. If set
Emeric Brun850efd52014-01-29 12:24:34 +010012420 to 'none', server certificate is not verified. In the other case, The
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012421 certificate provided by the server is verified using CAs from 'ca-file' and
12422 optional CRLs from 'crl-file' after having checked that the names provided in
Davor Ocelice9ed2812017-12-25 17:49:28 +010012423 the certificate's subject and subjectAlternateNames attributes match either
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012424 the name passed using the "sni" directive, or if not provided, the static
12425 host name passed using the "verifyhost" directive. When no name is found, the
12426 certificate's names are ignored. For this reason, without SNI it's important
12427 to use "verifyhost". On verification failure the handshake is aborted. It is
12428 critically important to verify server certificates when using SSL to connect
12429 to servers, otherwise the communication is prone to trivial man-in-the-middle
12430 attacks rendering SSL totally useless. Unless "ssl_server_verify" appears in
12431 the global section, "verify" is set to "required" by default.
Emeric Brunef42d922012-10-11 16:11:36 +020012432
Evan Broderbe554312013-06-27 00:05:25 -070012433verifyhost <hostname>
12434 This setting is only available when support for OpenSSL was built in, and
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012435 only takes effect if 'verify required' is also specified. This directive sets
12436 a default static hostname to check the server's certificate against when no
12437 SNI was used to connect to the server. If SNI is not used, this is the only
12438 way to enable hostname verification. This static hostname, when set, will
12439 also be used for health checks (which cannot provide an SNI value). If none
12440 of the hostnames in the certificate match the specified hostname, the
12441 handshake is aborted. The hostnames in the server-provided certificate may
12442 include wildcards. See also "verify", "sni" and "no-verifyhost" options.
Evan Broderbe554312013-06-27 00:05:25 -070012443
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012444weight <weight>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012445 The "weight" parameter is used to adjust the server's weight relative to
12446 other servers. All servers will receive a load proportional to their weight
12447 relative to the sum of all weights, so the higher the weight, the higher the
Willy Tarreau6704d672009-06-15 10:56:05 +020012448 load. The default weight is 1, and the maximal value is 256. A value of 0
12449 means the server will not participate in load-balancing but will still accept
12450 persistent connections. If this parameter is used to distribute the load
12451 according to server's capacity, it is recommended to start with values which
12452 can both grow and shrink, for instance between 10 and 100 to leave enough
12453 room above and below for later adjustments.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012454
12455
Cyril Bonté46175dd2015-07-02 22:45:32 +0200124565.3. Server IP address resolution using DNS
12457-------------------------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012458
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012459HAProxy allows using a host name on the server line to retrieve its IP address
12460using name servers. By default, HAProxy resolves the name when parsing the
12461configuration file, at startup and cache the result for the process' life.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012462This is not sufficient in some cases, such as in Amazon where a server's IP
12463can change after a reboot or an ELB Virtual IP can change based on current
12464workload.
12465This chapter describes how HAProxy can be configured to process server's name
12466resolution at run time.
12467Whether run time server name resolution has been enable or not, HAProxy will
12468carry on doing the first resolution when parsing the configuration.
12469
12470
Cyril Bonté46175dd2015-07-02 22:45:32 +0200124715.3.1. Global overview
12472----------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012473
12474As we've seen in introduction, name resolution in HAProxy occurs at two
12475different steps of the process life:
12476
12477 1. when starting up, HAProxy parses the server line definition and matches a
12478 host name. It uses libc functions to get the host name resolved. This
12479 resolution relies on /etc/resolv.conf file.
12480
Christopher Faulet67957bd2017-09-27 11:00:59 +020012481 2. at run time, HAProxy performs periodically name resolutions for servers
12482 requiring DNS resolutions.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012483
12484A few other events can trigger a name resolution at run time:
12485 - when a server's health check ends up in a connection timeout: this may be
12486 because the server has a new IP address. So we need to trigger a name
12487 resolution to know this new IP.
12488
Christopher Faulet67957bd2017-09-27 11:00:59 +020012489When using resolvers, the server name can either be a hostname, or a SRV label.
Davor Ocelice9ed2812017-12-25 17:49:28 +010012490HAProxy considers anything that starts with an underscore as a SRV label. If a
Christopher Faulet67957bd2017-09-27 11:00:59 +020012491SRV label is specified, then the corresponding SRV records will be retrieved
12492from the DNS server, and the provided hostnames will be used. The SRV label
12493will be checked periodically, and if any server are added or removed, haproxy
12494will automatically do the same.
Olivier Houchardecfa18d2017-08-07 17:30:03 +020012495
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012496A few things important to notice:
John Roeslerfb2fce12019-07-10 15:45:51 -050012497 - all the name servers are queried in the meantime. HAProxy will process the
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012498 first valid response.
12499
12500 - a resolution is considered as invalid (NX, timeout, refused), when all the
12501 servers return an error.
12502
12503
Cyril Bonté46175dd2015-07-02 22:45:32 +0200125045.3.2. The resolvers section
12505----------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012506
12507This section is dedicated to host information related to name resolution in
Christopher Faulet67957bd2017-09-27 11:00:59 +020012508HAProxy. There can be as many as resolvers section as needed. Each section can
12509contain many name servers.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012510
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012511When multiple name servers are configured in a resolvers section, then HAProxy
12512uses the first valid response. In case of invalid responses, only the last one
12513is treated. Purpose is to give the chance to a slow server to deliver a valid
12514answer after a fast faulty or outdated server.
12515
12516When each server returns a different error type, then only the last error is
Christopher Faulet67957bd2017-09-27 11:00:59 +020012517used by HAProxy. The following processing is applied on this error:
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012518
Christopher Faulet67957bd2017-09-27 11:00:59 +020012519 1. HAProxy retries the same DNS query with a new query type. The A queries are
12520 switch to AAAA or the opposite. SRV queries are not concerned here. Timeout
12521 errors are also excluded.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012522
Christopher Faulet67957bd2017-09-27 11:00:59 +020012523 2. When the fallback on the query type was done (or not applicable), HAProxy
12524 retries the original DNS query, with the preferred query type.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012525
Christopher Faulet67957bd2017-09-27 11:00:59 +020012526 3. HAProxy retries previous steps <resolve_retires> times. If no valid
12527 response is received after that, it stops the DNS resolution and reports
12528 the error.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012529
Christopher Faulet67957bd2017-09-27 11:00:59 +020012530For example, with 2 name servers configured in a resolvers section, the
12531following scenarios are possible:
12532
12533 - First response is valid and is applied directly, second response is
12534 ignored
12535
12536 - First response is invalid and second one is valid, then second response is
12537 applied
12538
12539 - First response is a NX domain and second one a truncated response, then
12540 HAProxy retries the query with a new type
12541
12542 - First response is a NX domain and second one is a timeout, then HAProxy
12543 retries the query with a new type
12544
12545 - Query timed out for both name servers, then HAProxy retries it with the
12546 same query type
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012547
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012548As a DNS server may not answer all the IPs in one DNS request, haproxy keeps
12549a cache of previous answers, an answer will be considered obsolete after
Christopher Faulet67957bd2017-09-27 11:00:59 +020012550<hold obsolete> seconds without the IP returned.
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012551
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012552
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012553resolvers <resolvers id>
Davor Ocelice9ed2812017-12-25 17:49:28 +010012554 Creates a new name server list labeled <resolvers id>
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012555
12556A resolvers section accept the following parameters:
12557
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020012558accepted_payload_size <nb>
Davor Ocelice9ed2812017-12-25 17:49:28 +010012559 Defines the maximum payload size accepted by HAProxy and announced to all the
Christopher Faulet67957bd2017-09-27 11:00:59 +020012560 name servers configured in this resolvers section.
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020012561 <nb> is in bytes. If not set, HAProxy announces 512. (minimal value defined
12562 by RFC 6891)
12563
Baptiste Assmann9d8dbbc2017-08-18 23:35:08 +020012564 Note: the maximum allowed value is 8192.
12565
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012566nameserver <id> <ip>:<port>
12567 DNS server description:
12568 <id> : label of the server, should be unique
12569 <ip> : IP address of the server
12570 <port> : port where the DNS service actually runs
12571
Ben Draut44e609b2018-05-29 15:40:08 -060012572parse-resolv-conf
12573 Adds all nameservers found in /etc/resolv.conf to this resolvers nameservers
12574 list. Ordered as if each nameserver in /etc/resolv.conf was individually
12575 placed in the resolvers section in place of this directive.
12576
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012577hold <status> <period>
12578 Defines <period> during which the last name resolution should be kept based
12579 on last resolution <status>
Baptiste Assmann987e16d2016-11-02 22:23:31 +010012580 <status> : last name resolution status. Acceptable values are "nx",
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012581 "other", "refused", "timeout", "valid", "obsolete".
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012582 <period> : interval between two successive name resolution when the last
12583 answer was in <status>. It follows the HAProxy time format.
12584 <period> is in milliseconds by default.
12585
Baptiste Assmann686408b2017-08-18 10:15:42 +020012586 Default value is 10s for "valid", 0s for "obsolete" and 30s for others.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012587
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012588resolve_retries <nb>
12589 Defines the number <nb> of queries to send to resolve a server name before
12590 giving up.
12591 Default value: 3
12592
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012593 A retry occurs on name server timeout or when the full sequence of DNS query
12594 type failover is over and we need to start up from the default ANY query
12595 type.
12596
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012597timeout <event> <time>
12598 Defines timeouts related to name resolution
12599 <event> : the event on which the <time> timeout period applies to.
12600 events available are:
Frédéric Lécaille93d33162019-03-06 09:35:59 +010012601 - resolve : default time to trigger name resolutions when no
12602 other time applied.
Christopher Faulet67957bd2017-09-27 11:00:59 +020012603 Default value: 1s
12604 - retry : time between two DNS queries, when no valid response
Frédéric Lécaille93d33162019-03-06 09:35:59 +010012605 have been received.
Christopher Faulet67957bd2017-09-27 11:00:59 +020012606 Default value: 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012607 <time> : time related to the event. It follows the HAProxy time format.
12608 <time> is expressed in milliseconds.
12609
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012610 Example:
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012611
12612 resolvers mydns
12613 nameserver dns1 10.0.0.1:53
12614 nameserver dns2 10.0.0.2:53
Ben Draut44e609b2018-05-29 15:40:08 -060012615 parse-resolv-conf
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012616 resolve_retries 3
Christopher Faulet67957bd2017-09-27 11:00:59 +020012617 timeout resolve 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012618 timeout retry 1s
Baptiste Assmann987e16d2016-11-02 22:23:31 +010012619 hold other 30s
12620 hold refused 30s
12621 hold nx 30s
12622 hold timeout 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012623 hold valid 10s
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012624 hold obsolete 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012625
12626
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200126276. Cache
12628---------
12629
12630HAProxy provides a cache, which was designed to perform cache on small objects
12631(favicon, css...). This is a minimalist low-maintenance cache which runs in
12632RAM.
12633
12634The cache is based on a memory which is shared between processes and threads,
12635this memory is split in blocks of 1k.
12636
12637If an object is not used anymore, it can be deleted to store a new object
12638independently of its expiration date. The oldest objects are deleted first
12639when we try to allocate a new one.
12640
12641The cache uses a hash of the host header and the URI as the key.
12642
12643It's possible to view the status of a cache using the Unix socket command
12644"show cache" consult section 9.3 "Unix Socket commands" of Management Guide
12645for more details.
12646
12647When an object is delivered from the cache, the server name in the log is
12648replaced by "<CACHE>".
12649
12650
126516.1. Limitation
12652----------------
12653
12654The cache won't store and won't deliver objects in these cases:
12655
12656- If the response is not a 200
12657- If the response contains a Vary header
12658- If the Content-Length + the headers size is greater than "max-object-size"
12659- If the response is not cacheable
12660
12661- If the request is not a GET
12662- If the HTTP version of the request is smaller than 1.1
12663- If the request contains an Authorization header
12664
12665
126666.2. Setup
12667-----------
12668
12669To setup a cache, you must define a cache section and use it in a proxy with
12670the corresponding http-request and response actions.
12671
12672
126736.2.1. Cache section
12674---------------------
12675
12676cache <name>
12677 Declare a cache section, allocate a shared cache memory named <name>, the
12678 size of cache is mandatory.
12679
12680total-max-size <megabytes>
12681 Define the size in RAM of the cache in megabytes. This size is split in
12682 blocks of 1kB which are used by the cache entries. Its maximum value is 4095.
12683
12684max-object-size <bytes>
12685 Define the maximum size of the objects to be cached. Must not be greater than
12686 an half of "total-max-size". If not set, it equals to a 256th of the cache size.
12687 All objects with sizes larger than "max-object-size" will not be cached.
12688
12689max-age <seconds>
12690 Define the maximum expiration duration. The expiration is set has the lowest
12691 value between the s-maxage or max-age (in this order) directive in the
12692 Cache-Control response header and this value. The default value is 60
12693 seconds, which means that you can't cache an object more than 60 seconds by
12694 default.
12695
12696
126976.2.2. Proxy section
12698---------------------
12699
12700http-request cache-use <name> [ { if | unless } <condition> ]
12701 Try to deliver a cached object from the cache <name>. This directive is also
12702 mandatory to store the cache as it calculates the cache hash. If you want to
12703 use a condition for both storage and delivering that's a good idea to put it
12704 after this one.
12705
12706http-response cache-store <name> [ { if | unless } <condition> ]
12707 Store an http-response within the cache. The storage of the response headers
12708 is done at this step, which means you can use others http-response actions
12709 to modify headers before or after the storage of the response. This action
12710 is responsible for the setup of the cache storage filter.
12711
12712
12713Example:
12714
12715 backend bck1
12716 mode http
12717
12718 http-request cache-use foobar
12719 http-response cache-store foobar
12720 server srv1 127.0.0.1:80
12721
12722 cache foobar
12723 total-max-size 4
12724 max-age 240
12725
12726
Willy Tarreau74ca5042013-06-11 23:12:07 +0200127277. Using ACLs and fetching samples
12728----------------------------------
12729
Davor Ocelice9ed2812017-12-25 17:49:28 +010012730HAProxy is capable of extracting data from request or response streams, from
Willy Tarreau74ca5042013-06-11 23:12:07 +020012731client or server information, from tables, environmental information etc...
12732The action of extracting such data is called fetching a sample. Once retrieved,
12733these samples may be used for various purposes such as a key to a stick-table,
12734but most common usages consist in matching them against predefined constant
12735data called patterns.
12736
12737
127387.1. ACL basics
12739---------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012740
12741The use of Access Control Lists (ACL) provides a flexible solution to perform
12742content switching and generally to take decisions based on content extracted
12743from the request, the response or any environmental status. The principle is
12744simple :
12745
Willy Tarreau74ca5042013-06-11 23:12:07 +020012746 - extract a data sample from a stream, table or the environment
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012747 - optionally apply some format conversion to the extracted sample
Willy Tarreau74ca5042013-06-11 23:12:07 +020012748 - apply one or multiple pattern matching methods on this sample
12749 - perform actions only when a pattern matches the sample
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012750
Willy Tarreau74ca5042013-06-11 23:12:07 +020012751The actions generally consist in blocking a request, selecting a backend, or
12752adding a header.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012753
12754In order to define a test, the "acl" keyword is used. The syntax is :
12755
Willy Tarreau74ca5042013-06-11 23:12:07 +020012756 acl <aclname> <criterion> [flags] [operator] [<value>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012757
12758This creates a new ACL <aclname> or completes an existing one with new tests.
12759Those tests apply to the portion of request/response specified in <criterion>
12760and may be adjusted with optional flags [flags]. Some criteria also support
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012761an operator which may be specified before the set of values. Optionally some
12762conversion operators may be applied to the sample, and they will be specified
12763as a comma-delimited list of keywords just after the first keyword. The values
12764are of the type supported by the criterion, and are separated by spaces.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012765
12766ACL names must be formed from upper and lower case letters, digits, '-' (dash),
12767'_' (underscore) , '.' (dot) and ':' (colon). ACL names are case-sensitive,
12768which means that "my_acl" and "My_Acl" are two different ACLs.
12769
12770There is no enforced limit to the number of ACLs. The unused ones do not affect
12771performance, they just consume a small amount of memory.
12772
Willy Tarreau74ca5042013-06-11 23:12:07 +020012773The criterion generally is the name of a sample fetch method, or one of its ACL
12774specific declinations. The default test method is implied by the output type of
12775this sample fetch method. The ACL declinations can describe alternate matching
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012776methods of a same sample fetch method. The sample fetch methods are the only
12777ones supporting a conversion.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012778
12779Sample fetch methods return data which can be of the following types :
12780 - boolean
12781 - integer (signed or unsigned)
12782 - IPv4 or IPv6 address
12783 - string
12784 - data block
12785
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012786Converters transform any of these data into any of these. For example, some
12787converters might convert a string to a lower-case string while other ones
12788would turn a string to an IPv4 address, or apply a netmask to an IP address.
12789The resulting sample is of the type of the last converter applied to the list,
12790which defaults to the type of the sample fetch method.
12791
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012792Each sample or converter returns data of a specific type, specified with its
12793keyword in this documentation. When an ACL is declared using a standard sample
12794fetch method, certain types automatically involved a default matching method
12795which are summarized in the table below :
12796
12797 +---------------------+-----------------+
12798 | Sample or converter | Default |
12799 | output type | matching method |
12800 +---------------------+-----------------+
12801 | boolean | bool |
12802 +---------------------+-----------------+
12803 | integer | int |
12804 +---------------------+-----------------+
12805 | ip | ip |
12806 +---------------------+-----------------+
12807 | string | str |
12808 +---------------------+-----------------+
12809 | binary | none, use "-m" |
12810 +---------------------+-----------------+
12811
12812Note that in order to match a binary samples, it is mandatory to specify a
12813matching method, see below.
12814
Willy Tarreau74ca5042013-06-11 23:12:07 +020012815The ACL engine can match these types against patterns of the following types :
12816 - boolean
12817 - integer or integer range
12818 - IP address / network
12819 - string (exact, substring, suffix, prefix, subdir, domain)
12820 - regular expression
12821 - hex block
12822
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012823The following ACL flags are currently supported :
12824
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012825 -i : ignore case during matching of all subsequent patterns.
12826 -f : load patterns from a file.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012827 -m : use a specific pattern matching method
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010012828 -n : forbid the DNS resolutions
Thierry FOURNIER9860c412014-01-29 14:23:29 +010012829 -M : load the file pointed by -f like a map file.
Thierry FOURNIER3534d882014-01-20 17:01:44 +010012830 -u : force the unique id of the ACL
Willy Tarreau6a06a402007-07-15 20:15:28 +020012831 -- : force end of flags. Useful when a string looks like one of the flags.
12832
Willy Tarreau74ca5042013-06-11 23:12:07 +020012833The "-f" flag is followed by the name of a file from which all lines will be
12834read as individual values. It is even possible to pass multiple "-f" arguments
12835if the patterns are to be loaded from multiple files. Empty lines as well as
12836lines beginning with a sharp ('#') will be ignored. All leading spaces and tabs
12837will be stripped. If it is absolutely necessary to insert a valid pattern
12838beginning with a sharp, just prefix it with a space so that it is not taken for
12839a comment. Depending on the data type and match method, haproxy may load the
12840lines into a binary tree, allowing very fast lookups. This is true for IPv4 and
12841exact string matching. In this case, duplicates will automatically be removed.
12842
Thierry FOURNIER9860c412014-01-29 14:23:29 +010012843The "-M" flag allows an ACL to use a map file. If this flag is set, the file is
12844parsed as two column file. The first column contains the patterns used by the
12845ACL, and the second column contain the samples. The sample can be used later by
12846a map. This can be useful in some rare cases where an ACL would just be used to
12847check for the existence of a pattern in a map before a mapping is applied.
12848
Thierry FOURNIER3534d882014-01-20 17:01:44 +010012849The "-u" flag forces the unique id of the ACL. This unique id is used with the
12850socket interface to identify ACL and dynamically change its values. Note that a
12851file is always identified by its name even if an id is set.
12852
Willy Tarreau74ca5042013-06-11 23:12:07 +020012853Also, note that the "-i" flag applies to subsequent entries and not to entries
12854loaded from files preceding it. For instance :
12855
12856 acl valid-ua hdr(user-agent) -f exact-ua.lst -i -f generic-ua.lst test
12857
12858In this example, each line of "exact-ua.lst" will be exactly matched against
12859the "user-agent" header of the request. Then each line of "generic-ua" will be
12860case-insensitively matched. Then the word "test" will be insensitively matched
12861as well.
12862
12863The "-m" flag is used to select a specific pattern matching method on the input
12864sample. All ACL-specific criteria imply a pattern matching method and generally
12865do not need this flag. However, this flag is useful with generic sample fetch
12866methods to describe how they're going to be matched against the patterns. This
12867is required for sample fetches which return data type for which there is no
Davor Ocelice9ed2812017-12-25 17:49:28 +010012868obvious matching method (e.g. string or binary). When "-m" is specified and
Willy Tarreau74ca5042013-06-11 23:12:07 +020012869followed by a pattern matching method name, this method is used instead of the
12870default one for the criterion. This makes it possible to match contents in ways
12871that were not initially planned, or with sample fetch methods which return a
12872string. The matching method also affects the way the patterns are parsed.
12873
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010012874The "-n" flag forbids the dns resolutions. It is used with the load of ip files.
12875By default, if the parser cannot parse ip address it considers that the parsed
12876string is maybe a domain name and try dns resolution. The flag "-n" disable this
12877resolution. It is useful for detecting malformed ip lists. Note that if the DNS
12878server is not reachable, the haproxy configuration parsing may last many minutes
John Roeslerfb2fce12019-07-10 15:45:51 -050012879waiting for the timeout. During this time no error messages are displayed. The
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010012880flag "-n" disable this behavior. Note also that during the runtime, this
12881function is disabled for the dynamic acl modifications.
12882
Willy Tarreau74ca5042013-06-11 23:12:07 +020012883There are some restrictions however. Not all methods can be used with all
12884sample fetch methods. Also, if "-m" is used in conjunction with "-f", it must
12885be placed first. The pattern matching method must be one of the following :
Willy Tarreau5adeda12013-03-31 22:13:34 +020012886
12887 - "found" : only check if the requested sample could be found in the stream,
12888 but do not compare it against any pattern. It is recommended not
Willy Tarreau74ca5042013-06-11 23:12:07 +020012889 to pass any pattern to avoid confusion. This matching method is
12890 particularly useful to detect presence of certain contents such
12891 as headers, cookies, etc... even if they are empty and without
12892 comparing them to anything nor counting them.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012893
12894 - "bool" : check the value as a boolean. It can only be applied to fetches
12895 which return a boolean or integer value, and takes no pattern.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012896 Value zero or false does not match, all other values do match.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012897
12898 - "int" : match the value as an integer. It can be used with integer and
Willy Tarreau74ca5042013-06-11 23:12:07 +020012899 boolean samples. Boolean false is integer 0, true is integer 1.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012900
12901 - "ip" : match the value as an IPv4 or IPv6 address. It is compatible
Willy Tarreau74ca5042013-06-11 23:12:07 +020012902 with IP address samples only, so it is implied and never needed.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012903
Davor Ocelice9ed2812017-12-25 17:49:28 +010012904 - "bin" : match the contents against a hexadecimal string representing a
Willy Tarreau5adeda12013-03-31 22:13:34 +020012905 binary sequence. This may be used with binary or string samples.
12906
12907 - "len" : match the sample's length as an integer. This may be used with
12908 binary or string samples.
12909
Willy Tarreau74ca5042013-06-11 23:12:07 +020012910 - "str" : exact match : match the contents against a string. This may be
12911 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012912
Willy Tarreau74ca5042013-06-11 23:12:07 +020012913 - "sub" : substring match : check that the contents contain at least one of
12914 the provided string patterns. This may be used with binary or
12915 string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012916
Willy Tarreau74ca5042013-06-11 23:12:07 +020012917 - "reg" : regex match : match the contents against a list of regular
12918 expressions. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012919
Willy Tarreau74ca5042013-06-11 23:12:07 +020012920 - "beg" : prefix match : check that the contents begin like the provided
12921 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012922
Willy Tarreau74ca5042013-06-11 23:12:07 +020012923 - "end" : suffix match : check that the contents end like the provided
12924 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012925
Willy Tarreau74ca5042013-06-11 23:12:07 +020012926 - "dir" : subdir match : check that a slash-delimited portion of the
12927 contents exactly matches one of the provided string patterns.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012928 This may be used with binary or string samples.
12929
Willy Tarreau74ca5042013-06-11 23:12:07 +020012930 - "dom" : domain match : check that a dot-delimited portion of the contents
12931 exactly match one of the provided string patterns. This may be
12932 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012933
12934For example, to quickly detect the presence of cookie "JSESSIONID" in an HTTP
12935request, it is possible to do :
12936
12937 acl jsess_present cook(JSESSIONID) -m found
12938
12939In order to apply a regular expression on the 500 first bytes of data in the
12940buffer, one would use the following acl :
12941
12942 acl script_tag payload(0,500) -m reg -i <script>
12943
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012944On systems where the regex library is much slower when using "-i", it is
12945possible to convert the sample to lowercase before matching, like this :
12946
12947 acl script_tag payload(0,500),lower -m reg <script>
12948
Willy Tarreau74ca5042013-06-11 23:12:07 +020012949All ACL-specific criteria imply a default matching method. Most often, these
12950criteria are composed by concatenating the name of the original sample fetch
12951method and the matching method. For example, "hdr_beg" applies the "beg" match
12952to samples retrieved using the "hdr" fetch method. Since all ACL-specific
12953criteria rely on a sample fetch method, it is always possible instead to use
12954the original sample fetch method and the explicit matching method using "-m".
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012955
Willy Tarreau74ca5042013-06-11 23:12:07 +020012956If an alternate match is specified using "-m" on an ACL-specific criterion,
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012957the matching method is simply applied to the underlying sample fetch method.
12958For example, all ACLs below are exact equivalent :
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012959
Willy Tarreau74ca5042013-06-11 23:12:07 +020012960 acl short_form hdr_beg(host) www.
12961 acl alternate1 hdr_beg(host) -m beg www.
12962 acl alternate2 hdr_dom(host) -m beg www.
12963 acl alternate3 hdr(host) -m beg www.
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012964
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012965
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012966The table below summarizes the compatibility matrix between sample or converter
12967types and the pattern types to fetch against. It indicates for each compatible
12968combination the name of the matching method to be used, surrounded with angle
12969brackets ">" and "<" when the method is the default one and will work by
12970default without "-m".
Willy Tarreau0ba27502007-12-24 16:55:16 +010012971
Willy Tarreau74ca5042013-06-11 23:12:07 +020012972 +-------------------------------------------------+
12973 | Input sample type |
12974 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012975 | pattern type | boolean | integer | ip | string | binary |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012976 +----------------------+---------+---------+---------+---------+---------+
12977 | none (presence only) | found | found | found | found | found |
12978 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012979 | none (boolean value) |> bool <| bool | | bool | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012980 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012981 | integer (value) | int |> int <| int | int | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012982 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012983 | integer (length) | len | len | len | len | len |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012984 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012985 | IP address | | |> ip <| ip | ip |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012986 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012987 | exact string | str | str | str |> str <| str |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012988 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012989 | prefix | beg | beg | beg | beg | beg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012990 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012991 | suffix | end | end | end | end | end |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012992 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012993 | substring | sub | sub | sub | sub | sub |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012994 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012995 | subdir | dir | dir | dir | dir | dir |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012996 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012997 | domain | dom | dom | dom | dom | dom |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012998 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012999 | regex | reg | reg | reg | reg | reg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013000 +----------------------+---------+---------+---------+---------+---------+
13001 | hex block | | | | bin | bin |
13002 +----------------------+---------+---------+---------+---------+---------+
Willy Tarreau6a06a402007-07-15 20:15:28 +020013003
13004
Willy Tarreau74ca5042013-06-11 23:12:07 +0200130057.1.1. Matching booleans
13006------------------------
13007
13008In order to match a boolean, no value is needed and all values are ignored.
13009Boolean matching is used by default for all fetch methods of type "boolean".
13010When boolean matching is used, the fetched value is returned as-is, which means
13011that a boolean "true" will always match and a boolean "false" will never match.
13012
13013Boolean matching may also be enforced using "-m bool" on fetch methods which
13014return an integer value. Then, integer value 0 is converted to the boolean
13015"false" and all other values are converted to "true".
13016
Willy Tarreau6a06a402007-07-15 20:15:28 +020013017
Willy Tarreau74ca5042013-06-11 23:12:07 +0200130187.1.2. Matching integers
13019------------------------
13020
13021Integer matching applies by default to integer fetch methods. It can also be
13022enforced on boolean fetches using "-m int". In this case, "false" is converted
13023to the integer 0, and "true" is converted to the integer 1.
13024
13025Integer matching also supports integer ranges and operators. Note that integer
13026matching only applies to positive values. A range is a value expressed with a
13027lower and an upper bound separated with a colon, both of which may be omitted.
Willy Tarreau6a06a402007-07-15 20:15:28 +020013028
13029For instance, "1024:65535" is a valid range to represent a range of
13030unprivileged ports, and "1024:" would also work. "0:1023" is a valid
13031representation of privileged ports, and ":1023" would also work.
13032
Willy Tarreau62644772008-07-16 18:36:06 +020013033As a special case, some ACL functions support decimal numbers which are in fact
13034two integers separated by a dot. This is used with some version checks for
13035instance. All integer properties apply to those decimal numbers, including
13036ranges and operators.
13037
Willy Tarreau6a06a402007-07-15 20:15:28 +020013038For an easier usage, comparison operators are also supported. Note that using
Willy Tarreau0ba27502007-12-24 16:55:16 +010013039operators with ranges does not make much sense and is strongly discouraged.
13040Similarly, it does not make much sense to perform order comparisons with a set
13041of values.
Willy Tarreau6a06a402007-07-15 20:15:28 +020013042
Willy Tarreau0ba27502007-12-24 16:55:16 +010013043Available operators for integer matching are :
Willy Tarreau6a06a402007-07-15 20:15:28 +020013044
13045 eq : true if the tested value equals at least one value
13046 ge : true if the tested value is greater than or equal to at least one value
13047 gt : true if the tested value is greater than at least one value
13048 le : true if the tested value is less than or equal to at least one value
13049 lt : true if the tested value is less than at least one value
13050
Willy Tarreau0ba27502007-12-24 16:55:16 +010013051For instance, the following ACL matches any negative Content-Length header :
Willy Tarreau6a06a402007-07-15 20:15:28 +020013052
13053 acl negative-length hdr_val(content-length) lt 0
13054
Willy Tarreau62644772008-07-16 18:36:06 +020013055This one matches SSL versions between 3.0 and 3.1 (inclusive) :
13056
13057 acl sslv3 req_ssl_ver 3:3.1
13058
Willy Tarreau6a06a402007-07-15 20:15:28 +020013059
Willy Tarreau74ca5042013-06-11 23:12:07 +0200130607.1.3. Matching strings
13061-----------------------
13062
13063String matching applies to string or binary fetch methods, and exists in 6
13064different forms :
13065
13066 - exact match (-m str) : the extracted string must exactly match the
Davor Ocelice9ed2812017-12-25 17:49:28 +010013067 patterns;
Willy Tarreau74ca5042013-06-11 23:12:07 +020013068
13069 - substring match (-m sub) : the patterns are looked up inside the
Davor Ocelice9ed2812017-12-25 17:49:28 +010013070 extracted string, and the ACL matches if any of them is found inside;
Willy Tarreau74ca5042013-06-11 23:12:07 +020013071
13072 - prefix match (-m beg) : the patterns are compared with the beginning of
13073 the extracted string, and the ACL matches if any of them matches.
13074
13075 - suffix match (-m end) : the patterns are compared with the end of the
13076 extracted string, and the ACL matches if any of them matches.
13077
Baptiste Assmann33db6002016-03-06 23:32:10 +010013078 - subdir match (-m dir) : the patterns are looked up inside the extracted
Willy Tarreau74ca5042013-06-11 23:12:07 +020013079 string, delimited with slashes ("/"), and the ACL matches if any of them
13080 matches.
13081
13082 - domain match (-m dom) : the patterns are looked up inside the extracted
13083 string, delimited with dots ("."), and the ACL matches if any of them
13084 matches.
Willy Tarreau6a06a402007-07-15 20:15:28 +020013085
13086String matching applies to verbatim strings as they are passed, with the
13087exception of the backslash ("\") which makes it possible to escape some
13088characters such as the space. If the "-i" flag is passed before the first
13089string, then the matching will be performed ignoring the case. In order
13090to match the string "-i", either set it second, or pass the "--" flag
Willy Tarreau0ba27502007-12-24 16:55:16 +010013091before the first string. Same applies of course to match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020013092
Mathias Weiersmuellercb250fc2019-12-02 09:43:40 +010013093Do not use string matches for binary fetches which might contain null bytes
13094(0x00), as the comparison stops at the occurrence of the first null byte.
13095Instead, convert the binary fetch to a hex string with the hex converter first.
13096
13097Example:
13098 # matches if the string <tag> is present in the binary sample
13099 acl tag_found req.payload(0,0),hex -m sub 3C7461673E
13100
Willy Tarreau6a06a402007-07-15 20:15:28 +020013101
Willy Tarreau74ca5042013-06-11 23:12:07 +0200131027.1.4. Matching regular expressions (regexes)
13103---------------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020013104
13105Just like with string matching, regex matching applies to verbatim strings as
13106they are passed, with the exception of the backslash ("\") which makes it
13107possible to escape some characters such as the space. If the "-i" flag is
13108passed before the first regex, then the matching will be performed ignoring
13109the case. In order to match the string "-i", either set it second, or pass
Willy Tarreau0ba27502007-12-24 16:55:16 +010013110the "--" flag before the first string. Same principle applies of course to
13111match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020013112
13113
Willy Tarreau74ca5042013-06-11 23:12:07 +0200131147.1.5. Matching arbitrary data blocks
13115-------------------------------------
13116
13117It is possible to match some extracted samples against a binary block which may
13118not safely be represented as a string. For this, the patterns must be passed as
13119a series of hexadecimal digits in an even number, when the match method is set
13120to binary. Each sequence of two digits will represent a byte. The hexadecimal
13121digits may be used upper or lower case.
13122
13123Example :
13124 # match "Hello\n" in the input stream (\x48 \x65 \x6c \x6c \x6f \x0a)
13125 acl hello payload(0,6) -m bin 48656c6c6f0a
13126
13127
131287.1.6. Matching IPv4 and IPv6 addresses
13129---------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020013130
13131IPv4 addresses values can be specified either as plain addresses or with a
13132netmask appended, in which case the IPv4 address matches whenever it is
13133within the network. Plain addresses may also be replaced with a resolvable
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010013134host name, but this practice is generally discouraged as it makes it more
Willy Tarreau0ba27502007-12-24 16:55:16 +010013135difficult to read and debug configurations. If hostnames are used, you should
13136at least ensure that they are present in /etc/hosts so that the configuration
13137does not depend on any random DNS match at the moment the configuration is
13138parsed.
Willy Tarreau6a06a402007-07-15 20:15:28 +020013139
Daniel Schnellereba56342016-04-13 00:26:52 +020013140The dotted IPv4 address notation is supported in both regular as well as the
13141abbreviated form with all-0-octets omitted:
13142
13143 +------------------+------------------+------------------+
13144 | Example 1 | Example 2 | Example 3 |
13145 +------------------+------------------+------------------+
13146 | 192.168.0.1 | 10.0.0.12 | 127.0.0.1 |
13147 | 192.168.1 | 10.12 | 127.1 |
13148 | 192.168.0.1/22 | 10.0.0.12/8 | 127.0.0.1/8 |
13149 | 192.168.1/22 | 10.12/8 | 127.1/8 |
13150 +------------------+------------------+------------------+
13151
13152Notice that this is different from RFC 4632 CIDR address notation in which
13153192.168.42/24 would be equivalent to 192.168.42.0/24.
13154
Willy Tarreauceb4ac92012-04-28 00:41:46 +020013155IPv6 may be entered in their usual form, with or without a netmask appended.
13156Only bit counts are accepted for IPv6 netmasks. In order to avoid any risk of
13157trouble with randomly resolved IP addresses, host names are never allowed in
13158IPv6 patterns.
13159
13160HAProxy is also able to match IPv4 addresses with IPv6 addresses in the
13161following situations :
13162 - tested address is IPv4, pattern address is IPv4, the match applies
13163 in IPv4 using the supplied mask if any.
13164 - tested address is IPv6, pattern address is IPv6, the match applies
13165 in IPv6 using the supplied mask if any.
13166 - tested address is IPv6, pattern address is IPv4, the match applies in IPv4
13167 using the pattern's mask if the IPv6 address matches with 2002:IPV4::,
13168 ::IPV4 or ::ffff:IPV4, otherwise it fails.
13169 - tested address is IPv4, pattern address is IPv6, the IPv4 address is first
13170 converted to IPv6 by prefixing ::ffff: in front of it, then the match is
13171 applied in IPv6 using the supplied IPv6 mask.
13172
Willy Tarreau74ca5042013-06-11 23:12:07 +020013173
131747.2. Using ACLs to form conditions
13175----------------------------------
13176
13177Some actions are only performed upon a valid condition. A condition is a
13178combination of ACLs with operators. 3 operators are supported :
13179
13180 - AND (implicit)
13181 - OR (explicit with the "or" keyword or the "||" operator)
13182 - Negation with the exclamation mark ("!")
Willy Tarreau6a06a402007-07-15 20:15:28 +020013183
Willy Tarreau74ca5042013-06-11 23:12:07 +020013184A condition is formed as a disjunctive form:
Willy Tarreau6a06a402007-07-15 20:15:28 +020013185
Willy Tarreau74ca5042013-06-11 23:12:07 +020013186 [!]acl1 [!]acl2 ... [!]acln { or [!]acl1 [!]acl2 ... [!]acln } ...
Willy Tarreaubef91e72013-03-31 23:14:46 +020013187
Willy Tarreau74ca5042013-06-11 23:12:07 +020013188Such conditions are generally used after an "if" or "unless" statement,
13189indicating when the condition will trigger the action.
Willy Tarreaubef91e72013-03-31 23:14:46 +020013190
Willy Tarreau74ca5042013-06-11 23:12:07 +020013191For instance, to block HTTP requests to the "*" URL with methods other than
13192"OPTIONS", as well as POST requests without content-length, and GET or HEAD
13193requests with a content-length greater than 0, and finally every request which
13194is not either GET/HEAD/POST/OPTIONS !
13195
13196 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013197 http-request deny if HTTP_URL_STAR !METH_OPTIONS || METH_POST missing_cl
13198 http-request deny if METH_GET HTTP_CONTENT
13199 http-request deny unless METH_GET or METH_POST or METH_OPTIONS
Willy Tarreau74ca5042013-06-11 23:12:07 +020013200
13201To select a different backend for requests to static contents on the "www" site
13202and to every request on the "img", "video", "download" and "ftp" hosts :
13203
13204 acl url_static path_beg /static /images /img /css
13205 acl url_static path_end .gif .png .jpg .css .js
13206 acl host_www hdr_beg(host) -i www
13207 acl host_static hdr_beg(host) -i img. video. download. ftp.
13208
Davor Ocelice9ed2812017-12-25 17:49:28 +010013209 # now use backend "static" for all static-only hosts, and for static URLs
Willy Tarreau74ca5042013-06-11 23:12:07 +020013210 # of host "www". Use backend "www" for the rest.
13211 use_backend static if host_static or host_www url_static
13212 use_backend www if host_www
13213
13214It is also possible to form rules using "anonymous ACLs". Those are unnamed ACL
13215expressions that are built on the fly without needing to be declared. They must
13216be enclosed between braces, with a space before and after each brace (because
13217the braces must be seen as independent words). Example :
13218
13219 The following rule :
13220
13221 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013222 http-request deny if METH_POST missing_cl
Willy Tarreau74ca5042013-06-11 23:12:07 +020013223
13224 Can also be written that way :
13225
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013226 http-request deny if METH_POST { hdr_cnt(Content-length) eq 0 }
Willy Tarreau74ca5042013-06-11 23:12:07 +020013227
13228It is generally not recommended to use this construct because it's a lot easier
13229to leave errors in the configuration when written that way. However, for very
13230simple rules matching only one source IP address for instance, it can make more
13231sense to use them than to declare ACLs with random names. Another example of
13232good use is the following :
13233
13234 With named ACLs :
13235
13236 acl site_dead nbsrv(dynamic) lt 2
13237 acl site_dead nbsrv(static) lt 2
13238 monitor fail if site_dead
13239
13240 With anonymous ACLs :
13241
13242 monitor fail if { nbsrv(dynamic) lt 2 } || { nbsrv(static) lt 2 }
13243
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013244See section 4.2 for detailed help on the "http-request deny" and "use_backend"
13245keywords.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013246
13247
132487.3. Fetching samples
13249---------------------
13250
13251Historically, sample fetch methods were only used to retrieve data to match
13252against patterns using ACLs. With the arrival of stick-tables, a new class of
13253sample fetch methods was created, most often sharing the same syntax as their
13254ACL counterpart. These sample fetch methods are also known as "fetches". As
13255of now, ACLs and fetches have converged. All ACL fetch methods have been made
13256available as fetch methods, and ACLs may use any sample fetch method as well.
13257
13258This section details all available sample fetch methods and their output type.
13259Some sample fetch methods have deprecated aliases that are used to maintain
13260compatibility with existing configurations. They are then explicitly marked as
13261deprecated and should not be used in new setups.
13262
13263The ACL derivatives are also indicated when available, with their respective
13264matching methods. These ones all have a well defined default pattern matching
13265method, so it is never necessary (though allowed) to pass the "-m" option to
13266indicate how the sample will be matched using ACLs.
13267
13268As indicated in the sample type versus matching compatibility matrix above,
13269when using a generic sample fetch method in an ACL, the "-m" option is
13270mandatory unless the sample type is one of boolean, integer, IPv4 or IPv6. When
13271the same keyword exists as an ACL keyword and as a standard fetch method, the
13272ACL engine will automatically pick the ACL-only one by default.
13273
13274Some of these keywords support one or multiple mandatory arguments, and one or
13275multiple optional arguments. These arguments are strongly typed and are checked
13276when the configuration is parsed so that there is no risk of running with an
Davor Ocelice9ed2812017-12-25 17:49:28 +010013277incorrect argument (e.g. an unresolved backend name). Fetch function arguments
13278are passed between parenthesis and are delimited by commas. When an argument
Willy Tarreau74ca5042013-06-11 23:12:07 +020013279is optional, it will be indicated below between square brackets ('[ ]'). When
13280all arguments are optional, the parenthesis may be omitted.
13281
13282Thus, the syntax of a standard sample fetch method is one of the following :
13283 - name
13284 - name(arg1)
13285 - name(arg1,arg2)
13286
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013287
132887.3.1. Converters
13289-----------------
13290
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013291Sample fetch methods may be combined with transformations to be applied on top
13292of the fetched sample (also called "converters"). These combinations form what
13293is called "sample expressions" and the result is a "sample". Initially this
13294was only supported by "stick on" and "stick store-request" directives but this
Davor Ocelice9ed2812017-12-25 17:49:28 +010013295has now be extended to all places where samples may be used (ACLs, log-format,
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013296unique-id-format, add-header, ...).
13297
13298These transformations are enumerated as a series of specific keywords after the
13299sample fetch method. These keywords may equally be appended immediately after
13300the fetch keyword's argument, delimited by a comma. These keywords can also
Davor Ocelice9ed2812017-12-25 17:49:28 +010013301support some arguments (e.g. a netmask) which must be passed in parenthesis.
Willy Tarreau0ba27502007-12-24 16:55:16 +010013302
Willy Tarreau97707872015-01-27 15:12:13 +010013303A certain category of converters are bitwise and arithmetic operators which
13304support performing basic operations on integers. Some bitwise operations are
13305supported (and, or, xor, cpl) and some arithmetic operations are supported
13306(add, sub, mul, div, mod, neg). Some comparators are provided (odd, even, not,
13307bool) which make it possible to report a match without having to write an ACL.
13308
Willy Tarreau74ca5042013-06-11 23:12:07 +020013309The currently available list of transformation keywords include :
Willy Tarreau0ba27502007-12-24 16:55:16 +010013310
Ben Shillitof25e8e52016-12-02 14:25:37 +00001331151d.single(<prop>[,<prop>*])
13312 Returns values for the properties requested as a string, where values are
13313 separated by the delimiter specified with "51degrees-property-separator".
13314 The device is identified using the User-Agent header passed to the
13315 converter. The function can be passed up to five property names, and if a
13316 property name can't be found, the value "NoData" is returned.
13317
13318 Example :
Davor Ocelice9ed2812017-12-25 17:49:28 +010013319 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request,
13320 # containing values for the three properties requested by using the
Ben Shillitof25e8e52016-12-02 14:25:37 +000013321 # User-Agent passed to the converter.
13322 frontend http-in
13323 bind *:8081
13324 default_backend servers
13325 http-request set-header X-51D-DeviceTypeMobileTablet \
13326 %[req.fhdr(User-Agent),51d.single(DeviceType,IsMobile,IsTablet)]
13327
Willy Tarreau97707872015-01-27 15:12:13 +010013328add(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013329 Adds <value> to the input value of type signed integer, and returns the
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013330 result as a signed integer. <value> can be a numeric value or a variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013331 name. The name of the variable starts with an indication about its scope. The
13332 scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013333 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013334 "sess" : the variable is shared with the whole session
13335 "txn" : the variable is shared with the transaction (request and response)
13336 "req" : the variable is shared only during request processing
13337 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013338 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013339 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013340
Nenad Merdanovicc31499d2019-03-23 11:00:32 +010013341aes_gcm_dec(<bits>,<nonce>,<key>,<aead_tag>)
13342 Decrypts the raw byte input using the AES128-GCM, AES192-GCM or
13343 AES256-GCM algorithm, depending on the <bits> parameter. All other parameters
13344 need to be base64 encoded and the returned result is in raw byte format.
13345 If the <aead_tag> validation fails, the converter doesn't return any data.
13346 The <nonce>, <key> and <aead_tag> can either be strings or variables. This
13347 converter requires at least OpenSSL 1.0.1.
13348
13349 Example:
13350 http-response set-header X-Decrypted-Text %[var(txn.enc),\
13351 aes_gcm_dec(128,txn.nonce,Zm9vb2Zvb29mb29wZm9vbw==,txn.aead_tag)]
13352
Willy Tarreau97707872015-01-27 15:12:13 +010013353and(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013354 Performs a bitwise "AND" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013355 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010013356 numeric value or a variable name. The name of the variable starts with an
13357 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013358 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013359 "sess" : the variable is shared with the whole session
13360 "txn" : the variable is shared with the transaction (request and response)
13361 "req" : the variable is shared only during request processing
13362 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013363 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013364 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013365
Holger Just1bfc24b2017-05-06 00:56:53 +020013366b64dec
13367 Converts (decodes) a base64 encoded input string to its binary
13368 representation. It performs the inverse operation of base64().
13369
Emeric Brun53d1a982014-04-30 18:21:37 +020013370base64
13371 Converts a binary input sample to a base64 string. It is used to log or
Davor Ocelice9ed2812017-12-25 17:49:28 +010013372 transfer binary content in a way that can be reliably transferred (e.g.
Emeric Brun53d1a982014-04-30 18:21:37 +020013373 an SSL ID can be copied in a header).
13374
Willy Tarreau97707872015-01-27 15:12:13 +010013375bool
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013376 Returns a boolean TRUE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010013377 non-null, otherwise returns FALSE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010013378 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010013379 presence of a flag).
13380
Emeric Brun54c4ac82014-11-03 15:32:43 +010013381bytes(<offset>[,<length>])
13382 Extracts some bytes from an input binary sample. The result is a binary
13383 sample starting at an offset (in bytes) of the original sample and
Tim Düsterhus4896c442016-11-29 02:15:19 +010013384 optionally truncated at the given length.
Emeric Brun54c4ac82014-11-03 15:32:43 +010013385
Willy Tarreau280f42b2018-02-19 15:34:12 +010013386concat([<start>],[<var>],[<end>])
13387 Concatenates up to 3 fields after the current sample which is then turned to
13388 a string. The first one, <start>, is a constant string, that will be appended
13389 immediately after the existing sample. It may be omitted if not used. The
13390 second one, <var>, is a variable name. The variable will be looked up, its
13391 contents converted to a string, and it will be appended immediately after the
13392 <first> part. If the variable is not found, nothing is appended. It may be
13393 omitted as well. The third field, <end> is a constant string that will be
13394 appended after the variable. It may also be omitted. Together, these elements
13395 allow to concatenate variables with delimiters to an existing set of
13396 variables. This can be used to build new variables made of a succession of
John Roeslerfb2fce12019-07-10 15:45:51 -050013397 other variables, such as colon-delimited values. Note that due to the config
Willy Tarreau280f42b2018-02-19 15:34:12 +010013398 parser, it is not possible to use a comma nor a closing parenthesis as
John Roeslerfb2fce12019-07-10 15:45:51 -050013399 delimiters.
Willy Tarreau280f42b2018-02-19 15:34:12 +010013400
13401 Example:
13402 tcp-request session set-var(sess.src) src
13403 tcp-request session set-var(sess.dn) ssl_c_s_dn
13404 tcp-request session set-var(txn.sig) str(),concat(<ip=,sess.ip,>),concat(<dn=,sess.dn,>)
13405 http-request set-header x-hap-sig %[var(txn.sig)]
13406
Willy Tarreau97707872015-01-27 15:12:13 +010013407cpl
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013408 Takes the input value of type signed integer, applies a ones-complement
13409 (flips all bits) and returns the result as an signed integer.
Willy Tarreau97707872015-01-27 15:12:13 +010013410
Willy Tarreau80599772015-01-20 19:35:24 +010013411crc32([<avalanche>])
13412 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32
13413 hash function. Optionally, it is possible to apply a full avalanche hash
13414 function to the output if the optional <avalanche> argument equals 1. This
13415 converter uses the same functions as used by the various hash-based load
13416 balancing algorithms, so it will provide exactly the same results. It is
13417 provided for compatibility with other software which want a CRC32 to be
13418 computed on some input keys, so it follows the most common implementation as
13419 found in Ethernet, Gzip, PNG, etc... It is slower than the other algorithms
13420 but may provide a better or at least less predictable distribution. It must
13421 not be used for security purposes as a 32-bit hash is trivial to break. See
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013422 also "djb2", "sdbm", "wt6", "crc32c" and the "hash-type" directive.
13423
13424crc32c([<avalanche>])
13425 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32C
13426 hash function. Optionally, it is possible to apply a full avalanche hash
13427 function to the output if the optional <avalanche> argument equals 1. This
13428 converter uses the same functions as described in RFC4960, Appendix B [8].
13429 It is provided for compatibility with other software which want a CRC32C to be
13430 computed on some input keys. It is slower than the other algorithms and it must
13431 not be used for security purposes as a 32-bit hash is trivial to break. See
13432 also "djb2", "sdbm", "wt6", "crc32" and the "hash-type" directive.
Willy Tarreau80599772015-01-20 19:35:24 +010013433
David Carlier29b3ca32015-09-25 14:09:21 +010013434da-csv-conv(<prop>[,<prop>*])
David Carlier4542b102015-06-01 13:54:29 +020013435 Asks the DeviceAtlas converter to identify the User Agent string passed on
13436 input, and to emit a string made of the concatenation of the properties
13437 enumerated in argument, delimited by the separator defined by the global
13438 keyword "deviceatlas-property-separator", or by default the pipe character
David Carlier840b0242016-03-16 10:09:55 +000013439 ('|'). There's a limit of 12 different properties imposed by the haproxy
David Carlier4542b102015-06-01 13:54:29 +020013440 configuration language.
13441
13442 Example:
13443 frontend www
Cyril Bonté307ee1e2015-09-28 23:16:06 +020013444 bind *:8881
13445 default_backend servers
David Carlier840b0242016-03-16 10:09:55 +000013446 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 +020013447
Willy Tarreau0851fd52019-12-17 10:07:25 +010013448debug([<prefix][,<destination>])
13449 This converter is used as debug tool. It takes a capture of the input sample
13450 and sends it to event sink <destination>, which may designate a ring buffer
13451 such as "buf0", as well as "stdout", or "stderr". Available sinks may be
13452 checked at run time by issuing "show events" on the CLI. When not specified,
13453 the output will be "buf0", which may be consulted via the CLI's "show events"
13454 command. An optional prefix <prefix> may be passed to help distinguish
13455 outputs from multiple expressions. It will then appear before the colon in
13456 the output message. The input sample is passed as-is on the output, so that
13457 it is safe to insert the debug converter anywhere in a chain, even with non-
13458 printable sample types.
13459
13460 Example:
13461 tcp-request connection track-sc0 src,debug(track-sc)
Thierry FOURNIER9687c772015-05-07 15:46:29 +020013462
Willy Tarreau97707872015-01-27 15:12:13 +010013463div(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013464 Divides the input value of type signed integer by <value>, and returns the
13465 result as an signed integer. If <value> is null, the largest unsigned
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013466 integer is returned (typically 2^63-1). <value> can be a numeric value or a
Daniel Schneller0b547052016-03-21 20:46:57 +010013467 variable name. The name of the variable starts with an indication about its
13468 scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013469 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013470 "sess" : the variable is shared with the whole session
13471 "txn" : the variable is shared with the transaction (request and response)
13472 "req" : the variable is shared only during request processing
13473 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013474 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013475 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013476
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013477djb2([<avalanche>])
13478 Hashes a binary input sample into an unsigned 32-bit quantity using the DJB2
13479 hash function. Optionally, it is possible to apply a full avalanche hash
13480 function to the output if the optional <avalanche> argument equals 1. This
13481 converter uses the same functions as used by the various hash-based load
13482 balancing algorithms, so it will provide exactly the same results. It is
13483 mostly intended for debugging, but can be used as a stick-table entry to
13484 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013485 32-bit hash is trivial to break. See also "crc32", "sdbm", "wt6", "crc32c",
13486 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013487
Willy Tarreau97707872015-01-27 15:12:13 +010013488even
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013489 Returns a boolean TRUE if the input value of type signed integer is even
Willy Tarreau97707872015-01-27 15:12:13 +010013490 otherwise returns FALSE. It is functionally equivalent to "not,and(1),bool".
13491
Marcin Deranek9631a282018-04-16 14:30:46 +020013492field(<index>,<delimiters>[,<count>])
13493 Extracts the substring at the given index counting from the beginning
13494 (positive index) or from the end (negative index) considering given delimiters
13495 from an input string. Indexes start at 1 or -1 and delimiters are a string
13496 formatted list of chars. Optionally you can specify <count> of fields to
13497 extract (default: 1). Value of 0 indicates extraction of all remaining
13498 fields.
13499
13500 Example :
13501 str(f1_f2_f3__f5),field(5,_) # f5
13502 str(f1_f2_f3__f5),field(2,_,0) # f2_f3__f5
13503 str(f1_f2_f3__f5),field(2,_,2) # f2_f3
13504 str(f1_f2_f3__f5),field(-2,_,3) # f2_f3_
13505 str(f1_f2_f3__f5),field(-3,_,0) # f1_f2_f3
Emeric Brunf399b0d2014-11-03 17:07:03 +010013506
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013507hex
Davor Ocelice9ed2812017-12-25 17:49:28 +010013508 Converts a binary input sample to a hex string containing two hex digits per
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013509 input byte. It is used to log or transfer hex dumps of some binary input data
Davor Ocelice9ed2812017-12-25 17:49:28 +010013510 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 +020013511 header).
Thierry FOURNIER2f49d6d2014-03-12 15:01:52 +010013512
Dragan Dosen3f957b22017-10-24 09:27:34 +020013513hex2i
13514 Converts a hex string containing two hex digits per input byte to an
John Roeslerfb2fce12019-07-10 15:45:51 -050013515 integer. If the input value cannot be converted, then zero is returned.
Dragan Dosen3f957b22017-10-24 09:27:34 +020013516
Cyril Bonté6bcd1822019-11-05 23:13:59 +010013517http_date([<offset],[<unit>])
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013518 Converts an integer supposed to contain a date since epoch to a string
13519 representing this date in a format suitable for use in HTTP header fields. If
Damien Claisseae6f1252019-10-30 15:57:28 +000013520 an offset value is specified, then it is added to the date before the
13521 conversion is operated. This is particularly useful to emit Date header fields,
13522 Expires values in responses when combined with a positive offset, or
13523 Last-Modified values when the offset is negative.
13524 If a unit value is specified, then consider the timestamp as either
13525 "s" for seconds (default behavior), "ms" for milliseconds, or "us" for
13526 microseconds since epoch. Offset is assumed to have the same unit as
13527 input timestamp.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013528
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013529in_table(<table>)
13530 Uses the string representation of the input sample to perform a look up in
13531 the specified table. If the key is not found in the table, a boolean false
13532 is returned. Otherwise a boolean true is returned. This can be used to verify
Davor Ocelice9ed2812017-12-25 17:49:28 +010013533 the presence of a certain key in a table tracking some elements (e.g. whether
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013534 or not a source IP address or an Authorization header was already seen).
13535
Tim Duesterhus1478aa72018-01-25 16:24:51 +010013536ipmask(<mask4>, [<mask6>])
13537 Apply a mask to an IP address, and use the result for lookups and storage.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013538 This can be used to make all hosts within a certain mask to share the same
Tim Duesterhus1478aa72018-01-25 16:24:51 +010013539 table entries and as such use the same server. The mask4 can be passed in
13540 dotted form (e.g. 255.255.255.0) or in CIDR form (e.g. 24). The mask6 can
13541 be passed in quadruplet form (e.g. ffff:ffff::) or in CIDR form (e.g. 64).
13542 If no mask6 is given IPv6 addresses will fail to convert for backwards
13543 compatibility reasons.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013544
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013545json([<input-code>])
Davor Ocelice9ed2812017-12-25 17:49:28 +010013546 Escapes the input string and produces an ASCII output string ready to use as a
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013547 JSON string. The converter tries to decode the input string according to the
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020013548 <input-code> parameter. It can be "ascii", "utf8", "utf8s", "utf8p" or
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013549 "utf8ps". The "ascii" decoder never fails. The "utf8" decoder detects 3 types
13550 of errors:
13551 - bad UTF-8 sequence (lone continuation byte, bad number of continuation
13552 bytes, ...)
13553 - invalid range (the decoded value is within a UTF-8 prohibited range),
13554 - code overlong (the value is encoded with more bytes than necessary).
13555
13556 The UTF-8 JSON encoding can produce a "too long value" error when the UTF-8
13557 character is greater than 0xffff because the JSON string escape specification
13558 only authorizes 4 hex digits for the value encoding. The UTF-8 decoder exists
13559 in 4 variants designated by a combination of two suffix letters : "p" for
13560 "permissive" and "s" for "silently ignore". The behaviors of the decoders
13561 are :
Davor Ocelice9ed2812017-12-25 17:49:28 +010013562 - "ascii" : never fails;
13563 - "utf8" : fails on any detected errors;
13564 - "utf8s" : never fails, but removes characters corresponding to errors;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013565 - "utf8p" : accepts and fixes the overlong errors, but fails on any other
Davor Ocelice9ed2812017-12-25 17:49:28 +010013566 error;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013567 - "utf8ps" : never fails, accepts and fixes the overlong errors, but removes
13568 characters corresponding to the other errors.
13569
13570 This converter is particularly useful for building properly escaped JSON for
Davor Ocelice9ed2812017-12-25 17:49:28 +010013571 logging to servers which consume JSON-formatted traffic logs.
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013572
13573 Example:
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013574 capture request header Host len 15
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020013575 capture request header user-agent len 150
13576 log-format '{"ip":"%[src]","user-agent":"%[capture.req.hdr(1),json(utf8s)]"}'
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013577
13578 Input request from client 127.0.0.1:
13579 GET / HTTP/1.0
13580 User-Agent: Very "Ugly" UA 1/2
13581
13582 Output log:
13583 {"ip":"127.0.0.1","user-agent":"Very \"Ugly\" UA 1\/2"}
13584
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013585language(<value>[,<default>])
13586 Returns the value with the highest q-factor from a list as extracted from the
13587 "accept-language" header using "req.fhdr". Values with no q-factor have a
13588 q-factor of 1. Values with a q-factor of 0 are dropped. Only values which
13589 belong to the list of semi-colon delimited <values> will be considered. The
13590 argument <value> syntax is "lang[;lang[;lang[;...]]]". If no value matches the
13591 given list and a default value is provided, it is returned. Note that language
13592 names may have a variant after a dash ('-'). If this variant is present in the
13593 list, it will be matched, but if it is not, only the base language is checked.
13594 The match is case-sensitive, and the output string is always one of those
Davor Ocelice9ed2812017-12-25 17:49:28 +010013595 provided in arguments. The ordering of arguments is meaningless, only the
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013596 ordering of the values in the request counts, as the first value among
13597 multiple sharing the same q-factor is used.
Thierry FOURNIERad903512014-04-11 17:51:01 +020013598
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013599 Example :
Thierry FOURNIERad903512014-04-11 17:51:01 +020013600
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013601 # this configuration switches to the backend matching a
13602 # given language based on the request :
Thierry FOURNIERad903512014-04-11 17:51:01 +020013603
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013604 acl es req.fhdr(accept-language),language(es;fr;en) -m str es
13605 acl fr req.fhdr(accept-language),language(es;fr;en) -m str fr
13606 acl en req.fhdr(accept-language),language(es;fr;en) -m str en
13607 use_backend spanish if es
13608 use_backend french if fr
13609 use_backend english if en
13610 default_backend choose_your_language
Thierry FOURNIERad903512014-04-11 17:51:01 +020013611
Willy Tarreau60a2ee72017-12-15 07:13:48 +010013612length
Etienne Carriereed0d24e2017-12-13 13:41:34 +010013613 Get the length of the string. This can only be placed after a string
13614 sample fetch function or after a transformation keyword returning a string
13615 type. The result is of type integer.
13616
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013617lower
13618 Convert a string sample to lower case. This can only be placed after a string
13619 sample fetch function or after a transformation keyword returning a string
13620 type. The result is of type string.
13621
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013622ltime(<format>[,<offset>])
13623 Converts an integer supposed to contain a date since epoch to a string
13624 representing this date in local time using a format defined by the <format>
13625 string using strftime(3). The purpose is to allow any date format to be used
13626 in logs. An optional <offset> in seconds may be applied to the input date
13627 (positive or negative). See the strftime() man page for the format supported
13628 by your operating system. See also the utime converter.
13629
13630 Example :
13631
13632 # Emit two colons, one with the local time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010013633 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013634 log-format %[date,ltime(%Y%m%d%H%M%S)]\ %ci:%cp
13635
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013636map(<map_file>[,<default_value>])
13637map_<match_type>(<map_file>[,<default_value>])
13638map_<match_type>_<output_type>(<map_file>[,<default_value>])
13639 Search the input value from <map_file> using the <match_type> matching method,
13640 and return the associated value converted to the type <output_type>. If the
13641 input value cannot be found in the <map_file>, the converter returns the
13642 <default_value>. If the <default_value> is not set, the converter fails and
13643 acts as if no input value could be fetched. If the <match_type> is not set, it
13644 defaults to "str". Likewise, if the <output_type> is not set, it defaults to
13645 "str". For convenience, the "map" keyword is an alias for "map_str" and maps a
13646 string to another string.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013647
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013648 It is important to avoid overlapping between the keys : IP addresses and
13649 strings are stored in trees, so the first of the finest match will be used.
13650 Other keys are stored in lists, so the first matching occurrence will be used.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013651
Tim Düsterhus4896c442016-11-29 02:15:19 +010013652 The following array contains the list of all map functions available sorted by
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013653 input type, match type and output type.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013654
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013655 input type | match method | output type str | output type int | output type ip
13656 -----------+--------------+-----------------+-----------------+---------------
13657 str | str | map_str | map_str_int | map_str_ip
13658 -----------+--------------+-----------------+-----------------+---------------
Willy Tarreau787a4c02014-05-10 07:55:30 +020013659 str | beg | map_beg | map_beg_int | map_end_ip
13660 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013661 str | sub | map_sub | map_sub_int | map_sub_ip
13662 -----------+--------------+-----------------+-----------------+---------------
13663 str | dir | map_dir | map_dir_int | map_dir_ip
13664 -----------+--------------+-----------------+-----------------+---------------
13665 str | dom | map_dom | map_dom_int | map_dom_ip
13666 -----------+--------------+-----------------+-----------------+---------------
13667 str | end | map_end | map_end_int | map_end_ip
13668 -----------+--------------+-----------------+-----------------+---------------
Ruoshan Huang3c5e3742016-12-02 16:25:31 +080013669 str | reg | map_reg | map_reg_int | map_reg_ip
13670 -----------+--------------+-----------------+-----------------+---------------
13671 str | reg | map_regm | map_reg_int | map_reg_ip
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013672 -----------+--------------+-----------------+-----------------+---------------
13673 int | int | map_int | map_int_int | map_int_ip
13674 -----------+--------------+-----------------+-----------------+---------------
13675 ip | ip | map_ip | map_ip_int | map_ip_ip
13676 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013677
Thierry Fournier8feaa662016-02-10 22:55:20 +010013678 The special map called "map_regm" expect matching zone in the regular
13679 expression and modify the output replacing back reference (like "\1") by
13680 the corresponding match text.
13681
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013682 The file contains one key + value per line. Lines which start with '#' are
13683 ignored, just like empty lines. Leading tabs and spaces are stripped. The key
13684 is then the first "word" (series of non-space/tabs characters), and the value
13685 is what follows this series of space/tab till the end of the line excluding
13686 trailing spaces/tabs.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013687
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013688 Example :
13689
13690 # this is a comment and is ignored
13691 2.22.246.0/23 United Kingdom \n
13692 <-><-----------><--><------------><---->
13693 | | | | `- trailing spaces ignored
13694 | | | `---------- value
13695 | | `-------------------- middle spaces ignored
13696 | `---------------------------- key
13697 `------------------------------------ leading spaces ignored
13698
Willy Tarreau97707872015-01-27 15:12:13 +010013699mod(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013700 Divides the input value of type signed integer by <value>, and returns the
13701 remainder as an signed integer. If <value> is null, then zero is returned.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013702 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013703 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013704 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013705 "sess" : the variable is shared with the whole session
13706 "txn" : the variable is shared with the transaction (request and response)
13707 "req" : the variable is shared only during request processing
13708 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013709 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013710 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013711
13712mul(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013713 Multiplies the input value of type signed integer by <value>, and returns
Thierry FOURNIER00c005c2015-07-08 01:10:21 +020013714 the product as an signed integer. In case of overflow, the largest possible
13715 value for the sign is returned so that the operation doesn't wrap around.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013716 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013717 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013718 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013719 "sess" : the variable is shared with the whole session
13720 "txn" : the variable is shared with the transaction (request and response)
13721 "req" : the variable is shared only during request processing
13722 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013723 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013724 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013725
Nenad Merdanovicb7e7c472017-03-12 21:56:55 +010013726nbsrv
13727 Takes an input value of type string, interprets it as a backend name and
13728 returns the number of usable servers in that backend. Can be used in places
13729 where we want to look up a backend from a dynamic name, like a result of a
13730 map lookup.
13731
Willy Tarreau97707872015-01-27 15:12:13 +010013732neg
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013733 Takes the input value of type signed integer, computes the opposite value,
13734 and returns the remainder as an signed integer. 0 is identity. This operator
13735 is provided for reversed subtracts : in order to subtract the input from a
13736 constant, simply perform a "neg,add(value)".
Willy Tarreau97707872015-01-27 15:12:13 +010013737
13738not
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013739 Returns a boolean FALSE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010013740 non-null, otherwise returns TRUE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010013741 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010013742 absence of a flag).
13743
13744odd
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013745 Returns a boolean TRUE if the input value of type signed integer is odd
Willy Tarreau97707872015-01-27 15:12:13 +010013746 otherwise returns FALSE. It is functionally equivalent to "and(1),bool".
13747
13748or(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013749 Performs a bitwise "OR" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013750 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010013751 numeric value or a variable name. The name of the variable starts with an
13752 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013753 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013754 "sess" : the variable is shared with the whole session
13755 "txn" : the variable is shared with the transaction (request and response)
13756 "req" : the variable is shared only during request processing
13757 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013758 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013759 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013760
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010013761protobuf(<field_number>,[<field_type>])
13762 This extracts the protocol buffers message field in raw mode of an input binary
13763 sample representation of a protocol buffer message with <field_number> as field
13764 number (dotted notation) if <field_type> is not present, or as an integer sample
13765 if this field is present (see also "ungrpc" below).
13766 The list of the authorized types is the following one: "int32", "int64", "uint32",
13767 "uint64", "sint32", "sint64", "bool", "enum" for the "varint" wire type 0
13768 "fixed64", "sfixed64", "double" for the 64bit wire type 1, "fixed32", "sfixed32",
13769 "float" for the wire type 5. Note that "string" is considered as a length-delimited
13770 type, so it does not require any <field_type> argument to be extracted.
13771 More information may be found here about the protocol buffers message field types:
13772 https://developers.google.com/protocol-buffers/docs/encoding
13773
Willy Tarreauc4dc3502015-01-23 20:39:28 +010013774regsub(<regex>,<subst>[,<flags>])
Willy Tarreau7eda8492015-01-20 19:47:06 +010013775 Applies a regex-based substitution to the input string. It does the same
13776 operation as the well-known "sed" utility with "s/<regex>/<subst>/". By
13777 default it will replace in the input string the first occurrence of the
13778 largest part matching the regular expression <regex> with the substitution
13779 string <subst>. It is possible to replace all occurrences instead by adding
13780 the flag "g" in the third argument <flags>. It is also possible to make the
13781 regex case insensitive by adding the flag "i" in <flags>. Since <flags> is a
13782 string, it is made up from the concatenation of all desired flags. Thus if
13783 both "i" and "g" are desired, using "gi" or "ig" will have the same effect.
13784 It is important to note that due to the current limitations of the
Baptiste Assmann66025d82016-03-06 23:36:48 +010013785 configuration parser, some characters such as closing parenthesis, closing
13786 square brackets or comma are not possible to use in the arguments. The first
13787 use of this converter is to replace certain characters or sequence of
13788 characters with other ones.
Willy Tarreau7eda8492015-01-20 19:47:06 +010013789
13790 Example :
13791
13792 # de-duplicate "/" in header "x-path".
13793 # input: x-path: /////a///b/c/xzxyz/
13794 # output: x-path: /a/b/c/xzxyz/
13795 http-request set-header x-path %[hdr(x-path),regsub(/+,/,g)]
13796
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020013797capture-req(<id>)
13798 Capture the string entry in the request slot <id> and returns the entry as
13799 is. If the slot doesn't exist, the capture fails silently.
13800
13801 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020013802 "http-response capture", "capture.req.hdr" and
13803 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020013804
13805capture-res(<id>)
13806 Capture the string entry in the response slot <id> and returns the entry as
13807 is. If the slot doesn't exist, the capture fails silently.
13808
13809 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020013810 "http-response capture", "capture.req.hdr" and
13811 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020013812
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013813sdbm([<avalanche>])
13814 Hashes a binary input sample into an unsigned 32-bit quantity using the SDBM
13815 hash function. Optionally, it is possible to apply a full avalanche hash
13816 function to the output if the optional <avalanche> argument equals 1. This
13817 converter uses the same functions as used by the various hash-based load
13818 balancing algorithms, so it will provide exactly the same results. It is
13819 mostly intended for debugging, but can be used as a stick-table entry to
13820 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013821 32-bit hash is trivial to break. See also "crc32", "djb2", "wt6", "crc32c",
13822 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013823
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013824set-var(<var name>)
Davor Ocelice9ed2812017-12-25 17:49:28 +010013825 Sets a variable with the input content and returns the content on the output
13826 as-is. The variable keeps the value and the associated input type. The name of
13827 the variable starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013828 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013829 "sess" : the variable is shared with the whole session
13830 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013831 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010013832 "req" : the variable is shared only during request processing,
13833 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013834 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013835 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013836
Dragan Dosen6e5a9ca2017-10-24 09:18:23 +020013837sha1
Tim Duesterhusd4376302019-06-17 12:41:44 +020013838 Converts a binary input sample to a SHA-1 digest. The result is a binary
Dragan Dosen6e5a9ca2017-10-24 09:18:23 +020013839 sample with length of 20 bytes.
13840
Tim Duesterhusd4376302019-06-17 12:41:44 +020013841sha2([<bits>])
13842 Converts a binary input sample to a digest in the SHA-2 family. The result
13843 is a binary sample with length of <bits>/8 bytes.
13844
13845 Valid values for <bits> are 224, 256, 384, 512, each corresponding to
13846 SHA-<bits>. The default value is 256.
13847
13848 Please note that this converter is only available when haproxy has been
13849 compiled with USE_OPENSSL.
13850
Nenad Merdanovic177adc92019-08-27 01:58:13 +020013851srv_queue
13852 Takes an input value of type string, either a server name or <backend>/<server>
13853 format and returns the number of queued sessions on that server. Can be used
13854 in places where we want to look up queued sessions from a dynamic name, like a
13855 cookie value (e.g. req.cook(SRVID),srv_queue) and then make a decision to break
13856 persistence or direct a request elsewhere.
13857
Tim Duesterhusca097c12018-04-27 21:18:45 +020013858strcmp(<var>)
13859 Compares the contents of <var> with the input value of type string. Returns
13860 the result as a signed integer compatible with strcmp(3): 0 if both strings
13861 are identical. A value less than 0 if the left string is lexicographically
13862 smaller than the right string or if the left string is shorter. A value greater
13863 than 0 otherwise (right string greater than left string or the right string is
13864 shorter).
13865
13866 Example :
13867
13868 http-request set-var(txn.host) hdr(host)
13869 # Check whether the client is attempting domain fronting.
13870 acl ssl_sni_http_host_match ssl_fc_sni,strcmp(txn.host) eq 0
13871
13872
Willy Tarreau97707872015-01-27 15:12:13 +010013873sub(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013874 Subtracts <value> from the input value of type signed integer, and returns
13875 the result as an signed integer. Note: in order to subtract the input from
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013876 a constant, simply perform a "neg,add(value)". <value> can be a numeric value
Daniel Schneller0b547052016-03-21 20:46:57 +010013877 or a variable name. The name of the variable starts with an indication about
13878 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013879 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013880 "sess" : the variable is shared with the whole session
13881 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013882 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010013883 "req" : the variable is shared only during request processing,
13884 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013885 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013886 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013887
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013888table_bytes_in_rate(<table>)
13889 Uses the string representation of the input sample to perform a look up in
13890 the specified table. If the key is not found in the table, integer value zero
13891 is returned. Otherwise the converter returns the average client-to-server
13892 bytes rate associated with the input sample in the designated table, measured
13893 in amount of bytes over the period configured in the table. See also the
13894 sc_bytes_in_rate sample fetch keyword.
13895
13896
13897table_bytes_out_rate(<table>)
13898 Uses the string representation of the input sample to perform a look up in
13899 the specified table. If the key is not found in the table, integer value zero
13900 is returned. Otherwise the converter returns the average server-to-client
13901 bytes rate associated with the input sample in the designated table, measured
13902 in amount of bytes over the period configured in the table. See also the
13903 sc_bytes_out_rate sample fetch keyword.
13904
13905table_conn_cnt(<table>)
13906 Uses the string representation of the input sample to perform a look up in
13907 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013908 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013909 connections associated with the input sample in the designated table. See
13910 also the sc_conn_cnt sample fetch keyword.
13911
13912table_conn_cur(<table>)
13913 Uses the string representation of the input sample to perform a look up in
13914 the specified table. If the key is not found in the table, integer value zero
13915 is returned. Otherwise the converter returns the current amount of concurrent
13916 tracked connections associated with the input sample in the designated table.
13917 See also the sc_conn_cur sample fetch keyword.
13918
13919table_conn_rate(<table>)
13920 Uses the string representation of the input sample to perform a look up in
13921 the specified table. If the key is not found in the table, integer value zero
13922 is returned. Otherwise the converter returns the average incoming connection
13923 rate associated with the input sample in the designated table. See also the
13924 sc_conn_rate sample fetch keyword.
13925
Thierry FOURNIER236657b2015-08-19 08:25:14 +020013926table_gpt0(<table>)
13927 Uses the string representation of the input sample to perform a look up in
13928 the specified table. If the key is not found in the table, boolean value zero
13929 is returned. Otherwise the converter returns the current value of the first
13930 general purpose tag associated with the input sample in the designated table.
13931 See also the sc_get_gpt0 sample fetch keyword.
13932
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013933table_gpc0(<table>)
13934 Uses the string representation of the input sample to perform a look up in
13935 the specified table. If the key is not found in the table, integer value zero
13936 is returned. Otherwise the converter returns the current value of the first
13937 general purpose counter associated with the input sample in the designated
13938 table. See also the sc_get_gpc0 sample fetch keyword.
13939
13940table_gpc0_rate(<table>)
13941 Uses the string representation of the input sample to perform a look up in
13942 the specified table. If the key is not found in the table, integer value zero
13943 is returned. Otherwise the converter returns the frequency which the gpc0
13944 counter was incremented over the configured period in the table, associated
13945 with the input sample in the designated table. See also the sc_get_gpc0_rate
13946 sample fetch keyword.
13947
Frédéric Lécaille6778b272018-01-29 15:22:53 +010013948table_gpc1(<table>)
13949 Uses the string representation of the input sample to perform a look up in
13950 the specified table. If the key is not found in the table, integer value zero
13951 is returned. Otherwise the converter returns the current value of the second
13952 general purpose counter associated with the input sample in the designated
13953 table. See also the sc_get_gpc1 sample fetch keyword.
13954
13955table_gpc1_rate(<table>)
13956 Uses the string representation of the input sample to perform a look up in
13957 the specified table. If the key is not found in the table, integer value zero
13958 is returned. Otherwise the converter returns the frequency which the gpc1
13959 counter was incremented over the configured period in the table, associated
13960 with the input sample in the designated table. See also the sc_get_gpc1_rate
13961 sample fetch keyword.
13962
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013963table_http_err_cnt(<table>)
13964 Uses the string representation of the input sample to perform a look up in
13965 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013966 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013967 errors associated with the input sample in the designated table. See also the
13968 sc_http_err_cnt sample fetch keyword.
13969
13970table_http_err_rate(<table>)
13971 Uses the string representation of the input sample to perform a look up in
13972 the specified table. If the key is not found in the table, integer value zero
13973 is returned. Otherwise the average rate of HTTP errors associated with the
13974 input sample in the designated table, measured in amount of errors over the
13975 period configured in the table. See also the sc_http_err_rate sample fetch
13976 keyword.
13977
13978table_http_req_cnt(<table>)
13979 Uses the string representation of the input sample to perform a look up in
13980 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013981 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013982 requests associated with the input sample in the designated table. See also
13983 the sc_http_req_cnt sample fetch keyword.
13984
13985table_http_req_rate(<table>)
13986 Uses the string representation of the input sample to perform a look up in
13987 the specified table. If the key is not found in the table, integer value zero
13988 is returned. Otherwise the average rate of HTTP requests associated with the
13989 input sample in the designated table, measured in amount of requests over the
13990 period configured in the table. See also the sc_http_req_rate sample fetch
13991 keyword.
13992
13993table_kbytes_in(<table>)
13994 Uses the string representation of the input sample to perform a look up in
13995 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013996 is returned. Otherwise the converter returns the cumulative number of client-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013997 to-server data associated with the input sample in the designated table,
13998 measured in kilobytes. The test is currently performed on 32-bit integers,
13999 which limits values to 4 terabytes. See also the sc_kbytes_in sample fetch
14000 keyword.
14001
14002table_kbytes_out(<table>)
14003 Uses the string representation of the input sample to perform a look up in
14004 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010014005 is returned. Otherwise the converter returns the cumulative number of server-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020014006 to-client data associated with the input sample in the designated table,
14007 measured in kilobytes. The test is currently performed on 32-bit integers,
14008 which limits values to 4 terabytes. See also the sc_kbytes_out sample fetch
14009 keyword.
14010
14011table_server_id(<table>)
14012 Uses the string representation of the input sample to perform a look up in
14013 the specified table. If the key is not found in the table, integer value zero
14014 is returned. Otherwise the converter returns the server ID associated with
14015 the input sample in the designated table. A server ID is associated to a
14016 sample by a "stick" rule when a connection to a server succeeds. A server ID
14017 zero means that no server is associated with this key.
14018
14019table_sess_cnt(<table>)
14020 Uses the string representation of the input sample to perform a look up in
14021 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010014022 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020014023 sessions associated with the input sample in the designated table. Note that
14024 a session here refers to an incoming connection being accepted by the
14025 "tcp-request connection" rulesets. See also the sc_sess_cnt sample fetch
14026 keyword.
14027
14028table_sess_rate(<table>)
14029 Uses the string representation of the input sample to perform a look up in
14030 the specified table. If the key is not found in the table, integer value zero
14031 is returned. Otherwise the converter returns the average incoming session
14032 rate associated with the input sample in the designated table. Note that a
14033 session here refers to an incoming connection being accepted by the
14034 "tcp-request connection" rulesets. See also the sc_sess_rate sample fetch
14035 keyword.
14036
14037table_trackers(<table>)
14038 Uses the string representation of the input sample to perform a look up in
14039 the specified table. If the key is not found in the table, integer value zero
14040 is returned. Otherwise the converter returns the current amount of concurrent
14041 connections tracking the same key as the input sample in the designated
14042 table. It differs from table_conn_cur in that it does not rely on any stored
14043 information but on the table's reference count (the "use" value which is
14044 returned by "show table" on the CLI). This may sometimes be more suited for
14045 layer7 tracking. It can be used to tell a server how many concurrent
14046 connections there are from a given address for example. See also the
14047 sc_trackers sample fetch keyword.
14048
Willy Tarreauffcb2e42014-07-10 16:29:08 +020014049upper
14050 Convert a string sample to upper case. This can only be placed after a string
14051 sample fetch function or after a transformation keyword returning a string
14052 type. The result is of type string.
14053
Thierry FOURNIER82ff3c92015-05-07 15:46:20 +020014054url_dec
14055 Takes an url-encoded string provided as input and returns the decoded
14056 version as output. The input and the output are of type string.
14057
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010014058ungrpc(<field_number>,[<field_type>])
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010014059 This extracts the protocol buffers message field in raw mode of an input binary
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010014060 sample representation of a gRPC message with <field_number> as field number
14061 (dotted notation) if <field_type> is not present, or as an integer sample if this
14062 field is present.
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010014063 The list of the authorized types is the following one: "int32", "int64", "uint32",
14064 "uint64", "sint32", "sint64", "bool", "enum" for the "varint" wire type 0
14065 "fixed64", "sfixed64", "double" for the 64bit wire type 1, "fixed32", "sfixed32",
14066 "float" for the wire type 5. Note that "string" is considered as a length-delimited
Frédéric Lécaille93d33162019-03-06 09:35:59 +010014067 type, so it does not require any <field_type> argument to be extracted.
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010014068 More information may be found here about the protocol buffers message field types:
14069 https://developers.google.com/protocol-buffers/docs/encoding
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010014070
14071 Example:
14072 // with such a protocol buffer .proto file content adapted from
14073 // https://github.com/grpc/grpc/blob/master/examples/protos/route_guide.proto
14074
14075 message Point {
14076 int32 latitude = 1;
14077 int32 longitude = 2;
14078 }
14079
14080 message PPoint {
14081 Point point = 59;
14082 }
14083
14084 message Rectangle {
14085 // One corner of the rectangle.
14086 PPoint lo = 48;
14087 // The other corner of the rectangle.
14088 PPoint hi = 49;
14089 }
14090
14091 let's say a body request is made of a "Rectangle" object value (two PPoint
14092 protocol buffers messages), the four protocol buffers fields could be
14093 extracted with these "ungrpc" directives:
14094
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010014095 req.body,ungrpc(48.59.1,int32) # "latitude" of "lo" first PPoint
14096 req.body,ungrpc(48.59.2,int32) # "longitude" of "lo" first PPoint
John Roeslerfb2fce12019-07-10 15:45:51 -050014097 req.body,ungrpc(49.59.1,int32) # "latitude" of "hi" second PPoint
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010014098 req.body,ungrpc(49.59.2,int32) # "longitude" of "hi" second PPoint
14099
Frédéric Lécaille93d33162019-03-06 09:35:59 +010014100 We could also extract the intermediary 48.59 field as a binary sample as follows:
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010014101
Frédéric Lécaille93d33162019-03-06 09:35:59 +010014102 req.body,ungrpc(48.59)
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010014103
John Roeslerfb2fce12019-07-10 15:45:51 -050014104 As a gRPC message is always made of a gRPC header followed by protocol buffers
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010014105 messages, in the previous example the "latitude" of "lo" first PPoint
14106 could be extracted with these equivalent directives:
14107
14108 req.body,ungrpc(48.59),protobuf(1,int32)
14109 req.body,ungrpc(48),protobuf(59.1,int32)
14110 req.body,ungrpc(48),protobuf(59),protobuf(1,int32)
14111
14112 Note that the first convert must be "ungrpc", the remaining ones must be
14113 "protobuf" and only the last one may have or not a second argument to
14114 interpret the previous binary sample.
14115
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010014116
Christopher Faulet85d79c92016-11-09 16:54:56 +010014117unset-var(<var name>)
14118 Unsets a variable if the input content is defined. The name of the variable
14119 starts with an indication about its scope. The scopes allowed are:
14120 "proc" : the variable is shared with the whole process
14121 "sess" : the variable is shared with the whole session
14122 "txn" : the variable is shared with the transaction (request and
14123 response),
14124 "req" : the variable is shared only during request processing,
14125 "res" : the variable is shared only during response processing.
14126 This prefix is followed by a name. The separator is a '.'. The name may only
14127 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
14128
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020014129utime(<format>[,<offset>])
14130 Converts an integer supposed to contain a date since epoch to a string
14131 representing this date in UTC time using a format defined by the <format>
14132 string using strftime(3). The purpose is to allow any date format to be used
14133 in logs. An optional <offset> in seconds may be applied to the input date
14134 (positive or negative). See the strftime() man page for the format supported
14135 by your operating system. See also the ltime converter.
14136
14137 Example :
14138
14139 # Emit two colons, one with the UTC time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010014140 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020014141 log-format %[date,utime(%Y%m%d%H%M%S)]\ %ci:%cp
14142
Marcin Deranek9631a282018-04-16 14:30:46 +020014143word(<index>,<delimiters>[,<count>])
14144 Extracts the nth word counting from the beginning (positive index) or from
14145 the end (negative index) considering given delimiters from an input string.
14146 Indexes start at 1 or -1 and delimiters are a string formatted list of chars.
14147 Optionally you can specify <count> of words to extract (default: 1).
14148 Value of 0 indicates extraction of all remaining words.
14149
14150 Example :
14151 str(f1_f2_f3__f5),word(4,_) # f5
14152 str(f1_f2_f3__f5),word(2,_,0) # f2_f3__f5
14153 str(f1_f2_f3__f5),word(3,_,2) # f3__f5
14154 str(f1_f2_f3__f5),word(-2,_,3) # f1_f2_f3
14155 str(f1_f2_f3__f5),word(-3,_,0) # f1_f2
Emeric Brunc9a0f6d2014-11-25 14:09:01 +010014156
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020014157wt6([<avalanche>])
14158 Hashes a binary input sample into an unsigned 32-bit quantity using the WT6
14159 hash function. Optionally, it is possible to apply a full avalanche hash
14160 function to the output if the optional <avalanche> argument equals 1. This
14161 converter uses the same functions as used by the various hash-based load
14162 balancing algorithms, so it will provide exactly the same results. It is
14163 mostly intended for debugging, but can be used as a stick-table entry to
14164 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010014165 32-bit hash is trivial to break. See also "crc32", "djb2", "sdbm", "crc32c",
14166 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020014167
Willy Tarreau97707872015-01-27 15:12:13 +010014168xor(<value>)
14169 Performs a bitwise "XOR" (exclusive OR) between <value> and the input value
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014170 of type signed integer, and returns the result as an signed integer.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014171 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010014172 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010014173 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010014174 "sess" : the variable is shared with the whole session
14175 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014176 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010014177 "req" : the variable is shared only during request processing,
14178 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014179 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014180 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010014181
Thierry FOURNIER01e09742016-12-26 11:46:11 +010014182xxh32([<seed>])
14183 Hashes a binary input sample into an unsigned 32-bit quantity using the 32-bit
14184 variant of the XXHash hash function. This hash supports a seed which defaults
14185 to zero but a different value maybe passed as the <seed> argument. This hash
14186 is known to be very good and very fast so it can be used to hash URLs and/or
14187 URL parameters for use as stick-table keys to collect statistics with a low
14188 collision rate, though care must be taken as the algorithm is not considered
14189 as cryptographically secure.
14190
14191xxh64([<seed>])
14192 Hashes a binary input sample into a signed 64-bit quantity using the 64-bit
14193 variant of the XXHash hash function. This hash supports a seed which defaults
14194 to zero but a different value maybe passed as the <seed> argument. This hash
14195 is known to be very good and very fast so it can be used to hash URLs and/or
14196 URL parameters for use as stick-table keys to collect statistics with a low
14197 collision rate, though care must be taken as the algorithm is not considered
14198 as cryptographically secure.
14199
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010014200
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200142017.3.2. Fetching samples from internal states
Willy Tarreau74ca5042013-06-11 23:12:07 +020014202--------------------------------------------
14203
14204A first set of sample fetch methods applies to internal information which does
14205not even relate to any client information. These ones are sometimes used with
14206"monitor-fail" directives to report an internal status to external watchers.
14207The sample fetch methods described in this section are usable anywhere.
14208
14209always_false : boolean
14210 Always returns the boolean "false" value. It may be used with ACLs as a
14211 temporary replacement for another one when adjusting configurations.
14212
14213always_true : boolean
14214 Always returns the boolean "true" value. It may be used with ACLs as a
14215 temporary replacement for another one when adjusting configurations.
14216
14217avg_queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010014218 Returns the total number of queued connections of the designated backend
Willy Tarreau74ca5042013-06-11 23:12:07 +020014219 divided by the number of active servers. The current backend is used if no
14220 backend is specified. This is very similar to "queue" except that the size of
14221 the farm is considered, in order to give a more accurate measurement of the
14222 time it may take for a new connection to be processed. The main usage is with
14223 ACL to return a sorry page to new users when it becomes certain they will get
14224 a degraded service, or to pass to the backend servers in a header so that
14225 they decide to work in degraded mode or to disable some functions to speed up
14226 the processing a bit. Note that in the event there would not be any active
14227 server anymore, twice the number of queued connections would be considered as
14228 the measured value. This is a fair estimate, as we expect one server to get
14229 back soon anyway, but we still prefer to send new traffic to another backend
14230 if in better shape. See also the "queue", "be_conn", and "be_sess_rate"
14231 sample fetches.
Krzysztof Piotr Oledzki346f76d2010-01-12 21:59:30 +010014232
Willy Tarreau74ca5042013-06-11 23:12:07 +020014233be_conn([<backend>]) : integer
Willy Tarreaua36af912009-10-10 12:02:45 +020014234 Applies to the number of currently established connections on the backend,
14235 possibly including the connection being evaluated. If no backend name is
14236 specified, the current one is used. But it is also possible to check another
14237 backend. It can be used to use a specific farm when the nominal one is full.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040014238 See also the "fe_conn", "queue", "be_conn_free", and "be_sess_rate" criteria.
14239
14240be_conn_free([<backend>]) : integer
14241 Returns an integer value corresponding to the number of available connections
14242 across available servers in the backend. Queue slots are not included. Backup
14243 servers are also not included, unless all other servers are down. If no
14244 backend name is specified, the current one is used. But it is also possible
14245 to check another backend. It can be used to use a specific farm when the
Patrick Hemmer155e93e2018-06-14 18:01:35 -040014246 nominal one is full. See also the "be_conn", "connslots", and "srv_conn_free"
14247 criteria.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040014248
14249 OTHER CAVEATS AND NOTES: if any of the server maxconn, or maxqueue is 0
14250 (meaning unlimited), then this fetch clearly does not make sense, in which
14251 case the value returned will be -1.
Willy Tarreau6a06a402007-07-15 20:15:28 +020014252
Willy Tarreau74ca5042013-06-11 23:12:07 +020014253be_sess_rate([<backend>]) : integer
14254 Returns an integer value corresponding to the sessions creation rate on the
14255 backend, in number of new sessions per second. This is used with ACLs to
14256 switch to an alternate backend when an expensive or fragile one reaches too
Davor Ocelice9ed2812017-12-25 17:49:28 +010014257 high a session rate, or to limit abuse of service (e.g. prevent sucking of an
Willy Tarreau74ca5042013-06-11 23:12:07 +020014258 online dictionary). It can also be useful to add this element to logs using a
14259 log-format directive.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014260
14261 Example :
14262 # Redirect to an error page if the dictionary is requested too often
14263 backend dynamic
14264 mode http
14265 acl being_scanned be_sess_rate gt 100
14266 redirect location /denied.html if being_scanned
Willy Tarreau0ba27502007-12-24 16:55:16 +010014267
Davor Ocelice9ed2812017-12-25 17:49:28 +010014268bin(<hex>) : bin
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014269 Returns a binary chain. The input is the hexadecimal representation
14270 of the string.
14271
14272bool(<bool>) : bool
14273 Returns a boolean value. <bool> can be 'true', 'false', '1' or '0'.
14274 'false' and '0' are the same. 'true' and '1' are the same.
14275
Willy Tarreau74ca5042013-06-11 23:12:07 +020014276connslots([<backend>]) : integer
14277 Returns an integer value corresponding to the number of connection slots
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014278 still available in the backend, by totaling the maximum amount of
Willy Tarreau74ca5042013-06-11 23:12:07 +020014279 connections on all servers and the maximum queue size. This is probably only
14280 used with ACLs.
Tait Clarridge7896d522012-12-05 21:39:31 -050014281
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014282 The basic idea here is to be able to measure the number of connection "slots"
Willy Tarreau55165fe2009-05-10 12:02:55 +020014283 still available (connection + queue), so that anything beyond that (intended
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014284 usage; see "use_backend" keyword) can be redirected to a different backend.
14285
Willy Tarreau55165fe2009-05-10 12:02:55 +020014286 'connslots' = number of available server connection slots, + number of
14287 available server queue slots.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014288
Willy Tarreaua36af912009-10-10 12:02:45 +020014289 Note that while "fe_conn" may be used, "connslots" comes in especially
Willy Tarreau55165fe2009-05-10 12:02:55 +020014290 useful when you have a case of traffic going to one single ip, splitting into
Willy Tarreau74ca5042013-06-11 23:12:07 +020014291 multiple backends (perhaps using ACLs to do name-based load balancing) and
Willy Tarreau55165fe2009-05-10 12:02:55 +020014292 you want to be able to differentiate between different backends, and their
Davor Ocelice9ed2812017-12-25 17:49:28 +010014293 available "connslots". Also, whereas "nbsrv" only measures servers that are
Willy Tarreau74ca5042013-06-11 23:12:07 +020014294 actually *down*, this fetch is more fine-grained and looks into the number of
Willy Tarreaua36af912009-10-10 12:02:45 +020014295 available connection slots as well. See also "queue" and "avg_queue".
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014296
Willy Tarreau55165fe2009-05-10 12:02:55 +020014297 OTHER CAVEATS AND NOTES: at this point in time, the code does not take care
14298 of dynamic connections. Also, if any of the server maxconn, or maxqueue is 0,
Willy Tarreau74ca5042013-06-11 23:12:07 +020014299 then this fetch clearly does not make sense, in which case the value returned
Willy Tarreau55165fe2009-05-10 12:02:55 +020014300 will be -1.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014301
Willy Tarreau70fe9442018-11-22 16:07:39 +010014302cpu_calls : integer
14303 Returns the number of calls to the task processing the stream or current
14304 request since it was allocated. This number is reset for each new request on
14305 the same connections in case of HTTP keep-alive. This value should usually be
14306 low and stable (around 2 calls for a typically simple request) but may become
14307 high if some processing (compression, caching or analysis) is performed. This
14308 is purely for performance monitoring purposes.
14309
14310cpu_ns_avg : integer
14311 Returns the average number of nanoseconds spent in each call to the task
14312 processing the stream or current request. This number is reset for each new
14313 request on the same connections in case of HTTP keep-alive. This value
14314 indicates the overall cost of processing the request or the connection for
14315 each call. There is no good nor bad value but the time spent in a call
14316 automatically causes latency for other processing (see lat_ns_avg below),
14317 and may affect other connection's apparent response time. Certain operations
14318 like compression, complex regex matching or heavy Lua operations may directly
14319 affect this value, and having it in the logs will make it easier to spot the
14320 faulty processing that needs to be fixed to recover decent performance.
14321 Note: this value is exactly cpu_ns_tot divided by cpu_calls.
14322
14323cpu_ns_tot : integer
14324 Returns the total number of nanoseconds spent in each call to the task
14325 processing the stream or current request. This number is reset for each new
14326 request on the same connections in case of HTTP keep-alive. This value
14327 indicates the overall cost of processing the request or the connection for
14328 each call. There is no good nor bad value but the time spent in a call
14329 automatically causes latency for other processing (see lat_ns_avg below),
14330 induces CPU costs on the machine, and may affect other connection's apparent
14331 response time. Certain operations like compression, complex regex matching or
14332 heavy Lua operations may directly affect this value, and having it in the
14333 logs will make it easier to spot the faulty processing that needs to be fixed
14334 to recover decent performance. The value may be artificially high due to a
14335 high cpu_calls count, for example when processing many HTTP chunks, and for
14336 this reason it is often preferred to log cpu_ns_avg instead.
14337
Cyril Bonté6bcd1822019-11-05 23:13:59 +010014338date([<offset>],[<unit>]) : integer
Willy Tarreau6236d3a2013-07-25 14:28:25 +020014339 Returns the current date as the epoch (number of seconds since 01/01/1970).
Damien Claisseae6f1252019-10-30 15:57:28 +000014340
14341 If an offset value is specified, then it is added to the current date before
14342 returning the value. This is particularly useful to compute relative dates,
14343 as both positive and negative offsets are allowed.
Willy Tarreau276fae92013-07-25 14:36:01 +020014344 It is useful combined with the http_date converter.
14345
Damien Claisseae6f1252019-10-30 15:57:28 +000014346 <unit> is facultative, and can be set to "s" for seconds (default behavior),
14347 "ms" for milliseconds or "us" for microseconds.
14348 If unit is set, return value is an integer reflecting either seconds,
14349 milliseconds or microseconds since epoch, plus offset.
14350 It is useful when a time resolution of less than a second is needed.
14351
Willy Tarreau276fae92013-07-25 14:36:01 +020014352 Example :
14353
14354 # set an expires header to now+1 hour in every response
14355 http-response set-header Expires %[date(3600),http_date]
Willy Tarreau6236d3a2013-07-25 14:28:25 +020014356
Damien Claisseae6f1252019-10-30 15:57:28 +000014357 # set an expires header to now+1 hour in every response, with
14358 # millisecond granularity
14359 http-response set-header Expires %[date(3600000,ms),http_date(0,ms)]
14360
Etienne Carrierea792a0a2018-01-17 13:43:24 +010014361date_us : integer
14362 Return the microseconds part of the date (the "second" part is returned by
14363 date sample). This sample is coherent with the date sample as it is comes
14364 from the same timeval structure.
14365
Willy Tarreaud716f9b2017-10-13 11:03:15 +020014366distcc_body(<token>[,<occ>]) : binary
14367 Parses a distcc message and returns the body associated to occurrence #<occ>
14368 of the token <token>. Occurrences start at 1, and when unspecified, any may
14369 match though in practice only the first one is checked for now. This can be
14370 used to extract file names or arguments in files built using distcc through
14371 haproxy. Please refer to distcc's protocol documentation for the complete
14372 list of supported tokens.
14373
14374distcc_param(<token>[,<occ>]) : integer
14375 Parses a distcc message and returns the parameter associated to occurrence
14376 #<occ> of the token <token>. Occurrences start at 1, and when unspecified,
14377 any may match though in practice only the first one is checked for now. This
14378 can be used to extract certain information such as the protocol version, the
14379 file size or the argument in files built using distcc through haproxy.
14380 Another use case consists in waiting for the start of the preprocessed file
14381 contents before connecting to the server to avoid keeping idle connections.
14382 Please refer to distcc's protocol documentation for the complete list of
14383 supported tokens.
14384
14385 Example :
14386 # wait up to 20s for the pre-processed file to be uploaded
14387 tcp-request inspect-delay 20s
14388 tcp-request content accept if { distcc_param(DOTI) -m found }
14389 # send large files to the big farm
14390 use_backend big_farm if { distcc_param(DOTI) gt 1000000 }
14391
Willy Tarreau595ec542013-06-12 21:34:28 +020014392env(<name>) : string
14393 Returns a string containing the value of environment variable <name>. As a
14394 reminder, environment variables are per-process and are sampled when the
14395 process starts. This can be useful to pass some information to a next hop
14396 server, or with ACLs to take specific action when the process is started a
14397 certain way.
14398
14399 Examples :
14400 # Pass the Via header to next hop with the local hostname in it
14401 http-request add-header Via 1.1\ %[env(HOSTNAME)]
14402
14403 # reject cookie-less requests when the STOP environment variable is set
14404 http-request deny if !{ cook(SESSIONID) -m found } { env(STOP) -m found }
14405
Willy Tarreau74ca5042013-06-11 23:12:07 +020014406fe_conn([<frontend>]) : integer
14407 Returns the number of currently established connections on the frontend,
Willy Tarreaud63335a2010-02-26 12:56:52 +010014408 possibly including the connection being evaluated. If no frontend name is
14409 specified, the current one is used. But it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020014410 frontend. It can be used to return a sorry page before hard-blocking, or to
14411 use a specific backend to drain new requests when the farm is considered
Davor Ocelice9ed2812017-12-25 17:49:28 +010014412 full. This is mostly used with ACLs but can also be used to pass some
Willy Tarreau74ca5042013-06-11 23:12:07 +020014413 statistics to servers in HTTP headers. See also the "dst_conn", "be_conn",
14414 "fe_sess_rate" fetches.
Willy Tarreaua36af912009-10-10 12:02:45 +020014415
Nenad Merdanovicad9a7e92016-10-03 04:57:37 +020014416fe_req_rate([<frontend>]) : integer
14417 Returns an integer value corresponding to the number of HTTP requests per
14418 second sent to a frontend. This number can differ from "fe_sess_rate" in
14419 situations where client-side keep-alive is enabled.
14420
Willy Tarreau74ca5042013-06-11 23:12:07 +020014421fe_sess_rate([<frontend>]) : integer
14422 Returns an integer value corresponding to the sessions creation rate on the
14423 frontend, in number of new sessions per second. This is used with ACLs to
14424 limit the incoming session rate to an acceptable range in order to prevent
14425 abuse of service at the earliest moment, for example when combined with other
14426 layer 4 ACLs in order to force the clients to wait a bit for the rate to go
14427 down below the limit. It can also be useful to add this element to logs using
14428 a log-format directive. See also the "rate-limit sessions" directive for use
14429 in frontends.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010014430
14431 Example :
14432 # This frontend limits incoming mails to 10/s with a max of 100
14433 # concurrent connections. We accept any connection below 10/s, and
14434 # force excess clients to wait for 100 ms. Since clients are limited to
14435 # 100 max, there cannot be more than 10 incoming mails per second.
14436 frontend mail
14437 bind :25
14438 mode tcp
14439 maxconn 100
14440 acl too_fast fe_sess_rate ge 10
14441 tcp-request inspect-delay 100ms
14442 tcp-request content accept if ! too_fast
14443 tcp-request content accept if WAIT_END
Willy Tarreaud72758d2010-01-12 10:42:19 +010014444
Nenad Merdanovic807a6e72017-03-12 22:00:00 +010014445hostname : string
14446 Returns the system hostname.
14447
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014448int(<integer>) : signed integer
14449 Returns a signed integer.
14450
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014451ipv4(<ipv4>) : ipv4
14452 Returns an ipv4.
14453
14454ipv6(<ipv6>) : ipv6
14455 Returns an ipv6.
14456
Willy Tarreau70fe9442018-11-22 16:07:39 +010014457lat_ns_avg : integer
14458 Returns the average number of nanoseconds spent between the moment the task
14459 handling the stream is woken up and the moment it is effectively called. This
14460 number is reset for each new request on the same connections in case of HTTP
14461 keep-alive. This value indicates the overall latency inflicted to the current
14462 request by all other requests being processed in parallel, and is a direct
14463 indicator of perceived performance due to noisy neighbours. In order to keep
14464 the value low, it is possible to reduce the scheduler's run queue depth using
14465 "tune.runqueue-depth", to reduce the number of concurrent events processed at
14466 once using "tune.maxpollevents", to decrease the stream's nice value using
14467 the "nice" option on the "bind" lines or in the frontend, or to look for
14468 other heavy requests in logs (those exhibiting large values of "cpu_ns_avg"),
14469 whose processing needs to be adjusted or fixed. Compression of large buffers
14470 could be a culprit, like heavy regex or long lists of regex.
14471 Note: this value is exactly lat_ns_tot divided by cpu_calls.
14472
14473lat_ns_tot : integer
14474 Returns the total number of nanoseconds spent between the moment the task
14475 handling the stream is woken up and the moment it is effectively called. This
14476 number is reset for each new request on the same connections in case of HTTP
14477 keep-alive. This value indicates the overall latency inflicted to the current
14478 request by all other requests being processed in parallel, and is a direct
14479 indicator of perceived performance due to noisy neighbours. In order to keep
14480 the value low, it is possible to reduce the scheduler's run queue depth using
14481 "tune.runqueue-depth", to reduce the number of concurrent events processed at
14482 once using "tune.maxpollevents", to decrease the stream's nice value using
14483 the "nice" option on the "bind" lines or in the frontend, or to look for
14484 other heavy requests in logs (those exhibiting large values of "cpu_ns_avg"),
14485 whose processing needs to be adjusted or fixed. Compression of large buffers
14486 could be a culprit, like heavy regex or long lists of regex. Note: while it
14487 may intuitively seem that the total latency adds to a transfer time, it is
14488 almost never true because while a task waits for the CPU, network buffers
14489 continue to fill up and the next call will process more at once. The value
14490 may be artificially high due to a high cpu_calls count, for example when
14491 processing many HTTP chunks, and for this reason it is often preferred to log
14492 lat_ns_avg instead, which is a more relevant performance indicator.
14493
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014494meth(<method>) : method
14495 Returns a method.
14496
Willy Tarreau0f30d262014-11-24 16:02:05 +010014497nbproc : integer
14498 Returns an integer value corresponding to the number of processes that were
14499 started (it equals the global "nbproc" setting). This is useful for logging
14500 and debugging purposes.
14501
Willy Tarreau74ca5042013-06-11 23:12:07 +020014502nbsrv([<backend>]) : integer
14503 Returns an integer value corresponding to the number of usable servers of
14504 either the current backend or the named backend. This is mostly used with
14505 ACLs but can also be useful when added to logs. This is normally used to
Willy Tarreaud63335a2010-02-26 12:56:52 +010014506 switch to an alternate backend when the number of servers is too low to
14507 to handle some load. It is useful to report a failure when combined with
14508 "monitor fail".
Willy Tarreau079ff0a2009-03-05 21:34:28 +010014509
Patrick Hemmerfabb24f2018-08-13 14:07:57 -040014510prio_class : integer
14511 Returns the priority class of the current session for http mode or connection
14512 for tcp mode. The value will be that set by the last call to "http-request
14513 set-priority-class" or "tcp-request content set-priority-class".
14514
14515prio_offset : integer
14516 Returns the priority offset of the current session for http mode or
14517 connection for tcp mode. The value will be that set by the last call to
14518 "http-request set-priority-offset" or "tcp-request content
14519 set-priority-offset".
14520
Willy Tarreau0f30d262014-11-24 16:02:05 +010014521proc : integer
14522 Returns an integer value corresponding to the position of the process calling
14523 the function, between 1 and global.nbproc. This is useful for logging and
14524 debugging purposes.
14525
Willy Tarreau74ca5042013-06-11 23:12:07 +020014526queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010014527 Returns the total number of queued connections of the designated backend,
14528 including all the connections in server queues. If no backend name is
14529 specified, the current one is used, but it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020014530 one. This is useful with ACLs or to pass statistics to backend servers. This
14531 can be used to take actions when queuing goes above a known level, generally
14532 indicating a surge of traffic or a massive slowdown on the servers. One
14533 possible action could be to reject new users but still accept old ones. See
14534 also the "avg_queue", "be_conn", and "be_sess_rate" fetches.
14535
Willy Tarreau84310e22014-02-14 11:59:04 +010014536rand([<range>]) : integer
14537 Returns a random integer value within a range of <range> possible values,
14538 starting at zero. If the range is not specified, it defaults to 2^32, which
14539 gives numbers between 0 and 4294967295. It can be useful to pass some values
14540 needed to take some routing decisions for example, or just for debugging
14541 purposes. This random must not be used for security purposes.
14542
Luca Schimweg8a694b82019-09-10 15:42:52 +020014543uuid([<version>]) : string
14544 Returns a UUID following the RFC4122 standard. If the version is not
14545 specified, a UUID version 4 (fully random) is returned.
14546 Currently, only version 4 is supported.
14547
Willy Tarreau74ca5042013-06-11 23:12:07 +020014548srv_conn([<backend>/]<server>) : integer
14549 Returns an integer value corresponding to the number of currently established
14550 connections on the designated server, possibly including the connection being
14551 evaluated. If <backend> is omitted, then the server is looked up in the
14552 current backend. It can be used to use a specific farm when one server is
14553 full, or to inform the server about our view of the number of active
Patrick Hemmer155e93e2018-06-14 18:01:35 -040014554 connections with it. See also the "fe_conn", "be_conn", "queue", and
14555 "srv_conn_free" fetch methods.
14556
14557srv_conn_free([<backend>/]<server>) : integer
14558 Returns an integer value corresponding to the number of available connections
14559 on the designated server, possibly including the connection being evaluated.
14560 The value does not include queue slots. If <backend> is omitted, then the
14561 server is looked up in the current backend. It can be used to use a specific
14562 farm when one server is full, or to inform the server about our view of the
14563 number of active connections with it. See also the "be_conn_free" and
14564 "srv_conn" fetch methods.
14565
14566 OTHER CAVEATS AND NOTES: If the server maxconn is 0, then this fetch clearly
14567 does not make sense, in which case the value returned will be -1.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014568
14569srv_is_up([<backend>/]<server>) : boolean
14570 Returns true when the designated server is UP, and false when it is either
14571 DOWN or in maintenance mode. If <backend> is omitted, then the server is
14572 looked up in the current backend. It is mainly used to take action based on
Davor Ocelice9ed2812017-12-25 17:49:28 +010014573 an external status reported via a health check (e.g. a geographical site's
Willy Tarreau74ca5042013-06-11 23:12:07 +020014574 availability). Another possible use which is more of a hack consists in
14575 using dummy servers as boolean variables that can be enabled or disabled from
14576 the CLI, so that rules depending on those ACLs can be tweaked in realtime.
14577
Willy Tarreauff2b7af2017-10-13 11:46:26 +020014578srv_queue([<backend>/]<server>) : integer
14579 Returns an integer value corresponding to the number of connections currently
14580 pending in the designated server's queue. If <backend> is omitted, then the
14581 server is looked up in the current backend. It can sometimes be used together
14582 with the "use-server" directive to force to use a known faster server when it
14583 is not much loaded. See also the "srv_conn", "avg_queue" and "queue" sample
14584 fetch methods.
14585
Willy Tarreau74ca5042013-06-11 23:12:07 +020014586srv_sess_rate([<backend>/]<server>) : integer
14587 Returns an integer corresponding to the sessions creation rate on the
14588 designated server, in number of new sessions per second. If <backend> is
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014589 omitted, then the server is looked up in the current backend. This is mostly
Willy Tarreau74ca5042013-06-11 23:12:07 +020014590 used with ACLs but can make sense with logs too. This is used to switch to an
14591 alternate backend when an expensive or fragile one reaches too high a session
Davor Ocelice9ed2812017-12-25 17:49:28 +010014592 rate, or to limit abuse of service (e.g. prevent latent requests from
Willy Tarreau74ca5042013-06-11 23:12:07 +020014593 overloading servers).
14594
14595 Example :
14596 # Redirect to a separate back
14597 acl srv1_full srv_sess_rate(be1/srv1) gt 50
14598 acl srv2_full srv_sess_rate(be1/srv2) gt 50
14599 use_backend be2 if srv1_full or srv2_full
14600
Willy Tarreau0f30d262014-11-24 16:02:05 +010014601stopping : boolean
14602 Returns TRUE if the process calling the function is currently stopping. This
14603 can be useful for logging, or for relaxing certain checks or helping close
14604 certain connections upon graceful shutdown.
14605
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014606str(<string>) : string
14607 Returns a string.
14608
Willy Tarreau74ca5042013-06-11 23:12:07 +020014609table_avl([<table>]) : integer
14610 Returns the total number of available entries in the current proxy's
14611 stick-table or in the designated stick-table. See also table_cnt.
14612
14613table_cnt([<table>]) : integer
14614 Returns the total number of entries currently in use in the current proxy's
14615 stick-table or in the designated stick-table. See also src_conn_cnt and
14616 table_avl for other entry counting methods.
14617
Christopher Faulet34adb2a2017-11-21 21:45:38 +010014618thread : integer
14619 Returns an integer value corresponding to the position of the thread calling
14620 the function, between 0 and (global.nbthread-1). This is useful for logging
14621 and debugging purposes.
14622
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014623var(<var-name>) : undefined
14624 Returns a variable with the stored type. If the variable is not set, the
Daniel Schneller0b547052016-03-21 20:46:57 +010014625 sample fetch fails. The name of the variable starts with an indication
14626 about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010014627 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010014628 "sess" : the variable is shared with the whole session
14629 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014630 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010014631 "req" : the variable is shared only during request processing,
14632 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014633 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014634 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014635
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200146367.3.3. Fetching samples at Layer 4
Willy Tarreau74ca5042013-06-11 23:12:07 +020014637----------------------------------
14638
14639The layer 4 usually describes just the transport layer which in haproxy is
14640closest to the connection, where no content is yet made available. The fetch
14641methods described here are usable as low as the "tcp-request connection" rule
14642sets unless they require some future information. Those generally include
14643TCP/IP addresses and ports, as well as elements from stick-tables related to
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014644the incoming connection. For retrieving a value from a sticky counters, the
14645counter number can be explicitly set as 0, 1, or 2 using the pre-defined
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020014646"sc0_", "sc1_", or "sc2_" prefix. These three pre-defined prefixes can only be
14647used if MAX_SESS_STKCTR value does not exceed 3, otherwise the counter number
14648can be specified as the first integer argument when using the "sc_" prefix.
14649Starting from "sc_0" to "sc_N" where N is (MAX_SESS_STKCTR-1). An optional
14650table may be specified with the "sc*" form, in which case the currently
14651tracked key will be looked up into this alternate table instead of the table
14652currently being tracked.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014653
Jérôme Magnin35e53a62019-01-16 14:38:37 +010014654bc_http_major : integer
Jérôme Magnin86577422018-12-07 09:03:11 +010014655 Returns the backend connection's HTTP major version encoding, which may be 1
14656 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
14657 encoding and not the version present in the request header.
14658
Willy Tarreau74ca5042013-06-11 23:12:07 +020014659be_id : integer
14660 Returns an integer containing the current backend's id. It can be used in
14661 frontends with responses to check which backend processed the request.
14662
Marcin Deranekd2471c22016-12-12 14:08:05 +010014663be_name : string
14664 Returns a string containing the current backend's name. It can be used in
14665 frontends with responses to check which backend processed the request.
14666
Willy Tarreau74ca5042013-06-11 23:12:07 +020014667dst : ip
14668 This is the destination IPv4 address of the connection on the client side,
14669 which is the address the client connected to. It can be useful when running
14670 in transparent mode. It is of type IP and works on both IPv4 and IPv6 tables.
14671 On IPv6 tables, IPv4 address is mapped to its IPv6 equivalent, according to
Willy Tarreau64ded3d2019-01-23 10:02:15 +010014672 RFC 4291. When the incoming connection passed through address translation or
14673 redirection involving connection tracking, the original destination address
14674 before the redirection will be reported. On Linux systems, the source and
14675 destination may seldom appear reversed if the nf_conntrack_tcp_loose sysctl
14676 is set, because a late response may reopen a timed out connection and switch
14677 what is believed to be the source and the destination.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014678
14679dst_conn : integer
14680 Returns an integer value corresponding to the number of currently established
14681 connections on the same socket including the one being evaluated. It is
14682 normally used with ACLs but can as well be used to pass the information to
14683 servers in an HTTP header or in logs. It can be used to either return a sorry
14684 page before hard-blocking, or to use a specific backend to drain new requests
14685 when the socket is considered saturated. This offers the ability to assign
14686 different limits to different listening ports or addresses. See also the
14687 "fe_conn" and "be_conn" fetches.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014688
Willy Tarreau16e01562016-08-09 16:46:18 +020014689dst_is_local : boolean
14690 Returns true if the destination address of the incoming connection is local
14691 to the system, or false if the address doesn't exist on the system, meaning
14692 that it was intercepted in transparent mode. It can be useful to apply
14693 certain rules by default to forwarded traffic and other rules to the traffic
Davor Ocelice9ed2812017-12-25 17:49:28 +010014694 targeting the real address of the machine. For example the stats page could
Willy Tarreau16e01562016-08-09 16:46:18 +020014695 be delivered only on this address, or SSH access could be locally redirected.
14696 Please note that the check involves a few system calls, so it's better to do
14697 it only once per connection.
14698
Willy Tarreau74ca5042013-06-11 23:12:07 +020014699dst_port : integer
14700 Returns an integer value corresponding to the destination TCP port of the
14701 connection on the client side, which is the port the client connected to.
14702 This might be used when running in transparent mode, when assigning dynamic
14703 ports to some clients for a whole application session, to stick all users to
14704 a same server, or to pass the destination port information to a server using
14705 an HTTP header.
14706
Willy Tarreau60ca10a2017-08-18 15:26:54 +020014707fc_http_major : integer
14708 Reports the front connection's HTTP major version encoding, which may be 1
14709 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
14710 encoding and not on the version present in the request header.
14711
Geoff Simmons7185b782019-08-27 18:31:16 +020014712fc_pp_authority : string
14713 Returns the authority TLV sent by the client in the PROXY protocol header,
14714 if any.
14715
Emeric Brun4f603012017-01-05 15:11:44 +010014716fc_rcvd_proxy : boolean
14717 Returns true if the client initiated the connection with a PROXY protocol
14718 header.
14719
Thierry Fournier / OZON.IO6310bef2016-07-24 20:16:50 +020014720fc_rtt(<unit>) : integer
14721 Returns the Round Trip Time (RTT) measured by the kernel for the client
14722 connection. <unit> is facultative, by default the unit is milliseconds. <unit>
14723 can be set to "ms" for milliseconds or "us" for microseconds. If the server
14724 connection is not established, if the connection is not TCP or if the
14725 operating system does not support TCP_INFO, for example Linux kernels before
14726 2.4, the sample fetch fails.
14727
14728fc_rttvar(<unit>) : integer
14729 Returns the Round Trip Time (RTT) variance measured by the kernel for the
14730 client connection. <unit> is facultative, by default the unit is milliseconds.
14731 <unit> can be set to "ms" for milliseconds or "us" for microseconds. If the
14732 server connection is not established, if the connection is not TCP or if the
14733 operating system does not support TCP_INFO, for example Linux kernels before
14734 2.4, the sample fetch fails.
14735
Christopher Fauletba0c53e2019-10-17 14:40:48 +020014736fc_unacked : integer
Joe Williams30fcd392016-08-10 07:06:44 -070014737 Returns the unacked counter measured by the kernel for the client connection.
14738 If the server connection is not established, if the connection is not TCP or
14739 if the operating system does not support TCP_INFO, for example Linux kernels
14740 before 2.4, the sample fetch fails.
14741
Christopher Fauletba0c53e2019-10-17 14:40:48 +020014742fc_sacked : integer
Joe Williams30fcd392016-08-10 07:06:44 -070014743 Returns the sacked counter measured by the kernel for the client connection.
14744 If the server connection is not established, if the connection is not TCP or
14745 if the operating system does not support TCP_INFO, for example Linux kernels
14746 before 2.4, the sample fetch fails.
14747
Christopher Fauletba0c53e2019-10-17 14:40:48 +020014748fc_retrans : integer
Joe Williams30fcd392016-08-10 07:06:44 -070014749 Returns the retransmits counter measured by the kernel for the client
14750 connection. If the server connection is not established, if the connection is
14751 not TCP or if the operating system does not support TCP_INFO, for example
14752 Linux kernels before 2.4, the sample fetch fails.
14753
Christopher Fauletba0c53e2019-10-17 14:40:48 +020014754fc_fackets : integer
Joe Williams30fcd392016-08-10 07:06:44 -070014755 Returns the fack counter measured by the kernel for the client
14756 connection. If the server connection is not established, if the connection is
14757 not TCP or if the operating system does not support TCP_INFO, for example
14758 Linux kernels before 2.4, the sample fetch fails.
14759
Christopher Fauletba0c53e2019-10-17 14:40:48 +020014760fc_lost : integer
Joe Williams30fcd392016-08-10 07:06:44 -070014761 Returns the lost counter measured by the kernel for the client
14762 connection. If the server connection is not established, if the connection is
14763 not TCP or if the operating system does not support TCP_INFO, for example
14764 Linux kernels before 2.4, the sample fetch fails.
14765
Christopher Fauletba0c53e2019-10-17 14:40:48 +020014766fc_reordering : integer
Joe Williams30fcd392016-08-10 07:06:44 -070014767 Returns the reordering counter measured by the kernel for the client
14768 connection. If the server connection is not established, if the connection is
14769 not TCP or if the operating system does not support TCP_INFO, for example
14770 Linux kernels before 2.4, the sample fetch fails.
14771
Marcin Deranek9a66dfb2018-04-13 14:37:50 +020014772fe_defbe : string
14773 Returns a string containing the frontend's default backend name. It can be
14774 used in frontends to check which backend will handle requests by default.
14775
Willy Tarreau74ca5042013-06-11 23:12:07 +020014776fe_id : integer
14777 Returns an integer containing the current frontend's id. It can be used in
Marcin Deranek6e413ed2016-12-13 12:40:01 +010014778 backends to check from which frontend it was called, or to stick all users
Willy Tarreau74ca5042013-06-11 23:12:07 +020014779 coming via a same frontend to the same server.
14780
Marcin Deranekd2471c22016-12-12 14:08:05 +010014781fe_name : string
14782 Returns a string containing the current frontend's name. It can be used in
14783 backends to check from which frontend it was called, or to stick all users
14784 coming via a same frontend to the same server.
14785
Cyril Bonté62ba8702014-04-22 23:52:25 +020014786sc_bytes_in_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014787sc0_bytes_in_rate([<table>]) : integer
14788sc1_bytes_in_rate([<table>]) : integer
14789sc2_bytes_in_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014790 Returns the average client-to-server bytes rate from the currently tracked
14791 counters, measured in amount of bytes over the period configured in the
14792 table. See also src_bytes_in_rate.
14793
Cyril Bonté62ba8702014-04-22 23:52:25 +020014794sc_bytes_out_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014795sc0_bytes_out_rate([<table>]) : integer
14796sc1_bytes_out_rate([<table>]) : integer
14797sc2_bytes_out_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014798 Returns the average server-to-client bytes rate from the currently tracked
14799 counters, measured in amount of bytes over the period configured in the
14800 table. See also src_bytes_out_rate.
14801
Cyril Bonté62ba8702014-04-22 23:52:25 +020014802sc_clr_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014803sc0_clr_gpc0([<table>]) : integer
14804sc1_clr_gpc0([<table>]) : integer
14805sc2_clr_gpc0([<table>]) : integer
Willy Tarreauf73cd112011-08-13 01:45:16 +020014806 Clears the first General Purpose Counter associated to the currently tracked
14807 counters, and returns its previous value. Before the first invocation, the
Willy Tarreau869948b2013-01-04 14:14:57 +010014808 stored value is zero, so first invocation will always return zero. This is
14809 typically used as a second ACL in an expression in order to mark a connection
14810 when a first ACL was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020014811
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014812 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020014813 # block if 5 consecutive requests continue to come faster than 10 sess
14814 # per second, and reset the counter as soon as the traffic slows down.
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014815 acl abuse sc0_http_req_rate gt 10
14816 acl kill sc0_inc_gpc0 gt 5
14817 acl save sc0_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020014818 tcp-request connection accept if !abuse save
14819 tcp-request connection reject if abuse kill
14820
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014821sc_clr_gpc1(<ctr>[,<table>]) : integer
14822sc0_clr_gpc1([<table>]) : integer
14823sc1_clr_gpc1([<table>]) : integer
14824sc2_clr_gpc1([<table>]) : integer
14825 Clears the second General Purpose Counter associated to the currently tracked
14826 counters, and returns its previous value. Before the first invocation, the
14827 stored value is zero, so first invocation will always return zero. This is
14828 typically used as a second ACL in an expression in order to mark a connection
14829 when a first ACL was verified.
14830
Cyril Bonté62ba8702014-04-22 23:52:25 +020014831sc_conn_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014832sc0_conn_cnt([<table>]) : integer
14833sc1_conn_cnt([<table>]) : integer
14834sc2_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014835 Returns the cumulative number of incoming connections from currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020014836 counters. See also src_conn_cnt.
14837
Cyril Bonté62ba8702014-04-22 23:52:25 +020014838sc_conn_cur(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014839sc0_conn_cur([<table>]) : integer
14840sc1_conn_cur([<table>]) : integer
14841sc2_conn_cur([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014842 Returns the current amount of concurrent connections tracking the same
14843 tracked counters. This number is automatically incremented when tracking
14844 begins and decremented when tracking stops. See also src_conn_cur.
14845
Cyril Bonté62ba8702014-04-22 23:52:25 +020014846sc_conn_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014847sc0_conn_rate([<table>]) : integer
14848sc1_conn_rate([<table>]) : integer
14849sc2_conn_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014850 Returns the average connection rate from the currently tracked counters,
14851 measured in amount of connections over the period configured in the table.
14852 See also src_conn_rate.
14853
Cyril Bonté62ba8702014-04-22 23:52:25 +020014854sc_get_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014855sc0_get_gpc0([<table>]) : integer
14856sc1_get_gpc0([<table>]) : integer
14857sc2_get_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014858 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014859 currently tracked counters. See also src_get_gpc0 and sc/sc0/sc1/sc2_inc_gpc0.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014860
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014861sc_get_gpc1(<ctr>[,<table>]) : integer
14862sc0_get_gpc1([<table>]) : integer
14863sc1_get_gpc1([<table>]) : integer
14864sc2_get_gpc1([<table>]) : integer
14865 Returns the value of the second General Purpose Counter associated to the
14866 currently tracked counters. See also src_get_gpc1 and sc/sc0/sc1/sc2_inc_gpc1.
14867
Thierry FOURNIER236657b2015-08-19 08:25:14 +020014868sc_get_gpt0(<ctr>[,<table>]) : integer
14869sc0_get_gpt0([<table>]) : integer
14870sc1_get_gpt0([<table>]) : integer
14871sc2_get_gpt0([<table>]) : integer
14872 Returns the value of the first General Purpose Tag associated to the
14873 currently tracked counters. See also src_get_gpt0.
14874
Cyril Bonté62ba8702014-04-22 23:52:25 +020014875sc_gpc0_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014876sc0_gpc0_rate([<table>]) : integer
14877sc1_gpc0_rate([<table>]) : integer
14878sc2_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014879 Returns the average increment rate of the first General Purpose Counter
14880 associated to the currently tracked counters. It reports the frequency
14881 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014882 src_gpc0_rate, sc/sc0/sc1/sc2_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
14883 that the "gpc0_rate" counter must be stored in the stick-table for a value to
14884 be returned, as "gpc0" only holds the event count.
Willy Tarreaue9656522010-08-17 15:40:09 +020014885
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014886sc_gpc1_rate(<ctr>[,<table>]) : integer
14887sc0_gpc1_rate([<table>]) : integer
14888sc1_gpc1_rate([<table>]) : integer
14889sc2_gpc1_rate([<table>]) : integer
14890 Returns the average increment rate of the second General Purpose Counter
14891 associated to the currently tracked counters. It reports the frequency
14892 which the gpc1 counter was incremented over the configured period. See also
14893 src_gpcA_rate, sc/sc0/sc1/sc2_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
14894 that the "gpc1_rate" counter must be stored in the stick-table for a value to
14895 be returned, as "gpc1" only holds the event count.
14896
Cyril Bonté62ba8702014-04-22 23:52:25 +020014897sc_http_err_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014898sc0_http_err_cnt([<table>]) : integer
14899sc1_http_err_cnt([<table>]) : integer
14900sc2_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014901 Returns the cumulative number of HTTP errors from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020014902 counters. This includes the both request errors and 4xx error responses.
14903 See also src_http_err_cnt.
14904
Cyril Bonté62ba8702014-04-22 23:52:25 +020014905sc_http_err_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014906sc0_http_err_rate([<table>]) : integer
14907sc1_http_err_rate([<table>]) : integer
14908sc2_http_err_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014909 Returns the average rate of HTTP errors from the currently tracked counters,
14910 measured in amount of errors over the period configured in the table. This
14911 includes the both request errors and 4xx error responses. See also
14912 src_http_err_rate.
14913
Cyril Bonté62ba8702014-04-22 23:52:25 +020014914sc_http_req_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014915sc0_http_req_cnt([<table>]) : integer
14916sc1_http_req_cnt([<table>]) : integer
14917sc2_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014918 Returns the cumulative number of HTTP requests from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020014919 counters. This includes every started request, valid or not. See also
14920 src_http_req_cnt.
14921
Cyril Bonté62ba8702014-04-22 23:52:25 +020014922sc_http_req_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014923sc0_http_req_rate([<table>]) : integer
14924sc1_http_req_rate([<table>]) : integer
14925sc2_http_req_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014926 Returns the average rate of HTTP requests from the currently tracked
14927 counters, measured in amount of requests over the period configured in
14928 the table. This includes every started request, valid or not. See also
14929 src_http_req_rate.
14930
Cyril Bonté62ba8702014-04-22 23:52:25 +020014931sc_inc_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014932sc0_inc_gpc0([<table>]) : integer
14933sc1_inc_gpc0([<table>]) : integer
14934sc2_inc_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014935 Increments the first General Purpose Counter associated to the currently
Willy Tarreau869948b2013-01-04 14:14:57 +010014936 tracked counters, and returns its new value. Before the first invocation,
14937 the stored value is zero, so first invocation will increase it to 1 and will
14938 return 1. This is typically used as a second ACL in an expression in order
14939 to mark a connection when a first ACL was verified :
Willy Tarreaue9656522010-08-17 15:40:09 +020014940
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014941 Example:
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014942 acl abuse sc0_http_req_rate gt 10
14943 acl kill sc0_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020014944 tcp-request connection reject if abuse kill
14945
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014946sc_inc_gpc1(<ctr>[,<table>]) : integer
14947sc0_inc_gpc1([<table>]) : integer
14948sc1_inc_gpc1([<table>]) : integer
14949sc2_inc_gpc1([<table>]) : integer
14950 Increments the second General Purpose Counter associated to the currently
14951 tracked counters, and returns its new value. Before the first invocation,
14952 the stored value is zero, so first invocation will increase it to 1 and will
14953 return 1. This is typically used as a second ACL in an expression in order
14954 to mark a connection when a first ACL was verified.
14955
Cyril Bonté62ba8702014-04-22 23:52:25 +020014956sc_kbytes_in(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014957sc0_kbytes_in([<table>]) : integer
14958sc1_kbytes_in([<table>]) : integer
14959sc2_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014960 Returns the total amount of client-to-server data from the currently tracked
14961 counters, measured in kilobytes. The test is currently performed on 32-bit
14962 integers, which limits values to 4 terabytes. See also src_kbytes_in.
Willy Tarreaue9656522010-08-17 15:40:09 +020014963
Cyril Bonté62ba8702014-04-22 23:52:25 +020014964sc_kbytes_out(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014965sc0_kbytes_out([<table>]) : integer
14966sc1_kbytes_out([<table>]) : integer
14967sc2_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014968 Returns the total amount of server-to-client data from the currently tracked
14969 counters, measured in kilobytes. The test is currently performed on 32-bit
14970 integers, which limits values to 4 terabytes. See also src_kbytes_out.
Willy Tarreaue9656522010-08-17 15:40:09 +020014971
Cyril Bonté62ba8702014-04-22 23:52:25 +020014972sc_sess_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014973sc0_sess_cnt([<table>]) : integer
14974sc1_sess_cnt([<table>]) : integer
14975sc2_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014976 Returns the cumulative number of incoming connections that were transformed
Willy Tarreaue9656522010-08-17 15:40:09 +020014977 into sessions, which means that they were accepted by a "tcp-request
14978 connection" rule, from the currently tracked counters. A backend may count
14979 more sessions than connections because each connection could result in many
Jamie Gloudonaaa21002012-08-25 00:18:33 -040014980 backend sessions if some HTTP keep-alive is performed over the connection
Willy Tarreaue9656522010-08-17 15:40:09 +020014981 with the client. See also src_sess_cnt.
14982
Cyril Bonté62ba8702014-04-22 23:52:25 +020014983sc_sess_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014984sc0_sess_rate([<table>]) : integer
14985sc1_sess_rate([<table>]) : integer
14986sc2_sess_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014987 Returns the average session rate from the currently tracked counters,
14988 measured in amount of sessions over the period configured in the table. A
14989 session is a connection that got past the early "tcp-request connection"
14990 rules. A backend may count more sessions than connections because each
14991 connection could result in many backend sessions if some HTTP keep-alive is
Jamie Gloudonaaa21002012-08-25 00:18:33 -040014992 performed over the connection with the client. See also src_sess_rate.
Willy Tarreaue9656522010-08-17 15:40:09 +020014993
Cyril Bonté62ba8702014-04-22 23:52:25 +020014994sc_tracked(<ctr>[,<table>]) : boolean
Willy Tarreau0f791d42013-07-23 19:56:43 +020014995sc0_tracked([<table>]) : boolean
14996sc1_tracked([<table>]) : boolean
14997sc2_tracked([<table>]) : boolean
Willy Tarreau6f1615f2013-06-03 15:15:22 +020014998 Returns true if the designated session counter is currently being tracked by
14999 the current session. This can be useful when deciding whether or not we want
15000 to set some values in a header passed to the server.
15001
Cyril Bonté62ba8702014-04-22 23:52:25 +020015002sc_trackers(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015003sc0_trackers([<table>]) : integer
15004sc1_trackers([<table>]) : integer
15005sc2_trackers([<table>]) : integer
Willy Tarreau2406db42012-12-09 12:16:43 +010015006 Returns the current amount of concurrent connections tracking the same
15007 tracked counters. This number is automatically incremented when tracking
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020015008 begins and decremented when tracking stops. It differs from sc0_conn_cur in
Willy Tarreau2406db42012-12-09 12:16:43 +010015009 that it does not rely on any stored information but on the table's reference
15010 count (the "use" value which is returned by "show table" on the CLI). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020015011 may sometimes be more suited for layer7 tracking. It can be used to tell a
15012 server how many concurrent connections there are from a given address for
15013 example.
Willy Tarreau2406db42012-12-09 12:16:43 +010015014
Willy Tarreau74ca5042013-06-11 23:12:07 +020015015so_id : integer
15016 Returns an integer containing the current listening socket's id. It is useful
15017 in frontends involving many "bind" lines, or to stick all users coming via a
15018 same socket to the same server.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015019
Willy Tarreau74ca5042013-06-11 23:12:07 +020015020src : ip
Davor Ocelice9ed2812017-12-25 17:49:28 +010015021 This is the source IPv4 address of the client of the session. It is of type
Willy Tarreau74ca5042013-06-11 23:12:07 +020015022 IP and works on both IPv4 and IPv6 tables. On IPv6 tables, IPv4 addresses are
15023 mapped to their IPv6 equivalent, according to RFC 4291. Note that it is the
15024 TCP-level source address which is used, and not the address of a client
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010015025 behind a proxy. However if the "accept-proxy" or "accept-netscaler-cip" bind
15026 directive is used, it can be the address of a client behind another
15027 PROXY-protocol compatible component for all rule sets except
Willy Tarreau64ded3d2019-01-23 10:02:15 +010015028 "tcp-request connection" which sees the real address. When the incoming
15029 connection passed through address translation or redirection involving
15030 connection tracking, the original destination address before the redirection
15031 will be reported. On Linux systems, the source and destination may seldom
15032 appear reversed if the nf_conntrack_tcp_loose sysctl is set, because a late
15033 response may reopen a timed out connection and switch what is believed to be
15034 the source and the destination.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015035
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010015036 Example:
15037 # add an HTTP header in requests with the originating address' country
15038 http-request set-header X-Country %[src,map_ip(geoip.lst)]
15039
Willy Tarreau74ca5042013-06-11 23:12:07 +020015040src_bytes_in_rate([<table>]) : integer
15041 Returns the average bytes rate from the incoming connection's source address
15042 in the current proxy's stick-table or in the designated stick-table, measured
15043 in amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015044 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_in_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015045
Willy Tarreau74ca5042013-06-11 23:12:07 +020015046src_bytes_out_rate([<table>]) : integer
15047 Returns the average bytes rate to the incoming connection's source address in
15048 the current proxy's stick-table or in the designated stick-table, measured in
Willy Tarreauc9705a12010-07-27 20:05:50 +020015049 amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015050 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_out_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015051
Willy Tarreau74ca5042013-06-11 23:12:07 +020015052src_clr_gpc0([<table>]) : integer
15053 Clears the first General Purpose Counter associated to the incoming
15054 connection's source address in the current proxy's stick-table or in the
15055 designated stick-table, and returns its previous value. If the address is not
15056 found, an entry is created and 0 is returned. This is typically used as a
15057 second ACL in an expression in order to mark a connection when a first ACL
15058 was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020015059
Jarno Huuskonen676f6222017-03-30 09:19:45 +030015060 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020015061 # block if 5 consecutive requests continue to come faster than 10 sess
15062 # per second, and reset the counter as soon as the traffic slows down.
15063 acl abuse src_http_req_rate gt 10
15064 acl kill src_inc_gpc0 gt 5
Willy Tarreau869948b2013-01-04 14:14:57 +010015065 acl save src_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020015066 tcp-request connection accept if !abuse save
15067 tcp-request connection reject if abuse kill
15068
Frédéric Lécaille6778b272018-01-29 15:22:53 +010015069src_clr_gpc1([<table>]) : integer
15070 Clears the second General Purpose Counter associated to the incoming
15071 connection's source address in the current proxy's stick-table or in the
15072 designated stick-table, and returns its previous value. If the address is not
15073 found, an entry is created and 0 is returned. This is typically used as a
15074 second ACL in an expression in order to mark a connection when a first ACL
15075 was verified.
15076
Willy Tarreau74ca5042013-06-11 23:12:07 +020015077src_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010015078 Returns the cumulative number of connections initiated from the current
Willy Tarreau74ca5042013-06-11 23:12:07 +020015079 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020015080 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015081 See also sc/sc0/sc1/sc2_conn_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015082
Willy Tarreau74ca5042013-06-11 23:12:07 +020015083src_conn_cur([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020015084 Returns the current amount of concurrent connections initiated from the
Willy Tarreau74ca5042013-06-11 23:12:07 +020015085 current incoming connection's source address in the current proxy's
15086 stick-table or in the designated stick-table. If the address is not found,
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015087 zero is returned. See also sc/sc0/sc1/sc2_conn_cur.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015088
Willy Tarreau74ca5042013-06-11 23:12:07 +020015089src_conn_rate([<table>]) : integer
15090 Returns the average connection rate from the incoming connection's source
15091 address in the current proxy's stick-table or in the designated stick-table,
15092 measured in amount of connections over the period configured in the table. If
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015093 the address is not found, zero is returned. See also sc/sc0/sc1/sc2_conn_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015094
Willy Tarreau74ca5042013-06-11 23:12:07 +020015095src_get_gpc0([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020015096 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau74ca5042013-06-11 23:12:07 +020015097 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020015098 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015099 See also sc/sc0/sc1/sc2_get_gpc0 and src_inc_gpc0.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015100
Frédéric Lécaille6778b272018-01-29 15:22:53 +010015101src_get_gpc1([<table>]) : integer
15102 Returns the value of the second General Purpose Counter associated to the
15103 incoming connection's source address in the current proxy's stick-table or in
15104 the designated stick-table. If the address is not found, zero is returned.
15105 See also sc/sc0/sc1/sc2_get_gpc1 and src_inc_gpc1.
15106
Thierry FOURNIER236657b2015-08-19 08:25:14 +020015107src_get_gpt0([<table>]) : integer
15108 Returns the value of the first General Purpose Tag associated to the
15109 incoming connection's source address in the current proxy's stick-table or in
15110 the designated stick-table. If the address is not found, zero is returned.
15111 See also sc/sc0/sc1/sc2_get_gpt0.
15112
Willy Tarreau74ca5042013-06-11 23:12:07 +020015113src_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020015114 Returns the average increment rate of the first General Purpose Counter
Willy Tarreau74ca5042013-06-11 23:12:07 +020015115 associated to the incoming connection's source address in the current proxy's
Willy Tarreauba2ffd12013-05-29 15:54:14 +020015116 stick-table or in the designated stick-table. It reports the frequency
15117 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015118 sc/sc0/sc1/sc2_gpc0_rate, src_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
15119 that the "gpc0_rate" counter must be stored in the stick-table for a value to
15120 be returned, as "gpc0" only holds the event count.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020015121
Frédéric Lécaille6778b272018-01-29 15:22:53 +010015122src_gpc1_rate([<table>]) : integer
15123 Returns the average increment rate of the second General Purpose Counter
15124 associated to the incoming connection's source address in the current proxy's
15125 stick-table or in the designated stick-table. It reports the frequency
15126 which the gpc1 counter was incremented over the configured period. See also
15127 sc/sc0/sc1/sc2_gpc1_rate, src_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
15128 that the "gpc1_rate" counter must be stored in the stick-table for a value to
15129 be returned, as "gpc1" only holds the event count.
15130
Willy Tarreau74ca5042013-06-11 23:12:07 +020015131src_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010015132 Returns the cumulative number of HTTP errors from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020015133 source address in the current proxy's stick-table or in the designated
Willy Tarreauc9705a12010-07-27 20:05:50 +020015134 stick-table. This includes the both request errors and 4xx error responses.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015135 See also sc/sc0/sc1/sc2_http_err_cnt. If the address is not found, zero is
Willy Tarreau74ca5042013-06-11 23:12:07 +020015136 returned.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015137
Willy Tarreau74ca5042013-06-11 23:12:07 +020015138src_http_err_rate([<table>]) : integer
15139 Returns the average rate of HTTP errors from the incoming connection's source
15140 address in the current proxy's stick-table or in the designated stick-table,
15141 measured in amount of errors over the period configured in the table. This
15142 includes the both request errors and 4xx error responses. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015143 not found, zero is returned. See also sc/sc0/sc1/sc2_http_err_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015144
Willy Tarreau74ca5042013-06-11 23:12:07 +020015145src_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010015146 Returns the cumulative number of HTTP requests from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020015147 source address in the current proxy's stick-table or in the designated stick-
15148 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015149 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015150
Willy Tarreau74ca5042013-06-11 23:12:07 +020015151src_http_req_rate([<table>]) : integer
15152 Returns the average rate of HTTP requests from the incoming connection's
15153 source address in the current proxy's stick-table or in the designated stick-
15154 table, measured in amount of requests over the period configured in the
Willy Tarreauc9705a12010-07-27 20:05:50 +020015155 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015156 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015157
Willy Tarreau74ca5042013-06-11 23:12:07 +020015158src_inc_gpc0([<table>]) : integer
15159 Increments the first General Purpose Counter associated to the incoming
15160 connection's source address in the current proxy's stick-table or in the
15161 designated stick-table, and returns its new value. If the address is not
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020015162 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc0.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015163 This is typically used as a second ACL in an expression in order to mark a
15164 connection when a first ACL was verified :
Willy Tarreauc9705a12010-07-27 20:05:50 +020015165
Jarno Huuskonen676f6222017-03-30 09:19:45 +030015166 Example:
Willy Tarreauc9705a12010-07-27 20:05:50 +020015167 acl abuse src_http_req_rate gt 10
Willy Tarreau869948b2013-01-04 14:14:57 +010015168 acl kill src_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020015169 tcp-request connection reject if abuse kill
Willy Tarreauc9705a12010-07-27 20:05:50 +020015170
Frédéric Lécaille6778b272018-01-29 15:22:53 +010015171src_inc_gpc1([<table>]) : integer
15172 Increments the second General Purpose Counter associated to the incoming
15173 connection's source address in the current proxy's stick-table or in the
15174 designated stick-table, and returns its new value. If the address is not
15175 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc1.
15176 This is typically used as a second ACL in an expression in order to mark a
15177 connection when a first ACL was verified.
15178
Willy Tarreau16e01562016-08-09 16:46:18 +020015179src_is_local : boolean
15180 Returns true if the source address of the incoming connection is local to the
15181 system, or false if the address doesn't exist on the system, meaning that it
15182 comes from a remote machine. Note that UNIX addresses are considered local.
15183 It can be useful to apply certain access restrictions based on where the
Davor Ocelice9ed2812017-12-25 17:49:28 +010015184 client comes from (e.g. require auth or https for remote machines). Please
Willy Tarreau16e01562016-08-09 16:46:18 +020015185 note that the check involves a few system calls, so it's better to do it only
15186 once per connection.
15187
Willy Tarreau74ca5042013-06-11 23:12:07 +020015188src_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020015189 Returns the total amount of data received from the incoming connection's
15190 source address in the current proxy's stick-table or in the designated
15191 stick-table, measured in kilobytes. If the address is not found, zero is
15192 returned. The test is currently performed on 32-bit integers, which limits
15193 values to 4 terabytes. See also sc/sc0/sc1/sc2_kbytes_in.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015194
Willy Tarreau74ca5042013-06-11 23:12:07 +020015195src_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020015196 Returns the total amount of data sent to the incoming connection's source
15197 address in the current proxy's stick-table or in the designated stick-table,
15198 measured in kilobytes. If the address is not found, zero is returned. The
15199 test is currently performed on 32-bit integers, which limits values to 4
15200 terabytes. See also sc/sc0/sc1/sc2_kbytes_out.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020015201
Willy Tarreau74ca5042013-06-11 23:12:07 +020015202src_port : integer
15203 Returns an integer value corresponding to the TCP source port of the
15204 connection on the client side, which is the port the client connected from.
15205 Usage of this function is very limited as modern protocols do not care much
15206 about source ports nowadays.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010015207
Willy Tarreau74ca5042013-06-11 23:12:07 +020015208src_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010015209 Returns the cumulative number of connections initiated from the incoming
Willy Tarreauc9705a12010-07-27 20:05:50 +020015210 connection's source IPv4 address in the current proxy's stick-table or in the
15211 designated stick-table, that were transformed into sessions, which means that
15212 they were accepted by "tcp-request" rules. If the address is not found, zero
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015213 is returned. See also sc/sc0/sc1/sc2_sess_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015214
Willy Tarreau74ca5042013-06-11 23:12:07 +020015215src_sess_rate([<table>]) : integer
15216 Returns the average session rate from the incoming connection's source
15217 address in the current proxy's stick-table or in the designated stick-table,
15218 measured in amount of sessions over the period configured in the table. A
15219 session is a connection that went past the early "tcp-request" rules. If the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015220 address is not found, zero is returned. See also sc/sc0/sc1/sc2_sess_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015221
Willy Tarreau74ca5042013-06-11 23:12:07 +020015222src_updt_conn_cnt([<table>]) : integer
15223 Creates or updates the entry associated to the incoming connection's source
15224 address in the current proxy's stick-table or in the designated stick-table.
15225 This table must be configured to store the "conn_cnt" data type, otherwise
15226 the match will be ignored. The current count is incremented by one, and the
15227 expiration timer refreshed. The updated count is returned, so this match
15228 can't return zero. This was used to reject service abusers based on their
15229 source address. Note: it is recommended to use the more complete "track-sc*"
15230 actions in "tcp-request" rules instead.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020015231
15232 Example :
15233 # This frontend limits incoming SSH connections to 3 per 10 second for
15234 # each source address, and rejects excess connections until a 10 second
15235 # silence is observed. At most 20 addresses are tracked.
15236 listen ssh
15237 bind :22
15238 mode tcp
15239 maxconn 100
Willy Tarreauc9705a12010-07-27 20:05:50 +020015240 stick-table type ip size 20 expire 10s store conn_cnt
Willy Tarreau74ca5042013-06-11 23:12:07 +020015241 tcp-request content reject if { src_updt_conn_cnt gt 3 }
Willy Tarreaua975b8f2010-06-05 19:13:27 +020015242 server local 127.0.0.1:22
15243
Willy Tarreau74ca5042013-06-11 23:12:07 +020015244srv_id : integer
15245 Returns an integer containing the server's id when processing the response.
15246 While it's almost only used with ACLs, it may be used for logging or
15247 debugging.
Hervé COMMOWICKdaa824e2011-08-05 12:09:44 +020015248
vkill1dfd1652019-10-30 16:58:14 +080015249srv_name : string
15250 Returns a string containing the server's name when processing the response.
15251 While it's almost only used with ACLs, it may be used for logging or
15252 debugging.
15253
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200152547.3.4. Fetching samples at Layer 5
Willy Tarreau74ca5042013-06-11 23:12:07 +020015255----------------------------------
Willy Tarreau0b1cd942010-05-16 22:18:27 +020015256
Willy Tarreau74ca5042013-06-11 23:12:07 +020015257The layer 5 usually describes just the session layer which in haproxy is
15258closest to the session once all the connection handshakes are finished, but
15259when no content is yet made available. The fetch methods described here are
15260usable as low as the "tcp-request content" rule sets unless they require some
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015261future information. Those generally include the results of SSL negotiations.
Willy Tarreauc735a072011-03-29 00:57:02 +020015262
Ben Shillitof25e8e52016-12-02 14:25:37 +00001526351d.all(<prop>[,<prop>*]) : string
15264 Returns values for the properties requested as a string, where values are
15265 separated by the delimiter specified with "51degrees-property-separator".
15266 The device is identified using all the important HTTP headers from the
15267 request. The function can be passed up to five property names, and if a
15268 property name can't be found, the value "NoData" is returned.
15269
15270 Example :
15271 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request
15272 # containing the three properties requested using all relevant headers from
15273 # the request.
15274 frontend http-in
15275 bind *:8081
15276 default_backend servers
15277 http-request set-header X-51D-DeviceTypeMobileTablet \
15278 %[51d.all(DeviceType,IsMobile,IsTablet)]
15279
Emeric Brun645ae792014-04-30 14:21:06 +020015280ssl_bc : boolean
15281 Returns true when the back connection was made via an SSL/TLS transport
15282 layer and is locally deciphered. This means the outgoing connection was made
15283 other a server with the "ssl" option.
15284
15285ssl_bc_alg_keysize : integer
15286 Returns the symmetric cipher key size supported in bits when the outgoing
15287 connection was made over an SSL/TLS transport layer.
15288
Olivier Houchard6b77f492018-11-22 18:18:29 +010015289ssl_bc_alpn : string
15290 This extracts the Application Layer Protocol Negotiation field from an
15291 outgoing connection made via a TLS transport layer.
Michael Prokop4438c602019-05-24 10:25:45 +020015292 The result is a string containing the protocol name negotiated with the
Olivier Houchard6b77f492018-11-22 18:18:29 +010015293 server. The SSL library must have been built with support for TLS
15294 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
15295 not advertised unless the "alpn" keyword on the "server" line specifies a
15296 protocol list. Also, nothing forces the server to pick a protocol from this
15297 list, any other one may be requested. The TLS ALPN extension is meant to
15298 replace the TLS NPN extension. See also "ssl_bc_npn".
15299
Emeric Brun645ae792014-04-30 14:21:06 +020015300ssl_bc_cipher : string
15301 Returns the name of the used cipher when the outgoing connection was made
15302 over an SSL/TLS transport layer.
15303
Patrick Hemmer65674662019-06-04 08:13:03 -040015304ssl_bc_client_random : binary
15305 Returns the client random of the back connection when the incoming connection
15306 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
15307 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
15308
Emeric Brun74f7ffa2018-02-19 16:14:12 +010015309ssl_bc_is_resumed : boolean
15310 Returns true when the back connection was made over an SSL/TLS transport
15311 layer and the newly created SSL session was resumed using a cached
15312 session or a TLS ticket.
15313
Olivier Houchard6b77f492018-11-22 18:18:29 +010015314ssl_bc_npn : string
15315 This extracts the Next Protocol Negotiation field from an outgoing connection
15316 made via a TLS transport layer. The result is a string containing the
Michael Prokop4438c602019-05-24 10:25:45 +020015317 protocol name negotiated with the server . The SSL library must have been
Olivier Houchard6b77f492018-11-22 18:18:29 +010015318 built with support for TLS extensions enabled (check haproxy -vv). Note that
15319 the TLS NPN extension is not advertised unless the "npn" keyword on the
15320 "server" line specifies a protocol list. Also, nothing forces the server to
15321 pick a protocol from this list, any other one may be used. Please note that
15322 the TLS NPN extension was replaced with ALPN.
15323
Emeric Brun645ae792014-04-30 14:21:06 +020015324ssl_bc_protocol : string
15325 Returns the name of the used protocol when the outgoing connection was made
15326 over an SSL/TLS transport layer.
15327
Emeric Brunb73a9b02014-04-30 18:49:19 +020015328ssl_bc_unique_id : binary
Emeric Brun645ae792014-04-30 14:21:06 +020015329 When the outgoing connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020015330 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
15331 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
Emeric Brun645ae792014-04-30 14:21:06 +020015332
Patrick Hemmer65674662019-06-04 08:13:03 -040015333ssl_bc_server_random : binary
15334 Returns the server random of the back connection when the incoming connection
15335 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
15336 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
15337
Emeric Brun645ae792014-04-30 14:21:06 +020015338ssl_bc_session_id : binary
15339 Returns the SSL ID of the back connection when the outgoing connection was
15340 made over an SSL/TLS transport layer. It is useful to log if we want to know
15341 if session was reused or not.
15342
Patrick Hemmere0275472018-04-28 19:15:51 -040015343ssl_bc_session_key : binary
15344 Returns the SSL session master key of the back connection when the outgoing
15345 connection was made over an SSL/TLS transport layer. It is useful to decrypt
15346 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
15347 BoringSSL.
15348
Emeric Brun645ae792014-04-30 14:21:06 +020015349ssl_bc_use_keysize : integer
15350 Returns the symmetric cipher key size used in bits when the outgoing
15351 connection was made over an SSL/TLS transport layer.
15352
Willy Tarreau74ca5042013-06-11 23:12:07 +020015353ssl_c_ca_err : integer
15354 When the incoming connection was made over an SSL/TLS transport layer,
15355 returns the ID of the first error detected during verification of the client
15356 certificate at depth > 0, or 0 if no error was encountered during this
15357 verification process. Please refer to your SSL library's documentation to
15358 find the exhaustive list of error codes.
Willy Tarreauc735a072011-03-29 00:57:02 +020015359
Willy Tarreau74ca5042013-06-11 23:12:07 +020015360ssl_c_ca_err_depth : integer
15361 When the incoming connection was made over an SSL/TLS transport layer,
15362 returns the depth in the CA chain of the first error detected during the
15363 verification of the client certificate. If no error is encountered, 0 is
15364 returned.
Willy Tarreau0ba27502007-12-24 16:55:16 +010015365
Emeric Brun43e79582014-10-29 19:03:26 +010015366ssl_c_der : binary
15367 Returns the DER formatted certificate presented by the client when the
15368 incoming connection was made over an SSL/TLS transport layer. When used for
15369 an ACL, the value(s) to match against can be passed in hexadecimal form.
15370
Willy Tarreau74ca5042013-06-11 23:12:07 +020015371ssl_c_err : integer
15372 When the incoming connection was made over an SSL/TLS transport layer,
15373 returns the ID of the first error detected during verification at depth 0, or
15374 0 if no error was encountered during this verification process. Please refer
15375 to your SSL library's documentation to find the exhaustive list of error
15376 codes.
Willy Tarreau62644772008-07-16 18:36:06 +020015377
Elliot Otchet71f82972020-01-15 08:12:14 -050015378ssl_c_i_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020015379 When the incoming connection was made over an SSL/TLS transport layer,
15380 returns the full distinguished name of the issuer of the certificate
15381 presented by the client when no <entry> is specified, or the value of the
15382 first given entry found from the beginning of the DN. If a positive/negative
15383 occurrence number is specified as the optional second argument, it returns
15384 the value of the nth given entry value from the beginning/end of the DN.
15385 For instance, "ssl_c_i_dn(OU,2)" the second organization unit, and
15386 "ssl_c_i_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050015387 The <format> parameter allows you to receive the DN suitable for
15388 consumption by different protocols. Currently supported is rfc2253 for
15389 LDAP v3.
15390 If you'd like to modify the format only you can specify an empty string
15391 and zero for the first two parameters. Example: ssl_c_i_dn(,0,rfc2253)
Willy Tarreau62644772008-07-16 18:36:06 +020015392
Willy Tarreau74ca5042013-06-11 23:12:07 +020015393ssl_c_key_alg : string
15394 Returns the name of the algorithm used to generate the key of the certificate
15395 presented by the client when the incoming connection was made over an SSL/TLS
15396 transport layer.
Willy Tarreau62644772008-07-16 18:36:06 +020015397
Willy Tarreau74ca5042013-06-11 23:12:07 +020015398ssl_c_notafter : string
15399 Returns the end date presented by the client as a formatted string
15400 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15401 transport layer.
Emeric Brunbede3d02009-06-30 17:54:00 +020015402
Willy Tarreau74ca5042013-06-11 23:12:07 +020015403ssl_c_notbefore : string
15404 Returns the start date presented by the client as a formatted string
15405 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15406 transport layer.
Willy Tarreaub6672b52011-12-12 17:23:41 +010015407
Elliot Otchet71f82972020-01-15 08:12:14 -050015408ssl_c_s_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020015409 When the incoming connection was made over an SSL/TLS transport layer,
15410 returns the full distinguished name of the subject of the certificate
15411 presented by the client when no <entry> is specified, or the value of the
15412 first given entry found from the beginning of the DN. If a positive/negative
15413 occurrence number is specified as the optional second argument, it returns
15414 the value of the nth given entry value from the beginning/end of the DN.
15415 For instance, "ssl_c_s_dn(OU,2)" the second organization unit, and
15416 "ssl_c_s_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050015417 The <format> parameter allows you to receive the DN suitable for
15418 consumption by different protocols. Currently supported is rfc2253 for
15419 LDAP v3.
15420 If you'd like to modify the format only you can specify an empty string
15421 and zero for the first two parameters. Example: ssl_c_s_dn(,0,rfc2253)
Willy Tarreaub6672b52011-12-12 17:23:41 +010015422
Willy Tarreau74ca5042013-06-11 23:12:07 +020015423ssl_c_serial : binary
15424 Returns the serial of the certificate presented by the client when the
15425 incoming connection was made over an SSL/TLS transport layer. When used for
15426 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun2525b6b2012-10-18 15:59:43 +020015427
Willy Tarreau74ca5042013-06-11 23:12:07 +020015428ssl_c_sha1 : binary
15429 Returns the SHA-1 fingerprint of the certificate presented by the client when
15430 the incoming connection was made over an SSL/TLS transport layer. This can be
15431 used to stick a client to a server, or to pass this information to a server.
Willy Tarreau2d0caa32014-07-02 19:01:22 +020015432 Note that the output is binary, so if you want to pass that signature to the
15433 server, you need to encode it in hex or base64, such as in the example below:
15434
Jarno Huuskonen676f6222017-03-30 09:19:45 +030015435 Example:
Willy Tarreau2d0caa32014-07-02 19:01:22 +020015436 http-request set-header X-SSL-Client-SHA1 %[ssl_c_sha1,hex]
Emeric Brun2525b6b2012-10-18 15:59:43 +020015437
Willy Tarreau74ca5042013-06-11 23:12:07 +020015438ssl_c_sig_alg : string
15439 Returns the name of the algorithm used to sign the certificate presented by
15440 the client when the incoming connection was made over an SSL/TLS transport
15441 layer.
Emeric Brun87855892012-10-17 17:39:35 +020015442
Willy Tarreau74ca5042013-06-11 23:12:07 +020015443ssl_c_used : boolean
15444 Returns true if current SSL session uses a client certificate even if current
15445 connection uses SSL session resumption. See also "ssl_fc_has_crt".
Emeric Brun7f56e742012-10-19 18:15:40 +020015446
Willy Tarreau74ca5042013-06-11 23:12:07 +020015447ssl_c_verify : integer
15448 Returns the verify result error ID when the incoming connection was made over
15449 an SSL/TLS transport layer, otherwise zero if no error is encountered. Please
15450 refer to your SSL library's documentation for an exhaustive list of error
15451 codes.
Emeric Brunce5ad802012-10-22 14:11:22 +020015452
Willy Tarreau74ca5042013-06-11 23:12:07 +020015453ssl_c_version : integer
15454 Returns the version of the certificate presented by the client when the
15455 incoming connection was made over an SSL/TLS transport layer.
Emeric Brunce5ad802012-10-22 14:11:22 +020015456
Emeric Brun43e79582014-10-29 19:03:26 +010015457ssl_f_der : binary
15458 Returns the DER formatted certificate presented by the frontend when the
15459 incoming connection was made over an SSL/TLS transport layer. When used for
15460 an ACL, the value(s) to match against can be passed in hexadecimal form.
15461
Elliot Otchet71f82972020-01-15 08:12:14 -050015462ssl_f_i_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020015463 When the incoming connection was made over an SSL/TLS transport layer,
15464 returns the full distinguished name of the issuer of the certificate
15465 presented by the frontend when no <entry> is specified, or the value of the
15466 first given entry found from the beginning of the DN. If a positive/negative
Emeric Brun87855892012-10-17 17:39:35 +020015467 occurrence number is specified as the optional second argument, it returns
Willy Tarreau74ca5042013-06-11 23:12:07 +020015468 the value of the nth given entry value from the beginning/end of the DN.
15469 For instance, "ssl_f_i_dn(OU,2)" the second organization unit, and
15470 "ssl_f_i_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050015471 The <format> parameter allows you to receive the DN suitable for
15472 consumption by different protocols. Currently supported is rfc2253 for
15473 LDAP v3.
15474 If you'd like to modify the format only you can specify an empty string
15475 and zero for the first two parameters. Example: ssl_f_i_dn(,0,rfc2253)
Emeric Brun87855892012-10-17 17:39:35 +020015476
Willy Tarreau74ca5042013-06-11 23:12:07 +020015477ssl_f_key_alg : string
15478 Returns the name of the algorithm used to generate the key of the certificate
15479 presented by the frontend when the incoming connection was made over an
15480 SSL/TLS transport layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020015481
Willy Tarreau74ca5042013-06-11 23:12:07 +020015482ssl_f_notafter : string
15483 Returns the end date presented by the frontend as a formatted string
15484 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15485 transport layer.
Emeric Brun2525b6b2012-10-18 15:59:43 +020015486
Willy Tarreau74ca5042013-06-11 23:12:07 +020015487ssl_f_notbefore : string
15488 Returns the start date presented by the frontend as a formatted string
15489 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15490 transport layer.
Emeric Brun87855892012-10-17 17:39:35 +020015491
Elliot Otchet71f82972020-01-15 08:12:14 -050015492ssl_f_s_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020015493 When the incoming connection was made over an SSL/TLS transport layer,
15494 returns the full distinguished name of the subject of the certificate
15495 presented by the frontend when no <entry> is specified, or the value of the
15496 first given entry found from the beginning of the DN. If a positive/negative
15497 occurrence number is specified as the optional second argument, it returns
15498 the value of the nth given entry value from the beginning/end of the DN.
15499 For instance, "ssl_f_s_dn(OU,2)" the second organization unit, and
15500 "ssl_f_s_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050015501 The <format> parameter allows you to receive the DN suitable for
15502 consumption by different protocols. Currently supported is rfc2253 for
15503 LDAP v3.
15504 If you'd like to modify the format only you can specify an empty string
15505 and zero for the first two parameters. Example: ssl_f_s_dn(,0,rfc2253)
Emeric Brunce5ad802012-10-22 14:11:22 +020015506
Willy Tarreau74ca5042013-06-11 23:12:07 +020015507ssl_f_serial : binary
15508 Returns the serial of the certificate presented by the frontend when the
15509 incoming connection was made over an SSL/TLS transport layer. When used for
15510 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun87855892012-10-17 17:39:35 +020015511
Emeric Brun55f4fa82014-04-30 17:11:25 +020015512ssl_f_sha1 : binary
15513 Returns the SHA-1 fingerprint of the certificate presented by the frontend
15514 when the incoming connection was made over an SSL/TLS transport layer. This
15515 can be used to know which certificate was chosen using SNI.
15516
Willy Tarreau74ca5042013-06-11 23:12:07 +020015517ssl_f_sig_alg : string
15518 Returns the name of the algorithm used to sign the certificate presented by
15519 the frontend when the incoming connection was made over an SSL/TLS transport
15520 layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020015521
Willy Tarreau74ca5042013-06-11 23:12:07 +020015522ssl_f_version : integer
15523 Returns the version of the certificate presented by the frontend when the
15524 incoming connection was made over an SSL/TLS transport layer.
15525
15526ssl_fc : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020015527 Returns true when the front connection was made via an SSL/TLS transport
15528 layer and is locally deciphered. This means it has matched a socket declared
15529 with a "bind" line having the "ssl" option.
15530
Willy Tarreau74ca5042013-06-11 23:12:07 +020015531 Example :
15532 # This passes "X-Proto: https" to servers when client connects over SSL
15533 listen http-https
15534 bind :80
15535 bind :443 ssl crt /etc/haproxy.pem
15536 http-request add-header X-Proto https if { ssl_fc }
15537
15538ssl_fc_alg_keysize : integer
15539 Returns the symmetric cipher key size supported in bits when the incoming
15540 connection was made over an SSL/TLS transport layer.
15541
15542ssl_fc_alpn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015543 This extracts the Application Layer Protocol Negotiation field from an
Willy Tarreau74ca5042013-06-11 23:12:07 +020015544 incoming connection made via a TLS transport layer and locally deciphered by
15545 haproxy. The result is a string containing the protocol name advertised by
15546 the client. The SSL library must have been built with support for TLS
15547 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
15548 not advertised unless the "alpn" keyword on the "bind" line specifies a
15549 protocol list. Also, nothing forces the client to pick a protocol from this
15550 list, any other one may be requested. The TLS ALPN extension is meant to
15551 replace the TLS NPN extension. See also "ssl_fc_npn".
15552
Willy Tarreau74ca5042013-06-11 23:12:07 +020015553ssl_fc_cipher : string
15554 Returns the name of the used cipher when the incoming connection was made
15555 over an SSL/TLS transport layer.
Willy Tarreauab861d32013-04-02 02:30:41 +020015556
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015557ssl_fc_cipherlist_bin : binary
15558 Returns the binary form of the client hello cipher list. The maximum returned
15559 value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010015560 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015561
15562ssl_fc_cipherlist_hex : string
15563 Returns the binary form of the client hello cipher list encoded as
15564 hexadecimal. The maximum returned value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010015565 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015566
15567ssl_fc_cipherlist_str : string
15568 Returns the decoded text form of the client hello cipher list. The maximum
15569 number of ciphers returned is according with the value of
15570 "tune.ssl.capture-cipherlist-size". Note that this sample-fetch is only
Davor Ocelice9ed2812017-12-25 17:49:28 +010015571 available with OpenSSL >= 1.0.2. If the function is not enabled, this
Emmanuel Hocdetddcde192017-09-01 17:32:08 +020015572 sample-fetch returns the hash like "ssl_fc_cipherlist_xxh".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015573
15574ssl_fc_cipherlist_xxh : integer
15575 Returns a xxh64 of the cipher list. This hash can be return only is the value
15576 "tune.ssl.capture-cipherlist-size" is set greater than 0, however the hash
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010015577 take in account all the data of the cipher list.
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015578
Patrick Hemmer65674662019-06-04 08:13:03 -040015579ssl_fc_client_random : binary
15580 Returns the client random of the front connection when the incoming connection
15581 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
15582 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
15583
Willy Tarreau74ca5042013-06-11 23:12:07 +020015584ssl_fc_has_crt : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020015585 Returns true if a client certificate is present in an incoming connection over
15586 SSL/TLS transport layer. Useful if 'verify' statement is set to 'optional'.
Emeric Brun9143d372012-12-20 15:44:16 +010015587 Note: on SSL session resumption with Session ID or TLS ticket, client
15588 certificate is not present in the current connection but may be retrieved
15589 from the cache or the ticket. So prefer "ssl_c_used" if you want to check if
15590 current SSL session uses a client certificate.
Emeric Brun2525b6b2012-10-18 15:59:43 +020015591
Olivier Houchardccaa7de2017-10-02 11:51:03 +020015592ssl_fc_has_early : boolean
15593 Returns true if early data were sent, and the handshake didn't happen yet. As
15594 it has security implications, it is useful to be able to refuse those, or
15595 wait until the handshake happened.
15596
Willy Tarreau74ca5042013-06-11 23:12:07 +020015597ssl_fc_has_sni : boolean
15598 This checks for the presence of a Server Name Indication TLS extension (SNI)
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020015599 in an incoming connection was made over an SSL/TLS transport layer. Returns
15600 true when the incoming connection presents a TLS SNI field. This requires
John Roeslerfb2fce12019-07-10 15:45:51 -050015601 that the SSL library is built with support for TLS extensions enabled (check
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020015602 haproxy -vv).
Willy Tarreau7875d092012-09-10 08:20:03 +020015603
Nenad Merdanovic1516fe32016-05-17 03:31:21 +020015604ssl_fc_is_resumed : boolean
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020015605 Returns true if the SSL/TLS session has been resumed through the use of
Jérôme Magnin4a326cb2018-01-15 14:01:17 +010015606 SSL session cache or TLS tickets on an incoming connection over an SSL/TLS
15607 transport layer.
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020015608
Willy Tarreau74ca5042013-06-11 23:12:07 +020015609ssl_fc_npn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015610 This extracts the Next Protocol Negotiation field from an incoming connection
Willy Tarreau74ca5042013-06-11 23:12:07 +020015611 made via a TLS transport layer and locally deciphered by haproxy. The result
15612 is a string containing the protocol name advertised by the client. The SSL
15613 library must have been built with support for TLS extensions enabled (check
15614 haproxy -vv). Note that the TLS NPN extension is not advertised unless the
15615 "npn" keyword on the "bind" line specifies a protocol list. Also, nothing
15616 forces the client to pick a protocol from this list, any other one may be
15617 requested. Please note that the TLS NPN extension was replaced with ALPN.
Willy Tarreaua33c6542012-10-15 13:19:06 +020015618
Willy Tarreau74ca5042013-06-11 23:12:07 +020015619ssl_fc_protocol : string
15620 Returns the name of the used protocol when the incoming connection was made
15621 over an SSL/TLS transport layer.
Willy Tarreau7875d092012-09-10 08:20:03 +020015622
Emeric Brunb73a9b02014-04-30 18:49:19 +020015623ssl_fc_unique_id : binary
David Sc1ad52e2014-04-08 18:48:47 -040015624 When the incoming connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020015625 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
15626 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
David Sc1ad52e2014-04-08 18:48:47 -040015627
Patrick Hemmer65674662019-06-04 08:13:03 -040015628ssl_fc_server_random : binary
15629 Returns the server random of the front connection when the incoming connection
15630 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
15631 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
15632
Willy Tarreau74ca5042013-06-11 23:12:07 +020015633ssl_fc_session_id : binary
15634 Returns the SSL ID of the front connection when the incoming connection was
15635 made over an SSL/TLS transport layer. It is useful to stick a given client to
15636 a server. It is important to note that some browsers refresh their session ID
15637 every few minutes.
Willy Tarreau7875d092012-09-10 08:20:03 +020015638
Patrick Hemmere0275472018-04-28 19:15:51 -040015639ssl_fc_session_key : binary
15640 Returns the SSL session master key of the front connection when the incoming
15641 connection was made over an SSL/TLS transport layer. It is useful to decrypt
15642 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
15643 BoringSSL.
15644
15645
Willy Tarreau74ca5042013-06-11 23:12:07 +020015646ssl_fc_sni : string
15647 This extracts the Server Name Indication TLS extension (SNI) field from an
15648 incoming connection made via an SSL/TLS transport layer and locally
15649 deciphered by haproxy. The result (when present) typically is a string
15650 matching the HTTPS host name (253 chars or less). The SSL library must have
15651 been built with support for TLS extensions enabled (check haproxy -vv).
15652
15653 This fetch is different from "req_ssl_sni" above in that it applies to the
15654 connection being deciphered by haproxy and not to SSL contents being blindly
15655 forwarded. See also "ssl_fc_sni_end" and "ssl_fc_sni_reg" below. This
John Roeslerfb2fce12019-07-10 15:45:51 -050015656 requires that the SSL library is built with support for TLS extensions
Cyril Bonté9c1eb1e2012-10-09 22:45:34 +020015657 enabled (check haproxy -vv).
Willy Tarreau62644772008-07-16 18:36:06 +020015658
Willy Tarreau74ca5042013-06-11 23:12:07 +020015659 ACL derivatives :
Willy Tarreau74ca5042013-06-11 23:12:07 +020015660 ssl_fc_sni_end : suffix match
15661 ssl_fc_sni_reg : regex match
Emeric Brun589fcad2012-10-16 14:13:26 +020015662
Willy Tarreau74ca5042013-06-11 23:12:07 +020015663ssl_fc_use_keysize : integer
15664 Returns the symmetric cipher key size used in bits when the incoming
15665 connection was made over an SSL/TLS transport layer.
Willy Tarreaub6fb4202008-07-20 11:18:28 +020015666
Willy Tarreaub6fb4202008-07-20 11:18:28 +020015667
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200156687.3.5. Fetching samples from buffer contents (Layer 6)
Willy Tarreau74ca5042013-06-11 23:12:07 +020015669------------------------------------------------------
Willy Tarreaub6fb4202008-07-20 11:18:28 +020015670
Willy Tarreau74ca5042013-06-11 23:12:07 +020015671Fetching samples from buffer contents is a bit different from the previous
15672sample fetches above because the sampled data are ephemeral. These data can
15673only be used when they're available and will be lost when they're forwarded.
15674For this reason, samples fetched from buffer contents during a request cannot
15675be used in a response for example. Even while the data are being fetched, they
15676can change. Sometimes it is necessary to set some delays or combine multiple
15677sample fetch methods to ensure that the expected data are complete and usable,
15678for example through TCP request content inspection. Please see the "tcp-request
15679content" keyword for more detailed information on the subject.
Willy Tarreau62644772008-07-16 18:36:06 +020015680
Willy Tarreau74ca5042013-06-11 23:12:07 +020015681payload(<offset>,<length>) : binary (deprecated)
Davor Ocelice9ed2812017-12-25 17:49:28 +010015682 This is an alias for "req.payload" when used in the context of a request (e.g.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015683 "stick on", "stick match"), and for "res.payload" when used in the context of
15684 a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010015685
Willy Tarreau74ca5042013-06-11 23:12:07 +020015686payload_lv(<offset1>,<length>[,<offset2>]) : binary (deprecated)
15687 This is an alias for "req.payload_lv" when used in the context of a request
Davor Ocelice9ed2812017-12-25 17:49:28 +010015688 (e.g. "stick on", "stick match"), and for "res.payload_lv" when used in the
Willy Tarreau74ca5042013-06-11 23:12:07 +020015689 context of a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010015690
Thierry FOURNIERd7d88812017-04-19 15:15:14 +020015691req.hdrs : string
15692 Returns the current request headers as string including the last empty line
15693 separating headers from the request body. The last empty line can be used to
15694 detect a truncated header block. This sample fetch is useful for some SPOE
15695 headers analyzers and for advanced logging.
15696
Thierry FOURNIER5617dce2017-04-09 05:38:19 +020015697req.hdrs_bin : binary
15698 Returns the current request headers contained in preparsed binary form. This
15699 is useful for offloading some processing with SPOE. Each string is described
15700 by a length followed by the number of bytes indicated in the length. The
15701 length is represented using the variable integer encoding detailed in the
15702 SPOE documentation. The end of the list is marked by a couple of empty header
15703 names and values (length of 0 for both).
15704
15705 *(<str:header-name><str:header-value>)<empty string><empty string>
15706
15707 int: refer to the SPOE documentation for the encoding
15708 str: <int:length><bytes>
15709
Willy Tarreau74ca5042013-06-11 23:12:07 +020015710req.len : integer
15711req_len : integer (deprecated)
15712 Returns an integer value corresponding to the number of bytes present in the
15713 request buffer. This is mostly used in ACL. It is important to understand
15714 that this test does not return false as long as the buffer is changing. This
15715 means that a check with equality to zero will almost always immediately match
15716 at the beginning of the session, while a test for more data will wait for
15717 that data to come in and return false only when haproxy is certain that no
15718 more data will come in. This test was designed to be used with TCP request
15719 content inspection.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015720
Willy Tarreau74ca5042013-06-11 23:12:07 +020015721req.payload(<offset>,<length>) : binary
15722 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020015723 in the request buffer. As a special case, if the <length> argument is zero,
15724 the the whole buffer from <offset> to the end is extracted. This can be used
15725 with ACLs in order to check for the presence of some content in a buffer at
15726 any location.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015727
Willy Tarreau74ca5042013-06-11 23:12:07 +020015728 ACL alternatives :
15729 payload(<offset>,<length>) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015730
Willy Tarreau74ca5042013-06-11 23:12:07 +020015731req.payload_lv(<offset1>,<length>[,<offset2>]) : binary
15732 This extracts a binary block whose size is specified at <offset1> for <length>
15733 bytes, and which starts at <offset2> if specified or just after the length in
15734 the request buffer. The <offset2> parameter also supports relative offsets if
15735 prepended with a '+' or '-' sign.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015736
Willy Tarreau74ca5042013-06-11 23:12:07 +020015737 ACL alternatives :
15738 payload_lv(<offset1>,<length>[,<offset2>]) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015739
Willy Tarreau74ca5042013-06-11 23:12:07 +020015740 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015741
Willy Tarreau74ca5042013-06-11 23:12:07 +020015742req.proto_http : boolean
15743req_proto_http : boolean (deprecated)
15744 Returns true when data in the request buffer look like HTTP and correctly
15745 parses as such. It is the same parser as the common HTTP request parser which
15746 is used so there should be no surprises. The test does not match until the
15747 request is complete, failed or timed out. This test may be used to report the
15748 protocol in TCP logs, but the biggest use is to block TCP request analysis
15749 until a complete HTTP request is present in the buffer, for example to track
15750 a header.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015751
Willy Tarreau74ca5042013-06-11 23:12:07 +020015752 Example:
15753 # track request counts per "base" (concatenation of Host+URL)
15754 tcp-request inspect-delay 10s
15755 tcp-request content reject if !HTTP
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020015756 tcp-request content track-sc0 base table req-rate
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015757
Willy Tarreau74ca5042013-06-11 23:12:07 +020015758req.rdp_cookie([<name>]) : string
15759rdp_cookie([<name>]) : string (deprecated)
15760 When the request buffer looks like the RDP protocol, extracts the RDP cookie
15761 <name>, or any cookie if unspecified. The parser only checks for the first
15762 cookie, as illustrated in the RDP protocol specification. The cookie name is
15763 case insensitive. Generally the "MSTS" cookie name will be used, as it can
15764 contain the user name of the client connecting to the server if properly
15765 configured on the client. The "MSTSHASH" cookie is often used as well for
15766 session stickiness to servers.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015767
Willy Tarreau74ca5042013-06-11 23:12:07 +020015768 This differs from "balance rdp-cookie" in that any balancing algorithm may be
15769 used and thus the distribution of clients to backend servers is not linked to
15770 a hash of the RDP cookie. It is envisaged that using a balancing algorithm
15771 such as "balance roundrobin" or "balance leastconn" will lead to a more even
15772 distribution of clients to backend servers than the hash used by "balance
15773 rdp-cookie".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015774
Willy Tarreau74ca5042013-06-11 23:12:07 +020015775 ACL derivatives :
15776 req_rdp_cookie([<name>]) : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015777
Willy Tarreau74ca5042013-06-11 23:12:07 +020015778 Example :
15779 listen tse-farm
15780 bind 0.0.0.0:3389
15781 # wait up to 5s for an RDP cookie in the request
15782 tcp-request inspect-delay 5s
15783 tcp-request content accept if RDP_COOKIE
15784 # apply RDP cookie persistence
15785 persist rdp-cookie
15786 # Persist based on the mstshash cookie
15787 # This is only useful makes sense if
15788 # balance rdp-cookie is not used
15789 stick-table type string size 204800
15790 stick on req.rdp_cookie(mstshash)
15791 server srv1 1.1.1.1:3389
15792 server srv1 1.1.1.2:3389
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015793
Willy Tarreau74ca5042013-06-11 23:12:07 +020015794 See also : "balance rdp-cookie", "persist rdp-cookie", "tcp-request" and the
15795 "req_rdp_cookie" ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015796
Willy Tarreau74ca5042013-06-11 23:12:07 +020015797req.rdp_cookie_cnt([name]) : integer
15798rdp_cookie_cnt([name]) : integer (deprecated)
15799 Tries to parse the request buffer as RDP protocol, then returns an integer
15800 corresponding to the number of RDP cookies found. If an optional cookie name
15801 is passed, only cookies matching this name are considered. This is mostly
15802 used in ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015803
Willy Tarreau74ca5042013-06-11 23:12:07 +020015804 ACL derivatives :
15805 req_rdp_cookie_cnt([<name>]) : integer match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015806
Alex Zorin4afdd132018-12-30 13:56:28 +110015807req.ssl_alpn : string
15808 Returns a string containing the values of the Application-Layer Protocol
15809 Negotiation (ALPN) TLS extension (RFC7301), sent by the client within the SSL
15810 ClientHello message. Note that this only applies to raw contents found in the
15811 request buffer and not to the contents deciphered via an SSL data layer, so
15812 this will not work with "bind" lines having the "ssl" option. This is useful
15813 in ACL to make a routing decision based upon the ALPN preferences of a TLS
Jarno Huuskonene504f812019-01-03 07:56:49 +020015814 client, like in the example below. See also "ssl_fc_alpn".
Alex Zorin4afdd132018-12-30 13:56:28 +110015815
15816 Examples :
15817 # Wait for a client hello for at most 5 seconds
15818 tcp-request inspect-delay 5s
15819 tcp-request content accept if { req_ssl_hello_type 1 }
Jarno Huuskonene504f812019-01-03 07:56:49 +020015820 use_backend bk_acme if { req.ssl_alpn acme-tls/1 }
Alex Zorin4afdd132018-12-30 13:56:28 +110015821 default_backend bk_default
15822
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020015823req.ssl_ec_ext : boolean
15824 Returns a boolean identifying if client sent the Supported Elliptic Curves
15825 Extension as defined in RFC4492, section 5.1. within the SSL ClientHello
Cyril Bonté307ee1e2015-09-28 23:16:06 +020015826 message. This can be used to present ECC compatible clients with EC
15827 certificate and to use RSA for all others, on the same IP address. Note that
15828 this only applies to raw contents found in the request buffer and not to
15829 contents deciphered via an SSL data layer, so this will not work with "bind"
15830 lines having the "ssl" option.
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020015831
Willy Tarreau74ca5042013-06-11 23:12:07 +020015832req.ssl_hello_type : integer
15833req_ssl_hello_type : integer (deprecated)
15834 Returns an integer value containing the type of the SSL hello message found
15835 in the request buffer if the buffer contains data that parse as a complete
15836 SSL (v3 or superior) client hello message. Note that this only applies to raw
15837 contents found in the request buffer and not to contents deciphered via an
15838 SSL data layer, so this will not work with "bind" lines having the "ssl"
15839 option. This is mostly used in ACL to detect presence of an SSL hello message
15840 that is supposed to contain an SSL session ID usable for stickiness.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015841
Willy Tarreau74ca5042013-06-11 23:12:07 +020015842req.ssl_sni : string
15843req_ssl_sni : string (deprecated)
15844 Returns a string containing the value of the Server Name TLS extension sent
15845 by a client in a TLS stream passing through the request buffer if the buffer
15846 contains data that parse as a complete SSL (v3 or superior) client hello
15847 message. Note that this only applies to raw contents found in the request
15848 buffer and not to contents deciphered via an SSL data layer, so this will not
15849 work with "bind" lines having the "ssl" option. SNI normally contains the
15850 name of the host the client tries to connect to (for recent browsers). SNI is
15851 useful for allowing or denying access to certain hosts when SSL/TLS is used
15852 by the client. This test was designed to be used with TCP request content
15853 inspection. If content switching is needed, it is recommended to first wait
15854 for a complete client hello (type 1), like in the example below. See also
15855 "ssl_fc_sni".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015856
Willy Tarreau74ca5042013-06-11 23:12:07 +020015857 ACL derivatives :
15858 req_ssl_sni : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015859
Willy Tarreau74ca5042013-06-11 23:12:07 +020015860 Examples :
15861 # Wait for a client hello for at most 5 seconds
15862 tcp-request inspect-delay 5s
15863 tcp-request content accept if { req_ssl_hello_type 1 }
15864 use_backend bk_allow if { req_ssl_sni -f allowed_sites }
15865 default_backend bk_sorry_page
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015866
Pradeep Jindalbb2acf52015-09-29 10:12:57 +053015867req.ssl_st_ext : integer
15868 Returns 0 if the client didn't send a SessionTicket TLS Extension (RFC5077)
15869 Returns 1 if the client sent SessionTicket TLS Extension
15870 Returns 2 if the client also sent non-zero length TLS SessionTicket
15871 Note that this only applies to raw contents found in the request buffer and
15872 not to contents deciphered via an SSL data layer, so this will not work with
15873 "bind" lines having the "ssl" option. This can for example be used to detect
15874 whether the client sent a SessionTicket or not and stick it accordingly, if
15875 no SessionTicket then stick on SessionID or don't stick as there's no server
15876 side state is there when SessionTickets are in use.
15877
Willy Tarreau74ca5042013-06-11 23:12:07 +020015878req.ssl_ver : integer
15879req_ssl_ver : integer (deprecated)
15880 Returns an integer value containing the version of the SSL/TLS protocol of a
15881 stream present in the request buffer. Both SSLv2 hello messages and SSLv3
15882 messages are supported. TLSv1 is announced as SSL version 3.1. The value is
15883 composed of the major version multiplied by 65536, added to the minor
15884 version. Note that this only applies to raw contents found in the request
15885 buffer and not to contents deciphered via an SSL data layer, so this will not
15886 work with "bind" lines having the "ssl" option. The ACL version of the test
Davor Ocelice9ed2812017-12-25 17:49:28 +010015887 matches against a decimal notation in the form MAJOR.MINOR (e.g. 3.1). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020015888 fetch is mostly used in ACL.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015889
Willy Tarreau74ca5042013-06-11 23:12:07 +020015890 ACL derivatives :
15891 req_ssl_ver : decimal match
Willy Tarreaud63335a2010-02-26 12:56:52 +010015892
Willy Tarreau47e8eba2013-09-11 23:28:46 +020015893res.len : integer
15894 Returns an integer value corresponding to the number of bytes present in the
15895 response buffer. This is mostly used in ACL. It is important to understand
15896 that this test does not return false as long as the buffer is changing. This
15897 means that a check with equality to zero will almost always immediately match
15898 at the beginning of the session, while a test for more data will wait for
15899 that data to come in and return false only when haproxy is certain that no
15900 more data will come in. This test was designed to be used with TCP response
15901 content inspection.
15902
Willy Tarreau74ca5042013-06-11 23:12:07 +020015903res.payload(<offset>,<length>) : binary
15904 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020015905 in the response buffer. As a special case, if the <length> argument is zero,
15906 the the whole buffer from <offset> to the end is extracted. This can be used
15907 with ACLs in order to check for the presence of some content in a buffer at
15908 any location.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015909
Willy Tarreau74ca5042013-06-11 23:12:07 +020015910res.payload_lv(<offset1>,<length>[,<offset2>]) : binary
15911 This extracts a binary block whose size is specified at <offset1> for <length>
15912 bytes, and which starts at <offset2> if specified or just after the length in
15913 the response buffer. The <offset2> parameter also supports relative offsets
15914 if prepended with a '+' or '-' sign.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015915
Willy Tarreau74ca5042013-06-11 23:12:07 +020015916 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015917
Willy Tarreau971f7b62015-09-29 14:06:59 +020015918res.ssl_hello_type : integer
15919rep_ssl_hello_type : integer (deprecated)
15920 Returns an integer value containing the type of the SSL hello message found
15921 in the response buffer if the buffer contains data that parses as a complete
15922 SSL (v3 or superior) hello message. Note that this only applies to raw
15923 contents found in the response buffer and not to contents deciphered via an
15924 SSL data layer, so this will not work with "server" lines having the "ssl"
15925 option. This is mostly used in ACL to detect presence of an SSL hello message
15926 that is supposed to contain an SSL session ID usable for stickiness.
15927
Willy Tarreau74ca5042013-06-11 23:12:07 +020015928wait_end : boolean
15929 This fetch either returns true when the inspection period is over, or does
15930 not fetch. It is only used in ACLs, in conjunction with content analysis to
Davor Ocelice9ed2812017-12-25 17:49:28 +010015931 avoid returning a wrong verdict early. It may also be used to delay some
Willy Tarreau74ca5042013-06-11 23:12:07 +020015932 actions, such as a delayed reject for some special addresses. Since it either
15933 stops the rules evaluation or immediately returns true, it is recommended to
Davor Ocelice9ed2812017-12-25 17:49:28 +010015934 use this acl as the last one in a rule. Please note that the default ACL
Willy Tarreau74ca5042013-06-11 23:12:07 +020015935 "WAIT_END" is always usable without prior declaration. This test was designed
15936 to be used with TCP request content inspection.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015937
Willy Tarreau74ca5042013-06-11 23:12:07 +020015938 Examples :
15939 # delay every incoming request by 2 seconds
15940 tcp-request inspect-delay 2s
15941 tcp-request content accept if WAIT_END
Willy Tarreaud63335a2010-02-26 12:56:52 +010015942
Willy Tarreau74ca5042013-06-11 23:12:07 +020015943 # don't immediately tell bad guys they are rejected
15944 tcp-request inspect-delay 10s
15945 acl goodguys src 10.0.0.0/24
15946 acl badguys src 10.0.1.0/24
15947 tcp-request content accept if goodguys
15948 tcp-request content reject if badguys WAIT_END
15949 tcp-request content reject
15950
15951
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200159527.3.6. Fetching HTTP samples (Layer 7)
Willy Tarreau74ca5042013-06-11 23:12:07 +020015953--------------------------------------
15954
15955It is possible to fetch samples from HTTP contents, requests and responses.
15956This application layer is also called layer 7. It is only possible to fetch the
15957data in this section when a full HTTP request or response has been parsed from
15958its respective request or response buffer. This is always the case with all
15959HTTP specific rules and for sections running with "mode http". When using TCP
15960content inspection, it may be necessary to support an inspection delay in order
15961to let the request or response come in first. These fetches may require a bit
15962more CPU resources than the layer 4 ones, but not much since the request and
15963response are indexed.
15964
15965base : string
15966 This returns the concatenation of the first Host header and the path part of
15967 the request, which starts at the first slash and ends before the question
15968 mark. It can be useful in virtual hosted environments to detect URL abuses as
15969 well as to improve shared caches efficiency. Using this with a limited size
15970 stick table also allows one to collect statistics about most commonly
15971 requested objects by host/path. With ACLs it can allow simple content
15972 switching rules involving the host and the path at the same time, such as
15973 "www.example.com/favicon.ico". See also "path" and "uri".
15974
15975 ACL derivatives :
15976 base : exact string match
15977 base_beg : prefix match
15978 base_dir : subdir match
15979 base_dom : domain match
15980 base_end : suffix match
15981 base_len : length match
15982 base_reg : regex match
15983 base_sub : substring match
15984
15985base32 : integer
15986 This returns a 32-bit hash of the value returned by the "base" fetch method
15987 above. This is useful to track per-URL activity on high traffic sites without
15988 having to store all URLs. Instead a shorter hash is stored, saving a lot of
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015989 memory. The output type is an unsigned integer. The hash function used is
15990 SDBM with full avalanche on the output. Technically, base32 is exactly equal
15991 to "base,sdbm(1)".
Willy Tarreau74ca5042013-06-11 23:12:07 +020015992
15993base32+src : binary
15994 This returns the concatenation of the base32 fetch above and the src fetch
15995 below. The resulting type is of type binary, with a size of 8 or 20 bytes
15996 depending on the source address family. This can be used to track per-IP,
15997 per-URL counters.
15998
William Lallemand65ad6e12014-01-31 15:08:02 +010015999capture.req.hdr(<idx>) : string
16000 This extracts the content of the header captured by the "capture request
16001 header", idx is the position of the capture keyword in the configuration.
16002 The first entry is an index of 0. See also: "capture request header".
16003
16004capture.req.method : string
16005 This extracts the METHOD of an HTTP request. It can be used in both request
16006 and response. Unlike "method", it can be used in both request and response
16007 because it's allocated.
16008
16009capture.req.uri : string
16010 This extracts the request's URI, which starts at the first slash and ends
16011 before the first space in the request (without the host part). Unlike "path"
16012 and "url", it can be used in both request and response because it's
16013 allocated.
16014
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020016015capture.req.ver : string
16016 This extracts the request's HTTP version and returns either "HTTP/1.0" or
16017 "HTTP/1.1". Unlike "req.ver", it can be used in both request, response, and
16018 logs because it relies on a persistent flag.
16019
William Lallemand65ad6e12014-01-31 15:08:02 +010016020capture.res.hdr(<idx>) : string
16021 This extracts the content of the header captured by the "capture response
16022 header", idx is the position of the capture keyword in the configuration.
16023 The first entry is an index of 0.
16024 See also: "capture response header"
16025
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020016026capture.res.ver : string
16027 This extracts the response's HTTP version and returns either "HTTP/1.0" or
16028 "HTTP/1.1". Unlike "res.ver", it can be used in logs because it relies on a
16029 persistent flag.
16030
Willy Tarreaua5910cc2015-05-02 00:46:08 +020016031req.body : binary
16032 This returns the HTTP request's available body as a block of data. It
16033 requires that the request body has been buffered made available using
16034 "option http-buffer-request". In case of chunked-encoded body, currently only
16035 the first chunk is analyzed.
16036
Thierry FOURNIER9826c772015-05-20 15:50:54 +020016037req.body_param([<name>) : string
16038 This fetch assumes that the body of the POST request is url-encoded. The user
16039 can check if the "content-type" contains the value
16040 "application/x-www-form-urlencoded". This extracts the first occurrence of the
16041 parameter <name> in the body, which ends before '&'. The parameter name is
16042 case-sensitive. If no name is given, any parameter will match, and the first
16043 one will be returned. The result is a string corresponding to the value of the
16044 parameter <name> as presented in the request body (no URL decoding is
16045 performed). Note that the ACL version of this fetch iterates over multiple
16046 parameters and will iteratively report all parameters values if no name is
16047 given.
16048
Willy Tarreaua5910cc2015-05-02 00:46:08 +020016049req.body_len : integer
16050 This returns the length of the HTTP request's available body in bytes. It may
16051 be lower than the advertised length if the body is larger than the buffer. It
16052 requires that the request body has been buffered made available using
16053 "option http-buffer-request".
16054
16055req.body_size : integer
16056 This returns the advertised length of the HTTP request's body in bytes. It
16057 will represent the advertised Content-Length header, or the size of the first
16058 chunk in case of chunked encoding. In order to parse the chunks, it requires
16059 that the request body has been buffered made available using
16060 "option http-buffer-request".
16061
Willy Tarreau74ca5042013-06-11 23:12:07 +020016062req.cook([<name>]) : string
16063cook([<name>]) : string (deprecated)
16064 This extracts the last occurrence of the cookie name <name> on a "Cookie"
16065 header line from the request, and returns its value as string. If no name is
16066 specified, the first cookie value is returned. When used with ACLs, all
16067 matching cookies are evaluated. Spaces around the name and the value are
16068 ignored as requested by the Cookie header specification (RFC6265). The cookie
16069 name is case-sensitive. Empty cookies are valid, so an empty cookie may very
16070 well return an empty value if it is present. Use the "found" match to detect
16071 presence. Use the res.cook() variant for response cookies sent by the server.
16072
16073 ACL derivatives :
16074 cook([<name>]) : exact string match
16075 cook_beg([<name>]) : prefix match
16076 cook_dir([<name>]) : subdir match
16077 cook_dom([<name>]) : domain match
16078 cook_end([<name>]) : suffix match
16079 cook_len([<name>]) : length match
16080 cook_reg([<name>]) : regex match
16081 cook_sub([<name>]) : substring match
Willy Tarreaud63335a2010-02-26 12:56:52 +010016082
Willy Tarreau74ca5042013-06-11 23:12:07 +020016083req.cook_cnt([<name>]) : integer
16084cook_cnt([<name>]) : integer (deprecated)
16085 Returns an integer value representing the number of occurrences of the cookie
16086 <name> in the request, or all cookies if <name> is not specified.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016087
Willy Tarreau74ca5042013-06-11 23:12:07 +020016088req.cook_val([<name>]) : integer
16089cook_val([<name>]) : integer (deprecated)
16090 This extracts the last occurrence of the cookie name <name> on a "Cookie"
16091 header line from the request, and converts its value to an integer which is
16092 returned. If no name is specified, the first cookie value is returned. When
16093 used in ACLs, all matching names are iterated over until a value matches.
Willy Tarreau0e698542011-09-16 08:32:32 +020016094
Willy Tarreau74ca5042013-06-11 23:12:07 +020016095cookie([<name>]) : string (deprecated)
16096 This extracts the last occurrence of the cookie name <name> on a "Cookie"
16097 header line from the request, or a "Set-Cookie" header from the response, and
16098 returns its value as a string. A typical use is to get multiple clients
16099 sharing a same profile use the same server. This can be similar to what
Willy Tarreau294d0f02015-08-10 19:40:12 +020016100 "appsession" did with the "request-learn" statement, but with support for
Willy Tarreau74ca5042013-06-11 23:12:07 +020016101 multi-peer synchronization and state keeping across restarts. If no name is
16102 specified, the first cookie value is returned. This fetch should not be used
16103 anymore and should be replaced by req.cook() or res.cook() instead as it
16104 ambiguously uses the direction based on the context where it is used.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016105
Willy Tarreau74ca5042013-06-11 23:12:07 +020016106hdr([<name>[,<occ>]]) : string
16107 This is equivalent to req.hdr() when used on requests, and to res.hdr() when
16108 used on responses. Please refer to these respective fetches for more details.
16109 In case of doubt about the fetch direction, please use the explicit ones.
16110 Note that contrary to the hdr() sample fetch method, the hdr_* ACL keywords
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016111 unambiguously apply to the request headers.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016112
Willy Tarreau74ca5042013-06-11 23:12:07 +020016113req.fhdr(<name>[,<occ>]) : string
16114 This extracts the last occurrence of header <name> in an HTTP request. When
16115 used from an ACL, all occurrences are iterated over until a match is found.
16116 Optionally, a specific occurrence might be specified as a position number.
16117 Positive values indicate a position from the first occurrence, with 1 being
16118 the first one. Negative values indicate positions relative to the last one,
16119 with -1 being the last one. It differs from req.hdr() in that any commas
16120 present in the value are returned and are not used as delimiters. This is
16121 sometimes useful with headers such as User-Agent.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016122
Willy Tarreau74ca5042013-06-11 23:12:07 +020016123req.fhdr_cnt([<name>]) : integer
16124 Returns an integer value representing the number of occurrences of request
16125 header field name <name>, or the total number of header fields if <name> is
16126 not specified. Contrary to its req.hdr_cnt() cousin, this function returns
16127 the number of full line headers and does not stop on commas.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016128
Willy Tarreau74ca5042013-06-11 23:12:07 +020016129req.hdr([<name>[,<occ>]]) : string
16130 This extracts the last occurrence of header <name> in an HTTP request. When
16131 used from an ACL, all occurrences are iterated over until a match is found.
16132 Optionally, a specific occurrence might be specified as a position number.
16133 Positive values indicate a position from the first occurrence, with 1 being
16134 the first one. Negative values indicate positions relative to the last one,
16135 with -1 being the last one. A typical use is with the X-Forwarded-For header
16136 once converted to IP, associated with an IP stick-table. The function
16137 considers any comma as a delimiter for distinct values. If full-line headers
Lukas Tribus23953682017-04-28 13:24:30 +000016138 are desired instead, use req.fhdr(). Please carefully check RFC7231 to know
Willy Tarreau74ca5042013-06-11 23:12:07 +020016139 how certain headers are supposed to be parsed. Also, some of them are case
Davor Ocelice9ed2812017-12-25 17:49:28 +010016140 insensitive (e.g. Connection).
Willy Tarreaud63335a2010-02-26 12:56:52 +010016141
Willy Tarreau74ca5042013-06-11 23:12:07 +020016142 ACL derivatives :
16143 hdr([<name>[,<occ>]]) : exact string match
16144 hdr_beg([<name>[,<occ>]]) : prefix match
16145 hdr_dir([<name>[,<occ>]]) : subdir match
16146 hdr_dom([<name>[,<occ>]]) : domain match
16147 hdr_end([<name>[,<occ>]]) : suffix match
16148 hdr_len([<name>[,<occ>]]) : length match
16149 hdr_reg([<name>[,<occ>]]) : regex match
16150 hdr_sub([<name>[,<occ>]]) : substring match
16151
16152req.hdr_cnt([<name>]) : integer
16153hdr_cnt([<header>]) : integer (deprecated)
16154 Returns an integer value representing the number of occurrences of request
16155 header field name <name>, or the total number of header field values if
16156 <name> is not specified. It is important to remember that one header line may
16157 count as several headers if it has several values. The function considers any
16158 comma as a delimiter for distinct values. If full-line headers are desired
16159 instead, req.fhdr_cnt() should be used instead. With ACLs, it can be used to
16160 detect presence, absence or abuse of a specific header, as well as to block
16161 request smuggling attacks by rejecting requests which contain more than one
16162 of certain headers. See "req.hdr" for more information on header matching.
16163
16164req.hdr_ip([<name>[,<occ>]]) : ip
16165hdr_ip([<name>[,<occ>]]) : ip (deprecated)
16166 This extracts the last occurrence of header <name> in an HTTP request,
16167 converts it to an IPv4 or IPv6 address and returns this address. When used
16168 with ACLs, all occurrences are checked, and if <name> is omitted, every value
16169 of every header is checked. Optionally, a specific occurrence might be
16170 specified as a position number. Positive values indicate a position from the
Davor Ocelice9ed2812017-12-25 17:49:28 +010016171 first occurrence, with 1 being the first one. Negative values indicate
Willy Tarreau74ca5042013-06-11 23:12:07 +020016172 positions relative to the last one, with -1 being the last one. A typical use
16173 is with the X-Forwarded-For and X-Client-IP headers.
16174
16175req.hdr_val([<name>[,<occ>]]) : integer
16176hdr_val([<name>[,<occ>]]) : integer (deprecated)
16177 This extracts the last occurrence of header <name> in an HTTP request, and
16178 converts it to an integer value. When used with ACLs, all occurrences are
16179 checked, and if <name> is omitted, every value of every header is checked.
16180 Optionally, a specific occurrence might be specified as a position number.
16181 Positive values indicate a position from the first occurrence, with 1 being
16182 the first one. Negative values indicate positions relative to the last one,
16183 with -1 being the last one. A typical use is with the X-Forwarded-For header.
16184
Frédéric Lécailleec891192019-02-26 15:02:35 +010016185
16186
Willy Tarreau74ca5042013-06-11 23:12:07 +020016187http_auth(<userlist>) : boolean
16188 Returns a boolean indicating whether the authentication data received from
16189 the client match a username & password stored in the specified userlist. This
16190 fetch function is not really useful outside of ACLs. Currently only http
16191 basic auth is supported.
16192
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010016193http_auth_group(<userlist>) : string
16194 Returns a string corresponding to the user name found in the authentication
16195 data received from the client if both the user name and password are valid
16196 according to the specified userlist. The main purpose is to use it in ACLs
16197 where it is then checked whether the user belongs to any group within a list.
Willy Tarreau74ca5042013-06-11 23:12:07 +020016198 This fetch function is not really useful outside of ACLs. Currently only http
16199 basic auth is supported.
16200
16201 ACL derivatives :
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010016202 http_auth_group(<userlist>) : group ...
16203 Returns true when the user extracted from the request and whose password is
16204 valid according to the specified userlist belongs to at least one of the
16205 groups.
Willy Tarreau74ca5042013-06-11 23:12:07 +020016206
Christopher Fauleta4063562019-08-02 11:51:37 +020016207http_auth_pass : string
16208 Returns the user's password found in the authentication data received from
16209 the client, as supplied in the Authorization header. Not checks are
16210 performed by this sample fetch. Only Basic authentication is supported.
16211
16212http_auth_type : string
16213 Returns the authentication method found in the authentication data received from
16214 the client, as supplied in the Authorization header. Not checks are
16215 performed by this sample fetch. Only Basic authentication is supported.
16216
16217http_auth_user : string
16218 Returns the user name found in the authentication data received from the
16219 client, as supplied in the Authorization header. Not checks are performed by
16220 this sample fetch. Only Basic authentication is supported.
16221
Willy Tarreau74ca5042013-06-11 23:12:07 +020016222http_first_req : boolean
Willy Tarreau7f18e522010-10-22 20:04:13 +020016223 Returns true when the request being processed is the first one of the
16224 connection. This can be used to add or remove headers that may be missing
Willy Tarreau74ca5042013-06-11 23:12:07 +020016225 from some requests when a request is not the first one, or to help grouping
16226 requests in the logs.
Willy Tarreau7f18e522010-10-22 20:04:13 +020016227
Willy Tarreau74ca5042013-06-11 23:12:07 +020016228method : integer + string
16229 Returns an integer value corresponding to the method in the HTTP request. For
16230 example, "GET" equals 1 (check sources to establish the matching). Value 9
16231 means "other method" and may be converted to a string extracted from the
16232 stream. This should not be used directly as a sample, this is only meant to
16233 be used from ACLs, which transparently convert methods from patterns to these
16234 integer + string values. Some predefined ACL already check for most common
16235 methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016236
Willy Tarreau74ca5042013-06-11 23:12:07 +020016237 ACL derivatives :
16238 method : case insensitive method match
Willy Tarreau6a06a402007-07-15 20:15:28 +020016239
Willy Tarreau74ca5042013-06-11 23:12:07 +020016240 Example :
16241 # only accept GET and HEAD requests
16242 acl valid_method method GET HEAD
16243 http-request deny if ! valid_method
Willy Tarreau6a06a402007-07-15 20:15:28 +020016244
Willy Tarreau74ca5042013-06-11 23:12:07 +020016245path : string
16246 This extracts the request's URL path, which starts at the first slash and
16247 ends before the question mark (without the host part). A typical use is with
16248 prefetch-capable caches, and with portals which need to aggregate multiple
16249 information from databases and keep them in caches. Note that with outgoing
16250 caches, it would be wiser to use "url" instead. With ACLs, it's typically
Davor Ocelice9ed2812017-12-25 17:49:28 +010016251 used to match exact file names (e.g. "/login.php"), or directory parts using
Willy Tarreau74ca5042013-06-11 23:12:07 +020016252 the derivative forms. See also the "url" and "base" fetch methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016253
Willy Tarreau74ca5042013-06-11 23:12:07 +020016254 ACL derivatives :
16255 path : exact string match
16256 path_beg : prefix match
16257 path_dir : subdir match
16258 path_dom : domain match
16259 path_end : suffix match
16260 path_len : length match
16261 path_reg : regex match
16262 path_sub : substring match
Willy Tarreau6a06a402007-07-15 20:15:28 +020016263
Willy Tarreau49ad95c2015-01-19 15:06:26 +010016264query : string
16265 This extracts the request's query string, which starts after the first
16266 question mark. If no question mark is present, this fetch returns nothing. If
16267 a question mark is present but nothing follows, it returns an empty string.
16268 This means it's possible to easily know whether a query string is present
Tim Düsterhus4896c442016-11-29 02:15:19 +010016269 using the "found" matching method. This fetch is the complement of "path"
Willy Tarreau49ad95c2015-01-19 15:06:26 +010016270 which stops before the question mark.
16271
Willy Tarreaueb27ec72015-02-20 13:55:29 +010016272req.hdr_names([<delim>]) : string
16273 This builds a string made from the concatenation of all header names as they
16274 appear in the request when the rule is evaluated. The default delimiter is
16275 the comma (',') but it may be overridden as an optional argument <delim>. In
16276 this case, only the first character of <delim> is considered.
16277
Willy Tarreau74ca5042013-06-11 23:12:07 +020016278req.ver : string
16279req_ver : string (deprecated)
16280 Returns the version string from the HTTP request, for example "1.1". This can
16281 be useful for logs, but is mostly there for ACL. Some predefined ACL already
16282 check for versions 1.0 and 1.1.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016283
Willy Tarreau74ca5042013-06-11 23:12:07 +020016284 ACL derivatives :
16285 req_ver : exact string match
Willy Tarreau0e698542011-09-16 08:32:32 +020016286
Willy Tarreau74ca5042013-06-11 23:12:07 +020016287res.comp : boolean
16288 Returns the boolean "true" value if the response has been compressed by
16289 HAProxy, otherwise returns boolean "false". This may be used to add
16290 information in the logs.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016291
Willy Tarreau74ca5042013-06-11 23:12:07 +020016292res.comp_algo : string
16293 Returns a string containing the name of the algorithm used if the response
16294 was compressed by HAProxy, for example : "deflate". This may be used to add
16295 some information in the logs.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016296
Willy Tarreau74ca5042013-06-11 23:12:07 +020016297res.cook([<name>]) : string
16298scook([<name>]) : string (deprecated)
16299 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
16300 header line from the response, and returns its value as string. If no name is
16301 specified, the first cookie value is returned.
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020016302
Willy Tarreau74ca5042013-06-11 23:12:07 +020016303 ACL derivatives :
16304 scook([<name>] : exact string match
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020016305
Willy Tarreau74ca5042013-06-11 23:12:07 +020016306res.cook_cnt([<name>]) : integer
16307scook_cnt([<name>]) : integer (deprecated)
16308 Returns an integer value representing the number of occurrences of the cookie
16309 <name> in the response, or all cookies if <name> is not specified. This is
16310 mostly useful when combined with ACLs to detect suspicious responses.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016311
Willy Tarreau74ca5042013-06-11 23:12:07 +020016312res.cook_val([<name>]) : integer
16313scook_val([<name>]) : integer (deprecated)
16314 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
16315 header line from the response, and converts its value to an integer which is
16316 returned. If no name is specified, the first cookie value is returned.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016317
Willy Tarreau74ca5042013-06-11 23:12:07 +020016318res.fhdr([<name>[,<occ>]]) : string
16319 This extracts the last occurrence of header <name> in an HTTP response, or of
16320 the last header if no <name> is specified. Optionally, a specific occurrence
16321 might be specified as a position number. Positive values indicate a position
16322 from the first occurrence, with 1 being the first one. Negative values
16323 indicate positions relative to the last one, with -1 being the last one. It
16324 differs from res.hdr() in that any commas present in the value are returned
16325 and are not used as delimiters. If this is not desired, the res.hdr() fetch
16326 should be used instead. This is sometimes useful with headers such as Date or
16327 Expires.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016328
Willy Tarreau74ca5042013-06-11 23:12:07 +020016329res.fhdr_cnt([<name>]) : integer
16330 Returns an integer value representing the number of occurrences of response
16331 header field name <name>, or the total number of header fields if <name> is
16332 not specified. Contrary to its res.hdr_cnt() cousin, this function returns
16333 the number of full line headers and does not stop on commas. If this is not
16334 desired, the res.hdr_cnt() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016335
Willy Tarreau74ca5042013-06-11 23:12:07 +020016336res.hdr([<name>[,<occ>]]) : string
16337shdr([<name>[,<occ>]]) : string (deprecated)
16338 This extracts the last occurrence of header <name> in an HTTP response, or of
16339 the last header if no <name> is specified. Optionally, a specific occurrence
16340 might be specified as a position number. Positive values indicate a position
16341 from the first occurrence, with 1 being the first one. Negative values
16342 indicate positions relative to the last one, with -1 being the last one. This
16343 can be useful to learn some data into a stick-table. The function considers
16344 any comma as a delimiter for distinct values. If this is not desired, the
16345 res.fhdr() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016346
Willy Tarreau74ca5042013-06-11 23:12:07 +020016347 ACL derivatives :
16348 shdr([<name>[,<occ>]]) : exact string match
16349 shdr_beg([<name>[,<occ>]]) : prefix match
16350 shdr_dir([<name>[,<occ>]]) : subdir match
16351 shdr_dom([<name>[,<occ>]]) : domain match
16352 shdr_end([<name>[,<occ>]]) : suffix match
16353 shdr_len([<name>[,<occ>]]) : length match
16354 shdr_reg([<name>[,<occ>]]) : regex match
16355 shdr_sub([<name>[,<occ>]]) : substring match
16356
16357res.hdr_cnt([<name>]) : integer
16358shdr_cnt([<name>]) : integer (deprecated)
16359 Returns an integer value representing the number of occurrences of response
16360 header field name <name>, or the total number of header fields if <name> is
16361 not specified. The function considers any comma as a delimiter for distinct
16362 values. If this is not desired, the res.fhdr_cnt() fetch should be used
16363 instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016364
Willy Tarreau74ca5042013-06-11 23:12:07 +020016365res.hdr_ip([<name>[,<occ>]]) : ip
16366shdr_ip([<name>[,<occ>]]) : ip (deprecated)
16367 This extracts the last occurrence of header <name> in an HTTP response,
16368 convert it to an IPv4 or IPv6 address and returns this address. Optionally, a
16369 specific occurrence might be specified as a position number. Positive values
16370 indicate a position from the first occurrence, with 1 being the first one.
16371 Negative values indicate positions relative to the last one, with -1 being
16372 the last one. This can be useful to learn some data into a stick table.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016373
Willy Tarreaueb27ec72015-02-20 13:55:29 +010016374res.hdr_names([<delim>]) : string
16375 This builds a string made from the concatenation of all header names as they
16376 appear in the response when the rule is evaluated. The default delimiter is
16377 the comma (',') but it may be overridden as an optional argument <delim>. In
16378 this case, only the first character of <delim> is considered.
16379
Willy Tarreau74ca5042013-06-11 23:12:07 +020016380res.hdr_val([<name>[,<occ>]]) : integer
16381shdr_val([<name>[,<occ>]]) : integer (deprecated)
16382 This extracts the last occurrence of header <name> in an HTTP response, and
16383 converts it to an integer value. Optionally, a specific occurrence might be
16384 specified as a position number. Positive values indicate a position from the
16385 first occurrence, with 1 being the first one. Negative values indicate
16386 positions relative to the last one, with -1 being the last one. This can be
16387 useful to learn some data into a stick table.
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010016388
Willy Tarreau74ca5042013-06-11 23:12:07 +020016389res.ver : string
16390resp_ver : string (deprecated)
16391 Returns the version string from the HTTP response, for example "1.1". This
16392 can be useful for logs, but is mostly there for ACL.
Willy Tarreau0e698542011-09-16 08:32:32 +020016393
Willy Tarreau74ca5042013-06-11 23:12:07 +020016394 ACL derivatives :
16395 resp_ver : exact string match
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010016396
Willy Tarreau74ca5042013-06-11 23:12:07 +020016397set-cookie([<name>]) : string (deprecated)
16398 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
16399 header line from the response and uses the corresponding value to match. This
Willy Tarreau294d0f02015-08-10 19:40:12 +020016400 can be comparable to what "appsession" did with default options, but with
Willy Tarreau74ca5042013-06-11 23:12:07 +020016401 support for multi-peer synchronization and state keeping across restarts.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010016402
Willy Tarreau74ca5042013-06-11 23:12:07 +020016403 This fetch function is deprecated and has been superseded by the "res.cook"
16404 fetch. This keyword will disappear soon.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010016405
Willy Tarreau74ca5042013-06-11 23:12:07 +020016406status : integer
16407 Returns an integer containing the HTTP status code in the HTTP response, for
16408 example, 302. It is mostly used within ACLs and integer ranges, for example,
16409 to remove any Location header if the response is not a 3xx.
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016410
Thierry Fournier0e00dca2016-04-07 15:47:40 +020016411unique-id : string
16412 Returns the unique-id attached to the request. The directive
16413 "unique-id-format" must be set. If it is not set, the unique-id sample fetch
16414 fails. Note that the unique-id is usually used with HTTP requests, however this
16415 sample fetch can be used with other protocols. Obviously, if it is used with
16416 other protocols than HTTP, the unique-id-format directive must not contain
16417 HTTP parts. See: unique-id-format and unique-id-header
16418
Willy Tarreau74ca5042013-06-11 23:12:07 +020016419url : string
16420 This extracts the request's URL as presented in the request. A typical use is
16421 with prefetch-capable caches, and with portals which need to aggregate
16422 multiple information from databases and keep them in caches. With ACLs, using
16423 "path" is preferred over using "url", because clients may send a full URL as
16424 is normally done with proxies. The only real use is to match "*" which does
16425 not match in "path", and for which there is already a predefined ACL. See
16426 also "path" and "base".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016427
Willy Tarreau74ca5042013-06-11 23:12:07 +020016428 ACL derivatives :
16429 url : exact string match
16430 url_beg : prefix match
16431 url_dir : subdir match
16432 url_dom : domain match
16433 url_end : suffix match
16434 url_len : length match
16435 url_reg : regex match
16436 url_sub : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016437
Willy Tarreau74ca5042013-06-11 23:12:07 +020016438url_ip : ip
16439 This extracts the IP address from the request's URL when the host part is
16440 presented as an IP address. Its use is very limited. For instance, a
16441 monitoring system might use this field as an alternative for the source IP in
16442 order to test what path a given source address would follow, or to force an
16443 entry in a table for a given source address. With ACLs it can be used to
16444 restrict access to certain systems through a proxy, for example when combined
16445 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016446
Willy Tarreau74ca5042013-06-11 23:12:07 +020016447url_port : integer
16448 This extracts the port part from the request's URL. Note that if the port is
16449 not specified in the request, port 80 is assumed. With ACLs it can be used to
16450 restrict access to certain systems through a proxy, for example when combined
16451 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016452
Willy Tarreau1ede1da2015-05-07 16:06:18 +020016453urlp([<name>[,<delim>]]) : string
16454url_param([<name>[,<delim>]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020016455 This extracts the first occurrence of the parameter <name> in the query
16456 string, which begins after either '?' or <delim>, and which ends before '&',
Willy Tarreau1ede1da2015-05-07 16:06:18 +020016457 ';' or <delim>. The parameter name is case-sensitive. If no name is given,
16458 any parameter will match, and the first one will be returned. The result is
16459 a string corresponding to the value of the parameter <name> as presented in
16460 the request (no URL decoding is performed). This can be used for session
Willy Tarreau74ca5042013-06-11 23:12:07 +020016461 stickiness based on a client ID, to extract an application cookie passed as a
16462 URL parameter, or in ACLs to apply some checks. Note that the ACL version of
Willy Tarreau1ede1da2015-05-07 16:06:18 +020016463 this fetch iterates over multiple parameters and will iteratively report all
16464 parameters values if no name is given
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016465
Willy Tarreau74ca5042013-06-11 23:12:07 +020016466 ACL derivatives :
16467 urlp(<name>[,<delim>]) : exact string match
16468 urlp_beg(<name>[,<delim>]) : prefix match
16469 urlp_dir(<name>[,<delim>]) : subdir match
16470 urlp_dom(<name>[,<delim>]) : domain match
16471 urlp_end(<name>[,<delim>]) : suffix match
16472 urlp_len(<name>[,<delim>]) : length match
16473 urlp_reg(<name>[,<delim>]) : regex match
16474 urlp_sub(<name>[,<delim>]) : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016475
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016476
Willy Tarreau74ca5042013-06-11 23:12:07 +020016477 Example :
16478 # match http://example.com/foo?PHPSESSIONID=some_id
16479 stick on urlp(PHPSESSIONID)
16480 # match http://example.com/foo;JSESSIONID=some_id
16481 stick on urlp(JSESSIONID,;)
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016482
Jarno Huuskonen676f6222017-03-30 09:19:45 +030016483urlp_val([<name>[,<delim>]]) : integer
Willy Tarreau74ca5042013-06-11 23:12:07 +020016484 See "urlp" above. This one extracts the URL parameter <name> in the request
16485 and converts it to an integer value. This can be used for session stickiness
16486 based on a user ID for example, or with ACLs to match a page number or price.
Willy Tarreaua9fddca2012-07-31 07:51:48 +020016487
Dragan Dosen0070cd52016-06-16 12:19:49 +020016488url32 : integer
16489 This returns a 32-bit hash of the value obtained by concatenating the first
16490 Host header and the whole URL including parameters (not only the path part of
16491 the request, as in the "base32" fetch above). This is useful to track per-URL
16492 activity. A shorter hash is stored, saving a lot of memory. The output type
16493 is an unsigned integer.
16494
16495url32+src : binary
16496 This returns the concatenation of the "url32" fetch and the "src" fetch. The
16497 resulting type is of type binary, with a size of 8 or 20 bytes depending on
16498 the source address family. This can be used to track per-IP, per-URL counters.
16499
Willy Tarreau198a7442008-01-17 12:05:32 +010016500
Christopher Fauletd47941d2020-01-08 14:40:19 +0100165017.3.7. Fetching samples for developers
16502---------------------------------------
16503
16504This set of sample fetch methods is reserved to developers and must never be
16505used on a production environment, except on developer demand, for debugging
16506purposes. Moreover, no special care will be taken on backwards compatibility.
16507There is no warranty the following sample fetches will never change, be renamed
16508or simply removed. So be really careful if you should use one of them. To avoid
16509any ambiguity, these sample fetches are placed in the dedicated scope "internal",
16510for instance "internal.strm.is_htx".
16511
16512internal.htx.data : integer
16513 Returns the size in bytes used by data in the HTX message associated to a
16514 channel. The channel is chosen depending on the sample direction.
16515
16516internal.htx.free : integer
16517 Returns the free space (size - used) in bytes in the HTX message associated
16518 to a channel. The channel is chosen depending on the sample direction.
16519
16520internal.htx.free_data : integer
16521 Returns the free space for the data in bytes in the HTX message associated to
16522 a channel. The channel is chosen depending on the sample direction.
16523
16524internal.htx.has_eom : boolean
16525 Returns true if the HTX message associated to a channel contains an
16526 end-of-message block (EOM). Otherwise, it returns false. The channel is
16527 chosen depending on the sample direction.
16528
16529internal.htx.nbblks : integer
16530 Returns the number of blocks present in the HTX message associated to a
16531 channel. The channel is chosen depending on the sample direction.
16532
16533internal.htx.size : integer
16534 Returns the total size in bytes of the HTX message associated to a
16535 channel. The channel is chosen depending on the sample direction.
16536
16537internal.htx.used : integer
16538 Returns the total size used in bytes (data + metadata) in the HTX message
16539 associated to a channel. The channel is chosen depending on the sample
16540 direction.
16541
16542internal.htx_blk.size(<idx>) : integer
16543 Returns the size of the block at the position <idx> in the HTX message
16544 associated to a channel or 0 if it does not exist. The channel is chosen
16545 depending on the sample direction. <idx> may be any positive integer or one
16546 of the special value :
16547 * head : The oldest inserted block
16548 * tail : The newest inserted block
16549 * fisrt : The first block where to (re)start the analysis
16550
16551internal.htx_blk.type(<idx>) : string
16552 Returns the type of the block at the position <idx> in the HTX message
16553 associated to a channel or "HTX_BLK_UNUSED" if it does not exist. The channel
16554 is chosen depending on the sample direction. <idx> may be any positive
16555 integer or one of the special value :
16556 * head : The oldest inserted block
16557 * tail : The newest inserted block
16558 * fisrt : The first block where to (re)start the analysis
16559
16560internal.htx_blk.data(<idx>) : binary
16561 Returns the value of the DATA block at the position <idx> in the HTX message
16562 associated to a channel or an empty string if it does not exist or if it is
16563 not a DATA block. The channel is chosen depending on the sample direction.
16564 <idx> may be any positive integer or one of the special value :
16565
16566 * head : The oldest inserted block
16567 * tail : The newest inserted block
16568 * fisrt : The first block where to (re)start the analysis
16569
16570internal.htx_blk.hdrname(<idx>) : string
16571 Returns the header name of the HEADER block at the position <idx> in the HTX
16572 message associated to a channel or an empty string if it does not exist or if
16573 it is not an HEADER block. The channel is chosen depending on the sample
16574 direction. <idx> may be any positive integer or one of the special value :
16575
16576 * head : The oldest inserted block
16577 * tail : The newest inserted block
16578 * fisrt : The first block where to (re)start the analysis
16579
16580internal.htx_blk.hdrval(<idx>) : string
16581 Returns the header value of the HEADER block at the position <idx> in the HTX
16582 message associated to a channel or an empty string if it does not exist or if
16583 it is not an HEADER block. The channel is chosen depending on the sample
16584 direction. <idx> may be any positive integer or one of the special value :
16585
16586 * head : The oldest inserted block
16587 * tail : The newest inserted block
16588 * fisrt : The first block where to (re)start the analysis
16589
16590internal.htx_blk.start_line(<idx>) : string
16591 Returns the value of the REQ_SL or RES_SL block at the position <idx> in the
16592 HTX message associated to a channel or an empty string if it does not exist
16593 or if it is not a SL block. The channel is chosen depending on the sample
16594 direction. <idx> may be any positive integer or one of the special value :
16595
16596 * head : The oldest inserted block
16597 * tail : The newest inserted block
16598 * fisrt : The first block where to (re)start the analysis
16599
16600internal.strm.is_htx : boolean
16601 Returns true if the current stream is an HTX stream. It means the data in the
16602 channels buffers are stored using the internal HTX representation. Otherwise,
16603 it returns false.
16604
16605
Willy Tarreau74ca5042013-06-11 23:12:07 +0200166067.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016607---------------------
Willy Tarreauced27012008-01-17 20:35:34 +010016608
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016609Some predefined ACLs are hard-coded so that they do not have to be declared in
16610every frontend which needs them. They all have their names in upper case in
Patrick Mézard2382ad62010-05-09 10:43:32 +020016611order to avoid confusion. Their equivalence is provided below.
Willy Tarreauced27012008-01-17 20:35:34 +010016612
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016613ACL name Equivalent to Usage
16614---------------+-----------------------------+---------------------------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016615FALSE always_false never match
Willy Tarreau2492d5b2009-07-11 00:06:00 +020016616HTTP req_proto_http match if protocol is valid HTTP
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016617HTTP_1.0 req_ver 1.0 match HTTP version 1.0
16618HTTP_1.1 req_ver 1.1 match HTTP version 1.1
Willy Tarreaud63335a2010-02-26 12:56:52 +010016619HTTP_CONTENT hdr_val(content-length) gt 0 match an existing content-length
16620HTTP_URL_ABS url_reg ^[^/:]*:// match absolute URL with scheme
16621HTTP_URL_SLASH url_beg / match URL beginning with "/"
16622HTTP_URL_STAR url * match URL equal to "*"
16623LOCALHOST src 127.0.0.1/8 match connection from local host
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016624METH_CONNECT method CONNECT match HTTP CONNECT method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020016625METH_DELETE method DELETE match HTTP DELETE method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016626METH_GET method GET HEAD match HTTP GET or HEAD method
16627METH_HEAD method HEAD match HTTP HEAD method
16628METH_OPTIONS method OPTIONS match HTTP OPTIONS method
16629METH_POST method POST match HTTP POST method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020016630METH_PUT method PUT match HTTP PUT method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016631METH_TRACE method TRACE match HTTP TRACE method
Emeric Brunbede3d02009-06-30 17:54:00 +020016632RDP_COOKIE req_rdp_cookie_cnt gt 0 match presence of an RDP cookie
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016633REQ_CONTENT req_len gt 0 match data in the request buffer
Willy Tarreaud63335a2010-02-26 12:56:52 +010016634TRUE always_true always match
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016635WAIT_END wait_end wait for end of content analysis
16636---------------+-----------------------------+---------------------------------
Willy Tarreauced27012008-01-17 20:35:34 +010016637
Willy Tarreaub937b7e2010-01-12 15:27:54 +010016638
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200166398. Logging
16640----------
Willy Tarreau844e3c52008-01-11 16:28:18 +010016641
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016642One of HAProxy's strong points certainly lies is its precise logs. It probably
16643provides the finest level of information available for such a product, which is
16644very important for troubleshooting complex environments. Standard information
16645provided in logs include client ports, TCP/HTTP state timers, precise session
16646state at termination and precise termination cause, information about decisions
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010016647to direct traffic to a server, and of course the ability to capture arbitrary
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016648headers.
16649
16650In order to improve administrators reactivity, it offers a great transparency
16651about encountered problems, both internal and external, and it is possible to
16652send logs to different sources at the same time with different level filters :
16653
16654 - global process-level logs (system errors, start/stop, etc..)
16655 - per-instance system and internal errors (lack of resource, bugs, ...)
16656 - per-instance external troubles (servers up/down, max connections)
16657 - per-instance activity (client connections), either at the establishment or
16658 at the termination.
Davor Ocelice9ed2812017-12-25 17:49:28 +010016659 - per-request control of log-level, e.g.
Jim Freeman9e8714b2015-05-26 09:16:34 -060016660 http-request set-log-level silent if sensitive_request
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016661
16662The ability to distribute different levels of logs to different log servers
16663allow several production teams to interact and to fix their problems as soon
16664as possible. For example, the system team might monitor system-wide errors,
16665while the application team might be monitoring the up/down for their servers in
16666real time, and the security team might analyze the activity logs with one hour
16667delay.
16668
16669
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200166708.1. Log levels
16671---------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016672
Simon Hormandf791f52011-05-29 15:01:10 +090016673TCP and HTTP connections can be logged with information such as the date, time,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016674source IP address, destination address, connection duration, response times,
Simon Hormandf791f52011-05-29 15:01:10 +090016675HTTP request, HTTP return code, number of bytes transmitted, conditions
16676in which the session ended, and even exchanged cookies values. For example
16677track a particular user's problems. All messages may be sent to up to two
16678syslog servers. Check the "log" keyword in section 4.2 for more information
16679about log facilities.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016680
16681
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200166828.2. Log formats
16683----------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016684
William Lallemand48940402012-01-30 16:47:22 +010016685HAProxy supports 5 log formats. Several fields are common between these formats
Simon Hormandf791f52011-05-29 15:01:10 +090016686and will be detailed in the following sections. A few of them may vary
16687slightly with the configuration, due to indicators specific to certain
16688options. The supported formats are as follows :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016689
16690 - the default format, which is very basic and very rarely used. It only
16691 provides very basic information about the incoming connection at the moment
16692 it is accepted : source IP:port, destination IP:port, and frontend-name.
16693 This mode will eventually disappear so it will not be described to great
16694 extents.
16695
16696 - the TCP format, which is more advanced. This format is enabled when "option
16697 tcplog" is set on the frontend. HAProxy will then usually wait for the
16698 connection to terminate before logging. This format provides much richer
16699 information, such as timers, connection counts, queue size, etc... This
16700 format is recommended for pure TCP proxies.
16701
16702 - the HTTP format, which is the most advanced for HTTP proxying. This format
16703 is enabled when "option httplog" is set on the frontend. It provides the
16704 same information as the TCP format with some HTTP-specific fields such as
16705 the request, the status code, and captures of headers and cookies. This
16706 format is recommended for HTTP proxies.
16707
Emeric Brun3a058f32009-06-30 18:26:00 +020016708 - the CLF HTTP format, which is equivalent to the HTTP format, but with the
16709 fields arranged in the same order as the CLF format. In this mode, all
16710 timers, captures, flags, etc... appear one per field after the end of the
16711 common fields, in the same order they appear in the standard HTTP format.
16712
William Lallemand48940402012-01-30 16:47:22 +010016713 - the custom log format, allows you to make your own log line.
16714
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016715Next sections will go deeper into details for each of these formats. Format
16716specification will be performed on a "field" basis. Unless stated otherwise, a
16717field is a portion of text delimited by any number of spaces. Since syslog
16718servers are susceptible of inserting fields at the beginning of a line, it is
16719always assumed that the first field is the one containing the process name and
16720identifier.
16721
16722Note : Since log lines may be quite long, the log examples in sections below
16723 might be broken into multiple lines. The example log lines will be
16724 prefixed with 3 closing angle brackets ('>>>') and each time a log is
16725 broken into multiple lines, each non-final line will end with a
16726 backslash ('\') and the next line will start indented by two characters.
16727
16728
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200167298.2.1. Default log format
16730-------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016731
16732This format is used when no specific option is set. The log is emitted as soon
16733as the connection is accepted. One should note that this currently is the only
16734format which logs the request's destination IP and ports.
16735
16736 Example :
16737 listen www
16738 mode http
16739 log global
16740 server srv1 127.0.0.1:8000
16741
16742 >>> Feb 6 12:12:09 localhost \
16743 haproxy[14385]: Connect from 10.0.1.2:33312 to 10.0.3.31:8012 \
16744 (www/HTTP)
16745
16746 Field Format Extract from the example above
16747 1 process_name '[' pid ']:' haproxy[14385]:
16748 2 'Connect from' Connect from
16749 3 source_ip ':' source_port 10.0.1.2:33312
16750 4 'to' to
16751 5 destination_ip ':' destination_port 10.0.3.31:8012
16752 6 '(' frontend_name '/' mode ')' (www/HTTP)
16753
16754Detailed fields description :
16755 - "source_ip" is the IP address of the client which initiated the connection.
16756 - "source_port" is the TCP port of the client which initiated the connection.
16757 - "destination_ip" is the IP address the client connected to.
16758 - "destination_port" is the TCP port the client connected to.
16759 - "frontend_name" is the name of the frontend (or listener) which received
16760 and processed the connection.
16761 - "mode is the mode the frontend is operating (TCP or HTTP).
16762
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016763In case of a UNIX socket, the source and destination addresses are marked as
16764"unix:" and the ports reflect the internal ID of the socket which accepted the
16765connection (the same ID as reported in the stats).
16766
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016767It is advised not to use this deprecated format for newer installations as it
16768will eventually disappear.
16769
16770
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200167718.2.2. TCP log format
16772---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016773
16774The TCP format is used when "option tcplog" is specified in the frontend, and
16775is the recommended format for pure TCP proxies. It provides a lot of precious
16776information for troubleshooting. Since this format includes timers and byte
16777counts, the log is normally emitted at the end of the session. It can be
16778emitted earlier if "option logasap" is specified, which makes sense in most
16779environments with long sessions such as remote terminals. Sessions which match
16780the "monitor" rules are never logged. It is also possible not to emit logs for
16781sessions for which no data were exchanged between the client and the server, by
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020016782specifying "option dontlognull" in the frontend. Successful connections will
16783not be logged if "option dontlog-normal" is specified in the frontend. A few
16784fields may slightly vary depending on some configuration options, those are
16785marked with a star ('*') after the field name below.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016786
16787 Example :
16788 frontend fnt
16789 mode tcp
16790 option tcplog
16791 log global
16792 default_backend bck
16793
16794 backend bck
16795 server srv1 127.0.0.1:8000
16796
16797 >>> Feb 6 12:12:56 localhost \
16798 haproxy[14387]: 10.0.1.2:33313 [06/Feb/2009:12:12:51.443] fnt \
16799 bck/srv1 0/0/5007 212 -- 0/0/0/0/3 0/0
16800
16801 Field Format Extract from the example above
16802 1 process_name '[' pid ']:' haproxy[14387]:
16803 2 client_ip ':' client_port 10.0.1.2:33313
16804 3 '[' accept_date ']' [06/Feb/2009:12:12:51.443]
16805 4 frontend_name fnt
16806 5 backend_name '/' server_name bck/srv1
16807 6 Tw '/' Tc '/' Tt* 0/0/5007
16808 7 bytes_read* 212
16809 8 termination_state --
16810 9 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 0/0/0/0/3
16811 10 srv_queue '/' backend_queue 0/0
16812
16813Detailed fields description :
16814 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016815 connection to haproxy. If the connection was accepted on a UNIX socket
16816 instead, the IP address would be replaced with the word "unix". Note that
16817 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016818 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010016819 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016820 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016821
16822 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016823 If the connection was accepted on a UNIX socket instead, the port would be
16824 replaced with the ID of the accepting socket, which is also reported in the
16825 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016826
16827 - "accept_date" is the exact date when the connection was received by haproxy
16828 (which might be very slightly different from the date observed on the
16829 network if there was some queuing in the system's backlog). This is usually
Willy Tarreau590a0512018-09-05 11:56:48 +020016830 the same date which may appear in any upstream firewall's log. When used in
16831 HTTP mode, the accept_date field will be reset to the first moment the
16832 connection is ready to receive a new request (end of previous response for
16833 HTTP/1, immediately after previous request for HTTP/2).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016834
16835 - "frontend_name" is the name of the frontend (or listener) which received
16836 and processed the connection.
16837
16838 - "backend_name" is the name of the backend (or listener) which was selected
16839 to manage the connection to the server. This will be the same as the
16840 frontend if no switching rule has been applied, which is common for TCP
16841 applications.
16842
16843 - "server_name" is the name of the last server to which the connection was
16844 sent, which might differ from the first one if there were connection errors
16845 and a redispatch occurred. Note that this server belongs to the backend
16846 which processed the request. If the connection was aborted before reaching
16847 a server, "<NOSRV>" is indicated instead of a server name.
16848
16849 - "Tw" is the total time in milliseconds spent waiting in the various queues.
16850 It can be "-1" if the connection was aborted before reaching the queue.
16851 See "Timers" below for more details.
16852
16853 - "Tc" is the total time in milliseconds spent waiting for the connection to
16854 establish to the final server, including retries. It can be "-1" if the
16855 connection was aborted before a connection could be established. See
16856 "Timers" below for more details.
16857
16858 - "Tt" is the total time in milliseconds elapsed between the accept and the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016859 last close. It covers all possible processing. There is one exception, if
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016860 "option logasap" was specified, then the time counting stops at the moment
16861 the log is emitted. In this case, a '+' sign is prepended before the value,
16862 indicating that the final one will be larger. See "Timers" below for more
16863 details.
16864
16865 - "bytes_read" is the total number of bytes transmitted from the server to
16866 the client when the log is emitted. If "option logasap" is specified, the
16867 this value will be prefixed with a '+' sign indicating that the final one
16868 may be larger. Please note that this value is a 64-bit counter, so log
16869 analysis tools must be able to handle it without overflowing.
16870
16871 - "termination_state" is the condition the session was in when the session
16872 ended. This indicates the session state, which side caused the end of
16873 session to happen, and for what reason (timeout, error, ...). The normal
16874 flags should be "--", indicating the session was closed by either end with
16875 no data remaining in buffers. See below "Session state at disconnection"
16876 for more details.
16877
16878 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016879 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016880 limits have been reached. For instance, if actconn is close to 512 when
16881 multiple connection errors occur, chances are high that the system limits
16882 the process to use a maximum of 1024 file descriptors and that all of them
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016883 are used. See section 3 "Global parameters" to find how to tune the system.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016884
16885 - "feconn" is the total number of concurrent connections on the frontend when
16886 the session was logged. It is useful to estimate the amount of resource
16887 required to sustain high loads, and to detect when the frontend's "maxconn"
16888 has been reached. Most often when this value increases by huge jumps, it is
16889 because there is congestion on the backend servers, but sometimes it can be
16890 caused by a denial of service attack.
16891
16892 - "beconn" is the total number of concurrent connections handled by the
16893 backend when the session was logged. It includes the total number of
16894 concurrent connections active on servers as well as the number of
16895 connections pending in queues. It is useful to estimate the amount of
16896 additional servers needed to support high loads for a given application.
16897 Most often when this value increases by huge jumps, it is because there is
16898 congestion on the backend servers, but sometimes it can be caused by a
16899 denial of service attack.
16900
16901 - "srv_conn" is the total number of concurrent connections still active on
16902 the server when the session was logged. It can never exceed the server's
16903 configured "maxconn" parameter. If this value is very often close or equal
16904 to the server's "maxconn", it means that traffic regulation is involved a
16905 lot, meaning that either the server's maxconn value is too low, or that
16906 there aren't enough servers to process the load with an optimal response
16907 time. When only one of the server's "srv_conn" is high, it usually means
16908 that this server has some trouble causing the connections to take longer to
16909 be processed than on other servers.
16910
16911 - "retries" is the number of connection retries experienced by this session
16912 when trying to connect to the server. It must normally be zero, unless a
16913 server is being stopped at the same moment the connection was attempted.
16914 Frequent retries generally indicate either a network problem between
16915 haproxy and the server, or a misconfigured system backlog on the server
16916 preventing new connections from being queued. This field may optionally be
16917 prefixed with a '+' sign, indicating that the session has experienced a
16918 redispatch after the maximal retry count has been reached on the initial
16919 server. In this case, the server name appearing in the log is the one the
16920 connection was redispatched to, and not the first one, though both may
16921 sometimes be the same in case of hashing for instance. So as a general rule
16922 of thumb, when a '+' is present in front of the retry count, this count
16923 should not be attributed to the logged server.
16924
16925 - "srv_queue" is the total number of requests which were processed before
16926 this one in the server queue. It is zero when the request has not gone
16927 through the server queue. It makes it possible to estimate the approximate
16928 server's response time by dividing the time spent in queue by the number of
16929 requests in the queue. It is worth noting that if a session experiences a
16930 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010016931 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016932 backend queue unless a redispatch occurs.
16933
16934 - "backend_queue" is the total number of requests which were processed before
16935 this one in the backend's global queue. It is zero when the request has not
16936 gone through the global queue. It makes it possible to estimate the average
16937 queue length, which easily translates into a number of missing servers when
16938 divided by a server's "maxconn" parameter. It is worth noting that if a
16939 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010016940 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016941 through both the server queue and the backend queue unless a redispatch
16942 occurs.
16943
16944
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200169458.2.3. HTTP log format
16946----------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016947
16948The HTTP format is the most complete and the best suited for HTTP proxies. It
16949is enabled by when "option httplog" is specified in the frontend. It provides
16950the same level of information as the TCP format with additional features which
16951are specific to the HTTP protocol. Just like the TCP format, the log is usually
16952emitted at the end of the session, unless "option logasap" is specified, which
16953generally only makes sense for download sites. A session which matches the
16954"monitor" rules will never logged. It is also possible not to log sessions for
16955which no data were sent by the client by specifying "option dontlognull" in the
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020016956frontend. Successful connections will not be logged if "option dontlog-normal"
16957is specified in the frontend.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016958
16959Most fields are shared with the TCP log, some being different. A few fields may
16960slightly vary depending on some configuration options. Those ones are marked
16961with a star ('*') after the field name below.
16962
16963 Example :
16964 frontend http-in
16965 mode http
16966 option httplog
16967 log global
16968 default_backend bck
16969
16970 backend static
16971 server srv1 127.0.0.1:8000
16972
16973 >>> Feb 6 12:14:14 localhost \
16974 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
16975 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 +010016976 {} "GET /index.html HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016977
16978 Field Format Extract from the example above
16979 1 process_name '[' pid ']:' haproxy[14389]:
16980 2 client_ip ':' client_port 10.0.1.2:33317
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016981 3 '[' request_date ']' [06/Feb/2009:12:14:14.655]
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016982 4 frontend_name http-in
16983 5 backend_name '/' server_name static/srv1
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016984 6 TR '/' Tw '/' Tc '/' Tr '/' Ta* 10/0/30/69/109
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016985 7 status_code 200
16986 8 bytes_read* 2750
16987 9 captured_request_cookie -
16988 10 captured_response_cookie -
16989 11 termination_state ----
16990 12 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 1/1/1/1/0
16991 13 srv_queue '/' backend_queue 0/0
16992 14 '{' captured_request_headers* '}' {haproxy.1wt.eu}
16993 15 '{' captured_response_headers* '}' {}
16994 16 '"' http_request '"' "GET /index.html HTTP/1.1"
Willy Tarreaud72758d2010-01-12 10:42:19 +010016995
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016996Detailed fields description :
16997 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016998 connection to haproxy. If the connection was accepted on a UNIX socket
16999 instead, the IP address would be replaced with the word "unix". Note that
17000 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010017001 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010017002 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010017003 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017004
17005 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010017006 If the connection was accepted on a UNIX socket instead, the port would be
17007 replaced with the ID of the accepting socket, which is also reported in the
17008 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017009
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017010 - "request_date" is the exact date when the first byte of the HTTP request
17011 was received by haproxy (log field %tr).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017012
17013 - "frontend_name" is the name of the frontend (or listener) which received
17014 and processed the connection.
17015
17016 - "backend_name" is the name of the backend (or listener) which was selected
17017 to manage the connection to the server. This will be the same as the
17018 frontend if no switching rule has been applied.
17019
17020 - "server_name" is the name of the last server to which the connection was
17021 sent, which might differ from the first one if there were connection errors
17022 and a redispatch occurred. Note that this server belongs to the backend
17023 which processed the request. If the request was aborted before reaching a
17024 server, "<NOSRV>" is indicated instead of a server name. If the request was
17025 intercepted by the stats subsystem, "<STATS>" is indicated instead.
17026
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017027 - "TR" is the total time in milliseconds spent waiting for a full HTTP
17028 request from the client (not counting body) after the first byte was
17029 received. It can be "-1" if the connection was aborted before a complete
John Roeslerfb2fce12019-07-10 15:45:51 -050017030 request could be received or a bad request was received. It should
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017031 always be very small because a request generally fits in one single packet.
17032 Large times here generally indicate network issues between the client and
Willy Tarreau590a0512018-09-05 11:56:48 +020017033 haproxy or requests being typed by hand. See section 8.4 "Timing Events"
17034 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017035
17036 - "Tw" is the total time in milliseconds spent waiting in the various queues.
17037 It can be "-1" if the connection was aborted before reaching the queue.
Willy Tarreau590a0512018-09-05 11:56:48 +020017038 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017039
17040 - "Tc" is the total time in milliseconds spent waiting for the connection to
17041 establish to the final server, including retries. It can be "-1" if the
Willy Tarreau590a0512018-09-05 11:56:48 +020017042 request was aborted before a connection could be established. See section
17043 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017044
17045 - "Tr" is the total time in milliseconds spent waiting for the server to send
17046 a full HTTP response, not counting data. It can be "-1" if the request was
17047 aborted before a complete response could be received. It generally matches
17048 the server's processing time for the request, though it may be altered by
17049 the amount of data sent by the client to the server. Large times here on
Willy Tarreau590a0512018-09-05 11:56:48 +020017050 "GET" requests generally indicate an overloaded server. See section 8.4
17051 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017052
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017053 - "Ta" is the time the request remained active in haproxy, which is the total
17054 time in milliseconds elapsed between the first byte of the request was
17055 received and the last byte of response was sent. It covers all possible
17056 processing except the handshake (see Th) and idle time (see Ti). There is
17057 one exception, if "option logasap" was specified, then the time counting
17058 stops at the moment the log is emitted. In this case, a '+' sign is
17059 prepended before the value, indicating that the final one will be larger.
Willy Tarreau590a0512018-09-05 11:56:48 +020017060 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017061
17062 - "status_code" is the HTTP status code returned to the client. This status
17063 is generally set by the server, but it might also be set by haproxy when
17064 the server cannot be reached or when its response is blocked by haproxy.
17065
17066 - "bytes_read" is the total number of bytes transmitted to the client when
17067 the log is emitted. This does include HTTP headers. If "option logasap" is
John Roeslerfb2fce12019-07-10 15:45:51 -050017068 specified, this value will be prefixed with a '+' sign indicating that
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017069 the final one may be larger. Please note that this value is a 64-bit
17070 counter, so log analysis tools must be able to handle it without
17071 overflowing.
17072
17073 - "captured_request_cookie" is an optional "name=value" entry indicating that
17074 the client had this cookie in the request. The cookie name and its maximum
17075 length are defined by the "capture cookie" statement in the frontend
17076 configuration. The field is a single dash ('-') when the option is not
17077 set. Only one cookie may be captured, it is generally used to track session
17078 ID exchanges between a client and a server to detect session crossing
17079 between clients due to application bugs. For more details, please consult
17080 the section "Capturing HTTP headers and cookies" below.
17081
17082 - "captured_response_cookie" is an optional "name=value" entry indicating
17083 that the server has returned a cookie with its response. The cookie name
17084 and its maximum length are defined by the "capture cookie" statement in the
17085 frontend configuration. The field is a single dash ('-') when the option is
17086 not set. Only one cookie may be captured, it is generally used to track
17087 session ID exchanges between a client and a server to detect session
17088 crossing between clients due to application bugs. For more details, please
17089 consult the section "Capturing HTTP headers and cookies" below.
17090
17091 - "termination_state" is the condition the session was in when the session
17092 ended. This indicates the session state, which side caused the end of
17093 session to happen, for what reason (timeout, error, ...), just like in TCP
17094 logs, and information about persistence operations on cookies in the last
17095 two characters. The normal flags should begin with "--", indicating the
17096 session was closed by either end with no data remaining in buffers. See
17097 below "Session state at disconnection" for more details.
17098
17099 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017100 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017101 limits have been reached. For instance, if actconn is close to 512 or 1024
17102 when multiple connection errors occur, chances are high that the system
17103 limits the process to use a maximum of 1024 file descriptors and that all
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017104 of them are used. See section 3 "Global parameters" to find how to tune the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017105 system.
17106
17107 - "feconn" is the total number of concurrent connections on the frontend when
17108 the session was logged. It is useful to estimate the amount of resource
17109 required to sustain high loads, and to detect when the frontend's "maxconn"
17110 has been reached. Most often when this value increases by huge jumps, it is
17111 because there is congestion on the backend servers, but sometimes it can be
17112 caused by a denial of service attack.
17113
17114 - "beconn" is the total number of concurrent connections handled by the
17115 backend when the session was logged. It includes the total number of
17116 concurrent connections active on servers as well as the number of
17117 connections pending in queues. It is useful to estimate the amount of
17118 additional servers needed to support high loads for a given application.
17119 Most often when this value increases by huge jumps, it is because there is
17120 congestion on the backend servers, but sometimes it can be caused by a
17121 denial of service attack.
17122
17123 - "srv_conn" is the total number of concurrent connections still active on
17124 the server when the session was logged. It can never exceed the server's
17125 configured "maxconn" parameter. If this value is very often close or equal
17126 to the server's "maxconn", it means that traffic regulation is involved a
17127 lot, meaning that either the server's maxconn value is too low, or that
17128 there aren't enough servers to process the load with an optimal response
17129 time. When only one of the server's "srv_conn" is high, it usually means
17130 that this server has some trouble causing the requests to take longer to be
17131 processed than on other servers.
17132
17133 - "retries" is the number of connection retries experienced by this session
17134 when trying to connect to the server. It must normally be zero, unless a
17135 server is being stopped at the same moment the connection was attempted.
17136 Frequent retries generally indicate either a network problem between
17137 haproxy and the server, or a misconfigured system backlog on the server
17138 preventing new connections from being queued. This field may optionally be
17139 prefixed with a '+' sign, indicating that the session has experienced a
17140 redispatch after the maximal retry count has been reached on the initial
17141 server. In this case, the server name appearing in the log is the one the
17142 connection was redispatched to, and not the first one, though both may
17143 sometimes be the same in case of hashing for instance. So as a general rule
17144 of thumb, when a '+' is present in front of the retry count, this count
17145 should not be attributed to the logged server.
17146
17147 - "srv_queue" is the total number of requests which were processed before
17148 this one in the server queue. It is zero when the request has not gone
17149 through the server queue. It makes it possible to estimate the approximate
17150 server's response time by dividing the time spent in queue by the number of
17151 requests in the queue. It is worth noting that if a session experiences a
17152 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010017153 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017154 backend queue unless a redispatch occurs.
17155
17156 - "backend_queue" is the total number of requests which were processed before
17157 this one in the backend's global queue. It is zero when the request has not
17158 gone through the global queue. It makes it possible to estimate the average
17159 queue length, which easily translates into a number of missing servers when
17160 divided by a server's "maxconn" parameter. It is worth noting that if a
17161 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010017162 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017163 through both the server queue and the backend queue unless a redispatch
17164 occurs.
17165
17166 - "captured_request_headers" is a list of headers captured in the request due
17167 to the presence of the "capture request header" statement in the frontend.
17168 Multiple headers can be captured, they will be delimited by a vertical bar
17169 ('|'). When no capture is enabled, the braces do not appear, causing a
17170 shift of remaining fields. It is important to note that this field may
17171 contain spaces, and that using it requires a smarter log parser than when
17172 it's not used. Please consult the section "Capturing HTTP headers and
17173 cookies" below for more details.
17174
17175 - "captured_response_headers" is a list of headers captured in the response
17176 due to the presence of the "capture response header" statement in the
17177 frontend. Multiple headers can be captured, they will be delimited by a
17178 vertical bar ('|'). When no capture is enabled, the braces do not appear,
17179 causing a shift of remaining fields. It is important to note that this
17180 field may contain spaces, and that using it requires a smarter log parser
17181 than when it's not used. Please consult the section "Capturing HTTP headers
17182 and cookies" below for more details.
17183
17184 - "http_request" is the complete HTTP request line, including the method,
17185 request and HTTP version string. Non-printable characters are encoded (see
17186 below the section "Non-printable characters"). This is always the last
17187 field, and it is always delimited by quotes and is the only one which can
17188 contain quotes. If new fields are added to the log format, they will be
17189 added before this field. This field might be truncated if the request is
17190 huge and does not fit in the standard syslog buffer (1024 characters). This
17191 is the reason why this field must always remain the last one.
17192
17193
Cyril Bontédc4d9032012-04-08 21:57:39 +0200171948.2.4. Custom log format
17195------------------------
William Lallemand48940402012-01-30 16:47:22 +010017196
Willy Tarreau2beef582012-12-20 17:22:52 +010017197The directive log-format allows you to customize the logs in http mode and tcp
William Lallemandbddd4fd2012-02-27 11:23:10 +010017198mode. It takes a string as argument.
William Lallemand48940402012-01-30 16:47:22 +010017199
Davor Ocelice9ed2812017-12-25 17:49:28 +010017200HAProxy understands some log format variables. % precedes log format variables.
William Lallemand48940402012-01-30 16:47:22 +010017201Variables can take arguments using braces ('{}'), and multiple arguments are
17202separated by commas within the braces. Flags may be added or removed by
17203prefixing them with a '+' or '-' sign.
17204
17205Special variable "%o" may be used to propagate its flags to all other
17206variables on the same format string. This is particularly handy with quoted
Dragan Dosen835b9212016-02-12 13:23:03 +010017207("Q") and escaped ("E") string formats.
William Lallemand48940402012-01-30 16:47:22 +010017208
Willy Tarreauc8368452012-12-21 00:09:23 +010017209If a variable is named between square brackets ('[' .. ']') then it is used
Willy Tarreaube722a22014-06-13 16:31:59 +020017210as a sample expression rule (see section 7.3). This it useful to add some
Willy Tarreauc8368452012-12-21 00:09:23 +010017211less common information such as the client's SSL certificate's DN, or to log
17212the key that would be used to store an entry into a stick table.
17213
William Lallemand48940402012-01-30 16:47:22 +010017214Note: spaces must be escaped. A space character is considered as a separator.
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017215In order to emit a verbatim '%', it must be preceded by another '%' resulting
Willy Tarreau06d97f92013-12-02 17:45:48 +010017216in '%%'. HAProxy will automatically merge consecutive separators.
William Lallemand48940402012-01-30 16:47:22 +010017217
Dragan Dosen835b9212016-02-12 13:23:03 +010017218Note: when using the RFC5424 syslog message format, the characters '"',
17219'\' and ']' inside PARAM-VALUE should be escaped with '\' as prefix (see
17220https://tools.ietf.org/html/rfc5424#section-6.3.3 for more details). In
17221such cases, the use of the flag "E" should be considered.
17222
William Lallemand48940402012-01-30 16:47:22 +010017223Flags are :
17224 * Q: quote a string
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017225 * X: hexadecimal representation (IPs, Ports, %Ts, %rt, %pid)
Dragan Dosen835b9212016-02-12 13:23:03 +010017226 * E: escape characters '"', '\' and ']' in a string with '\' as prefix
17227 (intended purpose is for the RFC5424 structured-data log formats)
William Lallemand48940402012-01-30 16:47:22 +010017228
17229 Example:
17230
17231 log-format %T\ %t\ Some\ Text
17232 log-format %{+Q}o\ %t\ %s\ %{-Q}r
17233
Dragan Dosen835b9212016-02-12 13:23:03 +010017234 log-format-sd %{+Q,+E}o\ [exampleSDID@1234\ header=%[capture.req.hdr(0)]]
17235
William Lallemand48940402012-01-30 16:47:22 +010017236At the moment, the default HTTP format is defined this way :
17237
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017238 log-format "%ci:%cp [%tr] %ft %b/%s %TR/%Tw/%Tc/%Tr/%Ta %ST %B %CC \
17239 %CS %tsc %ac/%fc/%bc/%sc/%rc %sq/%bq %hr %hs %{+Q}r"
William Lallemand48940402012-01-30 16:47:22 +010017240
William Lallemandbddd4fd2012-02-27 11:23:10 +010017241the default CLF format is defined this way :
William Lallemand48940402012-01-30 16:47:22 +010017242
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017243 log-format "%{+Q}o %{-Q}ci - - [%trg] %r %ST %B \"\" \"\" %cp \
17244 %ms %ft %b %s %TR %Tw %Tc %Tr %Ta %tsc %ac %fc \
17245 %bc %sc %rc %sq %bq %CC %CS %hrl %hsl"
William Lallemand48940402012-01-30 16:47:22 +010017246
William Lallemandbddd4fd2012-02-27 11:23:10 +010017247and the default TCP format is defined this way :
17248
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017249 log-format "%ci:%cp [%t] %ft %b/%s %Tw/%Tc/%Tt %B %ts \
17250 %ac/%fc/%bc/%sc/%rc %sq/%bq"
William Lallemandbddd4fd2012-02-27 11:23:10 +010017251
William Lallemand48940402012-01-30 16:47:22 +010017252Please refer to the table below for currently defined variables :
17253
William Lallemandbddd4fd2012-02-27 11:23:10 +010017254 +---+------+-----------------------------------------------+-------------+
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020017255 | R | var | field name (8.2.2 and 8.2.3 for description) | type |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017256 +---+------+-----------------------------------------------+-------------+
17257 | | %o | special variable, apply flags on all next var | |
17258 +---+------+-----------------------------------------------+-------------+
Willy Tarreau2beef582012-12-20 17:22:52 +010017259 | | %B | bytes_read (from server to client) | numeric |
17260 | H | %CC | captured_request_cookie | string |
17261 | H | %CS | captured_response_cookie | string |
William Lallemand5f232402012-04-05 18:02:55 +020017262 | | %H | hostname | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000017263 | H | %HM | HTTP method (ex: POST) | string |
17264 | H | %HP | HTTP request URI without query string (path) | string |
Andrew Hayworthe63ac872015-07-31 16:14:16 +000017265 | H | %HQ | HTTP request URI query string (ex: ?bar=baz) | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000017266 | H | %HU | HTTP request URI (ex: /foo?bar=baz) | string |
17267 | H | %HV | HTTP version (ex: HTTP/1.0) | string |
William Lallemanda73203e2012-03-12 12:48:57 +010017268 | | %ID | unique-id | string |
Willy Tarreau4bf99632014-06-13 12:21:40 +020017269 | | %ST | status_code | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020017270 | | %T | gmt_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017271 | | %Ta | Active time of the request (from TR to end) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017272 | | %Tc | Tc | numeric |
Willy Tarreau27b639d2016-05-17 17:55:27 +020017273 | | %Td | Td = Tt - (Tq + Tw + Tc + Tr) | numeric |
Yuxans Yao4e25b012012-10-19 10:36:09 +080017274 | | %Tl | local_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017275 | | %Th | connection handshake time (SSL, PROXY proto) | numeric |
17276 | H | %Ti | idle time before the HTTP request | numeric |
17277 | H | %Tq | Th + Ti + TR | numeric |
17278 | H | %TR | time to receive the full request from 1st byte| numeric |
17279 | H | %Tr | Tr (response time) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020017280 | | %Ts | timestamp | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017281 | | %Tt | Tt | numeric |
17282 | | %Tw | Tw | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010017283 | | %U | bytes_uploaded (from client to server) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017284 | | %ac | actconn | numeric |
17285 | | %b | backend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010017286 | | %bc | beconn (backend concurrent connections) | numeric |
17287 | | %bi | backend_source_ip (connecting address) | IP |
17288 | | %bp | backend_source_port (connecting address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017289 | | %bq | backend_queue | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010017290 | | %ci | client_ip (accepted address) | IP |
17291 | | %cp | client_port (accepted address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017292 | | %f | frontend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010017293 | | %fc | feconn (frontend concurrent connections) | numeric |
17294 | | %fi | frontend_ip (accepting address) | IP |
17295 | | %fp | frontend_port (accepting address) | numeric |
Willy Tarreau773d65f2012-10-12 14:56:11 +020017296 | | %ft | frontend_name_transport ('~' suffix for SSL) | string |
Willy Tarreau7346acb2014-08-28 15:03:15 +020017297 | | %lc | frontend_log_counter | numeric |
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020017298 | | %hr | captured_request_headers default style | string |
17299 | | %hrl | captured_request_headers CLF style | string list |
17300 | | %hs | captured_response_headers default style | string |
17301 | | %hsl | captured_response_headers CLF style | string list |
Willy Tarreau812c88e2015-08-09 10:56:35 +020017302 | | %ms | accept date milliseconds (left-padded with 0) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020017303 | | %pid | PID | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020017304 | H | %r | http_request | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017305 | | %rc | retries | numeric |
Willy Tarreau1f0da242014-01-25 11:01:50 +010017306 | | %rt | request_counter (HTTP req or TCP session) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017307 | | %s | server_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010017308 | | %sc | srv_conn (server concurrent connections) | numeric |
17309 | | %si | server_IP (target address) | IP |
17310 | | %sp | server_port (target address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017311 | | %sq | srv_queue | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020017312 | S | %sslc| ssl_ciphers (ex: AES-SHA) | string |
17313 | S | %sslv| ssl_version (ex: TLSv1) | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010017314 | | %t | date_time (with millisecond resolution) | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017315 | H | %tr | date_time of HTTP request | date |
17316 | H | %trg | gmt_date_time of start of HTTP request | date |
Jens Bissinger15c64ff2018-08-23 14:11:27 +020017317 | H | %trl | local_date_time of start of HTTP request | date |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017318 | | %ts | termination_state | string |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020017319 | H | %tsc | termination_state with cookie status | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017320 +---+------+-----------------------------------------------+-------------+
William Lallemand48940402012-01-30 16:47:22 +010017321
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020017322 R = Restrictions : H = mode http only ; S = SSL only
William Lallemand48940402012-01-30 16:47:22 +010017323
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010017324
173258.2.5. Error log format
17326-----------------------
17327
17328When an incoming connection fails due to an SSL handshake or an invalid PROXY
17329protocol header, haproxy will log the event using a shorter, fixed line format.
17330By default, logs are emitted at the LOG_INFO level, unless the option
17331"log-separate-errors" is set in the backend, in which case the LOG_ERR level
Davor Ocelice9ed2812017-12-25 17:49:28 +010017332will be used. Connections on which no data are exchanged (e.g. probes) are not
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010017333logged if the "dontlognull" option is set.
17334
17335The format looks like this :
17336
17337 >>> Dec 3 18:27:14 localhost \
17338 haproxy[6103]: 127.0.0.1:56059 [03/Dec/2012:17:35:10.380] frt/f1: \
17339 Connection error during SSL handshake
17340
17341 Field Format Extract from the example above
17342 1 process_name '[' pid ']:' haproxy[6103]:
17343 2 client_ip ':' client_port 127.0.0.1:56059
17344 3 '[' accept_date ']' [03/Dec/2012:17:35:10.380]
17345 4 frontend_name "/" bind_name ":" frt/f1:
17346 5 message Connection error during SSL handshake
17347
17348These fields just provide minimal information to help debugging connection
17349failures.
17350
17351
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200173528.3. Advanced logging options
17353-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017354
17355Some advanced logging options are often looked for but are not easy to find out
17356just by looking at the various options. Here is an entry point for the few
17357options which can enable better logging. Please refer to the keywords reference
17358for more information about their usage.
17359
17360
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200173618.3.1. Disabling logging of external tests
17362------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017363
17364It is quite common to have some monitoring tools perform health checks on
17365haproxy. Sometimes it will be a layer 3 load-balancer such as LVS or any
17366commercial load-balancer, and sometimes it will simply be a more complete
17367monitoring system such as Nagios. When the tests are very frequent, users often
17368ask how to disable logging for those checks. There are three possibilities :
17369
17370 - if connections come from everywhere and are just TCP probes, it is often
17371 desired to simply disable logging of connections without data exchange, by
17372 setting "option dontlognull" in the frontend. It also disables logging of
17373 port scans, which may or may not be desired.
17374
17375 - if the connection come from a known source network, use "monitor-net" to
17376 declare this network as monitoring only. Any host in this network will then
17377 only be able to perform health checks, and their requests will not be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017378 logged. This is generally appropriate to designate a list of equipment
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017379 such as other load-balancers.
17380
17381 - if the tests are performed on a known URI, use "monitor-uri" to declare
17382 this URI as dedicated to monitoring. Any host sending this request will
17383 only get the result of a health-check, and the request will not be logged.
17384
17385
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200173868.3.2. Logging before waiting for the session to terminate
17387----------------------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017388
17389The problem with logging at end of connection is that you have no clue about
17390what is happening during very long sessions, such as remote terminal sessions
17391or large file downloads. This problem can be worked around by specifying
Davor Ocelice9ed2812017-12-25 17:49:28 +010017392"option logasap" in the frontend. HAProxy will then log as soon as possible,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017393just before data transfer begins. This means that in case of TCP, it will still
17394log the connection status to the server, and in case of HTTP, it will log just
17395after processing the server headers. In this case, the number of bytes reported
17396is the number of header bytes sent to the client. In order to avoid confusion
17397with normal logs, the total time field and the number of bytes are prefixed
17398with a '+' sign which means that real numbers are certainly larger.
17399
17400
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200174018.3.3. Raising log level upon errors
17402------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020017403
17404Sometimes it is more convenient to separate normal traffic from errors logs,
17405for instance in order to ease error monitoring from log files. When the option
17406"log-separate-errors" is used, connections which experience errors, timeouts,
17407retries, redispatches or HTTP status codes 5xx will see their syslog level
17408raised from "info" to "err". This will help a syslog daemon store the log in
17409a separate file. It is very important to keep the errors in the normal traffic
17410file too, so that log ordering is not altered. You should also be careful if
17411you already have configured your syslog daemon to store all logs higher than
17412"notice" in an "admin" file, because the "err" level is higher than "notice".
17413
17414
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200174158.3.4. Disabling logging of successful connections
17416--------------------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020017417
17418Although this may sound strange at first, some large sites have to deal with
17419multiple thousands of logs per second and are experiencing difficulties keeping
17420them intact for a long time or detecting errors within them. If the option
17421"dontlog-normal" is set on the frontend, all normal connections will not be
17422logged. In this regard, a normal connection is defined as one without any
17423error, timeout, retry nor redispatch. In HTTP, the status code is checked too,
17424and a response with a status 5xx is not considered normal and will be logged
17425too. Of course, doing is is really discouraged as it will remove most of the
17426useful information from the logs. Do this only if you have no other
17427alternative.
17428
17429
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200174308.4. Timing events
17431------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017432
17433Timers provide a great help in troubleshooting network problems. All values are
17434reported in milliseconds (ms). These timers should be used in conjunction with
17435the session termination flags. In TCP mode with "option tcplog" set on the
17436frontend, 3 control points are reported under the form "Tw/Tc/Tt", and in HTTP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017437mode, 5 control points are reported under the form "TR/Tw/Tc/Tr/Ta". In
17438addition, three other measures are provided, "Th", "Ti", and "Tq".
17439
Guillaume de Lafondf27cddc2016-12-23 17:32:43 +010017440Timings events in HTTP mode:
17441
17442 first request 2nd request
17443 |<-------------------------------->|<-------------- ...
17444 t tr t tr ...
17445 ---|----|----|----|----|----|----|----|----|--
17446 : Th Ti TR Tw Tc Tr Td : Ti ...
17447 :<---- Tq ---->: :
17448 :<-------------- Tt -------------->:
17449 :<--------- Ta --------->:
17450
17451Timings events in TCP mode:
17452
17453 TCP session
17454 |<----------------->|
17455 t t
17456 ---|----|----|----|----|---
17457 | Th Tw Tc Td |
17458 |<------ Tt ------->|
17459
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017460 - Th: total time to accept tcp connection and execute handshakes for low level
Davor Ocelice9ed2812017-12-25 17:49:28 +010017461 protocols. Currently, these protocols are proxy-protocol and SSL. This may
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017462 only happen once during the whole connection's lifetime. A large time here
17463 may indicate that the client only pre-established the connection without
17464 speaking, that it is experiencing network issues preventing it from
Davor Ocelice9ed2812017-12-25 17:49:28 +010017465 completing a handshake in a reasonable time (e.g. MTU issues), or that an
Willy Tarreau590a0512018-09-05 11:56:48 +020017466 SSL handshake was very expensive to compute. Please note that this time is
17467 reported only before the first request, so it is safe to average it over
17468 all request to calculate the amortized value. The second and subsequent
17469 request will always report zero here.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017470
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017471 - Ti: is the idle time before the HTTP request (HTTP mode only). This timer
17472 counts between the end of the handshakes and the first byte of the HTTP
17473 request. When dealing with a second request in keep-alive mode, it starts
Willy Tarreau590a0512018-09-05 11:56:48 +020017474 to count after the end of the transmission the previous response. When a
17475 multiplexed protocol such as HTTP/2 is used, it starts to count immediately
17476 after the previous request. Some browsers pre-establish connections to a
17477 server in order to reduce the latency of a future request, and keep them
17478 pending until they need it. This delay will be reported as the idle time. A
17479 value of -1 indicates that nothing was received on the connection.
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017480
17481 - TR: total time to get the client request (HTTP mode only). It's the time
17482 elapsed between the first bytes received and the moment the proxy received
17483 the empty line marking the end of the HTTP headers. The value "-1"
17484 indicates that the end of headers has never been seen. This happens when
17485 the client closes prematurely or times out. This time is usually very short
17486 since most requests fit in a single packet. A large time may indicate a
17487 request typed by hand during a test.
17488
17489 - Tq: total time to get the client request from the accept date or since the
17490 emission of the last byte of the previous response (HTTP mode only). It's
Davor Ocelice9ed2812017-12-25 17:49:28 +010017491 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 +020017492 returns -1 as well. This timer used to be very useful before the arrival of
17493 HTTP keep-alive and browsers' pre-connect feature. It's recommended to drop
17494 it in favor of TR nowadays, as the idle time adds a lot of noise to the
17495 reports.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017496
17497 - Tw: total time spent in the queues waiting for a connection slot. It
17498 accounts for backend queue as well as the server queues, and depends on the
17499 queue size, and the time needed for the server to complete previous
17500 requests. The value "-1" means that the request was killed before reaching
17501 the queue, which is generally what happens with invalid or denied requests.
17502
17503 - Tc: total time to establish the TCP connection to the server. It's the time
17504 elapsed between the moment the proxy sent the connection request, and the
17505 moment it was acknowledged by the server, or between the TCP SYN packet and
17506 the matching SYN/ACK packet in return. The value "-1" means that the
17507 connection never established.
17508
17509 - Tr: server response time (HTTP mode only). It's the time elapsed between
17510 the moment the TCP connection was established to the server and the moment
17511 the server sent its complete response headers. It purely shows its request
17512 processing time, without the network overhead due to the data transmission.
17513 It is worth noting that when the client has data to send to the server, for
17514 instance during a POST request, the time already runs, and this can distort
17515 apparent response time. For this reason, it's generally wise not to trust
17516 too much this field for POST requests initiated from clients behind an
17517 untrusted network. A value of "-1" here means that the last the response
17518 header (empty line) was never seen, most likely because the server timeout
17519 stroke before the server managed to process the request.
17520
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017521 - Ta: total active time for the HTTP request, between the moment the proxy
17522 received the first byte of the request header and the emission of the last
17523 byte of the response body. The exception is when the "logasap" option is
17524 specified. In this case, it only equals (TR+Tw+Tc+Tr), and is prefixed with
17525 a '+' sign. From this field, we can deduce "Td", the data transmission time,
17526 by subtracting other timers when valid :
17527
17528 Td = Ta - (TR + Tw + Tc + Tr)
17529
17530 Timers with "-1" values have to be excluded from this equation. Note that
17531 "Ta" can never be negative.
17532
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017533 - Tt: total session duration time, between the moment the proxy accepted it
17534 and the moment both ends were closed. The exception is when the "logasap"
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017535 option is specified. In this case, it only equals (Th+Ti+TR+Tw+Tc+Tr), and
17536 is prefixed with a '+' sign. From this field, we can deduce "Td", the data
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017537 transmission time, by subtracting other timers when valid :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017538
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017539 Td = Tt - (Th + Ti + TR + Tw + Tc + Tr)
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017540
17541 Timers with "-1" values have to be excluded from this equation. In TCP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017542 mode, "Ti", "Tq" and "Tr" have to be excluded too. Note that "Tt" can never
17543 be negative and that for HTTP, Tt is simply equal to (Th+Ti+Ta).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017544
17545These timers provide precious indications on trouble causes. Since the TCP
17546protocol defines retransmit delays of 3, 6, 12... seconds, we know for sure
17547that timers close to multiples of 3s are nearly always related to lost packets
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017548due to network problems (wires, negotiation, congestion). Moreover, if "Ta" or
17549"Tt" is close to a timeout value specified in the configuration, it often means
17550that a session has been aborted on timeout.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017551
17552Most common cases :
17553
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017554 - If "Th" or "Ti" are close to 3000, a packet has probably been lost between
17555 the client and the proxy. This is very rare on local networks but might
17556 happen when clients are on far remote networks and send large requests. It
17557 may happen that values larger than usual appear here without any network
17558 cause. Sometimes, during an attack or just after a resource starvation has
17559 ended, haproxy may accept thousands of connections in a few milliseconds.
17560 The time spent accepting these connections will inevitably slightly delay
17561 processing of other connections, and it can happen that request times in the
17562 order of a few tens of milliseconds are measured after a few thousands of
17563 new connections have been accepted at once. Using one of the keep-alive
17564 modes may display larger idle times since "Ti" measures the time spent
Patrick Mezard105faca2010-06-12 17:02:46 +020017565 waiting for additional requests.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017566
17567 - If "Tc" is close to 3000, a packet has probably been lost between the
17568 server and the proxy during the server connection phase. This value should
17569 always be very low, such as 1 ms on local networks and less than a few tens
17570 of ms on remote networks.
17571
Willy Tarreau55165fe2009-05-10 12:02:55 +020017572 - If "Tr" is nearly always lower than 3000 except some rare values which seem
17573 to be the average majored by 3000, there are probably some packets lost
17574 between the proxy and the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017575
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017576 - If "Ta" is large even for small byte counts, it generally is because
17577 neither the client nor the server decides to close the connection while
17578 haproxy is running in tunnel mode and both have agreed on a keep-alive
17579 connection mode. In order to solve this issue, it will be needed to specify
17580 one of the HTTP options to manipulate keep-alive or close options on either
17581 the frontend or the backend. Having the smallest possible 'Ta' or 'Tt' is
17582 important when connection regulation is used with the "maxconn" option on
17583 the servers, since no new connection will be sent to the server until
17584 another one is released.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017585
17586Other noticeable HTTP log cases ('xx' means any value to be ignored) :
17587
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017588 TR/Tw/Tc/Tr/+Ta The "option logasap" is present on the frontend and the log
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017589 was emitted before the data phase. All the timers are valid
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017590 except "Ta" which is shorter than reality.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017591
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017592 -1/xx/xx/xx/Ta The client was not able to send a complete request in time
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017593 or it aborted too early. Check the session termination flags
17594 then "timeout http-request" and "timeout client" settings.
17595
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017596 TR/-1/xx/xx/Ta It was not possible to process the request, maybe because
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017597 servers were out of order, because the request was invalid
17598 or forbidden by ACL rules. Check the session termination
17599 flags.
17600
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017601 TR/Tw/-1/xx/Ta The connection could not establish on the server. Either it
17602 actively refused it or it timed out after Ta-(TR+Tw) ms.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017603 Check the session termination flags, then check the
17604 "timeout connect" setting. Note that the tarpit action might
17605 return similar-looking patterns, with "Tw" equal to the time
17606 the client connection was maintained open.
17607
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017608 TR/Tw/Tc/-1/Ta The server has accepted the connection but did not return
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017609 a complete response in time, or it closed its connection
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017610 unexpectedly after Ta-(TR+Tw+Tc) ms. Check the session
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017611 termination flags, then check the "timeout server" setting.
17612
17613
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200176148.5. Session state at disconnection
17615-----------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017616
17617TCP and HTTP logs provide a session termination indicator in the
17618"termination_state" field, just before the number of active connections. It is
176192-characters long in TCP mode, and is extended to 4 characters in HTTP mode,
17620each of which has a special meaning :
17621
17622 - On the first character, a code reporting the first event which caused the
17623 session to terminate :
17624
17625 C : the TCP session was unexpectedly aborted by the client.
17626
17627 S : the TCP session was unexpectedly aborted by the server, or the
17628 server explicitly refused it.
17629
17630 P : the session was prematurely aborted by the proxy, because of a
17631 connection limit enforcement, because a DENY filter was matched,
17632 because of a security check which detected and blocked a dangerous
17633 error in server response which might have caused information leak
Davor Ocelice9ed2812017-12-25 17:49:28 +010017634 (e.g. cacheable cookie).
Willy Tarreau570f2212013-06-10 16:42:09 +020017635
17636 L : the session was locally processed by haproxy and was not passed to
17637 a server. This is what happens for stats and redirects.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017638
17639 R : a resource on the proxy has been exhausted (memory, sockets, source
17640 ports, ...). Usually, this appears during the connection phase, and
17641 system logs should contain a copy of the precise error. If this
17642 happens, it must be considered as a very serious anomaly which
17643 should be fixed as soon as possible by any means.
17644
17645 I : an internal error was identified by the proxy during a self-check.
17646 This should NEVER happen, and you are encouraged to report any log
17647 containing this, because this would almost certainly be a bug. It
17648 would be wise to preventively restart the process after such an
17649 event too, in case it would be caused by memory corruption.
17650
Simon Horman752dc4a2011-06-21 14:34:59 +090017651 D : the session was killed by haproxy because the server was detected
17652 as down and was configured to kill all connections when going down.
17653
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070017654 U : the session was killed by haproxy on this backup server because an
17655 active server was detected as up and was configured to kill all
17656 backup connections when going up.
17657
Willy Tarreaua2a64e92011-09-07 23:01:56 +020017658 K : the session was actively killed by an admin operating on haproxy.
17659
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017660 c : the client-side timeout expired while waiting for the client to
17661 send or receive data.
17662
17663 s : the server-side timeout expired while waiting for the server to
17664 send or receive data.
17665
17666 - : normal session completion, both the client and the server closed
17667 with nothing left in the buffers.
17668
17669 - on the second character, the TCP or HTTP session state when it was closed :
17670
Willy Tarreauf7b30a92010-12-06 22:59:17 +010017671 R : the proxy was waiting for a complete, valid REQUEST from the client
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017672 (HTTP mode only). Nothing was sent to any server.
17673
17674 Q : the proxy was waiting in the QUEUE for a connection slot. This can
17675 only happen when servers have a 'maxconn' parameter set. It can
17676 also happen in the global queue after a redispatch consecutive to
17677 a failed attempt to connect to a dying server. If no redispatch is
17678 reported, then no connection attempt was made to any server.
17679
17680 C : the proxy was waiting for the CONNECTION to establish on the
17681 server. The server might at most have noticed a connection attempt.
17682
17683 H : the proxy was waiting for complete, valid response HEADERS from the
17684 server (HTTP only).
17685
17686 D : the session was in the DATA phase.
17687
17688 L : the proxy was still transmitting LAST data to the client while the
17689 server had already finished. This one is very rare as it can only
17690 happen when the client dies while receiving the last packets.
17691
17692 T : the request was tarpitted. It has been held open with the client
17693 during the whole "timeout tarpit" duration or until the client
17694 closed, both of which will be reported in the "Tw" timer.
17695
17696 - : normal session completion after end of data transfer.
17697
17698 - the third character tells whether the persistence cookie was provided by
17699 the client (only in HTTP mode) :
17700
17701 N : the client provided NO cookie. This is usually the case for new
17702 visitors, so counting the number of occurrences of this flag in the
17703 logs generally indicate a valid trend for the site frequentation.
17704
17705 I : the client provided an INVALID cookie matching no known server.
17706 This might be caused by a recent configuration change, mixed
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +020017707 cookies between HTTP/HTTPS sites, persistence conditionally
17708 ignored, or an attack.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017709
17710 D : the client provided a cookie designating a server which was DOWN,
17711 so either "option persist" was used and the client was sent to
17712 this server, or it was not set and the client was redispatched to
17713 another server.
17714
Willy Tarreau996a92c2010-10-13 19:30:47 +020017715 V : the client provided a VALID cookie, and was sent to the associated
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017716 server.
17717
Willy Tarreau996a92c2010-10-13 19:30:47 +020017718 E : the client provided a valid cookie, but with a last date which was
17719 older than what is allowed by the "maxidle" cookie parameter, so
17720 the cookie is consider EXPIRED and is ignored. The request will be
17721 redispatched just as if there was no cookie.
17722
17723 O : the client provided a valid cookie, but with a first date which was
17724 older than what is allowed by the "maxlife" cookie parameter, so
17725 the cookie is consider too OLD and is ignored. The request will be
17726 redispatched just as if there was no cookie.
17727
Willy Tarreauc89ccb62012-04-05 21:18:22 +020017728 U : a cookie was present but was not used to select the server because
17729 some other server selection mechanism was used instead (typically a
17730 "use-server" rule).
17731
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017732 - : does not apply (no cookie set in configuration).
17733
17734 - the last character reports what operations were performed on the persistence
17735 cookie returned by the server (only in HTTP mode) :
17736
17737 N : NO cookie was provided by the server, and none was inserted either.
17738
17739 I : no cookie was provided by the server, and the proxy INSERTED one.
17740 Note that in "cookie insert" mode, if the server provides a cookie,
17741 it will still be overwritten and reported as "I" here.
17742
Willy Tarreau996a92c2010-10-13 19:30:47 +020017743 U : the proxy UPDATED the last date in the cookie that was presented by
17744 the client. This can only happen in insert mode with "maxidle". It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017745 happens every time there is activity at a different date than the
Willy Tarreau996a92c2010-10-13 19:30:47 +020017746 date indicated in the cookie. If any other change happens, such as
17747 a redispatch, then the cookie will be marked as inserted instead.
17748
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017749 P : a cookie was PROVIDED by the server and transmitted as-is.
17750
17751 R : the cookie provided by the server was REWRITTEN by the proxy, which
17752 happens in "cookie rewrite" or "cookie prefix" modes.
17753
17754 D : the cookie provided by the server was DELETED by the proxy.
17755
17756 - : does not apply (no cookie set in configuration).
17757
Willy Tarreau996a92c2010-10-13 19:30:47 +020017758The combination of the two first flags gives a lot of information about what
17759was happening when the session terminated, and why it did terminate. It can be
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017760helpful to detect server saturation, network troubles, local system resource
17761starvation, attacks, etc...
17762
17763The most common termination flags combinations are indicated below. They are
17764alphabetically sorted, with the lowercase set just after the upper case for
17765easier finding and understanding.
17766
17767 Flags Reason
17768
17769 -- Normal termination.
17770
17771 CC The client aborted before the connection could be established to the
17772 server. This can happen when haproxy tries to connect to a recently
17773 dead (or unchecked) server, and the client aborts while haproxy is
17774 waiting for the server to respond or for "timeout connect" to expire.
17775
17776 CD The client unexpectedly aborted during data transfer. This can be
17777 caused by a browser crash, by an intermediate equipment between the
17778 client and haproxy which decided to actively break the connection,
17779 by network routing issues between the client and haproxy, or by a
17780 keep-alive session between the server and the client terminated first
17781 by the client.
Willy Tarreaud72758d2010-01-12 10:42:19 +010017782
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017783 cD The client did not send nor acknowledge any data for as long as the
17784 "timeout client" delay. This is often caused by network failures on
Cyril Bontédc4d9032012-04-08 21:57:39 +020017785 the client side, or the client simply leaving the net uncleanly.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017786
17787 CH The client aborted while waiting for the server to start responding.
17788 It might be the server taking too long to respond or the client
17789 clicking the 'Stop' button too fast.
17790
17791 cH The "timeout client" stroke while waiting for client data during a
17792 POST request. This is sometimes caused by too large TCP MSS values
17793 for PPPoE networks which cannot transport full-sized packets. It can
17794 also happen when client timeout is smaller than server timeout and
17795 the server takes too long to respond.
17796
17797 CQ The client aborted while its session was queued, waiting for a server
17798 with enough empty slots to accept it. It might be that either all the
17799 servers were saturated or that the assigned server was taking too
17800 long a time to respond.
17801
17802 CR The client aborted before sending a full HTTP request. Most likely
17803 the request was typed by hand using a telnet client, and aborted
17804 too early. The HTTP status code is likely a 400 here. Sometimes this
17805 might also be caused by an IDS killing the connection between haproxy
Willy Tarreau0f228a02015-05-01 15:37:53 +020017806 and the client. "option http-ignore-probes" can be used to ignore
17807 connections without any data transfer.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017808
17809 cR The "timeout http-request" stroke before the client sent a full HTTP
17810 request. This is sometimes caused by too large TCP MSS values on the
17811 client side for PPPoE networks which cannot transport full-sized
17812 packets, or by clients sending requests by hand and not typing fast
17813 enough, or forgetting to enter the empty line at the end of the
Willy Tarreau2705a612014-05-23 17:38:34 +020017814 request. The HTTP status code is likely a 408 here. Note: recently,
Willy Tarreau0f228a02015-05-01 15:37:53 +020017815 some browsers started to implement a "pre-connect" feature consisting
17816 in speculatively connecting to some recently visited web sites just
17817 in case the user would like to visit them. This results in many
17818 connections being established to web sites, which end up in 408
17819 Request Timeout if the timeout strikes first, or 400 Bad Request when
17820 the browser decides to close them first. These ones pollute the log
17821 and feed the error counters. Some versions of some browsers have even
17822 been reported to display the error code. It is possible to work
Davor Ocelice9ed2812017-12-25 17:49:28 +010017823 around the undesirable effects of this behavior by adding "option
Willy Tarreau0f228a02015-05-01 15:37:53 +020017824 http-ignore-probes" in the frontend, resulting in connections with
17825 zero data transfer to be totally ignored. This will definitely hide
17826 the errors of people experiencing connectivity issues though.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017827
17828 CT The client aborted while its session was tarpitted. It is important to
17829 check if this happens on valid requests, in order to be sure that no
Willy Tarreau55165fe2009-05-10 12:02:55 +020017830 wrong tarpit rules have been written. If a lot of them happen, it
17831 might make sense to lower the "timeout tarpit" value to something
17832 closer to the average reported "Tw" timer, in order not to consume
17833 resources for just a few attackers.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017834
Willy Tarreau570f2212013-06-10 16:42:09 +020017835 LR The request was intercepted and locally handled by haproxy. Generally
17836 it means that this was a redirect or a stats request.
17837
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010017838 SC The server or an equipment between it and haproxy explicitly refused
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017839 the TCP connection (the proxy received a TCP RST or an ICMP message
17840 in return). Under some circumstances, it can also be the network
Davor Ocelice9ed2812017-12-25 17:49:28 +010017841 stack telling the proxy that the server is unreachable (e.g. no route,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017842 or no ARP response on local network). When this happens in HTTP mode,
17843 the status code is likely a 502 or 503 here.
17844
17845 sC The "timeout connect" stroke before a connection to the server could
17846 complete. When this happens in HTTP mode, the status code is likely a
17847 503 or 504 here.
17848
17849 SD The connection to the server died with an error during the data
17850 transfer. This usually means that haproxy has received an RST from
17851 the server or an ICMP message from an intermediate equipment while
17852 exchanging data with the server. This can be caused by a server crash
17853 or by a network issue on an intermediate equipment.
17854
17855 sD The server did not send nor acknowledge any data for as long as the
17856 "timeout server" setting during the data phase. This is often caused
Davor Ocelice9ed2812017-12-25 17:49:28 +010017857 by too short timeouts on L4 equipment before the server (firewalls,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017858 load-balancers, ...), as well as keep-alive sessions maintained
17859 between the client and the server expiring first on haproxy.
17860
17861 SH The server aborted before sending its full HTTP response headers, or
17862 it crashed while processing the request. Since a server aborting at
17863 this moment is very rare, it would be wise to inspect its logs to
17864 control whether it crashed and why. The logged request may indicate a
17865 small set of faulty requests, demonstrating bugs in the application.
17866 Sometimes this might also be caused by an IDS killing the connection
17867 between haproxy and the server.
17868
17869 sH The "timeout server" stroke before the server could return its
17870 response headers. This is the most common anomaly, indicating too
17871 long transactions, probably caused by server or database saturation.
17872 The immediate workaround consists in increasing the "timeout server"
17873 setting, but it is important to keep in mind that the user experience
17874 will suffer from these long response times. The only long term
17875 solution is to fix the application.
17876
17877 sQ The session spent too much time in queue and has been expired. See
17878 the "timeout queue" and "timeout connect" settings to find out how to
17879 fix this if it happens too often. If it often happens massively in
17880 short periods, it may indicate general problems on the affected
17881 servers due to I/O or database congestion, or saturation caused by
17882 external attacks.
17883
17884 PC The proxy refused to establish a connection to the server because the
17885 process' socket limit has been reached while attempting to connect.
Cyril Bontédc4d9032012-04-08 21:57:39 +020017886 The global "maxconn" parameter may be increased in the configuration
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017887 so that it does not happen anymore. This status is very rare and
17888 might happen when the global "ulimit-n" parameter is forced by hand.
17889
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010017890 PD The proxy blocked an incorrectly formatted chunked encoded message in
17891 a request or a response, after the server has emitted its headers. In
17892 most cases, this will indicate an invalid message from the server to
Davor Ocelice9ed2812017-12-25 17:49:28 +010017893 the client. HAProxy supports chunk sizes of up to 2GB - 1 (2147483647
Willy Tarreauf3a3e132013-08-31 08:16:26 +020017894 bytes). Any larger size will be considered as an error.
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010017895
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017896 PH The proxy blocked the server's response, because it was invalid,
17897 incomplete, dangerous (cache control), or matched a security filter.
17898 In any case, an HTTP 502 error is sent to the client. One possible
17899 cause for this error is an invalid syntax in an HTTP header name
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010017900 containing unauthorized characters. It is also possible but quite
17901 rare, that the proxy blocked a chunked-encoding request from the
17902 client due to an invalid syntax, before the server responded. In this
17903 case, an HTTP 400 error is sent to the client and reported in the
17904 logs.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017905
17906 PR The proxy blocked the client's HTTP request, either because of an
17907 invalid HTTP syntax, in which case it returned an HTTP 400 error to
17908 the client, or because a deny filter matched, in which case it
17909 returned an HTTP 403 error.
17910
17911 PT The proxy blocked the client's request and has tarpitted its
17912 connection before returning it a 500 server error. Nothing was sent
17913 to the server. The connection was maintained open for as long as
17914 reported by the "Tw" timer field.
17915
17916 RC A local resource has been exhausted (memory, sockets, source ports)
17917 preventing the connection to the server from establishing. The error
17918 logs will tell precisely what was missing. This is very rare and can
17919 only be solved by proper system tuning.
17920
Willy Tarreau996a92c2010-10-13 19:30:47 +020017921The combination of the two last flags gives a lot of information about how
17922persistence was handled by the client, the server and by haproxy. This is very
17923important to troubleshoot disconnections, when users complain they have to
17924re-authenticate. The commonly encountered flags are :
17925
17926 -- Persistence cookie is not enabled.
17927
17928 NN No cookie was provided by the client, none was inserted in the
17929 response. For instance, this can be in insert mode with "postonly"
17930 set on a GET request.
17931
17932 II A cookie designating an invalid server was provided by the client,
17933 a valid one was inserted in the response. This typically happens when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017934 a "server" entry is removed from the configuration, since its cookie
Willy Tarreau996a92c2010-10-13 19:30:47 +020017935 value can be presented by a client when no other server knows it.
17936
17937 NI No cookie was provided by the client, one was inserted in the
17938 response. This typically happens for first requests from every user
17939 in "insert" mode, which makes it an easy way to count real users.
17940
17941 VN A cookie was provided by the client, none was inserted in the
17942 response. This happens for most responses for which the client has
17943 already got a cookie.
17944
17945 VU A cookie was provided by the client, with a last visit date which is
17946 not completely up-to-date, so an updated cookie was provided in
17947 response. This can also happen if there was no date at all, or if
17948 there was a date but the "maxidle" parameter was not set, so that the
17949 cookie can be switched to unlimited time.
17950
17951 EI A cookie was provided by the client, with a last visit date which is
17952 too old for the "maxidle" parameter, so the cookie was ignored and a
17953 new cookie was inserted in the response.
17954
17955 OI A cookie was provided by the client, with a first visit date which is
17956 too old for the "maxlife" parameter, so the cookie was ignored and a
17957 new cookie was inserted in the response.
17958
17959 DI The server designated by the cookie was down, a new server was
17960 selected and a new cookie was emitted in the response.
17961
17962 VI The server designated by the cookie was not marked dead but could not
17963 be reached. A redispatch happened and selected another one, which was
17964 then advertised in the response.
17965
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017966
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200179678.6. Non-printable characters
17968-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017969
17970In order not to cause trouble to log analysis tools or terminals during log
17971consulting, non-printable characters are not sent as-is into log files, but are
17972converted to the two-digits hexadecimal representation of their ASCII code,
17973prefixed by the character '#'. The only characters that can be logged without
17974being escaped are comprised between 32 and 126 (inclusive). Obviously, the
17975escape character '#' itself is also encoded to avoid any ambiguity ("#23"). It
17976is the same for the character '"' which becomes "#22", as well as '{', '|' and
17977'}' when logging headers.
17978
17979Note that the space character (' ') is not encoded in headers, which can cause
17980issues for tools relying on space count to locate fields. A typical header
17981containing spaces is "User-Agent".
17982
17983Last, it has been observed that some syslog daemons such as syslog-ng escape
17984the quote ('"') with a backslash ('\'). The reverse operation can safely be
17985performed since no quote may appear anywhere else in the logs.
17986
17987
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200179888.7. Capturing HTTP cookies
17989---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017990
17991Cookie capture simplifies the tracking a complete user session. This can be
17992achieved using the "capture cookie" statement in the frontend. Please refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017993section 4.2 for more details. Only one cookie can be captured, and the same
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017994cookie will simultaneously be checked in the request ("Cookie:" header) and in
17995the response ("Set-Cookie:" header). The respective values will be reported in
17996the HTTP logs at the "captured_request_cookie" and "captured_response_cookie"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017997locations (see section 8.2.3 about HTTP log format). When either cookie is
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017998not seen, a dash ('-') replaces the value. This way, it's easy to detect when a
17999user switches to a new session for example, because the server will reassign it
18000a new cookie. It is also possible to detect if a server unexpectedly sets a
18001wrong cookie to a client, leading to session crossing.
18002
18003 Examples :
18004 # capture the first cookie whose name starts with "ASPSESSION"
18005 capture cookie ASPSESSION len 32
18006
18007 # capture the first cookie whose name is exactly "vgnvisitor"
18008 capture cookie vgnvisitor= len 32
18009
18010
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200180118.8. Capturing HTTP headers
18012---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018013
18014Header captures are useful to track unique request identifiers set by an upper
18015proxy, virtual host names, user-agents, POST content-length, referrers, etc. In
18016the response, one can search for information about the response length, how the
18017server asked the cache to behave, or an object location during a redirection.
18018
18019Header captures are performed using the "capture request header" and "capture
18020response header" statements in the frontend. Please consult their definition in
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018021section 4.2 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018022
18023It is possible to include both request headers and response headers at the same
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010018024time. Non-existent headers are logged as empty strings, and if one header
18025appears more than once, only its last occurrence will be logged. Request headers
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018026are grouped within braces '{' and '}' in the same order as they were declared,
18027and delimited with a vertical bar '|' without any space. Response headers
18028follow the same representation, but are displayed after a space following the
18029request headers block. These blocks are displayed just before the HTTP request
18030in the logs.
18031
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020018032As a special case, it is possible to specify an HTTP header capture in a TCP
18033frontend. The purpose is to enable logging of headers which will be parsed in
18034an HTTP backend if the request is then switched to this HTTP backend.
18035
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018036 Example :
18037 # This instance chains to the outgoing proxy
18038 listen proxy-out
18039 mode http
18040 option httplog
18041 option logasap
18042 log global
18043 server cache1 192.168.1.1:3128
18044
18045 # log the name of the virtual server
18046 capture request header Host len 20
18047
18048 # log the amount of data uploaded during a POST
18049 capture request header Content-Length len 10
18050
18051 # log the beginning of the referrer
18052 capture request header Referer len 20
18053
18054 # server name (useful for outgoing proxies only)
18055 capture response header Server len 20
18056
18057 # logging the content-length is useful with "option logasap"
18058 capture response header Content-Length len 10
18059
Davor Ocelice9ed2812017-12-25 17:49:28 +010018060 # log the expected cache behavior on the response
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018061 capture response header Cache-Control len 8
18062
18063 # the Via header will report the next proxy's name
18064 capture response header Via len 20
18065
18066 # log the URL location during a redirection
18067 capture response header Location len 20
18068
18069 >>> Aug 9 20:26:09 localhost \
18070 haproxy[2022]: 127.0.0.1:34014 [09/Aug/2004:20:26:09] proxy-out \
18071 proxy-out/cache1 0/0/0/162/+162 200 +350 - - ---- 0/0/0/0/0 0/0 \
18072 {fr.adserver.yahoo.co||http://fr.f416.mail.} {|864|private||} \
18073 "GET http://fr.adserver.yahoo.com/"
18074
18075 >>> Aug 9 20:30:46 localhost \
18076 haproxy[2022]: 127.0.0.1:34020 [09/Aug/2004:20:30:46] proxy-out \
18077 proxy-out/cache1 0/0/0/182/+182 200 +279 - - ---- 0/0/0/0/0 0/0 \
18078 {w.ods.org||} {Formilux/0.1.8|3495|||} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010018079 "GET http://trafic.1wt.eu/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018080
18081 >>> Aug 9 20:30:46 localhost \
18082 haproxy[2022]: 127.0.0.1:34028 [09/Aug/2004:20:30:46] proxy-out \
18083 proxy-out/cache1 0/0/2/126/+128 301 +223 - - ---- 0/0/0/0/0 0/0 \
18084 {www.sytadin.equipement.gouv.fr||http://trafic.1wt.eu/} \
18085 {Apache|230|||http://www.sytadin.} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010018086 "GET http://www.sytadin.equipement.gouv.fr/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018087
18088
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200180898.9. Examples of logs
18090---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018091
18092These are real-world examples of logs accompanied with an explanation. Some of
18093them have been made up by hand. The syslog part has been removed for better
18094reading. Their sole purpose is to explain how to decipher them.
18095
18096 >>> haproxy[674]: 127.0.0.1:33318 [15/Oct/2003:08:31:57.130] px-http \
18097 px-http/srv1 6559/0/7/147/6723 200 243 - - ---- 5/3/3/1/0 0/0 \
18098 "HEAD / HTTP/1.0"
18099
18100 => long request (6.5s) entered by hand through 'telnet'. The server replied
18101 in 147 ms, and the session ended normally ('----')
18102
18103 >>> haproxy[674]: 127.0.0.1:33319 [15/Oct/2003:08:31:57.149] px-http \
18104 px-http/srv1 6559/1230/7/147/6870 200 243 - - ---- 324/239/239/99/0 \
18105 0/9 "HEAD / HTTP/1.0"
18106
18107 => Idem, but the request was queued in the global queue behind 9 other
18108 requests, and waited there for 1230 ms.
18109
18110 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.654] px-http \
18111 px-http/srv1 9/0/7/14/+30 200 +243 - - ---- 3/3/3/1/0 0/0 \
18112 "GET /image.iso HTTP/1.0"
18113
18114 => request for a long data transfer. The "logasap" option was specified, so
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010018115 the log was produced just before transferring data. The server replied in
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018116 14 ms, 243 bytes of headers were sent to the client, and total time from
18117 accept to first data byte is 30 ms.
18118
18119 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.925] px-http \
18120 px-http/srv1 9/0/7/14/30 502 243 - - PH-- 3/2/2/0/0 0/0 \
18121 "GET /cgi-bin/bug.cgi? HTTP/1.0"
18122
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020018123 => the proxy blocked a server response either because of an "http-response
18124 deny" rule, or because the response was improperly formatted and not
18125 HTTP-compliant, or because it blocked sensitive information which risked
18126 being cached. In this case, the response is replaced with a "502 bad
18127 gateway". The flags ("PH--") tell us that it was haproxy who decided to
18128 return the 502 and not the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018129
18130 >>> haproxy[18113]: 127.0.0.1:34548 [15/Oct/2003:15:18:55.798] px-http \
Willy Tarreaud72758d2010-01-12 10:42:19 +010018131 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 +010018132
18133 => the client never completed its request and aborted itself ("C---") after
18134 8.5s, while the proxy was waiting for the request headers ("-R--").
18135 Nothing was sent to any server.
18136
18137 >>> haproxy[18113]: 127.0.0.1:34549 [15/Oct/2003:15:19:06.103] px-http \
18138 px-http/<NOSRV> -1/-1/-1/-1/50001 408 0 - - cR-- 2/2/2/0/0 0/0 ""
18139
18140 => The client never completed its request, which was aborted by the
18141 time-out ("c---") after 50s, while the proxy was waiting for the request
Davor Ocelice9ed2812017-12-25 17:49:28 +010018142 headers ("-R--"). Nothing was sent to any server, but the proxy could
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018143 send a 408 return code to the client.
18144
18145 >>> haproxy[18989]: 127.0.0.1:34550 [15/Oct/2003:15:24:28.312] px-tcp \
18146 px-tcp/srv1 0/0/5007 0 cD 0/0/0/0/0 0/0
18147
18148 => This log was produced with "option tcplog". The client timed out after
18149 5 seconds ("c----").
18150
18151 >>> haproxy[18989]: 10.0.0.1:34552 [15/Oct/2003:15:26:31.462] px-http \
18152 px-http/srv1 3183/-1/-1/-1/11215 503 0 - - SC-- 205/202/202/115/3 \
Willy Tarreaud72758d2010-01-12 10:42:19 +010018153 0/0 "HEAD / HTTP/1.0"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018154
18155 => The request took 3s to complete (probably a network problem), and the
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018156 connection to the server failed ('SC--') after 4 attempts of 2 seconds
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018157 (config says 'retries 3'), and no redispatch (otherwise we would have
18158 seen "/+3"). Status code 503 was returned to the client. There were 115
18159 connections on this server, 202 connections on this proxy, and 205 on
18160 the global process. It is possible that the server refused the
18161 connection because of too many already established.
Willy Tarreau844e3c52008-01-11 16:28:18 +010018162
Willy Tarreau52b2d222011-09-07 23:48:48 +020018163
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200181649. Supported filters
18165--------------------
18166
18167Here are listed officially supported filters with the list of parameters they
18168accept. Depending on compile options, some of these filters might be
18169unavailable. The list of available filters is reported in haproxy -vv.
18170
18171See also : "filter"
18172
181739.1. Trace
18174----------
18175
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010018176filter trace [name <name>] [random-parsing] [random-forwarding] [hexdump]
Christopher Fauletc3fe5332016-04-07 15:30:10 +020018177
18178 Arguments:
18179 <name> is an arbitrary name that will be reported in
18180 messages. If no name is provided, "TRACE" is used.
18181
18182 <random-parsing> enables the random parsing of data exchanged between
18183 the client and the server. By default, this filter
18184 parses all available data. With this parameter, it
18185 only parses a random amount of the available data.
18186
Davor Ocelice9ed2812017-12-25 17:49:28 +010018187 <random-forwarding> enables the random forwarding of parsed data. By
Christopher Fauletc3fe5332016-04-07 15:30:10 +020018188 default, this filter forwards all previously parsed
18189 data. With this parameter, it only forwards a random
18190 amount of the parsed data.
18191
Davor Ocelice9ed2812017-12-25 17:49:28 +010018192 <hexdump> dumps all forwarded data to the server and the client.
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010018193
Christopher Fauletc3fe5332016-04-07 15:30:10 +020018194This filter can be used as a base to develop new filters. It defines all
18195callbacks and print a message on the standard error stream (stderr) with useful
18196information for all of them. It may be useful to debug the activity of other
18197filters or, quite simply, HAProxy's activity.
18198
18199Using <random-parsing> and/or <random-forwarding> parameters is a good way to
18200tests the behavior of a filter that parses data exchanged between a client and
18201a server by adding some latencies in the processing.
18202
18203
182049.2. HTTP compression
18205---------------------
18206
18207filter compression
18208
18209The HTTP compression has been moved in a filter in HAProxy 1.7. "compression"
18210keyword must still be used to enable and configure the HTTP compression. And
Christopher Fauletb30b3102019-09-12 23:03:09 +020018211when no other filter is used, it is enough. When used with the cache or the
18212fcgi-app enabled, it is also enough. In this case, the compression is always
18213done after the response is stored in the cache. But it is mandatory to
18214explicitly use a filter line to enable the HTTP compression when at least one
18215filter other than the cache or the fcgi-app is used for the same
18216listener/frontend/backend. This is important to know the filters evaluation
18217order.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020018218
Christopher Fauletb30b3102019-09-12 23:03:09 +020018219See also : "compression", section 9.4 about the cache filter and section 9.5
18220 about the fcgi-app filter.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020018221
18222
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +0200182239.3. Stream Processing Offload Engine (SPOE)
18224--------------------------------------------
18225
18226filter spoe [engine <name>] config <file>
18227
18228 Arguments :
18229
18230 <name> is the engine name that will be used to find the right scope in
18231 the configuration file. If not provided, all the file will be
18232 parsed.
18233
18234 <file> is the path of the engine configuration file. This file can
18235 contain configuration of several engines. In this case, each
18236 part must be placed in its own scope.
18237
18238The Stream Processing Offload Engine (SPOE) is a filter communicating with
18239external components. It allows the offload of some specifics processing on the
Davor Ocelice9ed2812017-12-25 17:49:28 +010018240streams in tiered applications. These external components and information
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020018241exchanged with them are configured in dedicated files, for the main part. It
18242also requires dedicated backends, defined in HAProxy configuration.
18243
18244SPOE communicates with external components using an in-house binary protocol,
18245the Stream Processing Offload Protocol (SPOP).
18246
Tim Düsterhus4896c442016-11-29 02:15:19 +010018247For all information about the SPOE configuration and the SPOP specification, see
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020018248"doc/SPOE.txt".
18249
18250Important note:
18251 The SPOE filter is highly experimental for now and was not heavily
18252 tested. It is really not production ready. So use it carefully.
18253
Christopher Faulet99a17a22018-12-11 09:18:27 +0100182549.4. Cache
18255----------
18256
18257filter cache <name>
18258
18259 Arguments :
18260
18261 <name> is name of the cache section this filter will use.
18262
18263The cache uses a filter to store cacheable responses. The HTTP rules
18264"cache-store" and "cache-use" must be used to define how and when to use a
John Roeslerfb2fce12019-07-10 15:45:51 -050018265cache. By default the corresponding filter is implicitly defined. And when no
Christopher Fauletb30b3102019-09-12 23:03:09 +020018266other filters than fcgi-app or compression are used, it is enough. In such
18267case, the compression filter is always evaluated after the cache filter. But it
18268is mandatory to explicitly use a filter line to use a cache when at least one
18269filter other than the compression or the fcgi-app is used for the same
Christopher Faulet27d93c32018-12-15 22:32:02 +010018270listener/frontend/backend. This is important to know the filters evaluation
18271order.
Christopher Faulet99a17a22018-12-11 09:18:27 +010018272
Christopher Fauletb30b3102019-09-12 23:03:09 +020018273See also : section 9.2 about the compression filter, section 9.5 about the
18274 fcgi-app filter and section 6 about cache.
18275
18276
182779.5. Fcgi-app
18278-------------
18279
18280filter fcg-app <name>
18281
18282 Arguments :
18283
18284 <name> is name of the fcgi-app section this filter will use.
18285
18286The FastCGI application uses a filter to evaluate all custom parameters on the
18287request path, and to process the headers on the response path. the <name> must
18288reference an existing fcgi-app section. The directive "use-fcgi-app" should be
18289used to define the application to use. By default the corresponding filter is
18290implicitly defined. And when no other filters than cache or compression are
18291used, it is enough. But it is mandatory to explicitly use a filter line to a
18292fcgi-app when at least one filter other than the compression or the cache is
18293used for the same backend. This is important to know the filters evaluation
18294order.
18295
18296See also: "use-fcgi-app", section 9.2 about the compression filter, section 9.4
18297 about the cache filter and section 10 about FastCGI application.
18298
18299
1830010. FastCGI applications
18301-------------------------
18302
18303HAProxy is able to send HTTP requests to Responder FastCGI applications. This
18304feature was added in HAProxy 2.1. To do so, servers must be configured to use
18305the FastCGI protocol (using the keyword "proto fcgi" on the server line) and a
18306FastCGI application must be configured and used by the backend managing these
18307servers (using the keyword "use-fcgi-app" into the proxy section). Several
18308FastCGI applications may be defined, but only one can be used at a time by a
18309backend.
18310
18311HAProxy implements all features of the FastCGI specification for Responder
18312application. Especially it is able to multiplex several requests on a simple
18313connection.
18314
1831510.1. Setup
18316-----------
18317
1831810.1.1. Fcgi-app section
18319--------------------------
18320
18321fcgi-app <name>
18322 Declare a FastCGI application named <name>. To be valid, at least the
18323 document root must be defined.
18324
18325acl <aclname> <criterion> [flags] [operator] <value> ...
18326 Declare or complete an access list.
18327
18328 See "acl" keyword in section 4.2 and section 7 about ACL usage for
18329 details. ACLs defined for a FastCGI application are private. They cannot be
18330 used by any other application or by any proxy. In the same way, ACLs defined
18331 in any other section are not usable by a FastCGI application. However,
18332 Pre-defined ACLs are available.
18333
18334docroot <path>
18335 Define the document root on the remote host. <path> will be used to build
18336 the default value of FastCGI parameters SCRIPT_FILENAME and
18337 PATH_TRANSLATED. It is a mandatory setting.
18338
18339index <script-name>
18340 Define the script name that will be appended after an URI that ends with a
18341 slash ("/") to set the default value of the FastCGI parameter SCRIPT_NAME. It
18342 is an optional setting.
18343
18344 Example :
18345 index index.php
18346
18347log-stderr global
18348log-stderr <address> [len <length>] [format <format>]
18349 [sample <ranges>:<smp_size>] <facility> [<level> [<minlevel>]]
18350 Enable logging of STDERR messages reported by the FastCGI application.
18351
18352 See "log" keyword in section 4.2 for details. It is an optional setting. By
18353 default STDERR messages are ignored.
18354
18355pass-header <name> [ { if | unless } <condition> ]
18356 Specify the name of a request header which will be passed to the FastCGI
18357 application. It may optionally be followed by an ACL-based condition, in
18358 which case it will only be evaluated if the condition is true.
18359
18360 Most request headers are already available to the FastCGI application,
18361 prefixed with "HTTP_". Thus, this directive is only required to pass headers
18362 that are purposefully omitted. Currently, the headers "Authorization",
18363 "Proxy-Authorization" and hop-by-hop headers are omitted.
18364
18365 Note that the headers "Content-type" and "Content-length" are never passed to
18366 the FastCGI application because they are already converted into parameters.
18367
18368path-info <regex>
18369 Define a regular expression to extract the script-name and the path-info
18370 from the URI. Thus, <regex> should have two captures: the first one to
18371 capture the script name and the second one to capture the path-info. It is an
18372 optional setting. If it is not defined, no matching is performed on the
18373 URI. and the FastCGI parameters PATH_INFO and PATH_TRANSLATED are not filled.
18374
18375 Example :
18376 path-info ^(/.+\.php)(/.*)?$
18377
18378option get-values
18379no option get-values
18380 Enable or disable the retrieve of variables about connection management.
18381
18382 HAproxy is able to send the record FCGI_GET_VALUES on connection
18383 establishment to retrieve the value for following variables:
18384
18385 * FCGI_MAX_REQS The maximum number of concurrent requests this
18386 application will accept.
18387
William Lallemand93e548e2019-09-30 13:54:02 +020018388 * FCGI_MPXS_CONNS "0" if this application does not multiplex connections,
18389 "1" otherwise.
Christopher Fauletb30b3102019-09-12 23:03:09 +020018390
18391 Some FastCGI applications does not support this feature. Some others close
18392 the connexion immediately after sending their response. So, by default, this
18393 option is disabled.
18394
18395 Note that the maximum number of concurrent requests accepted by a FastCGI
18396 application is a connection variable. It only limits the number of streams
18397 per connection. If the global load must be limited on the application, the
18398 server parameters "maxconn" and "pool-max-conn" must be set. In addition, if
18399 an application does not support connection multiplexing, the maximum number
18400 of concurrent requests is automatically set to 1.
18401
18402option keep-conn
18403no option keep-conn
18404 Instruct the FastCGI application to keep the connection open or not after
18405 sending a response.
18406
18407 If disabled, the FastCGI application closes the connection after responding
18408 to this request. By default, this option is enabled.
18409
18410option max-reqs <reqs>
18411 Define the maximum number of concurrent requests this application will
18412 accept.
18413
18414 This option may be overwritten if the variable FCGI_MAX_REQS is retrieved
18415 during connection establishment. Furthermore, if the application does not
18416 support connection multiplexing, this option will be ignored. By default set
18417 to 1.
18418
18419option mpxs-conns
18420no option mpxs-conns
18421 Enable or disable the support of connection multiplexing.
18422
18423 This option may be overwritten if the variable FCGI_MPXS_CONNS is retrieved
18424 during connection establishment. It is disabled by default.
18425
18426set-param <name> <fmt> [ { if | unless } <condition> ]
18427 Set a FastCGI parameter that should be passed to this application. Its
18428 value, defined by <fmt> must follows the log-format rules (see section 8.2.4
18429 "Custom Log format"). It may optionally be followed by an ACL-based
18430 condition, in which case it will only be evaluated if the condition is true.
18431
18432 With this directive, it is possible to overwrite the value of default FastCGI
18433 parameters. If the value is evaluated to an empty string, the rule is
18434 ignored. These directives are evaluated in their declaration order.
18435
18436 Example :
18437 # PHP only, required if PHP was built with --enable-force-cgi-redirect
18438 set-param REDIRECT_STATUS 200
18439
18440 set-param PHP_AUTH_DIGEST %[req.hdr(Authorization)]
18441
18442
1844310.1.2. Proxy section
18444---------------------
18445
18446use-fcgi-app <name>
18447 Define the FastCGI application to use for the backend.
18448
18449 Arguments :
18450 <name> is the name of the FastCGI application to use.
18451
18452 This keyword is only available for HTTP proxies with the backend capability
18453 and with at least one FastCGI server. However, FastCGI servers can be mixed
18454 with HTTP servers. But except there is a good reason to do so, it is not
18455 recommended (see section 10.3 about the limitations for details). Only one
18456 application may be defined at a time per backend.
18457
18458 Note that, once a FastCGI application is referenced for a backend, depending
18459 on the configuration some processing may be done even if the request is not
18460 sent to a FastCGI server. Rules to set parameters or pass headers to an
18461 application are evaluated.
18462
18463
1846410.1.3. Example
18465---------------
18466
18467 frontend front-http
18468 mode http
18469 bind *:80
18470 bind *:
18471
18472 use_backend back-dynamic if { path_reg ^/.+\.php(/.*)?$ }
18473 default_backend back-static
18474
18475 backend back-static
18476 mode http
18477 server www A.B.C.D:80
18478
18479 backend back-dynamic
18480 mode http
18481 use-fcgi-app php-fpm
18482 server php-fpm A.B.C.D:9000 proto fcgi
18483
18484 fcgi-app php-fpm
18485 log-stderr global
18486 option keep-conn
18487
18488 docroot /var/www/my-app
18489 index index.php
18490 path-info ^(/.+\.php)(/.*)?$
18491
18492
1849310.2. Default parameters
18494------------------------
18495
18496A Responder FastCGI application has the same purpose as a CGI/1.1 program. In
18497the CGI/1.1 specification (RFC3875), several variables must be passed to the
18498scipt. So HAProxy set them and some others commonly used by FastCGI
18499applications. All these variables may be overwritten, with caution though.
18500
18501 +-------------------+-----------------------------------------------------+
18502 | AUTH_TYPE | Identifies the mechanism, if any, used by HAProxy |
18503 | | to authenticate the user. Concretely, only the |
18504 | | BASIC authentication mechanism is supported. |
18505 | | |
18506 +-------------------+-----------------------------------------------------+
18507 | CONTENT_LENGTH | Contains the size of the message-body attached to |
18508 | | the request. It means only requests with a known |
18509 | | size are considered as valid and sent to the |
18510 | | application. |
18511 | | |
18512 +-------------------+-----------------------------------------------------+
18513 | CONTENT_TYPE | Contains the type of the message-body attached to |
18514 | | the request. It may not be set. |
18515 | | |
18516 +-------------------+-----------------------------------------------------+
18517 | DOCUMENT_ROOT | Contains the document root on the remote host under |
18518 | | which the script should be executed, as defined in |
18519 | | the application's configuration. |
18520 | | |
18521 +-------------------+-----------------------------------------------------+
18522 | GATEWAY_INTERFACE | Contains the dialect of CGI being used by HAProxy |
18523 | | to communicate with the FastCGI application. |
18524 | | Concretely, it is set to "CGI/1.1". |
18525 | | |
18526 +-------------------+-----------------------------------------------------+
18527 | PATH_INFO | Contains the portion of the URI path hierarchy |
18528 | | following the part that identifies the script |
18529 | | itself. To be set, the directive "path-info" must |
18530 | | be defined. |
18531 | | |
18532 +-------------------+-----------------------------------------------------+
18533 | PATH_TRANSLATED | If PATH_INFO is set, it is its translated version. |
18534 | | It is the concatenation of DOCUMENT_ROOT and |
18535 | | PATH_INFO. If PATH_INFO is not set, this parameters |
18536 | | is not set too. |
18537 | | |
18538 +-------------------+-----------------------------------------------------+
18539 | QUERY_STRING | Contains the request's query string. It may not be |
18540 | | set. |
18541 | | |
18542 +-------------------+-----------------------------------------------------+
18543 | REMOTE_ADDR | Contains the network address of the client sending |
18544 | | the request. |
18545 | | |
18546 +-------------------+-----------------------------------------------------+
18547 | REMOTE_USER | Contains the user identification string supplied by |
18548 | | client as part of user authentication. |
18549 | | |
18550 +-------------------+-----------------------------------------------------+
18551 | REQUEST_METHOD | Contains the method which should be used by the |
18552 | | script to process the request. |
18553 | | |
18554 +-------------------+-----------------------------------------------------+
18555 | REQUEST_URI | Contains the request's URI. |
18556 | | |
18557 +-------------------+-----------------------------------------------------+
18558 | SCRIPT_FILENAME | Contains the absolute pathname of the script. it is |
18559 | | the concatenation of DOCUMENT_ROOT and SCRIPT_NAME. |
18560 | | |
18561 +-------------------+-----------------------------------------------------+
18562 | SCRIPT_NAME | Contains the name of the script. If the directive |
18563 | | "path-info" is defined, it is the first part of the |
18564 | | URI path hierarchy, ending with the script name. |
18565 | | Otherwise, it is the entire URI path. |
18566 | | |
18567 +-------------------+-----------------------------------------------------+
18568 | SERVER_NAME | Contains the name of the server host to which the |
18569 | | client request is directed. It is the value of the |
18570 | | header "Host", if defined. Otherwise, the |
18571 | | destination address of the connection on the client |
18572 | | side. |
18573 | | |
18574 +-------------------+-----------------------------------------------------+
18575 | SERVER_PORT | Contains the destination TCP port of the connection |
18576 | | on the client side, which is the port the client |
18577 | | connected to. |
18578 | | |
18579 +-------------------+-----------------------------------------------------+
18580 | SERVER_PROTOCOL | Contains the request's protocol. |
18581 | | |
18582 +-------------------+-----------------------------------------------------+
18583 | HTTPS | Set to a non-empty value ("on") if the script was |
18584 | | queried through the HTTPS protocol. |
18585 | | |
18586 +-------------------+-----------------------------------------------------+
18587
18588
1858910.3. Limitations
18590------------------
18591
18592The current implementation have some limitations. The first one is about the
18593way some request headers are hidden to the FastCGI applications. This happens
18594during the headers analysis, on the backend side, before the connection
18595establishment. At this stage, HAProxy know the backend is using a FastCGI
18596application but it don't know if the request will be routed to a FastCGI server
18597or not. But to hide request headers, it simply removes them from the HTX
18598message. So, if the request is finally routed to an HTTP server, it never see
18599these headers. For this reason, it is not recommended to mix FastCGI servers
18600and HTTP servers under the same backend.
18601
18602Similarly, the rules "set-param" and "pass-header" are evaluated during the
18603request headers analysis. So the evaluation is always performed, even if the
18604requests is finally forwarded to an HTTP server.
18605
18606About the rules "set-param", when a rule is applied, a pseudo header is added
18607into the HTX message. So, the same way than for HTTP header rewrites, it may
18608fail if the buffer is full. The rules "set-param" will compete with
18609"http-request" ones.
18610
18611Finally, all FastCGI params and HTTP headers are sent into a unique record
18612FCGI_PARAM. Encoding of this record must be done in one pass, otherwise a
18613processing error is returned. It means the record FCGI_PARAM, once encoded,
18614must not exceeds the size of a buffer. However, there is no reserve to respect
18615here.
William Lallemand86d0df02017-11-24 21:36:45 +010018616
Willy Tarreau0ba27502007-12-24 16:55:16 +010018617/*
18618 * Local variables:
18619 * fill-column: 79
18620 * End:
18621 */